DISS. ETH NO. 26995
THE ROLE OF ENTROPY WAVES IN COMBUSTOR THERMOACOUSTICS
A thesis submitted to attain the degree of DOCTOR OF SCIENCES of ETH Z ¨URICH
(Dr. Sc. ETH Z¨urich)
presented by
MARKUS WEILENMANN
M. Sc. ETH Z¨urich
born on 01.03.1988
citizen of Winterthur (ZH), Switzerland
accepted on the recommendation of:
Prof. Dr. Nicolas Noiray Prof. Dr. Thomas Sattelmayer
Dr. Andrea Giusti
2020
In recent years, the effects and risks of climate change have become increas- ingly evident. Public awareness of the effects of man-made climate change has increased and there is common consensus that renewable energy sources have to be incorporated in the energy mix of the future. The transition to a fully renewable system is a long path and it is expected that gas turbines will continue to play an important role over the coming decades. The supply of energy from renewable sources is highly intermittent and gas turbines have proved to be a viable platform to tackle the challenge of stabilizing energy grids with a substantial proportion of renewables. Gas turbines allow highly efficient conversion of fossil energy and yield the lowest pollutant emission lev- els. Alternatives are scarce, especially where hydro power is not available and a nuclear phase-out is desired. In addition, promising research is ongoing re- garding the use of alternative fuels in aero engines and land based gas turbines.
The combustion of syngas or sustainably-produced fuels like H2, Ammonia and biofuels has great potential to reduce the environmental impact of these ma- chines in the future.
With exception of the recent crisis related to the Corona virus, international air travel is growing rapidly. In the coming decades, no viable alternative to combustion based propulsion is in sight and the contribution of the air trans- port sector to the global level of pollutant emissions is, and will be, substantial.
To reach low emission values in both land based gas turbines and aero engines, the application of the lean combustion concept is key. Unfortunately, asso- ciated designs are prone to thermoacoustic instabilities, which still remain a major challenge in the development process of new machines. The complex phenomenon of thermoacoustic instability is the result of constructive interac- tion between the sound field and a confined unsteady heat source. If it occurs in a practical system, the resulting acoustic pressure oscillations may reach dangerous levels that induce cyclic mechanical and thermal loads on engine components, which can ultimately lead to structural failure of the machine.
It is therefore safety critical to avoid thermoacoustic instabilities either by design or by restricting operation to conditions free of thermoacoustic insta- bilities, which is in contrast with the need for operational flexibility. Despite
intense research in the past, the prediction of thermoacoustic instabilities re- mains challenging and further work is required to understand and model the behaviour in real engines.
Entropy waves, which essentially are temperature fluctuations, can play an important role in thermoacoustic feedback, because these convective flow in- homogeneities produce sound when they are accelerated in the turbine inlet, and this sound can travel upstream and impact flame dynamics. This work is focused on the three key processes involving entropy waves in the context of thermoacoustic instabilities, which are 1) The generation of entropy waves in the reactive region, 2) the advection of entropy waves through the combustion chamber and their simultaneous decay, and 3), the sound generation resulting from the subsequent acceleration of entropy waves in the turbine inlet.
Quantitative measurements are presented of the production of entropy waves in a technically premixed turbulent combustor, subject to acoustic forcing.
Two burners of the same length with technical premixing of air and natural gas, operated at the same thermal power, were characterized: a matrix burner producing an array of turbulent jet flames, and a burner producing a single swirled turbulent flame. Significant differences are revealed between the two burners. The measurements can be used to derive predictive nonlinear models of thermoacoustic instabilities involving entropy wave feedback.
To study the advection and conversion into sound of entropy waves, non- reactive experiments were conducted, where a turbulent channel flow of air was subject to a temperature controlled pulsed jet-in-crossflow, generating en- tropy waves in a flow field that is more complex than previously studied chan- nel flows. The challenge of measuring low amplitude temperature fluctuations is tackled by employing a novel approach to background-oriented Schlieren (BOS) thermometry, which is complemented by large-eddy simulations (LES).
By comparison with a state-of-the-art decay model, the results show the strong dependency on the flow field. The challenge of isolating the contribution of accelerated entropy waves to the sound field is addressed by an experimental approach, which is based on the comparison of acoustic reflection coefficients to a reference case. A significant contribution was measured and is compared to a theoretical model for compact supercritical nozzles. Substantial discrep- ancies are revealed and explained.
A self-sustained thermoacoustic instability is studied in a sequential combus- tor, where the thermoacoustic feedback involves two flames: the perfectly pre- mixed swirled flame anchored in the first stage and the sequential flame estab- lished downstream of the mixing section, into which secondary fuel is injected in the vitiated stream from the first stage. It is shown that the large amplitude flapping of the secondary fuel jet in the mixing section plays a key role in the
in thermoacoustic feedback involving entropy waves by employing quantitative and qualitative experimental techniques.
Zusammenfassung
In den letzten Jahren sind die Effekte und Risiken der globalen Klimaerw¨armung immer sichtbarer geworden. Die Sensibilisierung und Aufkl¨arung der ¨Offentlichkeit bez¨uglich der durch den Menschen verursachten Klimawandel hat zugenommen und es besteht ein allgemeiner Konsens, dass erneuerbare Energiequellen eine wichtige Rolle spielen sollen im Energiemix der Zukunft. Der ¨Ubergang zu einer Energieversorgung, die ausschliesslich erneuerbare Energiequellen nutzt, ist jedoch ein langer Prozess und Gasturbinen werden in den n¨achsten Jahrzehn- ten voraussichtlich weiterhin eine wichtige Rolle spielen. Erneuerbare En- ergiequellen verhalten sich stark intermittierend und Gasturbinen haben sich f¨ur die Stabilisierung von Stromnetzen mit hohem Anteil an erneuerbaren En- ergietr¨agern als gute L¨osung etabliert. Gasturbinen sind mit grossem Abstand die effizienteste, auf fossilen Energiequellen beruhende, Kraftwerkstechnologie mit den tiefsten Emissionswerten. Alternativen sind rar, besonders wenn keine Wasserkraft verf¨ugbar ist und nukleare Stromerzeugung unerw¨unscht ist. Zu- dem wird intensiv geforscht an alternativen Brennstoffen f¨ur Gasturbinen und Flugzeugtriebwerke. Die Verbrennung von Syngas, oder erneuerbar hergestell- ten Brennstoffen wie Ammoniak, Wasserstoff und Biotreibstoff haben grosses Potential die Umweltvertr¨aglichkeit dieser Maschinen zu verbessern.
Der internationale Flugverkehr nimmt kontinuierlich zu, mit Ausnahme der Coronakrise im Jahr 2020. Zurzeit ist keine realistische Alternative in Sicht, um die klassischen Mantelstromtriebwerke in den kommenden Dekaden abzul¨osen.
Der Anteil der Aviatik an den global freigesetzten Schadstoffemissionen wird deshalb in der nahen Zukunft betr¨achtlich sein und eine weitere Optimierung der Technologie daher essenziell.
