UNIVERSITY OF CINCINNATI

Transcription

1 UNIVERSITY OF CINCINNATI Date: 14-Jan-2010 I, Bassam Abd El-Nabi, hereby submit this original work as part of the requirements for the degree of: Doctor of Philosophy in Aerospace Engineering It is entitled: Single Annular Combustor: Experimental investigations of Aerodynamics, Dynamics and Emissions Student Signature: Bassam Abd El-Nabi This work and its defense approved by: Committee Chair: San-Mou Jeng, PhD San-Mou Jeng, PhD Milind Jog, PhD Milind Jog, PhD Mustafa Furhan Andac, PhD Mustafa Furhan Andac, PhD Shaaban Abdallah, PhD Shaaban Abdallah, PhD 3/8/

2 Single Annular Combustor: Experimental investigations of Aerodynamics, Dynamics and Emissions A thesis submitted to the Division of Graduate Studies and Research of the University of Cincinnati in partial fulfillment of the requirements for the degree of DOCTORATE OF PHILOSOPHY (Ph.D.) in the Department of Aerospace Engineering and Engineering Mechanics of the College of Engineering March 2010 by Bassam Sabry Mohammad B.S., Mechanical Engineering, Cairo University, Egypt M.S., Mechanical Engineering, Cairo University, Egypt M.S., Nuclear Engineering, Missouri University of Science and Technology, USA Committee Chair: Dr. San-Mou Jeng i

3 Abstract The present work investigates the aerodynamics, dynamics and emissions of a Single Cup Combustor Sector. The Combustor resembles a real Gas Turbine Combustor with primary, secondary and dilution zones (also known as fuel rich dome combustor). The research is initiated by studying the effect of the combustor front end geometry on the flow field. Two different exit configurations (one causes a sudden expansion to the swirling flow while the other causes a gradual expansion), installed in a dump combustor, are tested using LDV. The results reveal that the expanding surface reduces the turbulence activities, eliminates the corner recirculation zone and increases the length of the CRZ appreciably. An asymmetry in the flow field is observed due to the asymmetry of the expanding surface. To study the effect of chamber geometry on the flow field, the dome configuration is tested in the combustor sector with the primary dilution jets blocked. The size of the CRZ is reduced significantly (40 % reduction in the height). With active primary jets, the CRZ is reconstructed in 3D by conducting several PIV measurements off-center. The confinement appears to significantly influence the shape of the CRZ such that the area ratio is similar for both the confinement and the CRZ (approximately 85%). The primary jets considerably contribute to the heat release process at high power conditions. Also, the primary jets drastically impact the flow field structure. Therefore, the parameters influencing the primary jets are studied using PIV (pressure drop, jets size, offcentering, interaction with convective cooling air, jet blockage and fuel injection). This study is referred to as a jet sensitivity study. The results indicate that the primary jets can be used effectively in controlling the flow field structure. A pressure drop of 4.3% and 7.6% result in ii

4 similar flows with no noticeable effect on the size of the CRZ and the four jets wake regions. On the other hand, the results show that the primary jets are very sensitive to perturbations. The cooling air interacts with the primary jet and influences the flow field although the momentum ratio has an order of magnitude of 100:1. The results also show that the big primary jets dictate the flow field in the primary zone as well as the secondary zone. However, relatively smaller jets mainly impact the primary zone. Also, the results point to the presence of a critical jet diameter beyond which the dilution jets have minimum impact on the secondary region. The jet off-centering shows significant effect on the flow field though it is on the order of 1.0 mm. The jet sensitivity study provides the combustion engineers with useful methods to control the flow field structure, an explanation for observed flow structure under different conditions and predictable flow field behavior with engine aging. All results obtained from the jet sensitivity study could be explained in terms of jet opposition. Hence, similar results are expected under reacting conditions even though the results presented here are obtained under isothermal conditions. The fuel injection is also shown to influence the flow field. High fuel flow rate is shown to have very strong impact on the flow field and thus results in a strong distortion of both the primary and secondary zones. The jets wake regions are shown to change in size with fuel injection. The left jet wake region continuously reduces in size with fuel injection while the right jet wake region does not. This offers a possible explanation for the observed combustion instabilities in the left primary jet region. The combustion instabilities are studied using a microphone, high speed camera and regular cameras. The frequency spectrum for the sector is established at different pressure drops (2, 4 and 6%) as well as different pre-heat temperatures (200, 400 and 600F). The iii

5 acoustic spectrum suggests that there are three frequencies of concern (280, 400 and 600 HZ). The high frequency appears to be related to the combustor ¼ longitudinal wave. The 280 Hz is due to a rotating instability while the 400 Hz is related to the primary jets. The emissions emanating from the combustor are studied using FTIR at pressure drop of 4% and different power conditions. The sector emissions characteristics are determined. Water injection is also used to control the pollutant emissions. Water fuel ratio of 100% and 50% results in a corresponding reduction in the NOx concentration with 50% and 22% (at high power conditions). No noticeable effects are observed on the NOx and CO at low power conditions. A high degree of homogeneity in the emissions contours is observed at the combustor exit at low power conditions (equivalence ratio of 0.3). However, this homogeneity is noticeably reduced at high power conditions (equivalence ratio of 0.6). iv

6 v

7 Acknowledgment All thanks are due to ALLAH (God) and may His peace and blessings be upon all His messengers and prophets including Moses, Jesus and Mohammad. It was only through ALLAH's help and support that this work was accomplished. I can never thank Him enough for all the blessing He bestowed upon me throughout my life and I seek His forgiveness for all my shortcomings. I would like to especially recognize and express my gratitude to my advisor, Dr. S. M. Jeng. It is certainly through his support, guidance and encouragement that this work is complete. He trusted me and gave me a room to make my own decisions. Dr. Jeng was always ready to give advice anytime. Dr. Jeng is one of the best things that happened in my life. Dr. Jeng taught me the basics of research, how to interact with people, how to think differently and how to solve problems. Also, my deepest appreciation to Dr. Abdallah Shabaan, Dr. Millind Jog and Dr. Gurhan Andac for their time and effort as my committee members. I am also very thankful to the people at GE aviation. This work was not possible without their help and support. Special thanks to all my colleagues at the Combustion Research Lab. Dr. Jun Cai taught me how to use and setup the PIV, LDV and FTIR systems. Mr. Curt Fox offered a great deal of help. He is the kind of person who would help you do anything, anytime in a timely manner! Also, I am very thankful to Samir, Fumi, and Kao. They helped me in experimental setup as well as running the experiments. I would especially like to thank my Mom and Dad. I achieved this work because of their prayers. My father supported me financially since I arrived to the United States. He paid a vii

8 huge amount of money to help me earn my degree. May ALLAH reward them for what they have done. Finally, there is a single person who has had a greater effect on this work than anyone else, and that is my wife. She has been patient and she supported and encouraged me. She was taking care of the family duties to provide me with the environment and the time necessary to complete this effort. We have four kids and there was no chance for me to earn my degree if she didn t sacrifice her time and effort. viii

9 Read: In the name of thy Lord Who createth, Createth man from a clot, Read: And thy Lord is the Most Bounteous, Who teacheth by the pen, Teacheth man that which he knew not. -Quran (20:114) My Lord! Advance me in knowledge. -Quran (96:1-5) vi

10 Table of Contents Acknowledgment... vii Table of Contents... ix List of Figures... xi Nomenclature... xvi Introduction... 1 Chapter 1. (Literature Review) Aerodynamics Flame stabilization and Generation of CRZ Swirl Cup Swirl Characterization Factors influencing the CRZ Size and strength Swirl Number Effects Confinement Effects Reynolds Number Effects Nozzle Insertion Effects Combustion Effects Realistic combustion chambers and effect of dilution jets Emissions Nitric Oxides (NOx) NOx categories Effect of pressure on NOx generation NOx reduction techniques Carbon Monoxide (CO) Dynamics and generation of Combustion Instabilities Chapter 2. (Dump Combustor Aerodynamic) Experimental setup, procedure and test conditions Swirl Cup and test section Experimental facility Diagnostics Test conditions, data acquisition and measurement Grid Results and Discussions Horizontal plane (X-Y) measurements for both configurations Vertical plane (X-Z) measurements for both configurations Axial velocity and turbulent fluctuating component Radial velocity and turbulent fluctuating component Vertical plane (Y-Z) measurements for both configurations Axial velocity and turbulent fluctuating component Tangential velocity and turbulent fluctuating component Confinement effect on the flow field Chapter 3. (SAC Aerodynamics) Experimental setup, procedure and test conditions ix

11 3.1.1 Swirler and SAC Combustor Experimental facility Diagnostics Test conditions, data acquisition and measurement grid Results and Discussions Mean flow measurements at the SAC mid plane using LDV Reacting flow vs. isothermal flow PIV results Effect of pressure drop on the flow structure Instantaneous PIV at SAC mid plane at pressure drop of 4.3% PIV off center measurements at pressure drop of 4.3% Effect of partially blocking the primary dilution jets strip Effect of blocking the cooling holes underneath the primary dilution jets Effect of primary jets off centering on the flow field Effect of primary jet size on the flow field Chapter 4. (SAC Dynamics) Experimental setup, procedure and test conditions Swirl cup and SAC Sector Diagnostics Test conditions and data acquisition Results and Discussions Chapter 5. (SAC Air/Fuel Mixing) Experimental setup, procedure and test conditions Swirl cup and SAC Sector Diagnostics Test conditions, data acquisition and measurement grid Results and Discussions Chapter 6. (Emissions) Experimental setup, procedure and test conditions Swirl cup and SAC Sector Diagnostics Test conditions and data acquisition Results and Discussions Species Measurements at the center of the SAC sector exit Propane results Methane results Species Contours at the SAC sector exit Chapter 7. (Conclusions) Chapter 8. (Proposed future work) References x

12 List of Figures 1.1 Typical coaxial Axial and radial swirlers Typical Swirl cup arrangement with counter rotating radial swirlers Typical SAC Combustor with swirl cup CFM56 swirl cup assembly Illustration of Ideal and Real combustion emissions [61] Order of magnitude of Gas Turbine Combustion emissions [61] Typical pollutant formation trends as function of the equivalence ratio [12] Typical variation of the equilibrium and thermal NOx vs. equivalence ratio Illustration of the effect of residence time on thermal NOx DLN combustor concepts [61] Multiple point LDI combustor developed by NASA [114] Illustration of LP, RQL and LDI concepts [94] Severity of Carbon Monoxide intoxications [115] Illustration of generation and growth of pressure oscillations known as thermo-acoustic instabilities [126] Different configurations of the swirl cup under investigation Rectangular Plexi-glass chamber (dump combustor) Schematic of the Experimental facility to study the dump combustor Aerodynamics Special flanges and adapters to allow installation of swirl cups to the vertical rig Laser Doppler Velocimetry system provided by Artium Technologies LDV setup on the vertical rig (plexi-glass chamber) Measurements Coordinate system for both configurations Axial velocity contours in the horizontal plane (configuration 1) Axial velocity contours in the horizontal plane (base configuration) Contours of the Root Mean Square of the axial velocity in the horizontal plane for configuration one Contours of the Root Mean Square of the axial velocity in the horizontal plane for the base configuration Axial velocity profiles in the X-Z plane (Config. 1 rectangular chamber) Axial velocity profiles in the X-Z plane (base Config. rectangular chamber)...57 xi

13 2.14 Contours of axial velocity and corresponding rms in X-Z plane (Config. 1 rectangular chamber) Contours of axial velocity and corresponding rms in the X-Z plane (base Config. rectangular chamber) Radial velocity profiles in the X-Z plane (Config. 1 rectangular chamber) Radial velocity profiles in the X-Z plane (base Config. rectangular chamber) Contours of radial velocity and corresponding rms for Config. 1 in the X-Z plane (rectangular chamber) Contours of radial velocity and corresponding rms for the base configuration in the X-Z plane (rectangular chamber) Difference between rms values of the axial and radial velocities for both configurations Axial velocity profiles in the Y-Z plane (Config. 1 rectangular chamber) Axial velocity profiles in the Y-Z plane (base Config. rectangular chamber) Contours of axial velocity and corresponding rms in the Y-Z plane (Config. 1 rectangular chamber) Tangential velocity profiles in the Y-Z plane (Config. 1 rectangular chamber) Tangential velocity profiles in the Y-Z plane (base Config. rectangular chamber) Contours of tangential velocity and corresponding rms in the Y-Z plane (Config. 1 rectangular chamber) Axial and radial velocity contours of the base configuration installed in the SAC sector Axial and radial rms contours of the base configuration installed in the SAC sector Assembly of the Single Annular Combustor Sector used in the study Single Annular Combustor Sector components Flange and adapter to enable installation of the SAC on the vertical rig PIV system provided by Lavision Mirror and sheet optics setup to generate the laser sheet at the measurement plane PIV setup on the vertical rig LDV set up on the vertical rig to study the SAC flow field SAC coordinate system SAC sector centerline total velocity contours using LDV at P/P = 4.9% SAC sector centerline radial velocity contours using LDV at P/P = 4.9% SAC sector centerline axial velocity contours using LDV at P/P = 4.9% SAC sector centerline axial velocity rms using LDV at P/P = 4.9% SAC sector centerline radial velocity rms using LDV at P/P = 4.9%...81 xii

14 3.14 SAC sector during ignition Reacting flow vs. Isothermal flow (N-Heptane) Reacting flow vs. Isothermal flow (Propane) Sequential images showing the progression of the reaction zone as the fuel flow rate increases SAC sector centerline total velocity contours using PIV at P/P = 4.3% SAC sector centerline total velocity contours using PIV at P/P = 7.6% Instantaneous PIV at 4.3% (Jet Impinge) Instantaneous PIV at 4.3% (Left Jet shooting down and right Jet shooting up) Instantaneous PIV at 4.3% (Right Jet shooting down and left Jet shooting up) PIV measurements at Y=0 at pressure drop of 4.3% PIV measurements at Y=0.65R at pressure drop of 4.3% PIV measurements at Y=1.1R at pressure drop of 4.3% PIV measurements at Y=1.3R at pressure drop of 4.3% D reconstructed CRZ over the flare Dimensions of the SAC sector CRZ SAC cross section schematic in the primary zone Illustration of the dilution jet strip with partial blocking Effect of partially blocking the primary dilution jets on the flow field Illustration of primary jets and cooling jets interaction Effect of blocking the cooling strip underneath the primary jets Illustration of primary jet holes off centering Effect of primary jets off centering on the SAC flow field Effect of primary jet size on the flow field Set up on the horizontal rig for Dynamics investigations Sample of the FFT to the acoustic signal Acoustic Spectrum using Methane at P/P=2% and T3=400F Acoustic Spectrum using Methane at P/P=4% and T3=200F Acoustic Spectrum using Methane at P/P=4% and T3=400F Acoustic Spectrum using Methane at P/P=4% and T3=600F Acoustic Spectrum using Methane at P/P=6% and T3=400F Acoustic Spectrum using Propane at P/P=2% and T3=400F xiii

15 4.9 Acoustic Spectrum using Propane at P/P=4% and T3=200F Acoustic Spectrum using Propane at P/P=4% and T3=400F Acoustic Spectrum using Propane at P/P=4% and T3=600F Acoustic Spectrum using Propane at P/P=6% and T3=400F Audio signal FFT at =0.23, P/P=2% and T3=200F Source of the heat release oscillations (280 Hz) at low equivalence ratio[125] Audio signal FFT at =0.6, P/P=2% and T3=200F Source of the heat release oscillations (400 Hz) at high equivalence ratio [125] Gaseous fuel nozzle tip Designed side plate to enable insertion of the sampling probe to the SAC Profiles of Fuel Air ratio (by mass) at different fuel flow rates Local Equivalence ratio contours and corresponding velocity field at fuel flow rate of 25 pphr (global equivalence ratio 0.375) Normalized equivalence ratio contours imposed on velocity vectors at fuel flow rate of 25 pphr velocity vectors and streamlines with and without fuel injection (fuel flow rate of 25 pphr) Difference in velocity magnitudes with and without fuel injection (fuel flow rate =25 pphr) Evaluation of PIV sensitivity Normalized local equivalence ratio profiles along the centerline at different fuel flow rates Normalized local equivalence ratio profiles along X/R=-0.8 at different fuel flow rates Normalized local equivalence ratio profiles along Z/R=1.725 at different fuel flow rates Effect of fuel injection on the flow field at fuel flow rate of 35 pphr Effect of fuel injection on the flow field at fuel flow rate of 45 pphr Effect of fuel injection on the flow field at fuel flow rate of 55 pphr Normalized local Equivalence ratio profiles at fuel flow rate of 55 pphr Effect of the fuel injection on the CRZ and Jets wake regions Schematic showing FTIR and emission probe setup [12] Typical absorption spectrum [132] Emission measurements in action at low power Emission measurements in action at high power xiv

