서지주요정보
Experimental and numerical investigations of multi-element lean-premixed hydrogen flame dynamics = 군집 희박 예혼합 수소 전소 화염의 동특성에 관한 실험 및 수치적 연구
서명 / 저자 Experimental and numerical investigations of multi-element lean-premixed hydrogen flame dynamics = 군집 희박 예혼합 수소 전소 화염의 동특성에 관한 실험 및 수치적 연구 / Hyebin Kang.
발행사항 [대전 : 한국과학기술원, 2024].
Online Access 원문보기 원문인쇄

소장정보

등록번호

8042698

소장위치/청구기호

학술문화관(도서관)2층 학위논문

DAE 24001

휴대폰 전송

도서상태

이용가능(대출불가)

사유안내

반납예정일

리뷰정보

초록정보

Hydrogen-fired gas turbine combustion technology is expected to play a key role in accelerating large-scale energy system decarbonization. However, the high reactivity of hydrogen poses significant challenges in designing safe and reliable gas turbine combustion systems. These challenges include increased NOx emissions, potential risks of detrimental flashback events, and modified thermoacoustic instabilities. All of these aspects are addressed in this doctoral dissertation. The present work aims to investigate combustion instabilities in lean-premixed hydrogen flames through both experimental and numerical approaches. To address flashback issues in ultrafast lean-premixed hydrogen flames, a multi-element nozzle array is employed, consisting of 293 small-scale injectors, each with a 3.0 mm inner diameter. Various measurement techniques, including high-speed OH$^*$ chemiluminescence imaging, OH planar laser-induced fluorescence, acoustic pressure measurements, photomultiplier tubes, and hot-wire anemometry, are utilized. Additionally, reduced-order thermoacoustic modeling analysis is applied. For the case of forced flame response, Large Eddy Simulation (LES) is performed using the Ansys Fluent CFD solver under both non-reacting and reacting flow conditions, and the results are then compared against experimental data. The first part of the study investigates self-excited instabilities of lean-premixed pure hydrogen flame ensemble under a broad range of operating conditions. The findings reveal that ultra-compact pure hydrogen flames generate high-amplitude pressure oscillations over a wide range of characteristic frequencies, from 400 to 1800 Hz, corresponding to the third- to tenth-order eigenmodes. Low-frequency flame dynamics exhibit complex interactions between vortices and periodic extinction-reignition processes, leading to large-scale asymmetric oscillations of the entire reaction zone. Intermediate-frequency dynamics display symmetric oscillations with flame merging and pinch-off, without interactions between constituent flames. High-frequency instabilities, surprisingly, are not influenced by structurally complex flame dynamics but instead exhibit a simple back-and-forth motion. This suggests that densely distributed lean-premixed hydrogen-air flames can sustain unstable combustion across a broad spectrum of time scales by modifying the spatiotemporal evolution of the flame ensemble. The second part of the study focuses on understanding the effect of hydrogen content on the thermoacoustic and emission characteristics of multi-element lean-premixed hydrogen/methane/air flames. The results indicate that the system’s response can be classified into several distinctive stages based on their static and dynamic stabilities. These stages include flame blowoff and thermoacoustically stable regions under relatively low hydrogen conditions, low-frequency instabilities at intermediate hydrogen concentration, and the triggering of intense pressure perturbations at approximately 1.7 kHz under high or pure hydrogen conditions. While the flame dynamics subjected to the lower hydrogen concentrations are described by axisymmetric longitudinal motions of parallel flame fronts, the response of higher hydrogen content flames is more pronounced in the transverse direction, accompanied by small-scale vortex roll-up and flame surface annihilation. The longitudinal-to-transverse dynamics plays a mechanistic role in accommodating higher-frequency heat release rate fluctuations, and this newly identified mechanism suggests the possibility of high-frequency transverse modes if such lateral motions are strong enough to induce inter-element flame interactions. Contrary to the substantial differences in thermoacoustic properties for different fuel compositions, the total nitrogen oxides emissions primarily depend on the adiabatic flame temperature. The final part of the study employs an integrative approach, combining loudspeaker-forced direct measurements and LES-based numerical simulations, to explore the dynamic response of multi-element lean-premixed hydrogen flames to harmonic velocity perturbations. The electronically-excited OH intensity distribution, generally assumed to be equivalent to the flame’s heat release rate, shows an anchored conical reaction zone. However, numerical simulations of heat release rate contours reveal the formation of a more concentrated thin annulus region created by preferential diffusion of hydrogen molecules. These results highlight discrepancies between OH intensity distribution and heat release rate contours, suggesting uncertainties in surrogate-dependent flame transfer function evaluations. By employing a reduced-order network modeling framework, the accuracy of predicting self-excited instabilities is tested, and LES-based transfer functions provide more accurate predictions of system stability compared to measurements. This study also identifies moderate transverse oscillation of local energy-concentrated regions as pivotal processes controlling high-frequency hydrogen combustion dynamics. Taken together, the combustion instability phenomenon arising from the coupling between turbulent flames and the acoustic field of the system exhibits significant nonlinear characteristics, emphasizing the importance of acquiring experimental data using a well-designed laboratory setup. The current experimental and numerical investigations provide valuable insights into the dynamics of clustered hydrogen flames under a wide range of operating conditions. These findings are expected to serve as fundamental knowledge for effectively managing high-frequency instabilities in future hydrogen-fired gas turbine engines.