Das Anwenden von mageren, vorgemischten Verbrennungskonzepten ist em- inent wichtig, um tiefe Emissionswerte in Flugzeugtriebwerken und Gastur- binenkraftwerken zu erreichen. Ungl¨ucklicherweise sind technische Realisierun- gen von Brennern, unter Anwendung dieses Konzepts, sehr anf¨allig auf ther- moakustische Instabilit¨aten, was bei der Entwicklung neuer Maschinen auch in Zukunft eine grosse H¨urde bleiben wird. Dieses komplexe Ph¨anomen ther- moakustischer Instabilit¨aten ist das Resultat konstruktiver Interaktion zwis- chen dem akustischen Feld und instation¨aren W¨armequellen in Brennkam-
mischen Belastungen der Maschinenkomponenten zum Totalausfall des Sys- tems f¨uhren k¨onnen. F¨ur den sicheren Betrieb ist es deshalb unabdingbar, thermoakustische Instabilit¨aten mit einer durchdaten Brennkammerarchitek- tur zu Vermeiden oder die Betriebsgrenzen auf ein Betriebsfenster, welches ein Auftreten thermoakustischer Instabilit¨aten ausschliesst einzugrenzen. Dies steht jedoch in Konflikt mit der ben¨otigten operationellen Flexibilit¨at. Die Vorhersage thermoakustischer Instabilit¨aten ist, trotz intensiver Forschung in der Vergangenheit, immer noch anspruchsvoll und mehr Forschungsarbeit ist n¨otig, um bessere Modelle f¨ur reale Systeme zu entwickeln.
Entropiewellen sind haupts¨achlich Temperaturfluktuationen und k¨onnen in thermoakustischen R¨uckkopplungsschleifen eine wichtige Rolle spielen, weil die Beschleunigung dieser konvektierten Inhomogenit¨aten im Turbineneintritt die Emission von Schallwellen verursacht, welche stromaufw¨arts propagieren und dort die Flammendynamik beeinflussen k¨onnen. Diese Arbeit behandelt drei Schl¨usselprozesse, welche f¨ur die Bedeutung von Entropiewellen im Kon- text von thermoakustischen Instabilit¨aten wichtig sind. 1) Die Produktion von Entropiewellen in der reaktiven Zone, 2) der konvektive Transport von Entropiewellen durch die Brennkammer und der gleichzeitige Zerfall und 3) die Emission von Schallwellen bei anschliessender Beschleunigung.
Es werden quantitative Messungen von Entropiewellen in einem technisch vorgemischten turbulenten Brenner, welcher akustisch angeregt wurde, pr¨asentiert.
Zwei Brenner identischer L¨ange mit technischer Vormischung von Erdgas und Luft wurden charakterisiert bei gleicher thermischer Leistung: Ein Matrix Brenner, welcher eine symmetrische Anordnung von turbulenten Flammen be- wirkt und ein Brenner, welcher eine einzelne, drall-stabilisierte, turbulente Flamme produziert. Es werden bedeutsame Unterschiede zwischen den zwei Brennertypen aufgezeigt. Die Messresultate k¨onnen zur Entwicklung nicht- linearer Modelle f¨ur die Vorhersage von thermoakustischen Instabilit¨aten mit involvierten Entropiewellen benutzt werden.
Der Zerfall von Entropiewellen w¨ahrend dem konvektiven Transport durch die Brennkammer, wurde in nicht-reaktiven Experimenten untersucht. Mittels einer Sirene wurde ein pulsierender, temperierter Luftstrahl in eine turbulente Kanalstr¨omung injiziert. Das resultierende Str¨omungsfeld mit Entropiewellen ist komplexer als in fr¨uheren Experimenten. Die Herausforderung diese Tem- peraturfluktuationen mit niedriger Amplitude zu messen, wurde mit einem neuartigen Ansatz zur Verwendung derbackground-Oriented Schlieren (BOS) Methode gemeistert. Die Resultate wurden erg¨anzt mit Daten vonLarge-Eddy Simulationen (LES). Der Vergleich mit einem aktuellen Zerfallsmodel zeigt die
starke Abh¨angigkeit vom Str¨omungsfeld.
Die Schwierigkeit den Anteil am akustischen Feld, welcher durch das Beschle- unigen von Entropiewellen entsteht, zu messen und zu isolieren, wurde mit einem experimentellen Verfahren gel¨ost, welches akustische Reflexionskoef- fizienten mit einem Referenzwert vergleicht. Es konnte ein signifikanter Beitrag gemessen werden, welcher mit einem theoretischen Modell f¨ur kompakte D¨usen, mit Schallgeschwindigkeit im engsten Querschnitt, verglichen wird.
In einem sequentiellen Brenner wurde eine selbst angeregte thermoakustische Instabilit¨at untersucht, bei der die thermoakustische R¨uckkopplungsschleife zwei Flammen umfasst: Eine perfekt vorgemischte drall-stabilisierte Flamme in der ersten Stufe und eine selbstz¨undende Flamme in der sequentiellen Stufe, welche sich stromabw¨arts der Mischungssektion befindet, in welcher der sekund¨are Brennstoff in das heisse Abgas der ersten Stufe ein ged¨ust wird. Es wird aufgezeigt, dass das starke Flattern des sekund¨aren Brennstoffstrahls einen wichtigen Mechanismus f¨ur diese Art von Instabilit¨at darstellt.
Im Allgemeinen ist das Ziel dieser Arbeit, mit Hilfe von qualitativen und quan- titativen experimentellen Methoden, n¨utzliche Erkenntnisse ¨uber den Einfluss von Entropiewellen auf thermoakustische R¨uckkopplungsschleifen zu gewin- nen.
The work leading to this thesis was carried out at the Combustion and Acous- tics for Power & Propulsion Systems Laboratory (CAPS), which is part of the Institute of Energy Technology and the Department of Mechanical and Process Engineering (D-MAVT) at ETH Z¨urich in Switzerland. During the past years, I experienced great teamwork and support by colleagues and faculty. Without the encouragement and inputs of numerous people, the presented results could not have been accomplished. For this reason, I would like to hereby express my profound gratitude to everybody who was involved in my PhD journey.
First of all, I would like to thank my supervisor, Prof. Dr. Nicolas Noiray.
I am thankful that you gave me the opportunity to join your team at a very early stage. I learned a lot during the setup process and enjoyed watching the lab grow. Also, I am proud that I was able to contribute to a lab that will hopefully help to solve current and future energy problems with its scientific contribution. Thanks a lot Nicolas, for your guidance and the countless hours of interesting discussions that made it possible to achieve what is documented in this thesis!
I would like to express my deep appreciation to the whole team at CAPS for making a bumpy journey a lot smoother. Special thanks go to Claire, Gia- como, Oli and Yuan, who were there almost throughout my entire time at the lab. You guys played a very important role in this endeavour. Thanks a lot for your support and friendship! My sincere thanks also go to all the other former or current team members: Abel, Roberto, Luigi, Tiemo, Bayu, Audrey, Francesco, Matteo, Georg, Sergey, Michael, Edouard, Jovo and Anna. It was a pleasure to work with all of you and many of you have become great friends!
I am grateful for the skilled mechanical support of Peter Feusi, Daniel Trottmann and Ren´e Pl¨uss. You were always willing to help me, even when quick solu- tions were required at an inconvenient time like Friday afternoon. Thank you very much!
Many thanks also to Ulrich Doll, Dominik Ebi and Rolf Bombach for their fruitful collaboration. I very much enjoyed working with you!
Thanks also to Andreas Aeschimann, Rudolf Tresch and the team at TKE. I
enjoyed the friendly and uncomplicated collaboration during all those years!
I would like to thank Prof. Dr. Thomas R¨osgen and Phillipp B¨uhlmann from the Institute of Fluid Dynamics. Thanks for your advice and the lending of equipment!
I would also like to thank Prof. Dr. Thomas Sattelmayer and Dr. Andrea Giusti for being co-referees of my PhD thesis. Thank you for the interesting comments and questions!
Last but not least, I would like to thank my family and friends for their per- sistent encouragement and support during this adventure!