16 6.5 Emissions profiles at combustor exit at different power conditions (fuel is propane and there is no water injection) Difference between measured CO and Thermal CO (predicted using CEA) Emissions profiles at combustor exit vs. predicted flame temperature (predicted using CEA) Effect of water injection on NOx generation (fuel is propane) Effect of water injection on CO generation (fuel is propane) Emissions profiles at combustor exit at different power conditions (fuel is Methane and there is no water injection) NOx contours at the SAC exit at an equivalence ratio of 0.6 with no water injection NOx contours at the SAC exit at an equivalence ratio of 0.6 with 50 % WFR CO contours at the SAC exit at an equivalence ratio of 0.6 with no water injection CO contours at the SAC exit at an equivalence ratio of 0.6 with 50 % WFR O 2 contours at the SAC exit at an equivalence ratio of 0.6 with no water injection O 2 contours at the SAC exit at an equivalence ratio of 0.6 with 50% WFR Local Equivalence ratio contours at the SAC exit at an equivalence ratio of 0.6 with no water injection Local Equivalence ratio contours at the SAC exit at an equivalence ratio of 0.6 with 50% WFR NOx contours at the SAC exit at an equivalence ratio of CO contours at the SAC exit at an equivalence ratio of O 2 contours at the SAC exit at an equivalence ratio of Local equivalence ratio contours at the SAC exit at a global equivalence ratio of xv

17 Nomenclature P/P = Pressure drop percentage V = Total velocity u' = Fluctuating component of the axial velocity Va = Axial velocity component Vr = Radial velocity component RMSA = Root mean square value of the axial velocity R = Flare exit radius CFD = Computational Fluid Dynamics CDL = Combustion Diagnostic Lab SAC = Single Annular Combustor DAC = Dual Annular Combustor PIV = Particle Image Velocimetry LDV = Laser Doppler Velocimetry DGV = Doppler Global Velocimetry GTC = Gas Turbine Combustor EPA = Environmental Protection Agency EGR = Exhaust Gas Recirculation UHC = Unburned Hydrocarbons LDI = Lean Direct Injection RQL = Rich Burn Quick Quench Lean Burn DLN = Dry Low NOx LP = Lean Premixed LPP = Lean Premixed Prevaporized EGR = Exhaust Gas Recirculation CRZ = Central Recirculation Zone PVC = Precessing Vortex Core X, Y, Z = Cartesian coordinate system with the origin defined at the flare center AF = Air to fuel ratio by mass D = Swirler tip diameter D t h G G a FA K b K f th CO 2 CO C H HO 2 N N 2 = Swirler hub diameter = Axial flux of angular momentum = Axial flux of axial momentum = Stoichiometric fuel air ratio by mass = Backward reaction rate = Forward reaction rate = Carbon dioxide = Carbon Monoxide = Carbon atom = Hydrogen atom = Water vapor = Monatomic Nitrogen = Nitrogen xvi

18 NO NO 2 NOx SOx O 2 O OH Re S x N i g l = Nitrogen monoxide = Nitrogen dioxide = Nitrous Oxides = Sulphur oxides = Oxygen = Monatomic oxygen = Hydroxyl = Reynolds number based on the fuel droplet diameter = Swirl number = Mole fraction of the species i = Global Equivalence ratio = Local Equivalence ratio = Swirler vane angle = Pressure exponent xvii

19 Introduction Human beings reliance on combustion devices has increased tremendously in the twentieth century. Combustion is a part of almost all conventional power generation applications. For example: Steam power cycles, gas power Cycles, vehicles engines, aircrafts propulsion systems, heating applications, etc. However, the current work is focused on the Gas Turbine Combustion Chambers. Gas Turbine Engines are relied on in several applications. They are used in power plants to generate electricity, they are used to propel aircrafts, they are used to power tanks and they are even used to power some vehicles (Mazda RX7 and RX8). The Gas Turbine Combustor is considered the heart of the Gas Turbine Engine. Combustion is a complex phenomenon that involves a strong interplay between fluid mechanics, heat transfer, mass transfer and chemical reactions. Until the 1960 s, the design objective of Gas Turbine Combustion Engineers was to design a combustor that produces the required thrust or power. Yet, the design requirements changed drastically in the last few decades. Pollutant Emissions emanating from the Gas Turbine Engine became of concern because of their considerable harmful impact on the environment. Emissions control techniques as well as new combustion chamber design concepts evolved to mitigate the emissions problem. However, these advancements were associated with the occurrence of sever stability problems due to some technical reasons (will be discussed later in the literature review section of the thesis). Moreover, new fuels (like Synthetic gas) were considered to alleviate the emissions problem as well as to moderate the consumption of oil which is expected to be depleted soon. New and challenging problems arose because of the 1

20 fuel variability and its effect on important characteristics like flashback, auto-ignition, lean blow out and generation of thermo-acoustic instabilities. This research is concentrated on three problems encountered in Gas Turbine Combustors. First problem is the problem of controlling the flow field structure. Second problem is development of thermo-acoustic instabilities and the parameters influencing such instabilities. The Third problem is the problem of the emissions emanating from the combustion chamber and how to effectively lessen such emissions. The research area of thermo-acoustic instabilities is a wide area and is an active field of research. The major motivation behind the current work is to investigate the relation between the flow field structure and thermo-acoustic instabilities. Several questions are posed: where and why are the combustion instabilities occurring? Are they linked to the swirling flow or the primary jets or both? If the jets are involved, then what parameters would affect the jets and result in the heat release fluctuations? Thermo-acoustic instabilities occur when fluctuations in heat release rate are coupled with one or more of the combustion chamber natural acoustic frequencies. On the other hand, heat release fluctuations are strongly tied to the combustor aerodynamics because an oscillation in a high turbulence region, involving chemical reaction, will yield an oscillation in the heat release rate. This thesis, based on the previous discussion, is divided into Eight Chapters. In Chapter 1, the literature review on aerodynamics, dynamics, and emissions is presented. In Chapter 2, the effect of the method of admission of the swirling flow to the combustion chamber on the turbulence activities and the flow structure is investigated using LDV, on a rectangular chamber. Two configurations are tested to conduct the study. The first 2

21 configuration provides the chamber with a swirling flow that expands suddenly while the second configuration provides the chamber with a swirling flow that expands gradually (on the dome surface). In Chapter 3, the base configuration is installed in a realistic Gas Turbine Combustor Sector and LDV as well as PIV are used to study the flow field structure under different operating conditions. The effects of the primary jets size, off-centering, partial blocking, overall pressure drop and interaction with convective cooling air on the flow field structure are presented. Also, a reconstruction of the CRZ in 3D is presented in order to help understand the effect of confinement on the volume of the CRZ. In chapter 4, the acoustic spectrum of the combustor is constructed using a microphone at three different pressure drops as well as three different pre-heat temperatures. Several captured videos, recorded using a high speed camera and regular cameras, are used to understand the progress of the reaction zone at different power conditions as well as the corresponding locations or modes of heat release oscillations. In Chapter 5, the fuel air mixing in the combustor sector is studied using PIV, probe and a paramagnetic analyzer. The influence of the fuel injection on the aerodynamics is presented. The impact of the fuel injection, at different fuel flow rates, on the primary jets wake regions, a potential source of ignition, is investigated. In Chapter 6, the emissions emanating from the sector at different power conditions, with propane and methane, are measured using an emission probe and FTIR. The effect of water injection on the generated NOx is studied at two different water flow rates. The exit section of the combustor is scanned at an equivalence ratio of 0.3 and 0.6 to determine the contours of different species as well as to study the degree of uniformity in mixing. 3

22 Finally, the conclusions of the current work are summarized in Chapter 7 and the future work is proposed in Chapter 8. 4

23 Chapter 1. (Literature Review) In the present chapter the literature review is presented. The review includes, mainly, three topics: aerodynamics, air fuel mixing and emissions. A short review on combustion instabilities will be presented because details are beyond the scope of the current research. The Aerodynamic literature review begins with a general description of the generation of the CRZ. The swirl cup, device commonly used to generate the CRZ, is discussed. Also, the Swirl Characterization and major factors affecting the CRZ are presented. This is followed by a review of the experiments conducted on more realistic combustion chambers with and without transverse dilution jets. The emissions literature review begins by a discussion of the mechanisms of generation of NOx and CO. Consequently, emissions control techniques that are currently used or being developed to control combustion emissions are presented. Techniques like Water and steam injection as well as Dry Low NOx (DLN) techniques are discussed. 1.1 Aerodynamics Researchers have been studying Aerodynamics in reacting and non-reacting flow fields for decades. There is no doubt that Aerodynamics play a vital rule in stability, emissions as well as dynamics issues related to GTC. If you try to light up a match or a cigarette lighter or even while barbequing you will come to know the importance of aerodynamics to the combustion process. Sir William Hawthorne said: With several others, I was sent on loan from the Royal Aircraft Establishment to lead a group on combustion at Power Jets [in 1940]. I had done my thesis on laminar and turbulent diffusion flames and knew the importance of aerodynamics in the combustion process. It 5

24 surprised me that others did not see that as much care was required in characterizing the aerodynamic features of a combustion chamber as in the design of a blade for a compressor or turbine. [1] A successful combustor or burner is the one that provides a stable flame. The word stable means a flame that is not extinguished over a wide range of operating condition. It is a flame that is maintained and doesn t blow off or flash back even if equivalence ratio, pressure or temperatures are varied (within an acceptable limit). Anchoring the flame (or stabilizing the flame) is commonly achieved by means of introducing a Swirl to the air flow. The idea depends on the generation of a Central Recirculation Zone (CRZ) which, as the name implies, recirculates a portion of the hot gases back to ignite the fresh coming mixture. Indeed, flame propagation depends on mass and heat transfer. Both mechanisms are enhanced significantly with the establishment of a CRZ. Once a CRZ is established we have better mixing, better combustion efficiency, shorter flame and improved blowout limits [2-4] Flame stabilization and Generation of CRZ Swirling flow is usually the method used to stabilize the flame in GTC. Swirling flow is, commonly, generated using vanes that impart a tangential component to the main flow velocity. There are two types of swirlers. They are classified according to the direction of flow (axial and radial swirlers). Figure 1.1 illustrates a typical coaxial axial and radial swirler. The term coaxial refers to two swirlers having the same axis as depicted in Figure 1.1. As the flow exits the swirler a vortex is formed and this vortex breaks down in different ways or modes. However, the bubble mode of break down is most common for the range of 6

25 Reynolds number and swirl numbers encountered in GTC [2, 3]. The flow structure, factors affecting the size of the CRZ and different methods of swirl generation for simple swirlers are well documented (for example, ref [2-4]). It was commonly believed that the CRZ is a closed loop which has minimum interaction with the surrounding flow [2, 5]. However, with the aid of flow visualization as well as instantaneous flow measurements, it became well established that the CRZ has a highly periodic asymmetric instantaneous structure (for example, ref [6-12]). In a pioneering paper, Sadanandan, R. [13] used simultaneous PLIF and PIV to study the relation between the location of the reaction zone and the flow field structure. He showed that the small scale instantaneous eddies that develop in the flow field strongly influences the orientation of the reaction zone. The burner he used was a non premixed combustor model. It is noteworthy that similar conclusions were deduced by U. Stopper et al. [14] while performing simultaneous PIV and LIF investigations on a premixed combustor model. The CRZ should be studied carefully for several reasons. Firstly, we need to understand the combustion phenomenon to help improve the performance of combustion devices. Secondly, the stability of the flame is influenced by the strength and size of the CRZ. Thirdly, the CRZ is the source of ignition and it impacts the heat release process. Consequently, all factors that impact the CRZ must be studied carefully. We cannot rely on CFD alone because of two reasons. First, the simulation of the complex, highly unsteady 3D, Reacting flow in a combustor is a challenging process and is still in the development phase. Even the good numerical studies that were conducted relied on experimental measurements to determine the order of magnitude of turbulence intensities and boundary conditions to 7

26 successfully model the problem! Secondly, the experimental results are essential to validate the Numerical schemes Swirl Cup 1.1 Typical coaxial Axial and radial swirlers. Few years ago, swirl cup was introduced as a unique device to prepare the air fuel mixture and generate a CRZ that is necessary for anchoring the flame. It is the typical fuel air preparation device used in several GEA combustion systems (T700/CT7, TFE738, CF34, F404/414, CFM56, CF6, GE90 and LMS100). The counter rotating coaxial radial swirl cup arrangement used in this study is illustrated in Figure 1.2. The inner swirler is connected to a venturi tube as demonstrated in Figure 1.2. The area ratio of the secondary to primary swirler is greater than 1. The secondary swirler is connected to a conical flare. A typical Single Annular Combustor (SAC) with swirl cup is shown in Figure 1.3. Many studies were conducted on counter-rotating swirl cup. The flow field has been studied by Mehta et al. [15]. The empirical design approach as well as CFD were conducted by Mongia et al. [16-22]. The Reynolds number effect on the flow structure of a counter rotating swirl cup was studied by Yongqiang Fu, et al. [23]. Effect of fuel and equivalence ratio on spray was studied by Yongqiang Fu, et al. [24]. In addition, Yongqiang Fu, et al. 8

27 investigated the Confinement effect on the flow field [25]. Hukam Mongia [26] discussed in details the bench mark experiments that were conducted on the swirl cup and compared the measurements to the CFD results. The swirl cup may have different configurations. The earlier swirl cup used by GEA had primary swirling jets rather than primary vanes. This swirl generation method is well established. The CFM56 engine uses this kind of swirl cup. Figure 2.4 shows the typical assembly of the CFM56 swirl cup. This specific hardware was subjected to numerous studies. Aerodynamics and effect of confinement were studied by Elkady [12]. The two phase swirling flow measurements, without reaction, for droplet characterization and droplet dynamics have been conducted by Wang et al. [27-31]. 1.2 Typical Swirl cup arrangement with counter rotating radial swirlers. 9

28 1.3 Typical SAC Combustor with swirl cup. 1.4 CFM56 swirl cup assembly Swirl Characterization The Swirl is usually characterized using a non dimensional number called the swirl number ( S N ). It is defined as the ratio between the axial flux of angular momentum (G ) and 10

29 the axial thrust ( G ) multiplied by the swirler exit radius ( R ) as shown in Eq. (1.1). As the x S N increase we have a stronger swirling flow. If The SN value is usually used to classify the flow as either a strong or weak swirling flow. S N > 0.6 the flow is considered as a strong swirling while if S N < 0.6 it is a weak swirling flow [2]. Weak or strong refers to whether a CRZ will develop or not. However, the 0.6 value of SN is not a universal value. The value of S N at which the CRZ will form depends on the flow rates, operating conditions and swirler configuration (axial, radial, helical etc). Usually swirlers for Fuel Rich GTC are designed with a value of S N around 1.0. For each swirler experimental measurements are vital because they will delineate the CRZ characteristics. The S N could be expressed in a simpler and more practical form based on some simplifying assumption. If we assume a constant axial velocity in the radial direction, neglect the static pressure gradient in the radial direction and assume a constant vane angle we can use Eq. (1.2) to obtain a good estimate for the value of S N. Furthermore, assuming a hubless swirler will result in Eq. (1.3) which provides a very simple and quick estimation for the value of S N. Some experiments were run on an axial swirler at the Combustion Diagnostic Lab (CDL) at the University of Cincinnati and it was found that the values of S N using Eq. (1.1), (1.2) and (1.3) were 1.175, and respectively. V t and V a are the tangential and axial velocity components respectively. D hs and respectively, P is the static pressure and v is the vane angle. D ts are the swirler hub and tip diameters 11

30 S G 0 t a N R R x 0 a a 0 R (V r ) V 2 dr GR V V 2 rdr P2 rdr (1.1) S D h / Dt tan (1.2) 3 1 D h / Dt N 2 S 2/ 3tan (1.3) N Factors influencing the CRZ Size and strength The CRZ is the source of continuous ignition in the combustor and it has to be studied carefully. The size and strength of the CRZ are directly related to flame static stability. In the following sections, different factors influencing the CRZ are discussed Swirl Number Effects Swirl number effects on the CRZ have been reported by several researchers (for example: [2-4, 32]). It is well known that increasing the value of S N increases the strength and size of the CRZ. Beer and Chigier [2] classified the flow field, based on the numerical value of the swirl number, to a flow with weak swirl and strong swirl. Fu et al. [32] reported that as the value of S N increases initially a weak lifted CRZ is established over the burner rim. Further increase of the S N (as it approaches 1.0) results in the establishment of strong CRZ close to the burner rim Confinement Effects The confinement size has a strong impact on the CRZ. Several experiments were conducted to study the effect of confinement on the CRZ [25, 33-36]. Beltagui and Maccallum [33-34] reported sharp changes in the dimensions of the CRZ with change in the furnace dimensions. Syred and Dahman [36] demonstrated that the confinement ratio 12