에너지 산업의 탄소중립 달성을 위한 핵심 기술로 수소 가스터빈이 주목받고 있다. 하지만 안정적으로 운용 가능한 시스템을 개발하기 위해서는 수소의 높은 반응성에서 비롯되는 여러 기술적 난제들을 극복해야 한다. 대표적으로 화염 역화, 열음향 불안정성, 질소산화물 증가와 같은 문제들을 학위 논문에서 중점적으로 다루었다. 본 연구에서는 군집 예혼합 수소 화염의 연소불안정 현상을 실험적/수치적 기법을 통해 조사하였다. 빠른 화염 전파속도를 갖는 수소의 역화 발생을 방지하기 위해 내경 3.0 mm의 소형 인젝터 293개로 구성된 다중노즐 배열을 설계하여 실험에 적용하였다. 고속 OH$^{*}$ 자발광 이미지, OH 평면 레이저 유도 형광법, 음향 압력 섭동, 광증배관, 열선 유속계 등의 다양한 실험적 기법과 더불어 저 차원 열음향 모델링 분석을 수행하였으며, 이를 통해 수소 화염의 자발 불안정 현상 및 외부 가진에 대한 화염 응답을 살펴보았다. 특히 가진 화염 응답의 경우 Ansys Fluent CFD 솔버를 이용한 대와류 모사(LES) 기반의 수치 해석을 통해 비반응 및 반응 유동장을 분석하고 이를 실험 결과와 비교하였다. 먼저 군집 예혼합 수소 화염의 자발 불안정성을 조사하기 위해 광범위한 운전 조건에서 실험을 수행하였다. 실험 결과, 수소 화염은 매우 좁은 반응 영역에서 존재하며, 종 방향 3차부터 10차 고유모드에 해당하는 400 부터 1800 Hz 사이의 넓은 주파수 범위에서 강한 압력 섭동을 나타냈다. 낮은 주파수에서는 와류 간의 복잡한 상호작용과 주기적인 소염 및 재점화 현상으로 인해 전체 반응장에 대규모 비대칭 거동이 관찰되었다. 불안정 주파수가 증가함에 따라 개별 화염들은 독립적으로 움직이기 시작하며, 양쪽 화염 면이 합쳐진 후 다시 분리되는 핀치 오프 현상이 모든 화염 면에 균일하게 나타났다. 수소 화염에서 특징적으로 관찰되는 고주파/고진폭 불안정 섭동은 구조적으로 복잡한 화염 거동 대신 길이 방향으로 가장 단순한 움직임을 통해 발생하였다. 해당 결과는 수소 화염이 다양한 형태의 군집 거동을 통해 넓은 주파수 스펙트럼에서 불안정을 유발할 수 있음을 보여주며, 특히 고주파 수소 화염의 동역학을 실제 실험을 통해 관찰한 것에 의의가 있다. 다음으로 군집 예혼합 수소/메탄/공기 화염의 수소 혼소율을 0\% 부터 100\% 까지 변화시키며 열음향 불안정성과 배기 배출 특성을 확인하였다. 