You are on page 1of 364

ACOUSTIC PERFORMANCE OF BIAS FLOW

LINERS IN GAS TURBINE COMBUSTORS

claus lahiri

D I S S E R TAT I O N

ACOUSTIC PERFORMANCE OF BIAS FLOW
LINERS IN GAS TURBINE COMBUSTORS

vorgelegt von
Dipl.-Ing. Claus Lahiri
geb. in Bergisch Gladbach

von der Fakultät V - Verkehrs- und Maschinensysteme
der Technischen Universität Berlin
zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften
– Dr.-Ing. –

genehmigte Dissertation

Berlin 2014

Promotionsausschuss:
Vorsitzender: Prof. Dr.-Ing. Dieter Peitsch
Gutachter: Prof. Dr. rer. nat. Lars Enghardt
Gutachter: Prof. Dr. Tech. Hans Bodén
Tag der wissenschaftlichen Aussprache: 25. November 2014

© claus lahiri, 2014

ABSTRACT

Thermoacoustic instabilities prevent the implementation of mod-
ern combustion concepts in gas turbines, which are essential for
higher efficiency and lower emissions. Bias flow liners are able to
suppress these instabilities by increasing the acoustic losses of the
system. However, it is an open question under which conditions
their full potential can be retrieved.
This thesis collects the available information concerning the
acoustic properties of bias flow liners and puts it into perspec-
tive for the application in a gas turbine combustor. The review
includes a rigorous assessment of the existing models. The mod-
els and previous findings are evaluated by comparing them to the
results of a comprehensive experimental study regarding the rel-
evant acoustic, geometric, thermodynamic, and flow parameters.
This includes for the first time the influence of pressure and tem-
perature.
The results reveal that there is a resonance dominated regime
at low bias flow Mach numbers with rather complex parameter
dependencies and a bias flow dominated regime which is mainly
dependent on three parameters only: the bias flow, the porosity,
and a resonance parameter.

Z U S A M M E N FA S S U N G

Der Einsatz moderner Verbrennungskonzepte zur Effizienzsteige-
rung und Schadstoffreduktion bei Gasturbinen wird oft durch
thermo-akustische Instabilitäten verhindert. Durchströmte Brenn-
kammerliner können die akustische Dämpfung des Systems er-
höhen und die Instabilitäten unterdrücken. Eine wesentliche Fra-
ge ist, bei welchen Parametereinstellungen eine optimale Dämp-
fungswirkung erzielt werden kann.
Diese Arbeit liefert eine umfassende Übersicht der bisherigen
Forschung. In einer experimentellen Studie werden akustische,
geometrische, thermodynamische und strömungsmechanische Pa-
rameter untersucht. Zum ersten Mal wird hier auch der Einfluss
von Druck und Temperatur durch eine Messung abgebildet. Diese
weitreichende Datenbasis ermöglicht eine detaillierte Bewertung
der einzelnen Parameter und der Qualität der Modelle.
Die Ergebnisse zeigen einen Betriebsbereich bei langsamen Ge-
schwindigkeiten der von Resonanzen dominiert ist und kompli-
zierte Parameterabhängigkeiten aufweist. Bei höheren Geschwin-
digkeiten dominiert die Durchströmung und die Schallabsorption
ist im Wesentlichen von nur drei Parametern abhängig: Porosität,
Durchströmungsgeschwindigkeit und einem Resonanzparameter.

.