31 strongly influences aerodynamics, heat transfer and stability limits. Fu et al. [25] studied the effect of confinement on the CRZ generated using a swirl cup with counter rotating co-axial radial swirlers. The results they presented showed that the increase of the chamber cross sectional area is associated with an enlargement in the size of the CRZ, but with a reduction of the CRZ strength (magnitude of axial velocities). Basically, more mass is recirculated as the CRZ size increases. Interestingly, they showed that there is a certain critical chamber dimension beyond which the CRZ noticeably reduced in size, but became very strong. As the confinement is totally removed the CRZ became short and strong because of the high rate of entrainment that takes place at the shear layer. Again, this is not a universal behavior because it depends on the type of swirler generating the swirling flow. Fu et al. [37] conducted similar experiments using a helical axial swirler. They reported that without confinement the size of the CRZ was enormous, but it was very weak. This emphasizes that the confinement impact on the characteristics of the CRZ is dependent on the method of swirl generation as well as the confinement ratio Reynolds Number Effects Sarpkaya, T. [38-39] studied the relationship between the Reynolds number and vortex breakdown. In his work, he presented a plot that linked the vortex break down to the Reynolds number and the Swirl number [39]. He reported three types of breakdown: spiral breakdown, helix breakdown and axisymmetric breakdown. Also, Sheen et al. [40] studied the effect of Reynolds number on the CRZ. They demonstrated that there are seven typical flow patterns that occur in a CRZ. The seven patterns are stable flow, vortex shedding, transition, penetration, vortex break down and attachment. Fu et al. [23] studied the effect of Reynolds number on the flow structure in a dump combustor. The Reynolds number was 13

32 increased by increasing the pressure drop across the chamber. They reported that the size of the CRZ remained almost constant and didn t depend on the Reynolds number. However, increasing the Reynolds number significantly escalated the strength of the CRZ (higher ve velocities). They tested the hardware at three different Reynolds numbers. The normalized velocities were almost identical though the Reynolds number was different. This appears to be usually the case at high Reynolds number Nozzle Insertion Effects The location of the fuel nozzle strongly influences the generated CRZ. Fu et al. showed that the effect of fuel insertion length is very similar to the effect of varying the swirl number [25]. As the fuel nozzle insertion length decreases, the adverse pressure gradient is augmented and the CRZ becomes larger and stronger Combustion Effects Syred and Beer [4] showed that the formation of a CRZ enhances the stability of the reacting flow in combustors. Syred et al. [41] conducted experimental investigations on a burner to study the effect of combustion on the CRZ. They concluded that for this specific combustor there is a slight difference in the flow field structure with and without combustion. On the other hand, several researchers reported that combustion has strong impact on the size of the CRZ. Chigier and Dvorak [42] used LDA to investigate the effect of combustion on a free swirling flow. They demonstrated that the combustion results in an enormous increase in the turbulent activities. Fuji et al. [43] studied the influence of combustion on an unconfined swirling flow. They demonstrated that the velocity profiles are significantly different in the reacting case when compared to the isothermal case. Gouldin et al. [44] studied the effect of combustion on the flow field using two different swirler configurations: counterswirl and co- 14

33 swirl. Interestingly, for cold flow the co-swirl didn t generate a CRZ. However, both configurations generated a CRZ under reacting conditions. Bulzan [45] used LDA to study the structure of a reacting spray swirl stabilized flame. He demonstrated that the gas velocity was higher for the reacting case because of the expansion associated with chemical reaction. Also, he showed that the length of the CRZ decreased, but the strength was augmented. El- Kady [12] conducted PIV measurements on a dump combustor and showed that combustion results in decrease of the size of the CRZ. Combustion is associated with expansion and increase in flow velocity (axial velocity). Consequently, the axial flux of momentum will increase and the Swirl number will decrease. Usually, this is referred to as the effective swirl number [12]. We have to keep in mind that several of the previous discussions applies to dump combustors with no primary dilution jets. When Primary jets are involved, they define the termination point of the CRZ Realistic combustion chambers and effect of dilution jets The word realistic refers to the presence of secondary and dilutions zones. In other words, it refers to the presence of primary and secondary or dilution jets. Interaction between swirling flow and dilution jets defines the flow field structure and influences the flame stability and emissions. Primary dilution Jets terminate the CRZ and complete the combustion process. It is a common practice to design the dilution jets to penetrate at least 40% of the primary zone width [46]. Secondary dilution jets follow for trimming the exit temperature profile and to produce a profile that matches the turbine thermal loading requirements (also known as required pattern factor). We have mentioned previously that the length of the CRZ depends on several parameters. However, in real GTC the transverse, primary, air jets define the termination point of the 15

34 primary zone [47]. Richards and Samuelsen reported that the location of the transverse jet dictates the behavior of the combustion process. They suggested that a placement of the transverse primary jets downstream of the swirler exit with half to one duct diameter will results in optimum penetration [48-49]. Stevens and Carrotte [50] conducted experimental investigations on the combustor dilution zone. They observed the generation of a complex flow field at the rear portion of each dilution jet. The flow field appeared to vary randomly at this location form one jet to another. This non uniformity resulted in an overall non uniform temperature distribution. Holdeman [51] studied the dilution jet region for a gas turbine combustor and found that the momentum flux ratio of the jets controlled the exit velocity and temperature profiles. Anacleto et al. [52] conducted experiments on a non-reacting water model of a can-type combustor. The combustor was a typical representation of a Rich-burn, Quick-quench, Lean burn (RQL) combustor. They deduced that the opposed jets in the center of the combustor gave rise to large velocity fluctuations. Doerr et al. [53] experimentally investigated an RQL combustor and demonstrated that there exist an optimum momentum flux ratio which will result in a homogenous temperature distribution at the combustor exit. Gritsch et al. [54] studied the effect of slot cooling on the dilution jets performance. They stated that the jet penetration was maximized at the highest momentum flux ratio and minimum slot-cooling blowing ratio. As the momentum flux ratio was reduced and the blowing ratio was increased, they noticed a reduction in the jet penetration. Goebel et al. [55] conducted velocity, turbulence and temperature measurements downstream of a reacting small scale combustor. The major finding of their work was that with an appreciable degree of swirl the dilution jets reduced the turbulence mixing and disrupted the flow significantly. 16

35 On the contrary, for the case with no swirl or weak swirl the jets enhanced the mixing. Also, they demonstrated that the exit temperature profiles are strongly dependent on the dilution jets. It is noteworthy that at low swirl the turbulence levels were roughly constant with and without combustion. Gulati et al. [56] conducted mean and fluctuating temperature measurements on a full scale DAC (Dual Annual Combustor). The dilution jets were found to increase the temperature fluctuations. Chirstoph Hassa et al [57] used LDV, Doppler Global Velocimetry (DGV) as well as PIV to investigate the flow field in a quadratic cross sectional combustion chamber. The test section didn t provide complete optical access in the dilution jet plane. Suad Jakirlic et al. [58] utilized PIV as well as time resolved flow visualization techniques to study the isothermal flow field in a rectangular SAC sector (dump combustor). P. Koutmos and J. J. McGuirk [59] conducted LDV measurements on a water model can-type combustor to study the interaction between the swirling flow and radially injected dilution jets. Chirstoph Hassa et al. [60] performed LDV measurements on a scaled down simple planar sector of an annular sector combustor (also known as Fuel Rich Dome Combustor). The sector they used was made from aluminum and fuel was introduced through an air blast atomizer. The dilution jets in their test section had a stagger arrangement. The aforementioned experiments conducted on reacting and non reacting flow show that there are several important common characteristics and few differences between both modes. Mainly, the differences are located in the primary zone close to the swirler exit. The secondary and dilution zone have several similarities. The turbulence levels in these regions appear to be very similar. Also, in both modes the dilution jets appeared to dictate the combustor exit velocity profiles. CFD plays an important role in decreasing the combustor 17

36 design time and design optimization. However, the computational models appear to lack the ability of modeling the dilution zone and combustor exit turbulence activities. 1.2 Emissions Hydrocarbon fuels are composed of Hydrogen (H) and Carbon (C), as the name implies. The energy of the fuel is released when it is mixed with an oxidizer and ignited in what is known as a combustion process. An ideal Combustion process, by definition, means that the C will be oxidized into Carbon Dioxide (CO 2 ) and the H will be converted into Water vapor (H 2 O). Air is usually the oxidizer used because it is available for free and in abundance. Air is composed of 21% Oxygen (O 2 ) and 79% Nitrogen (N 2 ), ratio by volume. The N 2, ideally, should go in and out without interacting or interfering with the combustion process. Figure 1.5 shows an illustration of the ideal and real combustion process. Real combustion is different than ideal combustion. The emissions resulting from the combustion process usually also include Unburned Hydro Carbons (UHC), NOx (Nitrogen Oxides are the sum of Nitrogen monoxide (NO) and Nitrogen Dioxide (NO 2 )), Carbon Monoxide (CO), Sulfur Oxides (SOx), Oxygen (O 2 ), and may be some soot (C) as illustrated in Figure 1.5. Not all emissions emanating from the combustor are toxic or considered as pollutants. CO 2, O 2, H 2 O, and N 2 are not toxic. However, CO 2 strongly influences the global warming. On the other hand, UHC, NOx, CO, SOx and soot are pollutants. It is worth mentioning that these pollutants form a very small percentage of the combustion emissions as depicted in Figure 1.6. Nevertheless, they have severe harmful effects on human beings as the environment. The harm on human beings includes lung cancer, lung disease and eye irritation and at high concentration CO and NOx might cause death. The impact of the environment is 18

37 mainly in the form of smog and acid rains. Acid rains take place at mountains and affect the water sources which result in a degradation of the water quality as well as strong impact on marine life. 1.5 Illustration of Ideal and Real combustion emissions [61]. 1.6 Order of magnitude of Gas Turbine Combustion emissions [61]. 19

38 In the last few decades, numerous amount of combustion application came to existence. For example: aircrafts, furnaces, Power plants, combined cycles and different types of transportation vehicles. Tons of toxic pollutants are released into the atmosphere every year. The Environmental Protection Agency (EPA) placed stringent regulations on combustion devices to reduce the impact on the environment. Different concepts to control combustion emissions have evolved. The regulations are becoming even more stringent and engineers and researchers are in a continuous challenge to develop new techniques and concepts to minimize the pollutant emissions and manufacture combustion engines that cope with the imposed regulations. In the current research we only focus on Gas Turbine Engines. They are usually involved in power generation, aircrafts and combined cycles. The problem with emissions control is that the reduction process is a tradeoff between CO, NOx and UHC as illustrated in Figure 1.7. The mixing process is of prime importance in the combustion process especially when it comes to Liquid fuels. High combustion efficiency is strongly related to the mixing process. Having mentioned air fuel mixing, It is worth mentioning the pioneering paper authored by V. G. Mcdonell and G. S. Samuelson in the field of fuel air mixing measurements and diagnostic [62]. They reviewed different measurement techniques for liquid fuel as well as gaseous fuel. It was mentioned earlier that usually simple combustion chambers are used in experimental investigations. Nevertheless, primary jets have significant effect on Lean Blow Out Limit (LBO) as well as the NOx and CO emissions. D. Shouse et al. [63] studied the aforementioned effects in the pioneering research they overtook in They showed that with no transverse jets the NOx as well as CO emissions emanating from the combustor are 20

39 the minimum possible. Once transverse jets are introduced the pollutent emissions increases. Also, they showed that there is an optimum diameter for the transverse jets at which the NOx and CO emissions could be brought to a minimum. However, this minimum is still considerably more than the one achieved with no transverse jets. They also showed that the best LBO limits are achieved at the same optimum diameter. 1.7 Typical pollutant formation trends as function of the equivalence ratio [12]. In the following sections the mechanisms as well as sources of generation of NOx and CO in GTC will be reviewed. Then, a review on the emissions control methods will be presented Nitric Oxides (NOx) The mixture of combustion products emanating from a combustion chamber consists of a considerable number of species as was discussed in the previous section. The easiest method to enable estimation of the species is to assume thermal equilibrium. This assumption assumes that the concentration of a certain species is constant and is not varying with time. This assumption is reasonable for several species, but is not for both NOx and CO. Both 21

40 should be modeled using chemical kinetics (reaction rates control the formation process). In other words, NOx and CO generation is kinetically controlled. The NOx variation vs. equivalence ratio is illustrated in Figure 1.8 Typical variation of the equilibrium and thermal NOx vs. equivalence ratio The figure shows the variation of NOx estimated using thermal equilibrium as well chemical kinetics. The curves are used to show trends and relative values only. It is clear that the thermal NOx is significantly higher than the thermal NOx. The chemical kinetics equations that enable correct modeling of NOx will be discussed in the following section. The importance of finite reaction rates or reaction time (also known as residence time) is illustrated as shown in Figure 1.9. As the reaction progresses the NOx concentration increases until it eventually reaches the equilibrium concentration Equilibrium NO Chemical Kinteics NO 3000 NO (PPM) Typical variation of the equilibrium and thermal NOx vs. equivalence ratio. 22

41 Thermal equilibrium NO 3000 NO (ppm) Chemical kinetics NO Residence time (ms) 1.9 Illustration of the effect of residence time on thermal NOx NOx categories The summation of NO and NO 2 is called NOx. NO 2 is a red brown gas that is irritating and very toxic. On the other hand, NO is colorless, odorless, but also a very toxic gas. There are three different categories of NOx: 1. Thermal NOx (also known as Zeldovich mechanism). 2. Fuel NOx. 3. Prompt NOx. Thermal NOx are formed due to the oxidation of the Nitrogen (N 2 ) (found in air) with the O 2 (also from air). Fuel NOx is a result of the oxidation of the Fuel Bond Nitrogen (FBN). Prompt NOx or Fenimore NOx occurs in Fuel Rich hydrocarbon flames due to a sudden generation mechanism, as the name implies. Rokke et al. [64] provided useful formulas that enable estimation of the local production rate of prompt NOx as well as thermal NOx for methane, propane and natural gas flames (buoyancy controlled turbulent diffusion flames). The equations provide very useful information regarding the relative contribution of each 23

42 type of NOx. However, Rokke et al. [64] stated that uncertainties in the estimation are about a factor of 2 for the case of thermal NOx and may reach up to 10 for the prompt NOx. Fernando Biagioli and Felix Güthe [65] presented a new classification of the NOx emissions that is useful for pre-mixed burners. There classification is based on Damkohler number criterion. NOx production is divided into two portions: First, a prompt (fast) contribution that occurs within the heat release region. Second, a postflame (slow) contribution that takes place in the combustion products. In the following paragraphs we proceed with the conventional classification and we discuss each mechanism separately. 1. Thermal NOx Zeldovich studied the mechanism of formation of NOx and he provided two equations (1.4 and 1.5) [66-67] that govern the formation process. These equations are well known as the Zeldovich mechanism or Zeldovich pathway. Equation 1.4 shows the first pathway of generation of direct thermal NOx. A molecule of the N 2 is broken into two free nitrogen atoms. One N atom reacts directly with the O atom to produce NO. Then, the other N atom is oxidized with O 2, following the pathway shown in equation (1.5). Lavoie et al. [68] extended the Zeldovich mechanism and provided equation (1.6) to enable better quantification of the generated thermal NOx. Equation (1.6) shows another possible pathway such that the N atom released from equation 1.4 is attacked by the OH radical and an NO atom is formed. The thermal NOx generation is initiated through equation (1.4). The forward and backward reaction rates are in given in equation (1.7) [69]. T is temperature in K and the reaction rates are expressed in cm mol s. Details about the possible solution to this set of equations are beyond the scope of this work. 24

43 O+N2 k f 1 NO+N (1.4) k b1 N+O2 k f 2 NO+O (1.5) k b 2 k f 3 N+OH NO+H (1.6) k b3 k k k k k k f 1 b1 f 2 b2 f 3 b exp( 38,370 / T) exp( 425 / T) exp( 4680 / T) exp( 20,820 / T) exp( 450 / T) exp( 24,560 / T) (1.7) Turns [70] provided an easy and practical way to enable understanding of the main parameters that influence thermal NOx generation. Equation (1.8) shows a one step global reaction for the formation of thermal NOx. If we only consider equations (1.4) and (1.5), assuming that N atoms have reached a steady concentration and assuming that O 2 and O are under equilibrium then we can deduce equation (1.9) which provides the rate of change of thermal NOx with time ([N 2 ] and [O 2 ] are the molecular concentration of Nitrogen and Oxygen respectively). The equation shows that the NOx rate of formation is strongly dependent on temperature (reaction rates has exponential dependence on temperature). Also, Oxygen and nitrogen concentration strongly influence the thermal NOx generation. Molecular concentration of a species i (N i ) is calculated using equation (1.10). P is the pressure, R c is a conversion constant and x i is the mole fraction of species i. Now we can also see that NOx formation rate is strongly dependent on pressure (this statement will be discussed later). K G N O NO (1.8)