혼합물의 수소 농도가 증가함에 따라 시스템은 여러 단계의 정적 및 동적 상태를 나타냈으며, 비교적 낮은 농도의 수소 혼소 시 발생하는 화염 날림 현상과 열음향 안정 구간으로 시작으로, 중간 수소 농도에서는 저주파 불안정성을, 고농도 혹은 순수 수소 조건에서는 약 1.7 kHz에 해당하는 높은 불안정 주파수에서 강한 압력 섭동을 발생시켰다. 혼합물의 수소 농도가 증가함에 따라 개별 화염의 동역학은 축 대칭의 종 방향 운동으로부터 작은 스케일의 와류 롤업과 화염 면의 소염을 동반한 횡 방향 움직임으로 변화하였다. 이를 통해 종 방향에서 횡 방향으로의 전환되는 수소 화염의 동역학이 고주파 열음향 불안정성을 야기시키는 주요 메커니즘임을 규명하였다. 또한 해당 결과는 시스템에 충분한 에너지가 공급될 경우 화염-화염 간 상호작용을 통해 kHz 단위의 횡 방향 불안정으로 성장할 수 있는 가능성을 시사한다. 수소/메탄 화염의 열음향 특성은 연료 조성에 따라 상당한 차이를 나타낸 반면 질소산화물 배출량은 단열 화염 온도에 지배적인 영향을 받는 것으로 확인되었다. 마지막으로 라우드스피커 기반의 가진 실험과 LES 기반의 수치 해석을 결합하여 외부 음향 가진에 대한 군집 수소 화염의 응답을 조사하였다. OH 강도는 일반적으로 화염의 열 방출 분포와 동일하다고 가정되며, 실험에서는 원뿔 형태의 반응 영역으로 확인되었다. 반면 수치해석을 통해 획득한 열 방출 분포는 수소 분자의 빠른 확산으로 인해, 얇은 두께의 반응 영역을 갖는, 화염 끝이 열린 원기둥 형태로 관찰되었다. 실험과 해석 간 이러한 열 방출 분포의 차이는 대리 변수를 통해 획득하는 화염전달함수 계측의 불확실성을 암시한다. 이후 저 차원 열음향 네트워크 모델링 기법을 도입하여 시스템의 불안정 구간을 예측하고, 실험을 통해 계측한 자발 불안정 구간과 비교하였다. 실험을 통해 계측한 화염전달함수보다 LES 기반의 화염전달함수가 더 높은 정확도로 불안정을 예측하였으며, 따라서 수치해석에서 관찰한 열 방출 분포가 실제에 더 가까운 것으로 사료된다. 화염의 열 방출 분포는 국부적 차이가 존재했지만 고주파 가진 환경에서 수소 화염의 응답은 좁은 영역에 집중된 화염의 횡 방향 움직임으로 실험과 해석에서 동일하게 관찰되었다. 난류 화염과 시스템 음향장의 공진 결합으로 발생하는 연소불안정 현상은 상당한 비선형 특성을 내포하고 있어 실험 데이터의 확보가 매우 중요하다. 본 연구 결과는 광범위한 운전조건에서 군집 수소 화염의 거동을 실험 및 수치해석을 통해 고찰하였다는 것에 의의가 있으며, 향후 수소 가스터빈 개발에 있어 고주파 불안정성 제어를 위한 기초 자료로 활용될 것으로 기대된다.