Z¨urich, September 2020 Markus Weilenmann
t Time [s]
T0 Acoustic period [s]
f Frequency [Hz], focal length [mm], f-number (aperture) f Downstream travelling Riemann invariant
g Upstream travelling Riemann invariant R Reflection coefficient [-]
Ra Purely acoustic reflection coefficient [-]
Re Reflection coefficient for entropy-wave-induced sound [-]
ω Angular frequency [rad/s]
M Mach number [-]
Mu Upstream Mach number [-]
St Strouhal number [-]
ϕ Phase [rad]
ϕg Phase at timetg[rad]
ϕe Phase difference between acoustics and entropy waves [rad]
tg Instant of generation of equivalence ratio fluctuation [s]
H Channel height [m]
Ht Nozzle throat height [m]
W Channel width [m]
cp Specific heat capacity constant pressure [J/(kg K)]
cv Specific heat capacity constant volume[J/(kg K)]
γ Ratio of specific heats [-]
T Temperature [K]
Tm Main flow temperature [K]
Tj Jet temperature [K]
∆Tjm Difference between jet and main flow temperatures [K]
q Heat release rate [W]
ρ Density [kg/m3]
p Pressure [Pa]
pt Pressure at nozzle throat [Pa]
φ Equivalence ratio [-]
Hea Relative entropic forcing strength [-]
˙
m Mass flow rate [kg/s]
˙
mm Mass flow rate main flow [kg/s]
˙
mj Mass flow rate of the jet [kg/s]
τc Time delay due to convective transport [s]
τa Time delay due to acoustic propagation [s]
G Gladstone-Dale constant [m3/kg]
R Gas constant [J/(kgK)]
I Intensity
IMie Mie scattering intensity n Refractive index [-]
δ BOS displacement [m]
Csetup BOS setup factor [-]
Deflection angle [rad]
C Histogram bin value W Histogram bin weight
fc Cut-off frequency [Hz]
K Relaxation coefficient (•) Mean quantity¯
(•) Zero-mean coherent fluctuations˜ (•) Stochastic fluctuations˘
(•)0 = ˜(•) + ˘(•) Zero-mean fluctuations h(•)i= ¯(•) + ˜(•) Phase-averaged quantity
(•) Laplace or Fourier transformationˆ (•)max Maximal quantity
(•)min Minimal quantity
List of Abbreviations
BOS Background oriented Schlieren PIV Particle image velocimetry
VG Vortex generator IRO Intensified relay optics FOV Field of view
ETF Entropy transfer function
PLIF Planar laser induced fluorescence SISO Single input/ single output MISO Multiple input/ single output
Abstract i
Zusammenfassung vi
List of symbols ix
Introduction 1
The energy challenge . . . 1
Thermoacoustics: A historical perspective . . . 4
Combustion noise . . . 6
The role of entropy waves in the termoacoustic feedback loop . . . 8
Entropy waves in thermoacoustics: current challenges . . . 12
Goals and Structure of the thesis . . . 13
1 Linear and nonlinear entropy-wave response of technically- premixed jet-flames-array and swirled flame to acoustic forc- ing 16 1.1 Introduction . . . 17
1.2 Experimental Setup . . . 19
1.2.1 Test rig . . . 19
1.2.2 OH PLIF-thermometry and chemiluminescence . . . 20
1.3 Methodology . . . 22
1.3.1 Quantification of acoustic velocity . . . 22
1.3.2 Thermometry and chemiluminescence . . . 23
1.4 Results and discussion . . . 24
1.4.1 Perfectly premixed vs technically premixed . . . 24
1.4.2 Nonlinearities at high forcing amplitude . . . 26
1.4.3 Entropy transfer function . . . 27
1.5 Conclusions . . . 29
2 On the dispersion of entropy waves in turbulent flows 30 2.1 Introduction . . . 31
Contents
2.2 Experimental setup and data processing method . . . 32
2.3 Numerical Setup . . . 36
2.4 Experimental and numerical results . . . 37
2.5 Conclusion . . . 42
3 Experiments on sound reflection and production by choked nozzle flows subject to acoustic and entropy waves 44 3.1 Introduction . . . 45
3.2 Setup . . . 48
3.3 Methodology . . . 50
3.3.1 Acoustic measurements . . . 51
3.3.2 BOS thermometry . . . 53
3.4 Results . . . 56
3.4.1 Choked nozzle subject to incident acoustic waves . . . . 56
3.4.2 Entropy-wave-induced backward acoustic waves . . . 58
3.4.3 BOS thermometry results . . . 62
3.4.4 Separation of acoustic and entropy-wave induced reflected- sound . . . 65
3.4.5 Decay of entropy waves . . . 67
3.5 Conclusion . . . 68
4 Background Oriented Schlieren of fuel jet flapping under ther- moacoustic oscillations in a sequential combustor 69 4.1 Introduction . . . 70
4.2 Experimental Setup . . . 72
4.2.1 BOS methodology . . . 72
4.2.2 Combustion test rig . . . 75
4.2.3 Diagnostics . . . 76
4.3 Results . . . 77
4.3.1 Acoustic mode . . . 77
4.3.2 Flapping fuel jet . . . 79
4.3.3 Velocity in mixing section . . . 80
4.3.4 Advection of equivalence ratio perturbations and au- toignition kernels . . . 85
4.4 Acknowledgement . . . 87
General Conclusions and Outlook 88
Bibliography 100
A BOS thermometry in reactive experiments 115 A.1 Setup and processing . . . 115 A.2 Results . . . 117 A.3 Conclusions and outlook . . . 118
B Overview of test conditions 119
List of Publications 120
Curriculum Vitae 121
The energy challenge
The slowing and limiting of man-made climate change is one of the hardest challenges that humanity will face in the coming decades. In Figure 1, the annual average temperature measured in Switzerland in recent years is com- pared to the mean of the period 1961-1990. A significant increase is evident over the last 50 years, which is a trend expected to continue [1], not only in Switzerland, but globally. The increase of mean temperature is detectable in all parts of the world, particularly so in the Arctic regions [2]. In Switzerland,
Deviation °C
Years above mean 1961-1990 Year Years below mean 1961-1990
Figure 1: Deviation of the mean annual temperature in Switzerland in com- parison with the reference period 1961-1990. ( cMeteo Schweiz)
the most visible effect of climate change is the continuous melting of glaciers.
Even within the last 10 years, an enormous loss of ice mass was observed at the Morteratsch glacier shown in Figure 2a. For many glaciers, annual shrinkage
can be easily perceived by the eye and this may lead to fresh water shortages in the future [3]. The warmer temperatures in mountainous areas also affects permafrost, which may lead to destabilization of slopes that can, among other consequences, cause rockfall on villages and traffic routes [4]. During the last few decades, the number of extreme weather phenomena drastically increased.
The scientific community generally agrees that these occurrences are triggered by the climatic change we are experiencing [5]. This includes floods, like for example the one experienced in Switzerland in 2005 (see Figure 2b), or also droughts, like in Vietnam 2016 (Figure 2c). The direct consequences are po- tential food and water scarcity [6, 7, 8], which can also lead to social tension [9]. More recent examples that took place in 2019 are the bush fires in Aus- tralia, which endangered the homes of millions of people, or hurricane Lorenzo which was the first storm of this category to almost reach continental Europe.