44 o 1/2 1 KP p 2 f1 O 2 eq 2 RuT d NO 2 k [ ] N dt (1.9) (1.10) N P/ R T *x i c i To estimate the NOx with the aforementioned mechanism, it is possible to assume that O, N, OH, and H are in equilibrium as suggested by Zeldovich [66-67]. This assumption is reasonable for N, OH and H, but not for O. Westenberg [71] and Fenimore [72] discussed that issue. Usually O exists for a short period of time in excess of equilibrium in the reaction zone. Fenimore [72] stated that Oxygen may be assumed in equilibrium in the hot products, but not in the low temperature flames. Iverach et al. [73-74] studied experimentally the NOx generation from fuel lean and fuel rich hydrocarbon flames as well as non hydrocarbon flames. The Zeldovich mechanism was accurate enough for NOx estimation in the case of non hydrocarbon flames. However, for hydrocarbon flames and under fuel rich conditions there were discrepancies. This is attributed to the presence of radicals in excess of equilibrium concentration ( Bowman [75]). Mohammad, B. S and S. M. Jeng [76] developed a computer code for the preliminary design of Single Annular Combustors (SAC). One of the design modules enables estimation of the NOx by solving the Zeldovich mechanism. Their code shows that slight change of the O concentration will result in strong impact on the estimated NOx, while a slight change of the O 2 concentration doesn t. This agrees well with the point addressed by Westenberg [71] and Fenimore [72]. Few studies were conducted to study the local formation of NOx in swirl stabilized combustors. The results of Elkady et al. [77] and P. Schmittel et al. [78] shows that NO generation occurs at regions of high temperature. The results they presented shows that the 26

45 CRZ is the source of generation of NOx at low equivalence ratio while the post recirculation zone is the source of generation for higher equivalence ratios that approach unity. However, the general trend of NOx emissions at the combustor exit is still the same. In other words, as the global equivalence ratio increases the NOx concentration at the combustor exit increases. 2. Fuel NOx Some fuels such as residual oil contain Nitrogen that is directly bonded with the fuel molecule (FBN). This FBN is the major source of NOx. One may think that there is no difference between thermal NOx and Fuel NOx because both depend on Nitrogen oxidation. This is true. However, the atmospheric Nitrogen responsible of the generation of thermal NOx is highly stable because of the triple atomic bond. On the other hand, FBN is directly related to the fuel through a single bond. Once the fuel burns, the nitrogen is released and it dissociates into highly unstable nitrogen which will either be directly oxidized with oxygen or will bond with another nitrogen atom to form N 2 [79]. 3. Prompt NOx This category of NOx is also referred to as flame NOx. The reason being the sudden or prompt NOx (as the name implies) that was observed at the flame front while studying this category of NOx. Fenimore [72] investigated fuel rich Hydro Carbon flames and found that NOx generation occurs suddenly in a short period of time. He suggested that equations (1.11) and (1.12) contribute to the formation of prompt NOx. CH radical in the flame region attacks the N 2 molecule and HCN and N atoms are formed as demonstrated by Equation (1.11). This reaction is followed by rapid oxidation of the N and HCN to NO as illustrated by equation (1.13). Equation (1.11) is considered the initiation reaction for prompt NOx generation. The N generated from equation (1.11) is oxidized as illustrated earlier by equation (1.5). Another 27

46 possible reaction could occur if the C 2 radical attacks the N 2 resulting in formation of 2CN which also rapidly forms NO due to oxidation as illustrated by Equation (1.13). Bowman reported that Fuel Rich premixed flames emit more than 50% of the NOx through the hydrocarbon radical-molecular nitrogen. Drake and Blint [80] reported that the prompt mechanism may account for more than 94% of the total NOx emissions. Miller and Bowman [81] summarized the set of equations that might contribute in the generation process. CH+N2 HCN N (1.11) C N 2CN 2 2 (1.12) HCN CN NCO NO (1.13) Effect of pressure on NOx generation Usually combustor testing is conducted under atmospheric pressure conditions because of the high complexity and cost associated with high pressure testing. A great deal of new combustor design performance is evaluated at atmospheric pressure conditions. The question is usually how to extrapolate the results at atmospheric pressure to high pressures. D. A. Sullivan [82] suggested that the NOx dependence on pressure is approximately P 0.5. One would speculate that the increase in pressure will be accompanied by an increase in the NOx concentration. However, several other researchers showed that an increase in pressure will yield a reduction in the NOx concentration [83-85]. Correa [86] reviewed the effect of pressure on NOx generation and proposed that the ratio of NOx concentration at high pressure to that at atmospheric pressure is better to be represented in the terms of ([NOx]/[NOx] atmospheric ) P. is a pressure parameter that varies from corresponds to very lean premixed conditions (low equivalence ratio) and increases as stoichiometry is 28

47 approached (high equivalence ratio). The results obtained later by Leonard [87] and Mongia et al. [88] agree very well with the conclusions of Correa [86]. According to the classification of Fernando Biagioli and Felix Güthe [65], NOx contribution in the heat release zone (fast NOx) is less sensitive to the flame temperature than the NOx released in the postflame zone (slow NOx). They also showed that the fast NOx decreases with increase in pressure (pressure exponent,, is -0.45) while the slow NOx increases with the increase in pressure ( is 0.67). It is noteworthy that they suggested that unmixdness of the fuel-air is the reason for the observed increase in NOx concentration with increase in pressure NOx reduction techniques Before going through details it is important to summarize the factors that strongly affect NOx generation (based on the previous discussion). These factors are: Temperature and residence time as illustrated by Figure 1.8 and 1.9 respectively (equivalence ratio is a representative of the flame temperature). So any attempt to reduce NOx will involve reduction of the flame temperature and/or residence time. A review of the literature shows that there are three major concepts that enable NOx reduction: Diluents injection concept (also known as wet combustion), Dry Low NOx (DLN) concept and Catalytic combustion concept. The three concepts are discussed in the next few paragraphs. The reduction technique known as flameless combustion was not included in the previous classification because this concept is similar to the diluents injection concept. It reduces the NOx based on low flame temperature through the recycling of a portion of the exhaust. The major difference between this method and Exhaust Gas Recirculation (EGR) is the large portion of the exhaust recycled in the flameless combustion technique. 29

48 1. Diluents injection This method involves injection of diluents to reduce the flame temperature and consequently reduce the generation of thermal NOx. The Diluents include water, steam and exhaust gas. All these diluents were introduced to the combustion chamber and a successfully reduction of the NOx was achieved. In a pioneering paper, Lefebvre [89] reviewed all three methods and discussed the pros and cons of each method. Water injection appears to be the most effective method in reducing NOx. The water is either injected directly in the flame region or in the air admitted to the dome. Reduction of the NOx concentration up to 80% was reported [90-91]. L. Daggett et al. showed that water injection is even possible in aircrafts [92]. They reported NOx reduction of 80%. However, the water injection was limited to takeoff only (or the high power conditions). Several problems are associated with water injection: generation of instabilities, the need of a costly water treatment system to reduce erosion effects and increased system maintenance. The steam injection effect on NOx is less pronounced than water injection. Steam injection was reported to decrease the NOx emissions with 60% [93]. The advantage of steam is in the uniformity achieved in the flow field with reduced impact on the combustor aerodynamics. It has to be mentioned that the significant reduction of NOx associated with water or steam injection was observed for thermal NOx only. It appears that the influence of water injection is significant on the initiation reaction (equation 1.4). EGR involves recirculation of a portion of the relatively cold combustion products and injecting it to the air prior to its admission to the combustion chamber. Reduction of the NOx up to 50% was reported. The influence of the EGR on the NOx is due to the reduction of O 2 as well as the increase in the heat capacity (due to the presence of H 2 O and CO 2 ). The major 30

49 disadvantage of this technique is the need of an intercooler to reduce the temperature of the exhaust gases before re-admission to the air. 2. DLN DLN concept was introduced by researchers in an attempt to provide the industry with low NOx combustors that are more robust and don t involve mechanical complication like those involved in the wet NOx combustors. The DLN concept is based on designing the combustor in a new fashion which results in a flame with relatively low temperature and consequently the generated pollutant emissions are brought to a minimum. In other words, to design a combustor that produces minimum emissions without addition of any external fluids. DLN is classified into three different concepts: Lean premixed (LP concept), Rich burn - Quick Quench-Lean burn (RQL concept) and Lean Direct Injection (LDI). The theory behind all three techniques is illustrated in Figure It is clear that all three methods avoid stoichiometry in order to reduce the flame temperature and hence reduce the thermal NOx generation. Figure 1.12 shows an illustration of all three methods with corresponding equivalence ratio and flame temperature [94]. LP concept, as the name implies, is a premixed type of combustion. The fuel and air are premixed in a mixer before they are introduced to the combustion chamber. This pre-mixing enables burning at low equivalence ratio and also results in better fuel air mixture homogeneity. The resulting flame is relatively clean and generates reduced amount of NOx. However, premixing means static stability problems due to the high risk of flash back to take place. Also, instabilities become an issue in this type of combustion. The LP combustor should be operated at the minimum possible equivalence ratio as depicted in Figure It is 31

50 noteworthy that if the utilized fuel is in the liquid form then the concept is known as Lean Premixed Prevaporized (LPP). The flow characterization, flame structure and flame velocities of LP combustors has been investigated by several researchers (for example Ref. [95-100] ). Even though the LP concept results in low NOx, but the system is well known to have problems with generation of instabilities as well as flashback problems. Both are considered stability problems with the former known as dynamics stability problem and the later known as static stability problem. N. Syred [101] mentioned that the high sensitivity of LP combustors to generation of instability is due to the absence of the inherently damping mechanism in non-premixed flame. A great deal of the research conducted on the LP concept involved investigations of the generated oscillation and possible methods to control and suppress such instabilities (for example Ref. [ ]). It is well established that the reason behind the generation of oscillation is the fluctuations in the equivalence ratio. The RQL concept was first introduced in 1980 by Mosier, S.A. and pierce, R. M. [106]. The combustor is divided into three stages. The combustion in the first stage takes place under fuel rich conditions. The fuel rich burning region increases the static stability of the combustor and significantly enhances LBO limit. Large amount of air is introduced abruptly in the secondary region to quench the reaction. Finally, lean burning takes place in the last stage. Many studies were conducted on the LPP concept [52, 89, ]. All conducted research showed that the success of the RQL combustor development is dependent on the design of the Quench region. The function of this region is to switch rich burn abruptly to lean burn without allowing the formation of hot streaks. The major disadvantage of this technique is the increased system size which makes it unsuitable for aviation applications. 32

51 Also, LPP was proven to result in lower NOx concentration than RQL. Nevertheless the RQL is safer and inherently stable when compared to LPP. LDI was introduced as a potential solution to the problems encountered in LPP and RQL. LDI concept is based on direct injection of the fuel in the combustion chamber in manner such that the resulting air/fuel mixture will be lean and homogenous. Hence, lower flame temperature and low NOx emissions will be achieved. The development of the LDI concept was overtook by Robert Tacina at NASA GRC [ ]. Usually, the combustor running under LDI concept is composed of multiple injection points as illustrated in Figure The figure shows that the combustor length is relatively short. This will result in reduced residence time which will result in lower overall NOx as illustrated in Figure DLN combustor concepts [61] Multiple point LDI combustor developed by NASA [114]. 33

52 1.12 Illustration of LP, RQL and LDI concepts [94] Carbon Monoxide (CO) CO is a colorless, odorless, tasteless and non irritating gas. It is a highly poisonous gas. An ideal combustion process takes place if all the Carbon atoms in the Hydrocarbon fuel are oxidized into CO 2. CO is very dangerous and it causes death at a concentration of 400 ppm (120 minutes exposure time). The severity of intoxication of CO is illustrated in The Carbon is considered the parent of CO. The Carbon, from the hydrocarbon fuel, goes through a sequence of reactions and it eventually transform into CO 2. CO may be present in the emissions because of one/or more of the following reasons: 1. The combustion takes place under fuel rich conditions and the available oxidizer is not sufficient to complete the oxidation process of C into CO High temperatures which increases the CO 2 dissociates into CO. 3. Significantly low temperatures at which the energy is not sufficient to complete the Carbon oxidation process. 34

53 1.13 Severity of Carbon Monoxide intoxications [115]. 1.3 Dynamics and generation of Combustion Instabilities A detailed review of combustion dynamics or acoustics and instabilities is beyond the scope of the current work. The literature review presented here will only address main aspects of generation of combustion instabilities. Combustion dynamics or thermo-acoustic instabilities (also known as combustion humming, Rumble, screech and growl) refer to the high pressure oscillation that develop in the combustion chamber, under some favorable conditions, and is associated with high amplitudes of sound level and vibrations that eventually result in hardware failure. The term thermo-acoustics refer to thermal effects coupled with system acoustics. The generation of instabilities follows the following sequence: oscillations in the heat release occur, the heat release oscillations are accompanied by pressure fluctuations, if the pressure fluctuations are in phase with one or more of the system (hardware or combustor) acoustic modes then the 35

54 instabilities are amplified resulting in the so called thermo-acoustic instabilities. Figure 1.14 shows an illustration of the generation and growth of such instabilities. The area of thermo-acoustic instabilities generation is an active field of research because of the devastating consequences of the development of such instabilities. Also, the new generation of combustors (LP) is prone to the generation of such instabilities. Tim Lieuwen [116] pointed out that there are three points of concern related to the generation of instabilities: what is the frequency of oscillations, under what conditions do they occur and what is the amplitude of the oscillations. Unfortunately, until the time of writing this thesis, prediction of the conditions under which they occur as well as predicting the amplitude of oscillations are in a preliminary phase. Candel, S. et al. [117] presented an overview of the combustion instabilities. They discussed numerical as well as experimental techniques implemented to diagnose generation of instabilities for non-premixed as well as premixed combustion modes. Similar effort was presented by Tim Lieuwen [116]. In an attempt to explore the factors affecting heat release oscillations as well as the regions susceptible for generation of instabilities, N. Syred [118] conducted several studies and suggested that heat release oscillations usually occur at the expanding swirling jet region close to the combustor front end. Also, N. Syred [101] studied the role of the Precessing Vortex Core (PVC) in generation of such instabilities (PVC is a common form of vortex break down that occurs in the primary region of the combustor). Tim Liewen [103] demonstrated that oscillations in the equivalence ratio has significant influence on heat release fluctuations and generation of instabilities. It is commonly believed that instabilities usually occur at regions of high turbulence activities. Having mentioned the rule of heat release oscillation on the development of thermo-acoustic instabilities, it is worth mentioning 36

55 that alleviating of such instabilities is primarily achieved by actively modulating the fuel flow rate [ ] (air flow modulation will also solve the problem of getting the waves out of phase, but it is not feasible). Diagnosis of dynamics is usually accomplished using a microphone accompanied by phase locked measurements (LDV, PIV and PLIF) to delineate the cycle of events (for example the research conducted by Mier [104]). One of the very successful techniques in diagnosing the amplitude and location of instabilities is the use of imaging techniques. Bruschi R. et al. [124] used a photodiode to diagnose the instabilities in a combustor model. An auto correlation of the pressure transducer signal and the photodiode signal revealed that the pressure transducer is not a suitable tool to diagnose the generation of instabilities. A breakthrough in the field of dynamics diagnostics was presented by Jun Cai et al. [125]. They applied fft to the high speed video and used the amplitude and phase information to diagnose the location, type and amplitude of oscillations. Normalized Pressure (p'/p) Number of Cycles 1.14 Illustration of generation and growth of pressure oscillations known as thermoacoustic instabilities [126]. 37