서지기타정보

서지기타정보
청구기호 {DAE 24001
형태사항 xiv, 150 p. : 삽도 ; 30 cm
언어 영어
일반주기 저자명의 한글표기 : 강혜빈
지도교수의 영문표기 : Kyu Tae Kim
지도교수의 한글표기 : 김규태
Including appendix
학위논문 학위논문(박사) - 한국과학기술원 : 항공우주공학과,
서지주기 References : p. 133-147
주제 Combustion dynamics
Fuel-flexible
Gas turbine combustion
Hydrogen
Multinozzle array
연소불안정
연료다변화
가스터빈 연소
수소
다중노즐
QR CODE

책소개

전체보기

목차

전체보기

이 주제의 인기대출도서

Changes in global energy sources over the last two decades [1.

Total primary energy supply by energy carrier group, 2020to 2050 under the 1.5'Cscenario Renewables include bioenergy, geothermal, hydropower, ocean, solar, and wind in all forms. Fossil fuels include coal, oil, and natural gas. p.p. = percentage point [3].

Power-to-X pathways to decarbonize energy with green hydrogen ㅣ11

Comparison of fuel properties between hydrogen, methane, and ammonia 5, 16].

(a) Flame stabilization and occurrence of flashback. Scaled temperature with respect to adiabatic flame temperature at various inlet velocities (W = 1 mm and 0 = 0.7 [22]). (b) Hardware damage due to flameholding in a Dry Low-NOx (DLN) fuel nozzle [31|.

(a) Diagram of a gas turbine engine and (b) different types ofgas turbine combustors 32

Aircraft and heavy-duty gas turbine engines. Box indicates combustor section 31

(a) Dependence of NO emissions upon residence time at different flame temperatures [36]. (b) Typical CO and NOx trends in a DLN combustion system as a function of combustion product temperature 32].

Paradigm shift in gas turbine combustion technology.

Current and future capabilities of OEMs in hydrogen gas turbine technology [45].

(a) Conventional large-scale swirl-stabilized nozzle for methane (or natural gas) combustion Full-scale multi-tube combustion systems for hydrogen fuel developed by (b) Mitsubishi Power (MHPS) (c) General Electric (GE), and (d) Kawasaki Heavy Industries (KHI). (e) 100% hydrogen-fueled dry- combustion gas turbine system developed by KHI equipped with nozzle array in Fig. 1.7d [2].

(a) Schematic of GE DLN1 fuel injection cross section. Multiple fuel injection points for fuel staging [57]. (b) Schematic of Ansaldo GT26 and GT36 sequential combustion schemes. Two combustion stages in series, the first based on flame propagation and second based on auto-ignition [58].

(a) Coupling between acoustics and heat release rate with a potential of driving thermoa- coustic instabilities. (b) Representation ofa single turbulent flame in a confined geometry.

Normalized combustor pressure amplitude plotted against the phase difference between pressure and heat release rate. A total of~ 15,000 data points is presented [71|.

(a) Time traces capturing the transition to a limit cycle in the combustor. (b) Stable operation at t = t1 and (c) limit cycle oscillations at t=t2.

Dependence of the acoustic driving, H(A), and damping, D(A), processes on the insta- bility amplitude A. (a) Linearly unstable system with nonlinear saturation at a limit cycle, ALC· (b) Nonlinearly unstable system with instability triggering point, AT, and a limit cycle, ALC [13, 72|.

Types of acoustic modes in combustors.

Types of combustion instabilities in rocket engines.

(a) Image ofkey external features of the F-1 engine. (b) Failure of liquid rocket engines due to high-frequency combustion instabilities. (c) Several baffle injector arrangements for F-1 engine test 63|.

Types of combustion instabilities in aircraft engines.

Combustion instabilities in gas turbine and aircraft engines [73]. (a) Helmholtz mode in a single combustor. (b) Longitudinal mode in a can-annular combustors [74]. (c) Azimuthal mode in annular combustors [61].