2011
2018
EU/ECHO a) b)
c)
d)
e)
Figure 2: Effects of climate change. a) Comparison of the Morteratsch Glacier in Switzerland between 2011 and 2018 ( cSammlung Gesellschaft f¨ur
¨
okologische Forschung / Sylvia Hamberger). b) Sarnen in 2005 during the se- vere flood in central europe ( cSRF). c) Exceptional drought in Vietnam 2016 ( cEU/ECHO) d) Severe bush fires in Australia 2019 ( cCNN). e) Hurricane Lorenzo in Autumn 2019 ( cNASA)
Partially due to these type of events, public awareness of climate change has substantially increased in recent time. The general consensus that the mat- ter has to be tackled urgently is also slowly leading to policies that intend to reduce the emission of greenhouse gases and therefore limit global warm-
ing. The ’Paris agreement’ [10], a treaty negotiated in 2015 in Paris during the 21st UNFCCC conference of parties (COP21) has, by early 2020, been ratified by 189 countries. Its intentions are to limit the increase of global av- erage temperature to well below 2 ◦C, compared to pre-industrial levels and to make an effort to keep the increase even below 1.5◦C. To achieve this goal, immediate action is required to limit and reduce the emission of greenhouse gases like CO2. Today, a majority of these emissions originate from the energy and transportation sectors through the combustion of hydrocarbon fuels [11].
It is therefore essential to improve our means of transportation and the way we generate electrical power, to achieve the goals of the ”Paris agreement”.
There are many challenges hidden in this task. In some cases, radical changes of technology are happening. For example the share of electrically powered vehicles has increased continuously in recent years. But in the growing field of air travel, there are no readily available technologies to substitute combustion based propulsion [12].
The share of solar, wind and hydro-power in the energy mix is being increased to replace coal power plants or nuclear power. Unfortunately, it is not feasible to switch entirely to renewable sources in the near future, because of several technological and political challenges. One of them being that power produc- tion of renewable sources is highly intermittent, which creates challenges in grid management [13, 14, 15]. Reliable, highly efficient, but also flexible sources of power are needed to ensure grid stability. Modern gas turbines are able to fulfill this need. Especially machines with recently designed staged combustion system which allow for very fast ramp time and high part-load efficiency [16, 17]. These wide operation windows also allow grid stabilization in power grids with a high concentration of renewables to be achieved. Compared to other combustion based sources, they emit almost no soot particles and very low NOx
[18]. Also, specific CO2emissions per MWh electric output of a combined cycle gas turbine power plant are less than half of plants relying on the combustion of coal [19]. Gas turbines are highly fuel flexible and can be operated with traditional gaseous fuels like natural gas, but also with bio-gas or liquid fu- els. Significant research is being undertaken regarding hydrogen combustion in gas turbines [20, 21, 22], which could play an important role in the future for a de-carbonized energy mix. Ammonia is also under investigation as a potential future fuel [23]. The projection of the energy mix in coming decades therefore predicts the significant usage of gas turbines [24]. The impact of improvements in combustion technology on global greenhouse gas emissions in the forthcoming decades are therefore crucial for the achievement of the ’Paris Agreement’ goals. This is the driving force behind combustion research today, especially in the field of gas turbine based systems for power production and
aerial transportation. The scientific progress made in recent decades has led to the development of very efficient systems, but many unanswered questions and technical challenges remain. Particularly in the case of aero-engines, which rely on the same basic operating principle as gas turbines, but are subject to more difficult boundary conditions, more challenging space and weight limitations, changing air inlet density and very strict safety requirements. The transition to efficient ultra-lean low NOx combustion is therefore very challenging [25].
The main reason is that such combustion architectures are highly prone to thermoacoustic instabilities, which is a phenomenon that can render safe and reliable operation impossible. This thesis focuses on the role and effect of so called entropy waves [26], which are essentially waves of fluctuating tempera- ture, originating from unsteady combustion, in the context of thermoacoustic instabilities [27]. This thesis aims to contribute to our understanding of the un- derlying mechanisms in order to allow more efficient, clean and safe operation of aero-engines and land-based gas turbines and therefore make a contribution to addressing the energy challenge.
Thermoacoustics: A historical perspective
In the year 1777, Byron Higgins made a fascinating discovery. He confined a small hydrogen flame in a glass tube and realized that the flame was producing sound [28]. He descibed the phenomenon as ”singing flames” and was the first to make a scientific observation of a thermoacoustic instability, where thermal energy is being converted to acoustic oscillations. In 1859, Dutch physicist Pieter Rijke placed a heated metal mesh in an open vertical tube at a quarter of the total height [29]. He observed that loud sound was produced as long as the tube stayed open, allowing convection to take place, which caused a fluc- tuating heat release. This setup is still well known today as ”Rijke-tube” and its behaviour is explained by an instability due to thermoacoustic coupling. In the 1870’s Georges Fr´ed´eric Eug`ene Kastner even saw artistic potential in this phenomenon and invented the pyrophone, an organ-like instrument depicted in Figure 3a. The theoretical explanation for the observed phenomena was first given by Lord Rayleigh [31, 32]. He formulated the Rayleigh-criterion:
If heat be communicated to, and abstracted from, a mass of air vibrating (for example) in a cylinder bounded by a piston, the effect produced will depend upon the phase of the vibration at which the transfer of heat takes place. If heat be given to the air at the moment of greatest condensation, or be taken from it at the moment of greatest rarefaction, the vibration will be encouraged.
On the other hand, if heat will be given at the moment of greatest rarefaction, or abstracted at the moment of greatest condensation, the vibration will be dis- couraged.
a) b) c)
d)
Figure 3: a) The pyrophone, developed in 1875 by Georges Fr´ed´eric Eug`ene Kastner [30] b) 5 F-1 rocket engines mounted on the booster stage of NASA’s Saturn V rocket ( cNASA). c) GT36 gas turbine ( cAnsaldo Energia) d) Trent-1000 turbofan aero-engine ( cRolls Royce)
This entertaining phenomena became a huge problem more than half a cen- tury later. If thermal power is increased, the potential strength of the acoustic pulsations also increases. Thermoacoustic instabilities can therefore reach dan- gerous levels that may lead to the destruction of thermal machines like rocket engines, jet engines or gas turbines. In the 1960’s, the United States were competing against the Soviet Union in space technology. To achieve the flight to the moon during the Apollo program of NASA, a very powerful rocket was needed, the Saturn V [33]. Its first stage was propelled by 5 F-1 engines (see Figure 3b), developed by the company Rocketdyne. The limited knowledge on thermoacoustics made it difficult to accurately predict the stability of a specific design. As a result, around 3200 full-scale test were performed of which roughly 2000 were dedicated to empirically finding an injector / baffle combination which could reliably avoid thermoacoustic instabilities [34]. This underlines the complexity of this phenomenon, which is very sensitive to op- erating conditions and will be explained in more detail in later sections.
For the development of today’s aircraft engines, gas turbines and also rocket engines, thermoacoustic instabilities are still a major challenge, but the ap- proach is quite different. Using numerical simulations and models, design
engineers aim at identifying and mitigating thermoacoustic problems already in the design phase [35]. Despite the immense progress in knowledge since the Apollo era, this remains challenging and is not always successful. The demand for ultra low emissions together with a flexible operating window, led to burner designs employing lean premixed flames [36, 37], which enable high part-load efficiency and operational flexibility with fast ramping, which is cru- cial for compensating the intermittent supply of renewable sources in power grids. Unfortunately, such modern burner architectures come at the cost of even higher propensity to exhibit thermoacoustic instabilities, which also hin- ders their desired incorporation in aero-engines.
The difficulty in the prediction of thermoacoustic instabilities lies in the com- plexity of the phenomenon in industrial applications. The complex architec- tures and flow fields result in unique behaviour from each combustor. For the development of tomorrow’s engines, it is therefore crucial to further improve predictive capabilities, which necessitates both numerical and experimental data to gain further insight into the mechanisms at play. Particularly in the case of feedback mechanisms related to entropy waves, the lack of experimental techniques to quantify and investigate the induced dynamics, has slowed the progress.