56 Chapter 2. (Dump Combustor Aerodynamic) In this chapter the aerodynamics of the Swirl cup installed in a dump combustor is presented. Two configurations are being studied. The major difference is in the degree of expansion that occurs with the flow admission to the combustion chamber. For the base configuration (we will use this name to refer to it through the thesis), the swirl cup flare is attached to a splash plate and a combustion dome. Hence, the swirling flow expands gradually. For the configuration one (this is the name we will use to refer to it), the flare is only attached to a splash plate. Therefore, the flow expands abruptly as it enters the combustion chamber. Later, in Chapter 3, we will be discussing the flow field with the base configuration installed in the SAC sector. 2.1 Experimental setup, procedure and test conditions Swirl Cup and test section The swirl cup used is similar to the one shown earlier in Figure 1.2. The effective area of the swirler cup is 0.62 in 2. The base configuration has a splash plate and a combustion dome attached to the exit as depicted in Figure 2.1. The dome expansion angle is not symmetric. There is a difference of 9 degrees in the expansion angle of both sides of the dome. Configuration 1 is also shown in Figure 2.1. It is only equipped with a splash plate. The rectangular plexi-glass chamber (dump combustor) is shown in Figure 2.2. The dimensions are 5.26 x 4.5 x 13 inch respectively. The thickness of the plexi-glass test section is 1/16 inch. 38

57 2.1 Different configurations of the swirl cup under investigation 2.2 Rectangular Plexi-glass chamber (dump combustor) 39

58 2.1.2 Experimental facility A schematic of the facility utilized is shown in Figure 2.3. The dump combustor is mounted on a flange above an 8 inch PVC pipe used as a manifold. The flange is designed in a manner to allow the combustor dome to be installed properly. Special adapters are designed as shown in Figure 2.4 to enable the setup of the hardware on the test rig. A honeycomb inside the manifold provides uniform flow to the SAC diffuser inlet. Pressure sensor and thermocouples are mounted inside the manifold for monitoring of pressure and temperature at the SAC sector inlet. Air enters the manifold through a 2 inch pipe which can provide pressurized air up to 2 lb/s at 100Psi. An atomizer generating olive oil seeding particles (1~5 μm) is connected to the manifold though a seeding port. A blower and an exhaust system are mounted above the SAC sector to collect aerosols from the combustor Diagnostics The flow field measurements are conducted using two component LDV provided by Artium Technologies Inc. system. The system is shown in Figure 2.5. The system uses two diode pumped solid-state (DPSS) lasers as the light source which doesn t require air or water cooling. The system comprises a transmitter, a receiver, two ASA signal processors and a computer. The 500 mm focal length transmitter and 300 mm focal length receiver were mounted on a computer controlled traverse as illustrated in Figure

59 . 2.3 Schematic of the Experimental facility to study the dump combustor Aerodynamics Test conditions, data acquisition and measurement Grid Experiments are conducted for isothermal flow condition under ambient pressure and temperature of 70±1ºF. LDV measurements are conducted at pressure drop of 4.9%. Data count was typically more than 7000 for the green laser (radial or tangential component of the velocity) and 3000 for the blue laser (axial component of the velocity). The measurements coordinate system is defined such that the axial direction of the swirler is designated as the Z direction with the zero set at the flare exit. The X direction is perpendicular to the Z and in the expanding dome plane. The Y direction is perpendicular to both Z and X. The coordinate system for both configurations under investigation is illustrated in Figure

60 Measurements are conducted in the axial radial plan (X-Z), the tangential radial plan (Y- Z) plane as well as the horizontal plane (X-Y). Both vertical measurements planes pass by the swirler centerline. The horizontal measurement plane is located 5mm from the flare. The measurement grid has a resolution of 2mm in both the X and Y directions. The vertical plane measurements are conducted at downstream locations (Z direction) of 5, 10, 20, 40, 50, 100, 150, 161 and 200 mm respectively. The measurement grid has a fine resolution of 1mm in both the radial and tangential directions. 2.4 Special flanges and adapters to allow installation of swirl cups to the vertical rig 42

61 2.5 Laser Doppler Velocimetry system provided by Artium Technologies 2.6 LDV setup on the vertical rig (plexi-glass chamber) 43

62 2.7 Measurements Coordinate system for both configurations 2.2 Results and Discussions Horizontal plane (X-Y) measurements for both configurations The axial velocity contours for configuration 1 and the base configuration are shown in Figure 2.8 and 2.9 respectively. The results are non-dimensionalized by the flare radius (R). There are two black contours. The inner one shows the edge of the CRZ (contour of zero axial velocity). The outer one shows the edge of the flare. Typically, the swirling flow expands as it leaves the swirler and a radial pressure gradient is developed. As the flow proceeds downstream, the pressure starts increasing and the adverse pressure gradient results in flow reversals commonly known as the CRZ. The CRZ is easily identified in both figures ( 2.8 and 2.9) as the region of ve axial velocities. The axial velocity magnitude is typically 10 m/s. Outside of the CRZ we notice high velocities indicating the presence of the high speed jet emanating from the swirl cup. The effect of the splash plate edge on the flow field is considerable. Figure 2.8 shows that the flow filed is axi-symmetric only in the central region and up to the axial velocity contour of 25 m/s. The outer region is not axi-symmetric because of the presence of the splash plate lip which is only present in the X-direction (see Figure 2.7). Consequently, the flow in the Y- direction is not bounded as it is in the X-direction. The lip of the splash plate has a height of 44

63 5mm. Yet, it affects the velocity magnitude significantly as depicted in Figure 2.7. The velocity in the bounded portion of the splash plate (X-Z plane) reaches roughly 20 m/s. However, the unbounded region (Y-Z plane) reaches velocities up to 30 m/s. In summary, the axial velocity is 50% higher in the unbounded region relative to the bounded region. The dome effect on the flow field is demonstrated in Figure 2.9. We notice that the flow field has high velocities in the Y-Z plane than the X-Z plane. Again, similar to configuration 1, the reason is the gradual expansion in the X-Z plane compared to the abrupt expansion in the Y- Z plane (see Figure 2.7). However, we also notice that along the X-direction the velocities are higher on the right hand side (+ve) than the left hand side (-ve). Mainly, this is due to the higher expansion on the +ve direction than the ve direction. This effect is going to be further illustrated when the measurements in the axial radial plane are presented. The root mean square value of the axial velocity (RMSA) in the horizontal plane for config 1 and the base configuration is shown in Figure 2.10 and 2.11 respectively. The RMS value indicates the oscillation and velocity fluctuations. The RMSA is basically the fluctuating component of the turbulent axial velocity (u' ). Regions with high turbulence become candidates for the development of periodic oscillations which results in generation of noise or thermo acoustic instabilities. 45

64 Y/R Va (m/s) X/R 2.8 Axial velocity contours in the horizontal plane (configuration 1) 1 Y/R Va (m/s) X/R 2.9 Axial velocity contours in the horizontal plane (base configuration) 46

65 The core of the CRZ is a region of low turbulence fluctuations as illustrated in Figure 2.10 and However, the edge of the CRZ is a region of high turbulence activities. The turbulence activities are maximized in the jet region as shown in Figure The presence of the dome damps the oscillations, relatively, as depicted in Figure Figure 2.7 and Figure 2.11 show that the turbulence activities are relatively higher on the unbounded regions than the bounded regions. We already showed that the velocities follow the same trend. It is worth mentioning that the mode of oscillation that commonly appears close to the flare exit is called Precessing Vortex Core (PVC). Jun Cai et.al performed a study on the generation of coherence structures or instability modes of a swirl cup assembly [127]. One of the modes reported was a PVC occurring around the edge of the CRZ. Y/R RMSA (m/s) X/R 2.10 Contours of the Root Mean Square of the axial velocity in the horizontal plane for configuration one 47

66 1 Y/R RMSA (m/s) X/R 2.11 Contours of the Root Mean Square of the axial velocity in the horizontal plane for the base configuration Vertical plane (X-Z) measurements for both configurations Axial velocity and turbulent fluctuating component The axial velocity profiles for configuration 1, at different axial distances from the swirler exit, are given in Figure The axial velocity profiles for the base configuration are shown in Figure The axial velocity profiles for configuration 1 (Figure 2.12) show typical double humped axial velocity profiles near the swirler exit. As the flow proceeds in the downstream direction, the humps move away from the centerline indicating the growth of the CRZ (profiles from Z=5 to Z=40 mm). A point is reached at which the maximum velocity occurs close to the dump combustor wall (Z=40mm). Further downstream, the CRZ starts closing until it disappears totally at around 150mm (no more negative velocities). 48

67 The contours of axial velocities for config 1 in the X-Z plane are shown in Figure The contour plot shows that the flow field is fairly symmetric. Also, this is illustrated by a plot of the streamlines as depicted in Figure The black contour is the contour of zero axial velocity. It defines the edge of the CRZ. We notice that the CRZ extends up to 6.5R. However, we have to keep in mind that we are only studying here the swirl cup in a dump combustor. In a realistic combustion chamber dilution jets are present and they define the termination point of the CRZ. This will be discussed in details later (Chapter 3) when the measurements in the realistic combustion chamber are presented. Based on the previous discussion, only the flow field near the swirl cup exit (up to 3R) is of importance. In the real combustor this is what we refer to as the primary zone or the dome region. In this region the swirling flow is the dominant flow. We notice that the typical magnitude of the axial velocity inside the CRZ in this region (up to 2R) is 10 m/s as shown in Figure The CRZ reaches a maximum width of 2R at a downstream distance of 1.8R. As illustrated in Figure 2.14, the turbulence activities are high in the jet region and close to the edge of the CRZ. There is no doubt that the edge of the CRZ is a very critical region as the flow is switching direction. Also, the turbulence activities are high close to the swirler exit. However, the core of the CRZ is a region of low turbulence. We also notice relatively high turbulence activities as the swirling jet impinge with the combustor walls. However, the turbulence decays downstream as the flow becomes fully developed. The high turbulence activities are essential in improving the mixing and enhancing the fuel air reaction. The axial velocity profiles for the base configuration are given in Figure The effect of the dome on the flow field is enormous. We notice that the profiles show no double humps along the expanding surfaces of the combustion dome. In other words, the combustion dome 49

68 eliminates the regions of low velocity close to the combustor wall. Careful examination of the velocity profiles at 5 and 10 mm (Figure 2.12 and 2.13), for both configurations, reveals that the combustion dome eliminated the corner recirculation zone. The contours of the axial velocities for the base configuration in the X-Z plane are given in Figure The contour plot shows that the flow field is not symmetric. The velocities on the right hand side are relatively higher than the left hand side. The right hand side is the one with higher expansion angle. In other words, the asymmetry of the dome plays a major rule in the asymmetry of the flow field as demonstrated in Figure The CRZ appears to be tilted toward the left hand side as illustrated in Figure It is interesting that only 9 difference would result in such noticeable effects. It is even more interesting that the effect was convected downstream up to the end of the combustor (see the velocity profile at 200 mm in Figure 2.13). The rms contours are given in Figure We notice that the dome presence results in relative decrease in the turbulence activities close to the swirler exit. This is due to the fact that the flow is being guided with the dome walls rather than undergoing free expansion. However, the turbulence doesn t decay early as it was the case for configuration 1. Figure 2.14 shows that the turbulence decays around a downstream distance of 4R for configuration 1. Nevertheless, Figure 2.15 shows that the turbulence activities for the base configuration extend downstream up to a distance of 7R Radial velocity and turbulent fluctuating component The radial velocity profiles corresponding to the axial velocity profiles discussed previously are given in Figure 2.16 and 2.17 for configuration 1 and the base configuration respectively. +ve velocity indicates that the velocity is in the direction of the +ve x-axis. 50

69 Configuration 1 velocity profiles show that as the flow emerges from the flare it expands suddenly and there is a corresponding tremendous increase in the radial velocity as shown in Figure We notice that for all axial locations, the radial velocity is roughly zero in the vicinity of the combustor centerline. Figure 2.16 shows that the radial velocity peaks starts moving away from the centerline as the flow progress downstream. This is accompanied by a decrease in the velocity magnitude as the flow spreads over the cross section. At a downstream distance of 20mm the expanding jet reaches the combustor walls. The jet impinges and bounces off the wall at a distance of around 40mm as depicted in Figure This is illustrated in Figure 3.16 as the radial velocity profiles switch direction at axial distances more than 40mm. Close to the combustor end, the radial velocity vanishes indicating that the flow becomes fully developed. The base configuration radial velocity profiles (Figure 2.17) show similar behavior to the configuration 1. However, the radial velocity is slightly higher on the dome side with the higher expansion angle (right hand side). Also, we notice that there are almost no radial velocity humps in the region close to the swirler exit. In other words, the expanding jet reaches its maximum velocity at the expanding surface of the dome. Again, this is a major difference between both configurations. Unlike the configuration1, at 200mm the radial velocity doesn t vanish. The radial velocity contours for configuration 1 and the base configuration are given in Figure 2.18 and 2.19 respectively. It is clear that for both configurations the high speed regions are the expanding jet regions close to the swirler exit. The radial velocity reduces significantly at around 1.5R for both configurations. 51

70 The radial velocity rms contours for configuration 1 and the base configuration are given in Figure 2.18 and 2.19 respectively. We notice that the turbulence activities are high for both configurations in the jet region as well as the edge of the CRZ. The dome appears to be damping the oscillations significantly in the region close to the swirler exit as illustrated in Figure 2.19 and Figure Again that is attributed to the guided expansion it provides to the flow. However, the turbulence activities extend downstream more than the configuration 1 case. Figure 2.18 shows that the relative high turbulence activities for the base configuration extend up to 4.5R rather than 3R for the configuration 1. An important parameter to keep an eye on is the difference between rms values for different velocity components. Existing turbulence model fail in predicting the swirling flow close to the swirler exit. That might be attributed to the high anisotropy of turbulence in that region. To emphasize that we show the difference between the rms values of the axial and radial velocities for configuration 1 and the base configuration in Figure As depicted in Figure 2.20, there is a remarkable difference between the rms values especially in the expanding jet region Vertical plane (Y-Z) measurements for both configurations Axial velocity and turbulent fluctuating component The axial velocity profiles for configuration 1, at different axial distances from the swirler exit, are given in Figure The axial velocity profiles for the base configuration are shown in Figure It is interesting to find out that both configurations have very similar axial velocity profiles in the Y-Z plane. That is attributed to the fact that both configurations have no restriction on the flow in the Y-Z plane (see Figure 2.7). Hence, similar velocity profiles in 52

71 that plane. Consequently, we limit the discussion in this section to configuration 1. Mainly, the flow field in the Y-Z to the X-Z plane will be compared. We notice that up to a distance of 2R the velocity inside the CRZ is typically 10 m/s (similar to the X-Z results). However, we notice that the jet velocity is relatively higher as depicted from the contours of axial velocity in Figure That might be attributed to two factors. First, the flow is not bounded in the Y-Z direction while it is bounded in the X-Z direction. Although we mentioned earlier that the splash plate edge has a height of around 5mm, but we have noticed a remarkable effect on the axial velocity in the horizontal plane as discussed earlier. Second, the chamber doesn t have a square cross section. It has a rectangular cross section with the short side in the Y-Z plane. In other words, the increase in velocity occurs also due to the relative reduction in the flow area Tangential velocity and turbulent fluctuating component The tangential velocity profiles for configuration 1, at different axial distances from the swirler exit, are given in Figure The tangential velocity profiles for the base configuration are shown in Figure Both configuration show similar behavior so we will be discussing configuration 1 only. The swirl cup has two radial swirlers with counter rotating vanes arrangement. However, the primary swirler has a larger effective area than the secondary swirler. It is noteworthy that at 5mm the central region of the combustor has almost zero tangential velocity as illustrated in Figure This is an indication that the mixing occurrs between the primary and secondary flow inside the venturi and before the flow reaches the 5mm downstream distance. However, we can still see that there is a small portion (from y=10 to 20mm) of the curve that has a counter rotation to the rest of the flow. Also, the tangential velocity profile at 53

72 5 mm shows a typical Rankin vortex behavior. That is; a forced or body vortex followed by a free vortex. As we proceed downstream we can see that the flow field is entirely rotating in the counter clockwise direction as illustrated in Figure We can see that up to 20 mm the tangential velocity profiles are not smooth. That is expected because of the highly 3D unsteady complex flow close to the swirler exit. Basically, this is the region which the CFD fails to capture. Also, the tangential velocity profiles continue to show a Rankin vortex behavior up to a downstream distance of 40mm. However, as we proceed downstream the swirling flow becomes fully developed and the tangential velocity profiles become smooth and follow a forced vortex behavior. The tangential velocity contours and corresponding rms values for configuration 1 are given in Figure The swirling flow is fairly symmetric around the combustor centerline. We can see that in the vicinity of the combustor centerline no swirl is present. High swirling velocities are present in the jet region and extend downstream up to a distance of 1R. High turbulence activities occur in the jet region and around the edge of the CRZ as shown in Figure Confinement effect on the flow field The axial and radial velocity contours of the base configuration installed in the SAC sector are presented in Figure 2.27 respectively. The CRZ shape is strongly influenced by the confinement. We notice that the CRZ height reduces from 8R (rectangular chamber) to 4.8R (SAC sector). The maximum width is almost the same. However, the CRZ is much fatter. The variation of the size of the CRZ will have a significant influence on the reacting flow characteristics. Recall that the size of the CRZ is directly related to the amount of mass being 54