(a) Normalized heat release response. (b) Gain offlame transfer function plotted as a function of forcing amplitude. Inlet conditions: Tin = 473K, V = 60 m/s, 0 = 0.60, XH2 = 0.30 [76].

Two-dimensional stability maps for (a) 100% NG and (b) 75%/25% NG/H2 with corre- sponding stable flame structures [101].

(a) Heat release scales with the maximum value of reference 1D flat flame for hydrogen (top), methane (middle), and propane (bottom) flames at constant SL and Vin [105]. (b) Visualization of the TurbUnstable flame represented by an iso-surface of temperature 24..

(a) Schematic and flame images ofcan-type multi-nozzle combustor system [50]. (b) Flame visualization using chemilurninescence images of annular-type multi-nozzle combustor system 136].

Comparison between lean-premixed hydrogen-air and methane-air flames 168]. (a) Stable flame structures, (b) calculated longitudinal-mode eigenfrequencies (curved lines for L1-L6 modes), and measurements of dominant instability frequencies (symbols) as a function of combustor length.

(a) NASA Lean Direction Injection (LDI) injector with its design principle [43], (b) 4x 4 mesoscale nozzle array |163, 164, 169], (c) hexagonal array [166], and (d) multislit nozzle ㅣ170, 171].

Overview of previous research on combustion characteristics of hydrogen.

Overview of previous research on flame-flame interactions.

Overview of previous research on clustered lean-premixed and diffusion flames.

Research overview.

3D CAD model of the test rig composed of laser-based diagnostic and gas sampling mea- surement setups. Direct flame photograph of clustered 293 hydrogen flames under adiabatic flame tem- perature of2000 K; digital SLR camera, Nikon D500, ISO5000, aperture f/3.5 with exposure time 1/13 second and color temperature of 4500 K.

Experimental setup for detailed investigation of self-excited instabilities, visualization of non-linear flame dynamics, and measurements ofexhaust gas emissions. (a) Cross-sectional view oflean- premixed combustion test rig along with OH PLIF diagnostic and gas sampling measurement setups. Length left to right is about 4880 mm when the piston is fully extended. (b) Spatial arrangements of multi-e

Experimental setup designed to investigate forced flame response. (a) Layout of lean- premixed combustion test facility, along with external acoustic forcing and flame imaging measurement configurations. (b) Spatial arrangement ofmulti-element injector array (N = 293) used for flame trans- fer function measurements. (c) Cutaway view of injector assembly, including two calibrated hot wire probes po

Emission spectra in typical hydrogen-air flames: (a) premixed H2-air jet flame and (b) diffusion H2-air jet flame. Note that the visible spectra extending from 350 nm to 850 nm in Fig. 2.4a was scaled by a factor of 6.5 with respect to the OH spectra in the ultraviolet [189].

(a) A schematic diagram ofthejet settling chamber to calibrate hot wire anemometer. (b) Calibration curve of the hot wire probe.

(a) Production ofcarbon dioxide (CO2), water (H2O),and oxygen (02) for 300 K reactants at 1 bar with adiabatic flame temperature of2000 K. (b) Dry NOx emissions (in ppmvd at 15% 02) for a constant temperature system operated at various hydrogen methane ratios relative to the dry NOx emissions from pure methane |190]

Schematic of imaging measurements setup.

(a) Schematic of radially symmetric optically thin flame [191]. (b) Line-of-sight image of typical swirl-stabilized flames and Abel-transformed OH* chemiluninescence image [192].

Schematic overview of dynamic mode decomposition (DMD on a fluid flow data set -195|.

Schematic of procedure for flame surface density (FSD) calculation.

Summary of instruments and corresponding uncertainties.

Schematic diagram of turbulent kinetic energy spectrum describing energy cascade of turbulent flows [197].