Combustion noise
The unsteady combustion process results in broadband sound generation known as combustion noise [38]. Combustion noise contributes significantly to the overall noise emitted by a machine to the surroundings, which is particularly relevant for aero-engines, since aircraft sometimes fly over densely populated areas and the main noise source in an aircraft are the engines. Aircraft noise can have severe impact on human health. It can cause community annoy- ance, disrupt sleep, adversely affect academic performance of children, and can increase the risk of cardiovascular disease in people living in the vicinity of airports [39]. With growing awareness, noise abatement is becoming an in- creasingly critical topic in the development of aero-engines. The total engine noise is caused by several noise sources explained in Figure 4, where combus- tion noise is an important contributor.
Combustion noise comprises two components: direct noise and indirect noise [40, 41, 38, 42]. Direct combustion noise is an immediate consequence of the unsteady heat release rate in unsteady flames. The fluctuating volumetric con- traction and expansion in the reactive region [43, 44], acts as a sound source.
In the case of broadband direct noise, the fluctuating heat release is usually induced by turbulence [45]. This type of noise also occurs in open flames, but
a) b)
Figure 4: a) Overview of noise sources in aero-engines by the example of a Rolls-Royce Trent 1000 ( cRolls-Royce plc) b) Typical noise distribution in aero-engines ( cSAFRAN Snecma)
in technical applications the combustion usually takes place in a confined en- vironment, known as a combustion chamber, which is schematically depicted in Figure 5 for a generic combustor typical for gas turbine applications. The fluctuating heat release of the flame q0 may also lead to the generation of in- homogeneities in gas temperature that are advected streamwise through the combustor with the bulk flow [47]. These hot or cold regions are also referred to as entropy spots. At the same time, the turbulent flame yields vortical structures [48] and pockets of altered gas composition (compositional spots) [49], which are also transported with the bulk flow. In the absence of axial mean velocity gradients, these convective disturbances are usually treated as passive variables that do not interact with the sound field field in the com- bustor. However, when and if, they reach the turbine inlet downstream of the combustor, a compensating pressure gradient is required to accelerate them [41, 50]. The resulting sound waves are known as indirect noise and travel both upstream toward the flame and downstream through the turbine stages to the exhaust. Indirect combustion noise comprises three components, named after their source, which areentropy noise[26], vorticity noise [51] andcompo- sitional noise [52, 49]. Under certain conditions, flames also yield strong, co- herent convective disturbances. The coherent entropy spots, or entropy waves, are essentially temperature waves, which are advected with the flow. These waves may greatly influence the sound field in the combustor, which is the driving force behind thermoacoustic instabilities.
Burner
Fuel injection holes Air
Air
Direct noise
Hot entropy spots
Vortex structures Compositional spots Turbine inlet
Indirect noise Combustion chamber
Cold entropy spots
Figure 5: Combustion noise sources in a generic technically premixed swirl stabilized combustor (adapted from [38, 46]). The wavelengths of the drawn noise components are not consistent with the drawn convective disturbances.
The role of entropy waves in the thermoacoustic feedback loop
Thermoacoustic instabilities arise due to constructive coupling between un- steady heat release rate and the acoustic field in a combustor [36], which is a result of the link between flame dynamics, combustor acoustics and burner aerodynamics in a confined combustion chamber, as illustrated in Figure 6.
The details of the interaction mechanisms are distinct for each combustion system and are strongly dependent on the combustor architecture and operat- ing window.
Figure 6: Thermoacoustic coupling in a combustion chamber. (˜•) denote coherent fluctuations, while (˘•) denote non-coherent fluctuations.
Rayleigh’s criterion states that the unsteady flame adds energy to the acoustic field if the unsteady heat release rate q0(x, t) is in phase with the acoustic pressurep0(x, t) at the location of the heat source [32]:
Z
p0(x, t)q0(x, t)dV >0 (1)
The fulfillment of this criteria is necessary for the occurence of thermoacoustic instabilities. The interaction mechanisms between the acoustic field and the flame, that encourage fluctuating heat release are various and include [53, 54, 55]:
• Equivalence ratio fluctuations – Inhomogeneities in the fuel to air ratio lead to locally hotter or colder flame temperatures [56]. This important mechanism is specifically pronounced in technically premixed combus- tors, where the fresh gas supply is affected by the acoustic field, while the gaseous fuel is injected at sonic velocity and is therefore decoupled from the acoustics.
• Flame surface wrinkling – Acoustic waves impinging on the flame lead to a change in flame surface area and therefore cause heat release rate fluctuations [57, 58].
• Flame-vortex interactions – Acoustic velocity fluctuations affect flame vortex roll-up, which may lead to unsteady heat release because of un- steady flame surface area. [59, 60].
• Flame-boundary interactions – The perturbed flame may interact with the walls of the combustion chamber which can lead to unsteady heat release [61].
In such confined combustion systems, there are also processes that lead to dissipation of acoustic energy and are therefore able to damp thermoacoustic oscillations. The main processes include [62, 55]:
• Vortex shedding – If the flow detaches near sharp edges, vortices are formed that allow the conversion of acoustic energy into vorticity [63, 64, 65].
• Transport – Some of the acoustic energy is transported to the exhaust and radiated to the environment [38].
• Boundary layers – Viscous and thermal dissipation of acoustic energy near the combustor walls [66].
• Dampers – Machines may be equipped with passive damping systems (e.g. Helmholtz dampers), that are capable of dissipating acoustic energy at a specific wavelength [67, 68].
Thermoacoustic instabilities may develop if the sources outweigh the damping.
The resulting, growing pulsations then usually experience non-linear satura- tion effects at an amplitude specific to the system, which is called a limit cycle.
In the previous section, the different mechanisms of combustion related sound production are discussed. There are direct and indirect sources, which all contribute to the overall sound field, which also heavily depends on the com- bustion chamber architecture, as sound waves are reflected at the walls and at the area contraction of the turbine inlet at the downstream end of the combus- tion chamber, but also upstream of the burner in the plenum. The Rayleigh criterion makes it clear that the timing plays an important role, to achieve constructive interaction between the resulting sound field and the flame. The phasing depends largely on the acoustic wavelength and thus on the oscillation frequency. Thermoacoustic instabilities therefore arise at distinct resonant fre- quencies, which exhibit modes, that are specific to the combustion system [69].
Some of these modes, normally at the lower end of the frequency spectrum, may be influenced by entropy wave feedback. Distinct entropy wave driven modes may also arise close to lean blow-off, or in aero-engines, where short
Burner
Fuel injection holes Air
Air
Coherent hot and cold entropy waves
Turbine inlet Combustion chamber
Figure 7: Coherent entropy waves in a generic technically premixed combustor.
The sound field in the combustion chamber is significantly affected by the sound resulting from accelerated entropy waves in the turbine inlet.
combustors lead to low residence times. The latter is often referred to as
”rumble” [70, 71].
These entropy waves are essentially coherent temperature fluctuations ˜T(x, t) resulting from the dynamics of the entire system, as shown in Figure 7. Recent work has shown that in real systems, in addition to the entropy waves origi- nating directly from the unsteady reaction zone, the mixing dynamics of the hot gas with cooling and dilution air flows significantly contributes to entropy wave production [47]. Unlike broadband indirect combustion noise, the accel- eration of coherent waves affects the sound field specifically at the frequency of the particular mode and is therefore most relevant in a thermoacoustic context. After their generation, entropy waves are convected towards the tur- bine where they are accelerated. The resulting coherent entropy-wave-induced sound waves may travel upstream, affect the flame dynamics and thus close the feedback loop. The experimental thermometry measurements presented in this thesis therefore focus on coherent temperature fluctuations.