73 recirculated back and consequently to the heat release rate. However, reacting flow measurements are needed to verify such statements. The axial velocity inside the CRZ in both confinements is around 10 m/s up to 3R. The expanding swirling jet radial velocity is similar in both confinements up to 1R. Usually real combustion chambers are designed with a contraction toward the combustor end such that the combustor exit will fit the turbine nozzle ring. The effect of the contraction on the flow field is apparent from the streamlines as depicted in Figure The flow field is tilted in the same direction of the inclined wall and that effect extends further away from the inclined way. It almost extends up to the other chamber wall as shown in Figure Radial velocity is high at the expanding jet region. It diminishes in other regions with exception to the area affected by the contracting combustor wall. The axial and radial rms contours for the same configuration are presented in Figure The highest turbulence activities are limited to the expanding jet region. It extends downstream up to a distance of 1.5R. As we proceed downstream, the rms value drops by about 50% in the vicinity of the CRZ edge. 55

74 mm 10mm 20mm 40mm A x ial v eloc ity (m /s ) x (mm) Axial velocity (m /s) mm 100mm 150mm 161mm 200mm x (mm) 2.12 Axial velocity profiles in the X-Z plane (Config. 1 rectangular chamber) 56

75 mm 10mm 20mm 40mm 15 Axial velocity (m/s) x (mm) mm 100mm 150mm 161mm 200mm Axial velocity (m/s) x (mm) 2.13 Axial velocity profiles in the X-Z plane (base Config. rectangular chamber) 57

76 2.14 Contours of axial velocity and corresponding rms in X-Z plane (Config. 1 rectangular chamber) 2.15 Contours of axial velocity and corresponding rms in the X-Z plane (base Config. rectangular chamber) 58

77 mm 10mm 20mm 20 Normal velocity (m/s) x (mm) Normal velocity (m/s) mm 50mm 100mm 150mm 161mm 200mm x (mm) 2.16 Radial velocity profiles in the X-Z plane (Config. 1 rectangular chamber) 59

78 mm 10mm 20mm 40mm Normal velocity (m/s) x (mm) mm 100mm 150mm 161mm 200mm Normal velocity (m/s) x (mm) 2.17 Radial velocity profiles in the X-Z plane (base Config. rectangular chamber) 60

79 2.18 Contours of radial velocity and corresponding rms for Config. 1 in the X-Z plane (rectangular chamber) 2.19 Contours of radial velocity and corresponding rms for the base configuration in the X-Z plane (rectangular chamber) 61

80 2.20 Difference between rms values of the axial and radial velocities for both configurations. 62

81 2.21 Axial velocity profiles in the Y-Z plane (Config. 1 rectangular chamber) 63

82 2.22 Axial velocity profiles in the Y-Z plane (base Config. rectangular chamber) 64

83 2.23 Contours of axial velocity and corresponding rms in the Y-Z plane (Config. 1 rectangular chamber) 65

84 2.24 Tangential velocity profiles in the Y-Z plane (Config. 1 rectangular chamber) 66

85 2.25 Tangential velocity profiles in the Y-Z plane (base Config. rectangular chamber) 67

86 2.26 Contours of tangential velocity and corresponding rms in the Y-Z plane (Config. 1 rectangular chamber) 2.27 Axial and radial velocity contours of the base configuration installed in the SAC sector. 68

87 2.28 Axial and radial rms contours of the base configuration installed in the SAC sector. 69

88 Chapter 3. (SAC Aerodynamics) Aerodynamics experimental studies have been commonly conducted on simplified geometries. Indeed, numerous publications are in existence for aerodynamic investigation using Laser Doppler Velocimetry (LDV) and/or Particle Image Velocimetry (PIV) on circular or rectangular geometries (dump combustor). In this chapter detailed aerodynamics investigations are reported on a realistic combustion chamber. Two component LDV as well as PIV measurements are conducted on a Single Annular Combustor (SAC) Sector (Fuel Rich Dome Combustor). The SAC sector comprises: swirl cup with counter rotating coaxial radial inlet swirlers, inlet diffuser, inlet cowl, film cooling strips as well as primary and secondary dilution jets. 3.1 Experimental setup, procedure and test conditions Swirler and SAC Combustor The counter rotating coaxial swirl cup mentioned in Chapter 3 and referred to as the base configuration is installed in the Single Annular Combustor (SAC) Sector. All the experiments in this chapter are conducted using the same swirl cup (base configuration). The SAC sector used is similar, in construction, to the typical combustor shown in Figure 1.3. The sector comprises two side quartz windows, primary and secondary dilution jets, eight film cooling strips on each side, combustion dome with cooling slots, inlet cowl and inlet diffuser. Two images of the SAC sector are given in Figure 3.1 and

89 3.1 Assembly of the Single Annular Combustor Sector used in the study 3.2 Single Annular Combustor Sector components 71

90 3.1.2 Experimental facility The same facility described in Chapter 3 is used to perform the study on the SAC sector. However, the sector assembly begins with a diffuser that has a relatively small window type inlet. Also, as the SAC sector is installed to the vertical rig, we notice that the splash plate has an angle of 7 with the horizontal plane. The flange and adapter shown in Figure 3.3 are designed to enable installation of the SAC sector to the rig and to level the splash plate with the horizontal plane Diagnostics The flow field measurements are conducted using PIV and LDV. The PIV system is provided by Lavision Inc. The system includs two 1376 x 1040 pixels 12 bit Lavision imager intense CCD cameras and two 120 mj 15 Hz pulse Nd-Yag lasers which are synchronized with CCD cameras. The PIV system is shown in Figure 3.4. The PIV laser has to enter the SAC sector from the exit. This is the only possible way to allow measurements in the dilution jet plane. Thus, a mirror is used to get the laser beam to the sheet optics and provide the laser sheet at the correct measurement plane as illustrated in Figure 3.5. Both cameras are placed on the top of one another to extend the measurement volume as demonstrated in Figure 3.6. Both cameras are placed perpendicular to the laser sheet. The set up is a real challenge since the exit port of the SAC sector has reduced width to fit the turbine nozzle ring. Figures 3.5 and 3.6 show how complex the PIV setup is. The two component LDV system used in this study was described in Chapter 3. The setup is illustrated in Figure

91 3.3 Flange and adapter to enable installation of the SAC on the vertical rig. 3.4 PIV system provided by Lavision 73

92 3.5 Mirror and sheet optics setup to generate the laser sheet at the measurement plane 3.6 PIV setup on the vertical rig 74

93 3.7 LDV set up on the vertical rig to study the SAC flow field Test conditions, data acquisition and measurement grid The measurements coordinate system is defined such that the axial direction of the swirler is designated as the Z direction with the zero set at the flare exit. The X direction is perpendicular to the Z and in the dilution jet plane. The origin of this coordinate system is located at the center of the flare at the swirl cup exit. The coordinate system is illustrated in Figure 3.8. Experiments are performed for isothermal flow condition under ambient pressure and temperature of 70±1ºF. LDV measurements are conducted at pressure drop of 4.9%. Pressure drop measurements are accurate to within ±0.1%. Data count is typically more than 7000 for the green laser (radial component of the velocity) and 3000 for the blue laser (axial 75

94 component of the velocity). LDV Measurements are conducted in the vertical plane (X-Z) passing by the SAC centerline (plane of the jets at Y=0). The flow field is scanned with a grid resolution of 2 mm by 2 mm. PIV measurements are conducted in the X-Z plane passing by the SAC center (Y=0) at pressure drops of 4.3% and 7.6% respectively to study the effect of pressure drop on the flow field. All other PIV experiments are conducted at pressure drop of 4.3%. The average flow field is obtained by averaging of 200 PIV instantaneous images. An interrogation area size of 32x32 pixels with a window overlap of 50% was used for the image processing algorithm. Also, several PIV measurements are conducted off center to study the flow filed behavior off center and to get an idea of the 3D structure of the CRZ. Namely, measurements are conducted in the X-Z plane at Y distances of 0.66R, 1.12R and 1.32R respectively. 3.8 SAC coordinate system 76

95 3.2 Results and Discussions Mean flow measurements at the SAC mid plane using LDV The total velocity contours (V ) are given in Figure 3.9. The total velocity defined here is 2 2 based on the radial and axial components only (i.e V V V 1/ 2 ). The swirling flow expands as it exits the flare and a radial pressure gradient is set. As the flow proceeds downstream the pressure increases in the vortex core. Therefore, flow reversals occur and a typical CRZ is set as illustrated in Figure 3.9. Also, we notice that the expanding jet velocity on the ve x direction (close to dome edge) is larger than that on the +ve x direction. The reason is the asymmetry in the dome expansion angle. As the main flow proceeds downstream it interacts with the primary dilution jets and the CRZ is closed as demonstrated in Figure 3.9. The black contours are contours of zero axial velocity. a r As mentioned earlier, the combustor comprises a set of primary as well as secondary dilution holes. There are four dilution jets in the measurement plane. The upper left jet is shooting down at an angle. The other three jets have radial exits. The four are clearly identified from their high total exit velocity as shown in Figure 3.9. The jet trajectories are not radial. That indicates the pressure imposed from the main flow on the jets. This is a typical jet in cross flow configuration. As the dilution jets mix with the main flow, we notice the formation of four jet wake regions. The size of the primary dilution jet wake regions are more than twice the secondary dilution jet wake regions. 77

96 3.9 SAC sector centerline total velocity contours using LDV at P/P = 4.9% The total velocity contours reveals that the high speed regions are the dilution jets as well as the jet mixing regions. The remaining regions of the combustor are relatively low speed regions. The radial velocity contours are given in Figure We notice high radial velocities occur as the swirling flow expands once it exits the flare. The dilution jets are easily identified on the figure. The axial velocity contours are given in Figure The axial velocity in all wake regions as well as the CRZ is around 13 m/s as illustrated in Figure High axial velocity is noticed post the recirculation zone in the vicinity of the combustor centerline. The axial and radial rms contours are given in Figure 3.12 and 3.13 respectively. The red contours on both plots enclose regions of high rms values. High turbulence activities are noticed in all jet regions (shear layer regions) as depicted in Figure 3.12 and

97 respectively. However, the central region of the combustor where the primary jets mix with the main flow is the region of maximum turbulence activities. Such regions become very sensitive and susceptible to periodic oscillations. Indeed, we have noticed, while studying the SAC reacting flow dynamics, the generation of such instabilities [125]. Also, we have noticed that, during the SAC sector ignition, the same regions may even ignite before other regions as illustrated in Figure 3.14 (compare to Figure 3.13). The image is captured by a Phantom high speed camera during ignition at 6600 frames/second). Video 1 shows the SAC sector while conducting reacting flow experiments. In that video, the fuel is being increased steadily from an equivalence ratio of 0.2 to 0.9. The blue color is an indication of reaction taking place. We notice that initially the reaction takes place at the CRZ edge. Then, as fuel increases it progresses downstream and reaches the primary jets lower boundary. Then, it proceeds to the primary jets upper boundary. Then, it reaches the secondary jets lower boundary and progresses to the secondary jet upper boundary. The progression of the reaction is illustrated in Figure Reacting flow vs. isothermal flow We made extensive use of the High Speed Camera (HSC) to study the detailed behavior of the reacting flow. More details will be given later in Chapter 4. Reacting HSC images are presented in Figure 3.15 and 3.16 (N-Heptane and Propane). We notice that the dilution jet angle increases slightly than the isothermal case as illustrated in Figures 3.15 and That is expected because of the expansion (higher velocities) that associates heat release. The main flow momentum increases and more pressure is exerted on the primary jets resulting in a change in the jet trajectory as shown in Figures 3.15 and

98 3.10 SAC sector centerline radial velocity contours using LDV at P/P = 4.9% 3.11 SAC sector centerline axial velocity contours using LDV at P/P = 4.9% 80

99 3.12 SAC sector centerline axial velocity rms using LDV at P/P = 4.9% 3.13 SAC sector centerline radial velocity rms using LDV at P/P = 4.9% 81

100 3.14 SAC sector during ignition 3.15 Reacting flow vs. Isothermal flow (N-Heptane) 82

101 3.16 Reacting flow vs. Isothermal flow (Propane) PIV results Effect of pressure drop on the flow structure The total velocity contours at pressure drop of 4.3 and 7.6% are shown in Figure 3.18 and 3.19 respectively. We notice that the CRZ as well as all wake regions maintained, roughly, their absolute size. The termination point of the CRZ is determined by the dilution jets location. We conclude that the flow structure is independent of the pressure drop. This is the typical flow behavior at such high Reynolds number. However, we notice that the velocity magnitude varies proportional to the square root of the pressure drop as illustrated in both figures. 83

102 Video 1 Fuel increased steadily from equivalence ratio of 0.2 to Sequential images showing the progression of the reaction zone as the fuel flow rate increases 84

103 3.18 SAC sector centerline total velocity contours using PIV at P/P = 4.3% 3.19 SAC sector centerline total velocity contours using PIV at P/P = 7.6% 85

104 Instantaneous PIV at SAC mid plane at pressure drop of 4.3% Instantaneous PIV results are important as they reveal the true behavior of the flow field. Figures 3.20 to 3.22 show three instantaneous PIV results at pressure drop of 4.3%. We may classify the instantaneous PIV results into three scenarios. The first scenario is when both jets impinge as depicted in Figure When this takes place, the flow is fairly symmetric and we see that around the combustor centerline a portion of the flow is going up and the other portion goes down toward the flare. The second scenario is when the Right dilution Jet is shooting up and at the same instant the left dilution jet is shooting down as demostrated in Figure At this point, the flow field structure in the primary zone is controlled by the left dilution jet which is shooting down and feeding the CRZ with air. On the contrary, the post recirculation zone structure is controlled by the right dilution jet which is shooting up and interacting strongly with the secondary region flow. The third scenario is when the right dilution jet is shooting down and the left dilution jet is shooting up. This could be explained in a similar fashion to the second scenario. The instantaneous results reveal the true behavior of the dilution jets and how they are in a continuous up and down motion. Also, they show the strong interaction and influence of the dilution jets on the flow field. The series question now is: if the dilution jets have a strong influence on the flow field then how sensitive are they to disturbances in the flow field? This is the topic we will be discussing in the next section. We have to keep in mind that these are heat release regions and oscillations in such regions is associated with oscillations in heat release rate which may result in generation of combustion instabilities. 86

105 3.20 Instantaneous PIV at 4.3% (Jet Impinge) 3.21 Instantaneous PIV at 4.3% (Left Jet shooting down and right Jet shooting up) 87

106 3.22 Instantaneous PIV at 4.3% (Right Jet shooting down and left Jet shooting up) PIV off center measurements at pressure drop of 4.3% The PIV total velocity contours (V ) at Y=0, 0.65R, 1.1R and 1.3R are presented in Figure 3.23, 3.24, 3.25 and 3.26 respectively. The contours show the velocities in the primary region only. This is the zone of action in the Fuel rich combustors. The measurement plane at 0.65R is far away from the primary holes. Nevertheless, we can see that two relatively weak jets still show in the velocity contours as shown in Figure The air in this region is basically entrained by the primary jets located at the SAC sector centerline. Also, some air is entrained by the jets on the other side of the SAC sector. Moreover, we notice that the left jet wake region disappear totally at 0.65R. However, the right jet wake region still exists. The reason might be the higher velocities on the Left hand side than the right hand side as illustrated in Figure

107 At 1.1R we are measuring at the plane passing by the second primary dilution jet. Consequently, we can identify a high speed jet region on the left hand side of the SAC Sector as demonstrated in Figure Recall that the jet is not a complete jet. The hole producing the jet is not a complete circle, but rather one half of a circle. Similar to our previous explanation, we can identify two jet wake regions as illustrated in Figure At 1.3R we are measuring at a plane in the vicinity of the second primary jets. That s why high velocities are observed in the jet regions. Also, two wake regions exist for the two jets as depicted in Figure However, most of the flow in this plan is moving upward. Very small portion is being recirculated. It is of interest to make use of the previous data and reconstruct the CRZ in three dimensions. In every measurement plane we have the contours identifying the edge of the CRZ. The coordinates are recorded and Solidworks [128] is used to reconstruct the CRZ in 3D as demonstrated in Figure There is a strong asymmetry in the CRZ. The dimensions of the CRZ are presented in Figure The height of the CRZ is roughly constant and equals to 2.7R. In the X-direction the CRZ extends up to 1.6R. However, in the Y-direction the CRZ extends up to 1.3R. The CRZ extends slightly over the size of the flare in the X direction, but extends significantly in the Y direction. We believe that this is mainly the effect of confinement. Recall that the SAC sector is not a square chamber. As the term implies it is a sector from an annular combustor. The primary zone of the SAC sector is illustrated with the dimensions in Figure b and w are the breadth and width at the centerline respectively. The ratio of b/w is around The ratio of the CRZ b/w (1.3R/1.6R) is around It is even more interesting to repeat the same calculations for the CRZ measured earlier in Chapter 2 for the configuration 1. Recall that the configuration 1 89