Fundamental difference among the three modeling approaches (RANS/LES/DNS) lies in which length scales are modeled and resolved.

Governing equations for Reynolds Averaged Navier-Stokes (RANS).

Governing equations for Large Eddy Simulation (LES).

Method of LES-based flame transfer function analysis.

Reduced-order network model for multi-element combustion test rig

Experimental conditions.

Normalized pressure amplitude spectra for combustor lengths between 1200 and 1800 mm for adiabatic flame temperatures of1400, 1600, 1800, 2000 K (differentiated by colors) and mean nozzle velocity conditions of30 m/s (left column) and 40 m/s (right column). Vertical bands with light green and light magenta indicate third and fourth longitudinal eigenfrequencies calculated using a low-order thermoa

Self-excited instability measurement data under mean nozzle velocity of30 m/s. Measured instability frequency plotted against combustor length (top). Curved line indicate longitudinal-mode eigenfrequencies obtained from low-order network model (bottom). Dependence ofnormalized dynamic pressure and heat release rate fluctuation amplitudes on combustor length. Error bars show one standard deviation.

Self-excited instability measurement data under mean nozzle velocity of40 m/s. Measured instability frequency plotted against combustor length (top). Curved line indicate longitudinal-mode eigenfrequencies obtained from low-order network model (bottom). Dependence of normalized dynamic pressure and heat release rate fluctuation amplitudes on combustor length. Error bars show one standard deviation

Comparison between calculated longitudinal-mode eigenfrequencies (curved lines for L1-L5 modes) and measurements of dominant instability frequencies (symbols) for lean-premixed hydrogen-air flames as a function of combustor length between 600 and 1800 mm. Test conditions: Tad = 2000 K and Unozzle = 30 m/S. Cutoff pressure fluctuation amplitude is 0.05 kPa (zero-to-peak).

Illustration of isocontour stability maps in the (lc,Unozzle) domain for constant adiabatic temperature of 1600 K, in terms of (a, b) normalized combustor pressure amplitude at dominant and harmonic frequencies, (c) normalized amplitude offlame's global heat release rate fluctuations, (d) con- tribution of harmonic amplitude with respect to dominant frequency, (e) limit-cycle frequency, and (f) ph

Thermoacoustically stable structures of an ensemble of 293 lean-premixed pure H2/air flames under adiabatic flame temperature of 1400, 1600, 1800, 2000 K. (a) Time-averaged line-of-sight integrated OH* chemiluminescence emission images. (b) Instantaneous OH PLIF images for 60 X 60 mm2 region ofinterest (ROI) marked. Spatial resolution of OH* chemilurninescence and OH PLIF mea- surements is 0.129 a

Phase-resolved flame dynamics for Case A during a period of limit cycle oscillation at 597 Hz. (a) Line-of-sight integrated OH* chemiluminescence intensity distributions. (b) Instantaneous OH PLIF images for 60 X 60 mm2 region ofinterest. (c) Ensemble-averaged OH PLIF images corresponding to six phase angles.

Overview of flame surface area calculation procedure. (a) Maximum-local-gradient-based flame front calculation using phase-resolved single-shot OH PLIF images presented in Fig. 3.7b. (b) Binarization of individual flame region along with close-up of boxed region. (c) Flame area calculation through averaging left- and right-hand-side contours after revolving both fronts about flame axis (x- axis).

Phase-resolved flame dynamics for Cade B during a period of limit cycle oscillation at 421 Hz. (a) Line-of-sight integrated OH* chemiluminescence intensity distributions. (b) Instantaneous OH PLIF images for 60 X 60 mm2 region ofinterest. (c) Ensemble-averaged OH PLIF images corresponding to six phase angles. (d) Time traces of combustor dynamic pressure and global heat release rate oscilla- tions

Phase-resolved flame dynamics for Cade C duringa period oflimitcycle oscillation at 1717 Hz. (a) Line-of-sight integrated OH* chemiluninescence intensity distributions. (b) Instantaneous OH PLIF images for 60 X 60 mm2 region ofinterest. (c) Ensemble-averaged OHPLIF images corresponding to six phase angles.