Entropy waves in thermoacoustics: current chal- lenges
The mechanisms involved in triggering entropy wave driven thermoacoustic instabilities are still not fully revealed and understood [26, 35, 72]. A reli- able prediction remains challenging because of the high sensitivity of the phe- nomenon to the geometry of the combustion chamber, operating conditions and resulting flow field. This calls for further insight and the development of more advanced models that are capable of predicting entropy wave generation, transport and conversion, in technically relevant environments. Experimental data is scarce and mostly limited to academic configurations, which makes it difficult to properly validate numerical results and models in realistic environ- ments. The main reasons for this are:
• Lack of experimental thermometry methods. Low amplitude tempera- ture oscillations at frequencies of 50 to 1000 Hz are very challenging to measure, especially in the harsh environment of combusting flow. Tradi- tional methods like thermocouples and RTD’s fail because of their ther- mal inertia. Therefore, non-trivial optical methods need to be applied, which come with challenging technical difficulties and limitations.
• The separation ofdirect and indirect sound components is difficult. It is usually achieved in the time domain by taking advantage of the different propagation speeds of entropy and acoustic waves. However, this only works well with single pulses or at very low frequency. Therefore, to gain insight in more realistic configurations, new approaches are required.
To avoid these problems, experiments have been conducted mostly using sim- plified non-reactive test rigs, where entropy waves are synthetically generated, e.g. by using a heating grid operated in pulsed manner. The experiments have confirmed the existence of non-negligible contributions of entropy and compo- sitional waves to the sound field. The channel flows in these setups are rather simple, therefore experiments with more complex flows, that are closer to the situation in real engines, are needed.
In more complex flows, the production and decay rates of entropy waves are highly dependent on the flow field. Each burner type exhibits unique be- haviour. In order to predict and mitigate the impact of entropy waves on thermoacoustic instabilities, these processes need to be understood and char- acterized. The decay of the entropy wave amplitude is thereby of equal im- portance than the actual generation, as the sound field is only affected by entropy waves if they don’t disperse fully during advection through the com- bustion chamber. This has lead to conflicting conclusions on the relevance
of entropy wave feedback to the thermoacoustic stability in real engines. Ex- perimental data on decay rates in harsh environments are therefore needed to validate and develop suitable models. Simultaneously, a reliable way of quantifying the contribution of the transported and subsequently accelerated disturbances to the sound field is needed for the validation of entropy-to-sound conversion models at frequencies relevant for entropy modes. The situation in real engines is more complex compared to laboratory experiments and the dif- ferences, induced by the added complexity, may crucially impact the overall stability. Experimentally, the main challenge lies within the task of separating this entropy-wave-induced contribution from the rest of the sound field. Each chapter of this thesis deals with a different aspect related to entropy waves in thermoacoustic feedback. A literature review is included as part of the introduction to each chapter.
Goals and structure of the thesis
The main goal of this thesis is to contribute to the understanding and char- acterization of the processes related to entropy waves in the thermoacoustic feedback loop, by employing new experimental approaches in flows that are more complex than in previous experimental studies.
In the context of thermoacoustics, three main processes, involving entropy waves, need to be understood:
1) The generation of entropy waves by unsteady combustion, 2) the transport of entropy waves through the combustion chamber and 3) the conversion into sound waves at the combustor exit, which can then travel upstream to close the feedback loop. The acoustic network shown in Figure 8 graphically explains the role of each process. The area of contribution of the individual chapters is indicated by the color code.
Chapter 1: Generation of entropy waves
This chapter aims to bring insight to the generation of entropy waves by inves- tigating the characteristics of entropy wave production in two fundamentally different burners architectures, a 4×4 matrix configuration of a turbulent jet flame array and a classical single swirled flame. The burners were operated at a thermoacoustically stable operating condition, burning natural gas under technical premixing with air and were artificially forced using loudspeakers.
The acoustic input to the flame was captured by microphones and the co- herent temperature response downstream of the reacting region was measured using OH-LIF thermometry. The resulting single input/ single output (SISO) entropy transfer function (ETF) gives insight into the fundamental character- istics of these burner types regarding entropy wave production.
Chapter 2: Transport and decay of entropy waves
This chapter addresses the challenge of quantifying the decay of entropy waves during advection. Experiments were conducted in a high Reynolds number flow (Re > 1000000), subject to a pulsed hot jet-in-crossflow. The resulting flow field is a step towards more complex flows, as they exist in real engines. The en- tropy waves are measured during their advection using Background-Oriented Schlieren (BOS) thermometry, employing a novel post-processing approach.
The experimental data is complemented with Large-Eddy simulations (LES) and the characterized decay of the coherent entropy wave amplitude is com- pared to predictions from a state-of-the-art low order model.
Chapter 3: Conversion of entropy waves to sound waves
In chapter 3, the influence of accelerated coherent temperature fluctuations is investigated experimentally. A turbulent flow at 3 bars total pressure was subject to a pulsed jet-in-crossflow, where the jet was at higher, lower or the same temperature as the main flow. The generated hot or cold waves were accelerated in a nozzle after advection with the main flow. The isolation of the coherent entropy wave contribution to the sound field was achieved by em- ploying a method, based on the comparison of acoustic reflection coefficients to reference values taken in absence of entropy waves. The advected temper- ature fluctuations are quantified using background-oriented Schlieren (BOS) thermometry, employing the same approach developed in chapter 2. By per- forming a Mie scattering signal analysis of seeded experiments, the phasing between the acoustic field and the entropy waves was determined. The sub- sequent analysis allows for comparison with theoretical models and leads to conclusions for the observed differences.
Chapter 4: Thermoacoustic instability
This chapter deals with a self-sustained thermoacoustic instability in a generic sequential combustor. The mechanisms at play are studied using high-speed Background Oriented Schlieren (BOS), particle image velocimetry (PIV), OH chemiluminescence and pressure time traces. The discovered fuel jet flapping mechanism leads to compositional and entropy waves, which periodically trig- ger auto-ignition kernels. Although not quantified, entropy waves from the first stage and dilution air actively participate in the thermoacoustic instabil- ity in this chapter. The importance of entropy waves for a similar operating condition is shown in [73] by direct quantification from LES data.
Chapter 1:
Generation of Entropy waves
Chapter 2:
Decay of entropy waves during transport
Chapter 3:
Conversion of entropy waves to sound waves
Chapter 4:
Self-sustained
thermoacoustic instability in a sequential
combustion system
++
++
A B F
C T
Plenum A B
F
Transport
Transport Dispersion
1 2 3 4
Turbine inlet T Combustion
chamber C a)
b)
c)
Figure 8: a) Schematic drawing of a generic combustor with an area contraction at the exit. b) Thermacoustic network of the combustor drawn in a). τadenotes the acoustic time delay,τcthe convective one andαdthe decay of entropy waves by dispersion. c) Chapters of this thesis, covering the areas indicated by the color code.