108 is tested in a rectangular chamber with the b/w around The CRZ b/w (1.6R/1.9R) is around The previous discussion shows that the CRZ shape is strongly dependent on the geometry of the confinement Effect of partially blocking the primary dilution jets strip The SAC sector has two strips with holes that provide the primary dilution jets. There are two strips that oppose one another as illustrated in Figure 3.2. The primary jets angle with the horizontal is 18. We partially blocked each strip, one at a time, and conducted PIV on the SAC sector to study how sensitive is the flow field to primary jets perturbations and to evaluate the effectiveness of the blocking process in controlling the flow field structure. The blocking process is illustrated in Figure The velocity contours and streamlines with and without blockage are presented in Figure The contours on the left hand side show the flow field that results when the left primary dilution jet is partially blocked. The contours on the middle correspond to the flow field with all jets fully open. The contours on the right hand side present the flow field that corresponds to partially blocking the right primary dilution jet strip. With the left primary dilution jets partially blocked, we notice that the left jet becomes weak relative to right jet and thus the impingement point is shifted from the SAC centerline to the left half of the combustor as depicted in Figure A strong asymmetry occurs in the SAC primary zone. As the right primary jet penetrates more in the primary zone it controls the structure of the SAC primary zone as shown in Figure It is noteworthy that as the left dilution jet becomes weaker, the pressure applied from the swirling flow result in more curvature in the jet trajectory. In other words, with the jet partially blocked the jet angle 90

109 with the horizontal increases. However, we notice that the size of the CRZ is roughly the same with or without blockage. It is interesting that the effect of partially blocking the left dilution jet extends to the secondary region or to the post recirculation region as shown in Figure This is apparent when we compare the case with and without blockage. Figure 3.31 shows that without blockage the swirling flow mix with the primary jets and is almost shooting vertical in the secondary region in the vicinity of the SAC Sector centerline. However, with the left jet partially blocked, the flow in the secondary region tilts toward the left hand side of the SAC sector and is not shooting vertical anymore. In summary, partially blocking of the left jet results in an increase in the left primary jet angle with 6.5, the right jet penetration increases by 9.2% and the jet impingement region tilts with an angle of -7 relative to the original position. The ve signs means in anticlockwise direction. The effects discussed previously will have a strong impact on the emissions, stability and dynamics characteristics of the SAC sector. However, more reacting flow experiments are needed to study such effects. The effect of partially blocking the right dilution jet on the flow field could be explained in a similar manner. The effect of partially blocking of the right jet is summarized as follows: an increase in the right primary jet angle with 3.5, the left jet penetration increases by 7.5% and the jet impingement region tilts with an angle of +20 relative to the original position. +ve refers to rotation in the clockwise direction. 91

110 4 Z/R V(m/s) X/R 3.23 PIV measurements at Y=0 at pressure drop of 4.3%. 4 Z/R 3 2 V(m/s) X/R 3.24 PIV measurements at Y=0.65R at pressure drop of 4.3%. 92

111 4 Z/R V(m/s) X/R 3.25 PIV measurements at Y=1.1R at pressure drop of 4.3%. 4 Z/R V(m/s) X/R 3.26 PIV measurements at Y=1.3R at pressure drop of 4.3%. 93

112 3.27 3D reconstructed CRZ over the flare 3.28 Dimensions of the SAC sector CRZ 94

113 3.29 SAC cross section schematic in the primary zone Illustration of the dilution jet strip with partial blocking 95

114 3.31 Effect of partially blocking the primary dilution jets on the flow field 96

115 Effect of blocking the cooling holes underneath the primary dilution jets The primary dilution jets are provided through holes drilled on the liner cooling strip. The cooling strip is a strip that is cooled by convective cooling. Holes with a diameter in the order of magnitude of 1 mm provide the cooling air necessary to protect the liner from melting as depicted in Figure The concern is raised about what kind of interaction takes place between the primary jets and the cooling jet if any? This is important because the cooling jet is shooting perpendicular to the primary dilution jet at the point where it leaves the hole as illustrated in Figure However, we have to keep in mind that the diameter of the hole providing the primary jet is around 15 times the hole providing the cooling jet. Also, the couple of holes underneath the primary jets are useless from the point of view of convective cooling. Figure 3.33 presents the velocity contours and streamlines with and without blockage of the cooling strip underneath the primary dilution jets. The plot on the left shows the case with the left cooling strip is blocked. The middle plot shows the flow field without blockage and the right plot shows the flow filed with the right cooling strip blocked. The cooling air jet interacts with the primary jet and pushes against it resulting in a slight increase of the primary jet angle with the horizontal. As the left cooling jet is blocked, we notice a slight decrease in the jet angle and a corresponding change in the flow field in the primary as well as the secondary zone as shown in Figure Without blockage, both primary jets impinge and a portion is recirculated back in the CRZ and the other portion shoots up in the secondary region. With blockages, both portions noticeably change directions as demonstrated in Figure The events associated with the blockage are 97

116 summarized as follows: the CRZ centerline tilts with an angle of +8 relative to the original direction. The jet impingement region angle increases with + 6. Although the change is not significant, we are surprised by the effect the cooling jet imposed on the primary jet even though the momentum of the primary jet could be 100 times more than the cooling jet. The effect of blocking the right cooling strip could be explained in a similar manner as shown in Figure The effects on the flow field are summarized as follows: the CRZ centerline tilts with an angle of relative to the original direction. The jet impingement region angle increases with Illustration of primary jets and cooling jets interaction Effect of primary jets off centering on the flow field We have noticed from the previous sections that the primary jets have a strong influence on the flow field structure. The aim of the current section is to study the effect of the slight 98

117 primary jets off centering on the flow field. The hole diameter is also reduced from 0.65 inch to 0.4 inch. The off-centering is simply achieved using a standard washer. The primary jet holes are both located on the SAC centerline in the original assembly of the SAC sector as demonstrated in Figure We off centered the jets asymmetrically such that the inner hole was placed off center by 1.4 mm and the outer hole is placed off center by 1.0 mm as depicted in Figure The off centering results are presented in Figure The figure shows the total velocity contours with the jets centered and off centered. With the jets centered, we notice that the flow field is fairly symmetric as depicted in Figure However, with the jets off centered, the right jet is slightly larger in diameter than the left jet in the measurements plane (recall the off centering is not symmetric. Please see Figure 3.34). It is very interesting to see that the higher strength of the right jet is manifested in more penetration and more influence on the flow field as illustrated in Figure There is not much opposition from the left jet, hence the flow field in the primary zone is strongly influenced by the right jet as depicted in Figure The effect of off centering on the flow field is summarized as follows: the CRZ centerline tilts with an angle of relative to the original location. The Jet impingement region tilts with an angle of relative to the original direction. The right jet penetration increases with 11% while the left jet penetration decreases by 8.5%. As engines age such asymmetry in the geometry is inevitable. This emphasizes the importance of the current study for combustor designers Effect of primary jet size on the flow field The original size of the primary jet holes were 0.65 inch. We conducted the measurements using jets with diameters of 0.4 and 0.31 inch to study the effect of primary jet size on the 99

118 SAC flow field. The corresponding decrease in the jet area is 62% and 77% respectively. The total velocity contours and streamlines for all three cases are presented in Figure As the jet diameter decreases the momentum decreases significantly. We notice that less penetration is achieved as the diameter decreases as shown in Figure For the big jets we notice that a small portion of the jet is recirculated back in the CRZ. A big portion of the jets mix in the secondary region and appreciably influences the flow field in the secondary region. The SAC sector cross sectional area starts decreasing toward the outlet after approximately 4R as depicted in Figure 3.9. However, this contraction in the area doesn t affect the flow field in the vicinity in the combustor centerline in the secondary region for the big jet s case as depicted in Figure We can see that the flow resulting from the jet impinge is almost shooting up vertically in the big jets case. As the jet size is reduced we notice the portion of the primary jets that is not recirculated decrease drastically and so does the influence on the primary jets on the secondary region. For the 0.31 inch jets we notice that the flow field in the secondary region is tilted toward the right hand side. This appears to be the effect of the contraction which now has more influence on the secondary region than the primary jets that control the flow structure in the primary region. The previous results could be quantified as follows: 62% reduction in area results in a tilt in the CRZ with +7, jet impingement region tilts with +6 and the penetration is reduced with 11.5%. 77% reduction in the jet area results in a tilt in the CRZ with +16.5, jet impingement region tilts with +16 and the penetration is reduced with 25%. The previous results emphasize the importance of the careful selection of the dilution jet s size. It is not only the location of the jets that is of high importance, but the size of the jets 100

119 will have significant effects on the flow structure and behavior in the primary as well as the secondary zone. 101

120 3.33 Effect of blocking the cooling strip underneath the primary jets 102

121 3.34 Illustration of primary jet holes off centering 3.35 Effect of primary jets off centering on the SAC flow field 103

122 3.36 Effect of primary jet size on the flow field 104

123 Chapter 4. (SAC Dynamics) While running reacting flow experiments on the SAC sector we noticed the generation of thermo-acoustic instabilities. The aim of the current chapter is to construct the acoustic spectrum of the SAC under investigations. Also, several high speed videos are recorded to help provide more insight about the thermo-acoustic instability phenomenon. Prof. S. M. Jeng and the research team at the University of Cincinnati Combustion Research Lab (CDL) developed a new diagnostic technique to diagnose the combustion instabilities from the high speed videos. However, this effort is discussed in details elsewhere [125, 129]. 4.1 Experimental setup, procedure and test conditions Swirl cup and SAC Sector The Single Annular Combustor (SAC) Sector presented earlier in Chapter 4 is used to conduct the dynamics investigations. Optical access to the combustion zone is provided by the means of a 5x7.5x inch quartz window Diagnostics The pressure fluctuations (acoustics) are recorded using an AKG D112 dynamic cardioids microphone. At selected test conditions, a High Speed Phantom Camera (HSC) V7.3 is used to capture the flow field at a speed of 6600 frames per second (fps). The signal form the microphone is also fed to the HSC through a Data Acquisition (DT9800 provided by Vision Research) and a Signal Acquisition Module (SAM). Also, a network of regular speed cameras is used to record the flame from different view angels. The air and natural gas flow 105

124 rate are recorded by the Micromotion F300 and F25 flow meters. Type K thermocouple is used to measure the temperature at the air heater inlet and exit. Pressure drop is monitored using a Meriam pressure gauge meter. The burner is mounted on a horizontal test rig which includes a 72kw inline heater to heat the air up to 700F. A 8 inch diameter, 2 feet long plenum pipe is located upstream of the burner. The setup on the horizontal test rig is illustrated in Figure Test conditions and data acquisition To obtain the acoustic spectrum we do the following: 1. The fuel flow is increased from equivalence ratio ( ) of 0.2 to 0.9 in steps of 0.03 at fixed pressure drop and pre-heat temperature. 2. The experiment is conducted at pressure drops ( P/P) of 2%, 4% and 6%. At each pressure drop, the experiment is conducted at air preheat temperatures (T3) of 200F, 400F and 600F. 3. The previous test conditions are repeated using two different gaseous fuels (CH 4 and C 3 H 8 ). 4. FFT is applied to the audio signal corresponding to each test condition to identify the frequencies of concern. We have around 18 different test conditions at every pressure drop and air preheat temperature. Thus, the total number of test conditions per fuel is around 160. This totals to around 360 test conditions for both fuels that we used. This is around 360 FFT similar to the one (just a sample) present in Figure

125 4.1 Set up on the horizontal rig for Dynamics investigations. 107

126 4.2 Sample of the FFT to the acoustic signal 4.2 Results and Discussions The SAC sector acoustic spectrums with CH 4 as fuel at different operating conditions are presented in Figure 4.3 to 4.7. The corresponding acoustic spectrums with C 3 H 8 as fuel are presented in Figure 4.8 to The reason of plotting such acoustic spectrum is simply to determine the frequencies of concern and understand the conditions under which they appear. From the previous figures we identified three frequencies of concern. The first frequency is a low frequency of around 280 Hz. The second frequency is a frequency of around 400 Hz. The third frequency is a frequency of around 600 Hz. It is now important to try to investigate the modes of oscillations associated with the frequencies of concern. The NASA CEA code [130] is used to estimate the speed of sound at different test conditions and therefore we have an estimate for the quarter axial wave of the combustor. The ¼ wave is numerically between Hz. 108

127 The acoustic spectrums show that at low power (low equivalence ratio) the low frequency dominates. However, at high power (high equivalence ratio) the high frequency dominates. There is a region of interference were both modes exist. More insight about the modes involved could be obtained from the HSV. The blue color of the flame is an indication of the presence of CH which is a good indication of the chemical reaction and the energy release rate. Video 2 is a HSV captured at 6600 fps with the equivalence ratio of 0.23 and at a pressure drop of 2%. The frequency that corresponds to the HSV is around 280 Hz as shown in Figure We can see that most of the chemical reaction is taking place in the primary zone. The dilution jets are not involved in the chemical reaction. Earlier in Chapter 4, we discussed that there are two regions that are good candidates for thermo-acoustic instabilities to develop. The first region is in the primary zone close to the swirler exit where the swirling flow dominates. The second region is the primary jet regions. Thus, the two hundred something Hz that occur at low power is more likely attributed to the swirling flow as depicted in Video 2. This was asserted from an advanced FFT applied to the video as discussed elsewhere [125]. Figure 4.14 shows the regions in the flow field where the 280 Hz oscillations take place. Video 3 is a HSV captured at 6600 fps with an equivalence ratio of 0.6. The video corresponds to instability with a frequency of 414 Hz as depicted in Figure We notice that the jet region is contributing appreciably to the heat release process. It is more likely that the mode of oscillation encountered at high power is attributed to the primary jets. This is also verified from an advanced video FFT as discussed elsewhere [125]. Figure 4.16 shows the regions that oscillate at 414 Hz. 109

128 4.3 Acoustic Spectrum using Methane at P/P=2% and T3=400F. 4.4 Acoustic Spectrum using Methane at P/P=4% and T3=200F. 110

129 4.5 Acoustic Spectrum using Methane at P/P=4% and T3=400F. 4.6 Acoustic Spectrum using Methane at P/P=4% and T3=600F. 111

130 4.7 Acoustic Spectrum using Methane at P/P=6% and T3=400F. 4.8 Acoustic Spectrum using Propane at P/P=2% and T3=400F. 112

131 4.9 Acoustic Spectrum using Propane at P/P=4% and T3=200F Acoustic Spectrum using Propane at P/P=4% and T3=400F. 113

132 4.11 Acoustic Spectrum using Propane at P/P=4% and T3=600F Acoustic Spectrum using Propane at P/P=6% and T3=400F. 114

133 Amplitude Frequency 4.13 Audio signal FFT at =0.23, P/P=2% and T3=200F 4.14 Source of the heat release oscillations (280 Hz) at low equivalence ratio[125]. 115

134 Amplitude Frequency 4.15 Audio signal FFT at =0.6, P/P=2% and T3=200F 4.16 Source of the heat release oscillations (400 Hz) at high equivalence ratio [125]. 116

135 Video 2 High Speed Video showing oscillation with a frequency of 240 Hz (2%, 200F, =0.23) Video 3 High Speed Video showing oscillation with a frequency of 414 Hz (2%, 200F, =0.61) 117

136 Chapter 5. (SAC Air/Fuel Mixing) In the current chapter we study the process of air fuel mixing inside the SAC. The study is limited to isothermal mixing. N 2 is injected through the fuel nozzle port (to represent fuel). A probe is used to scan the flow field and measure the fuel concentration. The probe is connected to an Oxygen paramagnetic analyser to measure the O 2 concentration. 5.1 Experimental setup, procedure and test conditions Swirl cup and SAC Sector The fuel nozzle through which the N 2 (fuel substitute) is admitted has a tip similar to the one shown in Figure 5.1. The same SAC sector we described earlier in Chapter 4 is used to conduct the study. However, the SAC side wall has to be modified drastically to enable admission of the sampling probe to the combustion zone. N 2 could sufficiently simulate the fuel in this diffusion flame arrangement because the molecular weight is 28 (comparable to the molecular weight of C 3 H 8 (44) and CH 4 (16)). The probe couldn t be inserted from exit flanged because of the contraction that reduces the cross section significantly toward the SAC exit as depicted in Figure 3.1. We decided to introduce the sampling probe through the side wall. In order to be able to scan the flow field in the X-Z plane, the side plate is designed as shown in Figure 5.2. The side plate is composed from aluminum strips that are held together using two aluminum holders. Then, the sampling probe is inserted through a hole in a relatively longer movable aluminum strip as illustrated in Figure 5.2. The movable strip is attached to the computer controlled traverse to enable scanning of the flow field in the X-direction. This basically means that we have to scan a complete line in the X-direction and then switch the 118