Temporal and spatial evolution patterns of (a) 597 Hz (Case A) and (b) 1717 Hz (Case C) self-excited hydrogen flame dynamics. (c, d) Time traces of combustor dynamic pressure, global heat release rate oscillations (upper panel), and phase difference between two properties (lower panel) overlaid with 0o to 360o phase angles (filled circles). Fast Fourier transform (FFT) ofpressure signal and probab

Effect ofinjector diameter on the occurrence of combustion instabilities oflean-premixed hydrogen-air flames in two different multi-element injector arrays. Experimental data for Configuration B are reproduced from [168]. (a, b) Normalized combustor pressure and global heat release rate oscillation amplitude against instability frequencies. Color-coded symbols are used to differentiate the order o

Comparison of combustion instabilities of lean-premixed hydrogen-air flames in four different patterns of multi-element nozzle arrays [168-171].

Experimental conditions.

Variation of (a) equivalence ratio, (b) laminar flame speed, and (c) effective Lewis number from pure methane to pure hydrogen for Cases A and B.

Dependence of self-excited pressure oscillations on hydrogen mole fraction under constant adiabatic flame temperature conditions of (a) 1600 K and (b) 2000 K, corresponding to Cases A and B respectively. Insets fL2 - fLs stand for eigenfrequencies calculated by low-order network modeling.

Thermoacoustically stable structures oflean-premixed H2 CH4 airflames with different fuel compositions between XH2 = 0.5 and 1.0 (Case A). (a) Representative flame luminescence photographs for qualitative comparison, (b) line-of-sight integrated OH* chemiluminescence images ofentire reaction zone, and (c) instantaneous OH fluorescence images for 60 X 60 mm2 region of interest marked in Fig. 4.3b.

Power spectral density and representative time traces of pressure oscillations along with corresponding probability density functions for Case B. From top to bottom, hydrogen mole fraction is increased from 0.1 to 1.0. Colors are used to differentiate the order ofeach mode: blue (L2), green (L3) and yellow (L8). Vertical gray bands represent eigenfrequencies obtained from low-order thermoacoustic

(a) Sequence of phase-synchronized single-shot OH PLIF images during a period oflimit cycle oscillations at 630 and 1680 Hzfor XH2 = 0.60 and 0.90, respectively (Case B). (b) Spatiotemporal evolution of center flame fronts calculated from instantaneous OH fluorescence images marked as green rectangles in Fig. 4.5a (rotated by 90이). (c) Comparison of acoustic pressure and global heat release rate f

(a, b) Time traces of OH* chemiluminescence intensity fluctuations and (c, d) high-speed flame images during a period of limit cycle oscillations at 630 and 1680 Hz for XH2 = 0.60 and 0.90, respectively (Case B). A total of 32,000 images were obtained at a sampling rate of 16,000 frames per second for 2 S At = 1/16,000 = 62.5 lis). Flow direction is from top to bottom in positive c-direction, and

DMD analysis offlame dynamics at 630 Hz for XH2 = 60%. (a) DMD spectrum of mode amplitude (top) and growth rate (bottom). Colored open circles denote dominant modes ofinterest. (b Spatial distribution offirst 5 dominant DMD modes.

DMD analysis offlame dynamics at 1680 Hz for XH2 = 90%. (a) DMD spectrum ofmode amplitude (top) and growth rate (bottom). Colored open circles denote dominant modes ofinterest. (b) Spatial distribution of first 6 dominant DMD modes.