Chapter 1
Linear and nonlinear
entropy-wave response of technically-premixed
jet-flames-array and swirled flame to acoustic forcing
The content of this chapter is published in:
Weilenmann, M., Doll, U., Bombach, R., Blond´e, A., Ebi, D., Xiong, Y.
and Noiray, N. ”Linear and nonlinear entropy-wave response of technically- premixed jet-flames-array and swirled flame to acoustic forcing”. In: Proceed- ings of the Combustion Instititute (2020)
The goal of this chapter is to gain insight into the generation of entropy waves in acoustically perturbed flames. Quantitative measurements of the produc- tion of entropy waves in a technically premixed turbulent combustor, subject to acoustic forcing, are presented. Entropy transfer function (ETF) relating acoustic input, obtained with microphones, to entropy wave output, obtained with OH-LIF thermometry at a distance of four flame heights from the burner outlet, were measured between 40 Hz and 90 Hz. These ETF were obtained using two burners of same length with technical premixing of air and natu- ral gas, operated at the same thermal power: a matrix burner producing an array of turbulent jet flames, and a burner producing a single swirled turbu- lent flame. It is found that the ETF of the matrix burner exhibits a low-pass behavior, with a gain ranging from approximately 0.7 at 40 Hz down to 0.25
at 90 Hz, while the gain of the swirled flame ETF was not exceeding 0.2. It is also demonstrated that entropy wave production with the matrix burner is highly nonlinear, with a dramatic drop of the ETF gain occurring beyond a certain level of acoustic forcing amplitude. These measurements can be used to derive predictive nonlinear models of thermoacoustic instabilities involving entropy wave feedback.
1.1 Introduction
Decreasing temperature fluctuations and composition inhomogeneities at the outlet of aeroengine combustion chambers is becoming more and more im- portant for the development of new combustor technologies. There are two reasons for this: First, these entropy and compositional waves contribute sig- nificantly to the noise pollution from modern aircraft [38, 26, 52] in the form of so-called indirect combustion noise, which originates from the acceleration of entropy or compositional fluctuations in the turbine stages [41, 49]. Second, in the race for development of low-pollutant-emissions combustors, the man- ufacturers face the problem of thermoacoustic instabilities which damage the mechanical parts, and which involve constructive interaction between acoustic waves, combustion dynamics and sometimes entropy waves [35].
Entropy waves are coherent temperature fluctuations produced by unsteady combustion which are advected through the combustion chamber. As shown in this paper, they can appear when an intense acoustic field in the combustor modulates the air mass flow through the burner, which leads to equivalence ratio oscillations that transform into a modulation of the hot products temper- ature. If dispersion of these waves by turbulent mixing is slow with respect to the residence time of the combustion chamber, acoustic pressure fluctuations are generated when the entropy waves are accelerated in the turbine. An un- stable feedback can therefore establish between combustion dynamics, entropy waves and acoustic field, and the development of new combustor technologies necessitates models to predict this type of thermoacoustic instabilities. Three major processes have to be understood: 1) The generation of entropy waves by unsteady combustion, 2) the decay of entropy waves due to shear dispersion, 3) the conversion of entropy waves to entropy noise.
Recently, studies have been performed on all 3 problems. Regarding the decay of entropy waves during advection, Eckstein et al. concluded in 2004 that in their experimental configuration, entropy waves are significantly attenuated during the advection process [74]. Recent studies aimed at deriving scaling laws for the spatial decay of the entropy waves, [75, 76, 77, 78] but most of them considered very simplified configurations, which calls for further investi-
1.1. Introduction
a) Matrix
b) Swirled
100mm
62mm
Air
Air NG
NG
Plenum Microphones
Fuel injection holes
Ø6mm
13.5mm
20 mm 34 mm
Figure 1.1: Sketches of the generic combustor equipped with a) the matrix burner, and with b) the axial-swirler burner. Pictures and main dimensions of the burners are shown. Stiff-injection of the natural gas (NG) in the air flow is done inside the burners (technical premixing).
gations with turbulent reacting flows in complex geometries that are relevant for practical systems. Studies that focus on the conversion of entropy waves to entropy noise [79, 80, 81] usually rely on synthetically generated temper- ature waves, using, for instance, electric heaters. Recently, several numerical and experimental studies dealing with the production of entropy waves, were conducted [82, 83, 84, 85, 86, 87]. However, there is still a compelling need for further experimental data, especially with turbulent technically premixed flames under technical premixing of fuel and air, for which temperature fluctu- ation measurements are very challenging. This study aims to contribute to fill this gap with quantitative temperature measurements by OH-PLIF thermom- etry [88]. The next section describes the combustor and the diagnostics used in this study. Section 3 provides information on the methods for postprocess- ing the acoustic and optical data. Section 1.4 presents the entropy transfer function (ETF) measured for an array of turbulent jet flames and for a single turbulent swirled flame at low and high forcing amplitudes.
1.2 Experimental Setup
1.2.1 Test rig
The experiments were conducted with a modular combustor operated at at- mospheric pressure (see Fig. 1.1). It consists of a plenum and a combustion chamber (62×62 mm2 cross section) terminated by an orifice plate to ensure thermo-acoustically stable operation for all the conditions considered in this work. Exchangeable side walls for optical access (quartz windows), or instru- mentation (ceramic or water-cooled aluminium plates with microphone ports, hot gas probe, igniter, etc) are mounted on a water-cooled aluminium frame made of a series of 250 mm long modules. The plenum features an acoustically stiff main air injection and sealed loudspeaker modules that are connected to the plenum with ceramic honeycombs. Natural gas is either directly injected with the air into the plenum in order to achieve perfectly premixed conditions at the burner inlet, or within the burner for technical premixing. Three mi- crophones (G.R.A.S. 46BD-S2) were placed between the loudspeakers module and the burner to measure forward and backward acoustic travelling waves.
Two burners of 10 cm length were considered: a matrix burner shown in Fig. 1.1a featuring 4×4 channels producing 16 turbulent jet flames, and a burner with an axial swirler producing a single swirled turbulent flame which is depicted in Fig. 1.1b. The matrix burner made of Hastelloy X was additively manufactured with selective laser melting to form a hollow cavity from which natural gas is injected via 16 small holes (one per channel) of diameter 0.8 mm in order to achieve acoustically stiff injection at nominal condition. The holes are axially staged (8 at 70 mm and 8 at 20 mm from the burner outlet) for im- proved thermoacoustic stability. The gas injection holes staging distance was defined in the same way as the vortex generators staging in [89]. The second burner features a central lance (20 mm) with an axial swirler, inserted in a hole (34 mm) in order to form an annular channel. The lance is terminated by an ogive shaped tip protruding into the combustion chamber. The 8 fuel injection holes are evenly spread around the circumference 20 mm downstream of the swirler at the same axial position. The nominal conditions of 30 kW thermal power and equivalence ratio φ = 0.76, were maintained throughout the study with the exception of the experiments investigating the effect of φ and the comparison between technically and perfectly premixed conditions.
For the latter, the thermal power was increased to 40 kW, because stable op- eration was not possible at 30 kW under perfectly premixed conditions.
The first module of the combustion chamber is equipped with 4 quartz win- dows, which provides optical access to the flame and identical heat losses on the four sides. A ceramic plate equipped with an electric igniter and three
1.2. Experimental Setup
HS camera
Excitation camera
Absorption camera
LIF Lasers Intensifier
Mirror
Dye cell Beam splitter Sheet optics
a) Top view
b) Side view
LIF camera 253 mm
62 mm 50 mm
Figure 1.2: Schematical arrangement of the optical components and cameras used for combined OH-PLIF/OH absorption thermometry and OH chemilu- minescence measurements of the flame.
quartz windows are mounted as side walls of the second module of the com- bustion chamber such that OH-PLIF in the hot products could be performed (see Fig.1.2).
1.2.2 OH PLIF-thermometry and chemiluminescence
Planar temperature distributions downstream of the combustion chamber for the matrix as well as the axial-swirler burner configuration were acquired by means of simultaneous OH-PLIF and OH laser absorption measurements [88].
By combining both methods, the local, instantaneous OH concentration can be determined from a single laser pulse. The laser light absorption measurement is used for both the absorption correction of the LIF images as well as the abso- lute calibration of the OH concentration. By assuming chemical equilibrium in the hot products and under lean conditions (φ <0.9), the resulting single-pulse OH concentration field is almost independent of the equivalence ratio [88] and, therefore, can unambiguously be converted into an instantaneous temperature distribution. Due to the rapid decrease of OH concentration with temperature, the method has a lower temperature detection limit of approximately 1300 K.