137 movable strip with the upper or lower strip to be able to scan the flow field at a different Z. The designed side plate is illustrated in Figure Gaseous fuel nozzle tip. 5.2 Designed side plate to enable insertion of the sampling probe to the SAC Diagnostics The sampling probe is connected to the O 2 paramagnetic analyzer for O 2 measurement. The measurement theory of the analyzer is based on the relatively high magnetic 119

138 susceptibility of the O 2 compared to other gases. The device provides excellent precision in the range of 1-100%. We also used 2D PIV to study the flow field. The PIV system is described in Chapter 4. The flow field is obtained by averaging of 150 images Test conditions, data acquisition and measurement grid The same coordinate system defined earlier in Chapter 4 is used in this Chapter. Experiments are performed for isothermal flow condition under ambient pressure and temperature of 70±1ºF. All measurements are conducted at pressure drop of 4.0%. Pressure drop measurements are accurate to within ±0.1%. The corresponding air flow rate is roughly 1000 pphr. The air fuel mixing measurements are conducted at a fuel flow rate of 25 pphr. The SAC sector is scanned with a resolution of 3mm in the X direction. Measurements are conducted at Z/R= 0.2, 0.58, 0.96, 1.34, 1.72, 2.49 and 3.25 respectively. Also, the air fuel mixing measurements is performed at fuel flow rates of 35, 45 and 55 pphr. However, the measurements are conducted at selected locations only as follows: 1. At Z /R= 1.725, the measurements are employed at X/R=0, -0.8, -1.4 and At X/R=0.8 and X/R=0 (centerline), the measurements are conducted at Z/R=0.2, 0.772, and respectively. PIV measurements are performed in the X-Z plane passing by the SAC center (Y=0) with and without fuel at fuel flow rates of 25, 35, 45 and 55 pphr. The average flow field is obtained by averaging of 150 PIV instantaneous images. An interrogation area size of 32x32 pixels with a window overlap of 50% is used for the image processing algorithm. 120

139 5.2 Results and Discussions The profiles of Fuel to air ratio by mass (F/A) at different downstream distances are presented in Figure 5.3. The measurements are conducted at 25pphr (Global F/A by mass is 0.025). The fuel concentration profile close to the flare exit (Z/R=0.2) show sharp double humped profile as demonstrated in Figure 5.3. At the same axial distance (Z/R=0.2) the measurements extend on the X direction from -1.2R to 1.2R (physical domain boundaries). In other words, beyond ±1.2R the dome surface is present. We notice that at this axial location the minimum F/A ratio is 0.05 (twice the global ratio). The concentration peaks and reach 0.12 (5 times the global ratio) at a radial distance of around 0.7R indicating the maximum fuel concentration. As we proceed downstream to an axial distance of 0.581R the fuel spread over the crosssection and peaks at a radial distance of ±1.2R as shown in Figure 5.3. The minimum F/A is still The profiles extend between a radial distance of ±1.8R (edge of the dome). The maximum F/A is around 0.09 (3.5 times of the global ratio). The F/A profiles flatten at a downstream distance of 0.962R and show weak peaks at radial distances of -1.8R and 2.0R respectively. The asymmetry may be attributed to the fuel nozzle location (the fuel nozzle was slightly off center of the swirl cup arrangement). We notice that the profiles are similar at distances of 0.96, 1.34 and 1.72 as depicted in Figure 5.3. This indicates that the most of the mixing takes place in a downstream distance of around 1.0R. However, as the dilution jets are encountered (Z/R=2.49 and 3.25) the F/A profiles drops significantly as tremendous amount of air is admitted through the holes. 121

140 FA ratio by mass (FA global=0.025) FA ratio by mass Z/R=0.2 Z/R=0.581 Z/R=0.962 Z/R=1.343 Z/R=1.724 Z/R=2.486 Z/R= X/R 5.3 Profiles of Fuel Air ratio (by mass) at different fuel flow rates. The local equivalence ratio ( ) contours are presented in Figure 5.4 with the l corresponding velocity vectors. The corresponding global equivalence ratio at 25 pphr is The contours are plotted based on a typical stoichiometric air to fuel ratio of 15:1. The dome region is a region of prime importance in Fuel Rich Dome Combustors. As predicted from the name the dome region is fuel rich and this is of prime importance for stability reasons. However, we high fuel concentration is present in the vicinity of the swirling jet region as illustrated in Figure 5.4. As expected, the l drops to zero in the primary jets region. The value of l reaches that of the as the swirling flow interacts with the primary g jets. 122

141 5.4 Local Equivalence ratio contours and corresponding velocity field at fuel flow rate of 25 pphr (global equivalence ratio 0.375) 123

142 Figure 5.5 shows the contours of normalized l (normalized by the g ) imposed on the flow field velocity vectors. The Black contours represent the edge of the CRZ and the primary jets wake regions respectively. Such a plot is of prime importance to delineate the stability characteristic of the Combustor [131]. We notice that the fuel spreads in the vicinity of the edge of the CRZ. However, it looks like the spreading angle (45 ) is more than the one required to achieve high degree of stability (55-60 ). A reduced spreading angle would result in more concentration of the fuel inside the CRZ. The velocities are lower and this will result in a highly sustained flame. Nevertheless, the results have to be interpreted carefully as we are only discussing isothermal conditions. However, for the reacting case, we are expecting the width of the CRZ to increase slightly because of the expansion associated with fuel reaction. 4 Z/R FI)Norm X/R 5.5 Normalized equivalence ratio contours imposed on velocity vectors at fuel flow rate of 25 pphr 124

143 It is interesting to see what kind of impact results from the fuel injection on the aerodynamics of the combustor. Recall that in Chapter 4 we discussed the sensitivity of the primary jets to perturbations. We drew the general conclusion that the flow field is sensitive especially when it comes to perturbations that directly affect the primary jets. We have to keep in mind that this might not be a universal behavior because the size of the jets is relatively big in the combustor under study. This was illustrated in the study we conducted on the effect of the jet size on the flow field (see Figure 3.36). The total air flow rate at pressure drop of 4% was around 1000 pphr. Usually Fuel Rich Dome Combustors are designed such that 30% of the total air flow rate flows through the combustion dome (roughly 350 pphr). If the fuel flow rate is 25 pphr then the percentage of the fuel flow rate represents around 7% of the air flow rate. Figure 5.6 presents the velocity vectors and streamlines for the primary region with and without fuel injection. The fuel flow rate is 25 pphr. We notice that there is a noticeable change in the flow field structure with fuel injection. To be able to provide better explanation of the effect of fuel injection, we construct a plot of the difference in velocity magnitudes with and without fuel injection. Figure 5.7 presents the value of Delta V defined as V (with injection)-v (without injection). We notice that the velocity increases in the swirling jet region because most of the fuel is present in this region as depicted in Figure 5.5. It is interesting to see that the velocity decrease significantly in the vicinity of the primary jets upstream region and increase significantly in the vicinity of the primary jets downstream region. This is due to the change in the primary jets angle. It is clear from Figure 5.6 and 5.7 that the jet angle increased with fuel injection. The reason behind that lies in the increase in the pressure that the swirling flow imposes on the primary jets with 125

144 fuel injection. The fuel flow rate only represents 7% of the dome flow rate, but the fuel mixes with the air and increase the momentum in the swirling jet region. In other words, the momentum is increased with fuel injection in locations that are critical to the primary jets. Recall that we showed in Chapter 4 that the small cooling holes underneath the primary jets do have a noticeable effect on the flow field. This is only due to their critical location. Likewise, the injected fuel is concentrated in the swirling jet region which impinges with the primary jets close to the primary holes. This is a critical location because the primary dilution air is admitted through the holes. To assess and evaluate the PIV sensitivity and accuracy, two groups of images of the flow field are captured with 25 pphr and 55phr fuel flow rate. Each group has 75 images. The flow field is derived by processing the images. If the PIV is accurate enough the results should be replicated. In other words, the two 75 images groups should produce very similar flow field. Figure 5.8 show the difference between the flow field derived from both groups at 25 and 55 pphr respectively. The results show excellent agreement (the difference in velocity magnitude is roughly zero). We have to keep in mind that we are only comparing the flow field resulting from processing of 75 images. All the results we showed earlier are derived from processing of 150 images. As the number of processed images increases we should expect more accurate results. Figure 5.8 shows that the uncertainty is restricted to the jet regions because this is a region of very high gradients. Also, while conducting LDV measurements, same regions showed significant fluctuations. Figure 5.8 shows that the PIV could be sufficiently and successfully used in the sensitivity study. 126

145 5.6 velocity vectors and streamlines with and without fuel injection (fuel flow rate of 25 pphr) 127

146 Z/R 3 2 Delta V X/R 5.7 Difference in velocity magnitudes with and without fuel injection (fuel flow rate =25 pphr) 5.8 Evaluation of PIV sensitivity 128

147 The fuel concentration measurements are conducted at 35, 45 and 55 pphr, at selected locations in order to study the effect of increasing the fuel flow rate on the mixing process. Figure 5.9 presents the normalized equivalence ratio centerline profile measurements. The normalized contours are expected to be similar. However, it is clear that the 55 pphr profile is showing different behavior. In other words, the equivalence ratio for the 55 pphr is less than the one expected. Figure 5.10 presents the normalized equivalence ratio profiles along the line X/R=-0.8. Again, we notice that the 25, 35 and 45 pphr profiles show similar behavior. However, the 55 pphr behaves differently. Figure 5.11 presents the normalized equivalence ratio profiles at Z/R= Similar conclusions are deduced from the figure. To be able to better interpret the results, PIV measurements are conducted with and without fuel injection at fuel flow rates of 35, 45 and 55 pphr as presented in Figure 5.12, 5.13 and 5.14 respectively. Interestingly, the PIV results at 25, 35 and 45 pphr are similar. However, a fuel flow rate of 55 pphr results in a significantly different flow field. Such conclusion is in agreement with the odd trends of the F/A concentration obtained at a fuel flow rate of 55 pphr. The high fuel flow rate results in a strong distortion of the flow field. 55 pphr is approximately 16% of the air flow rate in the primary region. The high fuel flow rate appears to have a strong influence on the left primary jet which, in turn, strongly impact the flow field structure. It appears that the high fuel flow rate imposes an appreciable amount of pressure on the left primary jet as illustrated in Figure This results in a significant increase in the jet angle, reduction in the jet momentum and penetration. Meanwhile, the penetration of the right jet increases due to the reduced opposition of the left jet as depicted in Figure

148 The F/A mixing at 55 pphr is further studied. Figure 5.15 shows the normalized equivalence ratio profiles at Z/R=0.2, 0.96 and 1.7 respectively. Sharp variation in the profile occurs between Z/R= 0.2 and 0.96 indicating that most of the mixing process takes place in an axial distance of approximately 1R. However, the strong disruption noticed previously doesn t influence the fuel spreading as shown in Figure It is of importance to study the effect of fuel injection on the size of the CRZ. We have to keep in mind that the CRZ is the source of ignition and the upstream region of the CRZ is very sensitive and a region of generation of instabilities. Figure 5.16 shows the size of the CRZ as well as the jet wake regions at different fuel flow rates. It appears that the size of the CRZ is approximately constant with slight variation in the upstream region close to the flare exit. On the other hand, significant reduction of the jets wake region takes place with fuel injection especially for the left primary jet. 5.9 Normalized local equivalence ratio profiles along the centerline at different fuel flow rates. 130

149 5.10 Normalized local equivalence ratio profiles along X/R=-0.8 at different fuel flow rates Normalized local equivalence ratio profiles along Z/R=1.725 at different fuel flow rates. 131

150 5.12 Effect of fuel injection on the flow field at fuel flow rate of 35 pphr. 132

151 5.13 Effect of fuel injection on the flow field at fuel flow rate of 45 pphr. 133

152 5.14 Effect of fuel injection on the flow field at fuel flow rate of 55 pphr. 134

153 5.15 Normalized local Equivalence ratio profiles at fuel flow rate of 55 pphr. 4 Fuel = 0 Fuel = 25 pphr Fuel = 35 pphr Fuel = 45 pphr Fuel = 55 pphr 3 Z/R X/R 5.16 Effect of the fuel injection on the CRZ and Jets wake regions. 135

154 Chapter 6. (Emissions) For several years researchers focused on the emissions generation and control in GTC. Regulations are becoming more stringent and efficient methods for emission control are needed. In the current chapter, the emissions the SAC sector emissions characteristics and control are discussed. 6.1 Experimental setup, procedure and test conditions Swirl cup and SAC Sector The Single Annular Combustor (SAC) Sector presented earlier in Chapter 4 is used to conduct the emissions investigations Diagnostics The SAC sector is installed on the test rig described previously in Chapter 5 (see Figure 4.1). A MultiGas 2030 Fourier Transform Infra Red (FTIR) analyzer is used to conduct the species measurement at the SAC sector exit. A schematic of the FTIR and the emission probe setup are illustrated in Figure 6.1. The FTIR theory of operation is based on the different absorption characteristics every molecule poses when subjected to infra red radiation. Exceptions apply to noble gases and homonuclear diatoms (such as O 2 and N 2 ) as they are infrared inactive. The wave number is an indication of the type of the species detected. Wave number here becomes similar to a human being finger print. The absorption magnitude is an indication of the quantity of that specific species. Typical absorption spectrum is presented in Figure

155 6.1 Schematic showing FTIR and emission probe setup [12]. 137

156 6.2 Typical absorption spectrum [132]. The gas sample is extracted from the SAC sector using a gas sampling probe with liquid cooling jacket. The sampling probe size is designed with small diameter to minimize the impact on the flow field [133]. The sampling probe is constructed from Stainless steel. The outside diameter is 6.25 mm and the tip diameter is 0.7 mm. The probe consists of three concentric tubes. The inner tube is to allow the gas sample to the FTIR. The other two tubes are used for cooling purposes. The line connecting the probe to the FTIR is electrically heated to maintain the sample temperature at 150 C to prevent water condensation. Figure 6.3 shows the setup on the horizontal rig with the emission probe placed at the SAC sector exit. Figure 6.4 shows the emissions measurement in action at high power conditions Test conditions and data acquisition Different species concentrations are measured at a pressure drop and a pre-heat temperature of 4% and 600F respectively. The FTIR reports the measurements in wet basis. The measurements are conducted as follows: 1. The emission probe is placed at the geometrical center at the SAC sector exit. Using Propane as fuel, the power level is varied from low to high power. This is 138

157 accomplished by increasing the fuel flow rate from 8 pphr to 40 pphr in steps of 2 pphr. Roughly the measurements correspond to 0.15 < g < Measurements are repeated with water injection to study the effect of water injection on the generated species. Water to Fuel Ratio (WFR) of 50% and 100% are studied. The measurements are conducted at fuel flow rates of 8 to 40 pphr at steps of 4 pphr (approximately). 3. The exit section is scanned with a x and y resolutions of 8.4 mm and 10 mm respectively. The scanning of the exit cross section is conducted at three conditions. First, low power conditions at fuel low rate of 18 pphr with no water injection. Second. High power condition at fuel flow rate of 36 pphr with WFR of 50%. 4. Finally, Methane is used as a fuel and the centerline measurements are repeated with no water injection. 6.3 Emission measurements in action at low power. 139

158 6.4 Emission measurements in action at high power. 6.2 Results and Discussions The concentrations reported by the FTIR are based on wet analysis. Equation 7.1 is used to convert the species concentration from wet to dry basis. Y and Y are the dry and wet molar concentration of the species i respectively. The concentrations measurements (mole fraction) include Oxygen ( Y O2 i dry %), Carbon Dioxide ( Y CO2 %), Carbon Monoxide ( Y CO %), Hydrogen Cyanide ( Y HCN ppm), Propane ( Y C3H8 ppm), Nitric Oxide also known as Nitrogen monoxide ( Y NO ppm) and Nitrogen dioxide ( Y NO2 ppm). i wet The local air fuel ratio A F local is estimated from the measured concentrations using Equation 7.2 (for propane). The theoretical or stoichiometric air fuel ratio A F sti (propane) is computed using Equation 7.3 and it is approximately