Left /right columns: 630 and 1680 Hz for XH2 = 0.60 and 0.90 (Case B). (a, b) Normalized OH* chemiluminescence emission intensities corresponding to fundamental mode, with cutoff amplitude being set to 10% of maximum amplitude. (c, d) Phase difference between acoustic pressure and heat release rate fluctuations for fundamental mode. Spatial distribution of(e, f) first DMD mode and (g, h) second DM

Influence of adiabatic temperature on emission characteristics ofhydrogen fuel blends: (a) Nitrogen oxides and (b) carbon monoxide emissions by volume on a dry basis, corrected to 15% oxygen (reported in units ofppmvd). (c) Measured oxygen concentrations presented in vo1% and (d) calculated correction factors for hydrogen fuel blends.

Dry NOx emissions for a constant temperature system operated at various hydro- gen/methane ratios relative to the dry NOx emissions from pure methane [190].

Influence of adiabatic temperature on total (a) nitrogen oxides and (b) carbon monoxide emissions by volume on a dry basis corrected to 15% oxygen (reported in unit of ppmvd). Additional corrections for hydrogen-containing fuel blends were made for all test cases. Correction factors are summarized in Fig. 4.10d. Exponential function-based least-squares fit of experimental data provides squared lin

Experimental conditions.

(a) Spatial arrangement of multi-element injector (N = 293) array along with 5-injector elements for numerical simulation. (b) Illustration ofcomputational domain and corresponding boundary conditions. (c) Unstructured meshed overlaid with calculated heat release rate contours (about 4 million cells). All dimensions in millimeters.

(a) Laminar burning velocity and adiabatic flame temperature. (b) Profiles of mass fractions of major species and temperature of a stoichiometric 1-D propagating premixed flame [213].

H2/O2 reaction mechanism utilized in this study [213].

Influence ofgridrefinement evaluated in terms of mean axial velocity and mean temperature variations: (a, b) property variations at nozzle center plane, and (c, d) representative velocity and temperature profiles at four equally spaced axial locations, marked as horizontal dashed lines in Figs. 5.3a and 5.3b.

Numerical models and boundary conditions.

Turbulent kinetic energy (TKE) spectrum in the near field ofthe injector exit

Overview of experimental and numerical studies carried out in Chapter 5.

(a) Two-dimensional velocity distribution for center injector element measured under non- reacting flow conditions. Linear interpolation was used to represent the velocity field in the x - y plane. (b) Comparison ofradial velocity profiles (mean and fluctuation values) between experiment and numerical simulations at X/D =1,2, 3, 4, and 5 marked with white dashed lines. D is inner diameter of singl

Time-averaged fields on a vertical plane along the burner axis. (a) H2 mole fraction, (b O2 mole fraction, (c) OH mole fraction, (d) heat release rate (HRR), and (e) temperature.

Experimental and numerical simulation results for center nozzle flame structures under unperturbed conditions. From left to right: (a) Abel-deconvoluted time-averaged OH* emission intensity profile, (b) computed heat release rate (HRR) contour, with averaging time for LES being 120 ms, (c) time-averaged OHfluorescence intensity distribution, (d) mean OH mass fraction (LES), (e) flame surface densi

(a) Transfer functions of upstream and downstream velocity disturbances across whole injector assembly, measured under isothermal flow conditions. Red solid line marks average values of measured quantities between 60 and 1200 Hz. Data points with different colors correspond to different amplitudes of downstream velocity fluctuations up to ~ 40% ofmean nozzle velocity. (b) Flame transfer functions

(a) Longitudinal-mode eigenfrequencies calculated using experimental (solid lines) and numerical (dashed lines) flame transfer functions. Experimentally measured instability frequencies noted (diamond symbols). (b) Top: self-sustained pressure amplitude from model validation measurements Middle and bottom: Growth rates ofL3-mode, calculated from experimental and numerical FTF data, plotted as a fu

(a, b) Time traces of normalized amplitude of acoustic velocity and heat release rate fluctuations in response to single-frequency excitation of 600 and 1350 Hz (LES). (c to f) Forced flame dynamics represented in terms of Abel-deconvoluted OH* chemilunninescence emission intensity (experi- ment) and calculated heat release rate contours (simulation): only center flame is shown here for clarity, a