Taking into account the accuracy considerations summarized in [88], the mea- surement uncertainty of the method is below 5 % in the observed temperature range of 1600 K to 2100 K. Recently, the technique has been sucessfully applied to measure planar temperature distributions in high pressure combustion envi-
ronments related to aeroengine [90] and stationary gas turbine research [91]. In Figure 1.2, the optical setup of the combined OH-PLIF/OH absorption mea- surements is depicted. The LIF signal was excited using a frequency-doubled dye laser (Quantel TDL 90 pumped with YG981,∼20 mJ per pulse at 20 Hz, 0.08 cm-1 FWHM) tuned to the Q1(8) transition of the A-X (ν0 = 1, ν00 = 0) band near 283.55 nm. The laser beam was formed into a collimated light sheet by using a combination of cylindrical lenses (f = −25 mm, f = 400 mm).
To avoid saturation effects on the OH fluorescence signal, the resulting sheet width was twice as large as required and only the central 45 mm were used for the measurements. Likewise, the pulse energy was attenuated to about 50 % and a third cylindrical lens (f = 1000 mm) was used to position the beam waist outside the test section. The resulting sheet thickness was of the order of 1 mm throughout the imaged domain. Prior to the actual experimental campaign, LIF saturation was investigated in a matrix burner laboratory experiment (φ
= 0.8), using the same light sheet and signal collection optics. Images of the OH-PLIF signal were acquired at pulse energies of 100 %, 50 % and 25 % and OH concentrations as well as temperatures were derived. A saturation effect was observed for the unattenuated beam, whereas the curves for 50 % and 25
% converged to similar values. Due to the higher signal to noise ratio, the 50 % setting is used in the measurements. The OH-fluorescence emission from the horizontally oriented sheet was captured from above with an intensified CCD camera (PCO Dicam Pro), equipped with an UV lens (Cerco, 100 mm f/2.8) and a bandpass interference filter (Chroma, transmission > 70 % at 310 nm, FWHM 10 nm). The resulting camera field-of-view was 62×62 mm2 at a spa- tial resolution of 120µm per pixel. To perform the simultaneous measurement of OH laser light absorption, part of the incoming and outgoing light sheet was deflected by beam splitters onto two cuvettes filled with dye solution. The re- sulting intensity profiles before and behind the test section were recorded with two CCD cameras of the same type (PCO Pixelfly equipped with Schneider- Kreuznach 25mm f/.95 lens).
The cameras were synchronized at 10 Hz to capture every second light pulse.
To resolve the fast coherent oscillations of hot gas temperature in response to the acoustic forcing imposed with the loudspeaker (40 – 90 Hz), phase-locked averaging was performed by synchronizing the onset of the forcing sinusoidal signal with the 10 Hz acquisition rate of the PLIF system. For excitation frequencies being integer multiples of 10 Hz, phase-sweeps of the loudspeaker forcing could then be performed by merely modifying the phase of their input signal. While this approach results in very accurate phase-locking, it is also very time consuming, because the data has to be measured, stored and pro- cessed at each phase angle separately. Convergence of the spatially integrated
1.3. Methodology
phase-locked averaged temperature fields within one percent was reached after 100 cycles, by which we mean that the addition of the 101st cycle changed the result by less than 1 %.
High speed OH* chemiluminescence of the flame was recorded using a high- speed camera (Photron HSSX) and a highspeed intensifier (Lavision highspeed IRO), as depicted in Fig. 1.2 a). A UV lens (Cerco, 100 mm f/2.8) was used together with an optical narrow bandwidth (10 nm) filter centered around 308 nm. Due to space limitations, the camera setup was mounted parallel to the test rig and a UV mirror was used to observe the flame.
1.3 Methodology
The goal of this paper is to investigate the entropy wave generation by an ar- ray of 4×4 turbulent jet flames and by a single turbulent swirled flame, which are operated at the same thermal power and equivalence ratio with technical premixing of air and NG, when they are submitted to an acoustic forcing. To that end, we consider the single input/single output (SISO) Entropy Transfer Function (ETF), which relates the coherent velocity fluctuations ˜u just up- stream of the burners to the coherent averaged temperature fluctuations ˜T at a distance of 4 flame heights from the burners outlet, where ¯uand ¯T represent the mean absolute axial velocity and temperature:
ET F = T /˜ T¯
˜
u/¯u (1.1)
These positions for the measurements of mean and oscillating velocity and temperature were chosen for the following reasons: First, the length of the burners Lb = 0.1 m is small compared to the shortest acoustic wavelength considered in this work (λmin = c/fmax = 340/120 = 2.8 m), and one can expect a quasi-incompressible motion of the air in the compact burner where
˜
u/¯ujust upstream of the burner inlet is the same as ˜u/¯u at the axial position of the NG holes. Second, the quantitative measurement of temperature using LIF thermometry requires the mixture to be at equilibrium, which is not the case in the direct vicinity of the flame, and which motivated our decision to measure the temperature field significantly downstream of the flame, in the second combustion chamber module.
1.3.1 Quantification of acoustic velocity
The acoustic velocity at the inlet of the matrix is reconstructed from the pressure signals of the three microphones placed upstream of the burner using the Multi-Microphone Method. With this method, the acoustic pressure is
min max
1600 2100
OH Chemiluminescence[counts] Temperature [K]
a) b) c) d) e) f) g) h)
Matrix: f = 40 Hz = 12.6 % Swirled: f = 40 Hz = 38.4%
Figure 1.3: OH* Chemiluminescence (OH CHEM) and LIF thermometry at a forcing frequency of 40Hz for the configurations with matrix burner (a-d) and for the burner with swirler (e-h).. a,e) Instantaneous OH CHEM ; b,f) instantaneous LIF results; c,g) phase-averaged OH CHEM d,h) phase-averaged LIF results. The green rectangles show the ROI at 40 Hz.
expressed as the solution of the one-dimensional wave equation with mean flow:
ˆ
p(ω, x) =f(ω)e−iωc1+Mx +g(ω)eiωc1−Mx (1.2) where ˆ· denotes the Laplace transformation, and where the Riemann invari- antsf andgare determined from an overdetermined system using least-square inversion [92]. The acoustic velocity ˆu(ω) at the burner inlet is then recon- structed as :
ˆ
u(ω) = ρc(f(ω)−g(ω)) (1.3) The mean flow velocity ¯u is determined from the air mass flow measurement and the plenum cross section area (62×62−π×102/4 = 3765 mm2) using a Bronkhorst mass flow meter.
1.3.2 Thermometry and chemiluminescence
The high-speed OH chemiluminescence recordings of both flame types showed significant fluctuation in intensity and flame length. Using the purely sinu- soidal loudspeaker excitation signal as a reference, a phase angle was assigned to each recorded frame. The high frame-rate allows the observation of in- staneous cycles, as shown in Figure 1.3a (matrix) and 1.3e (swirled) for an excitation frequency of 40 Hz for both the matrix and swirled type flame. By
![Figure 3: a) The pyrophone, developed in 1875 by Georges Fr´ ed´ eric Eug` ene Kastner [30] b) 5 F-1 rocket engines mounted on the booster stage of NASA’s Saturn V rocket ( c NASA)](https://thumb-eu.123doks.com/thumbv2/1library_info/5357898.1683318/22.892.174.711.218.430/figure-pyrophone-developed-georges-kastner-engines-mounted-booster.webp)