. . . . .2 Outline . . . . .1 Temperature 26 – 2. . .2.4. . . . . . . . . .3 Thermodynamic Parameters . . . . . . . . . . . .1 Motivation . . . 8 2 bias flow liners 11 2. . . .2 Amplitude / Sound Pressure Level 28 2. . 11 2. . . . . . . . . . . .3 Howe Rayleigh Conductivity Model . . .CONTENTS 1 introduction 1 1. 57 4.1 Rayleigh Conductivity . . . . . . . . . . . .2. .3 Cavity Geometry 24 2. . .2 Perforation Geometry 20 – 2. . . . . .4 Jing Model (Modified Howe Model) . . . . . . . . . . . . . 16 2. . 52 4 modeling of bias flow liners 57 4. . . . . . .4. .2 Geometry Parameters . .2. 27 2. . . . . . .4 Chronological Overview . .2 Acoustic Impedance . . . .1 Parameter Overview 15 2. . . . . . . . . . . 59 4. . .1 Impedance Modeling of Perforations 61 – 4. . . . . . . . . . . . . . . . . . . .3. . . . . . .1 Grazing Flow 32 – 2. . . . .2 Pressure 27 2. . . . . . . . . . . . 71 4. . .5. .2. . . . . . . . . . . . .3 Bias Flow Impedance 67 4. .1. . . . . . . . .5 Luong Model (Simplified Howe Model) . . .2. .1 Combustor Liner . . . . . . . . .1 Frequency 27 – 2. . . .1 Bias Flow as a Concept of Impedance Control . . . . . . .2 Sound Absorption due to Vorticity Shedding . . 7 1. . . . . . . .4 Acoustic Parameters .3 Recent Developments .3. 26 2. . . .5. .2 Bias Flow 33 3 literature on bias flow liners 43 3.5 Flow Parameters . .1 Orifice Geometry 17 – 2. .2. . . . . . 32 2. . . 68 4. . . . . 45 3. . . . . . . . . . . .2 Grazing Flow Impedance 67 – 4. 73 . 43 3. . . . . . . . 49 3. .

.2 Evanescent Modes 105 – 5. . . . . . 76 4. . .3 Compared to the Absorption Coefficient 156 6. . . . . . .1 Spectral Analysis . . . . . . . . . . 97 5. .2.1 Cut-On Frequency 104 – 5. . .8 Bellucci Impedance Model . . . . .2 Dissipation Error 156 – 6. .3 Losses Within the Fluid 130 6 experimental method & analysis 139 6. . . . . . . . . 159 6.9. 79 4. .5. . .5 Accuracy 170 6. . . . . .4. . .6. . 75 4. 144 6.5. . . . .1 Average Dissipation Coefficient 155 – 6. . .3 End-Reflections 165 – 6. .4 Estimation of the Humidity 189 – 6.1 Losses at the Wall 114 – 5. . .4 Attenuation of Sound . .x contents 4. . .5 Attenuation at Elevated Pressure and Temperature 190 – 6. .1.2 Transfer Matrix Method vs.1 Welch Method 140 – 6. . . .1 Acoustic Wave Equation . . .4. . . .6. .3 Plane Waves . . . . . . . . . . . .6.1 Setup & Instrumentation 159 – 6.10. 151 6. . . . . .6. . . Eldredge & Dowling Method 93 – 4. . .10. . . . . . . .2 Air Supply 177 – 6.4 Dissipation Coefficient . . . .2 Eldredge & Dowling Method 85 4. . . .1. . . .4. . . . . . . . . . . . . . . .1 Impedance Models 89 – 4. . .1 Setup & Instrumentation 172 – 6.3 Reflection and Transmission Coefficients .7 Betts Impedance Model . . . . . . . . .6. .5 Duct Acoustic Test Rig . . .5. . . . . . . . .2 Microphone Calibra- tion 162 – 6. . . . . . 111 5. . . . . . . . . Luong 95 5 duct acoustics 97 5. . . . . . . . . 89 4.10.5. . . .5.2. . . . .4. . . . . . . .2 Rejection of Flow Noise 143 6. . . 171 6. . . . . .2 Losses due to Turbulent Flow 126 – 5. . . . .3 Resonances in an Annular Cavity 108 5. . . . .4. .2. 154 6.6 Bauer Impedance Model . . .4 Influence of Evanes- cent Modes 165 – 6. . . .9 Application to a Cylindrical Geometry .1 Transfer Matrix Method 80 – 4.10 Some Comparisons . . 73 4. .2 Plane Wave Decomposition . .3 Howe vs. . .4.6. . . . . . . . . .2 Three-Dimensional Waves . .9. . . .6 Accuracy 193 . 109 5. . . . .6 Hot Acoustic Test Rig . . 140 6. . . . . . . 101 5. . . . . . . . .3 Microphone Probes 183 – 6. .

. . . . . . . . . . . . . . . . . . . . . . .16 Pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247 7. .6 Wall Thickness . . . . 260 7. . . . . . . . . . . . . 293 B. . . . 271 8 concluding remarks 275 A properties of air 283 A. . . . . . . . .3 Grazing Flow . . . contents xi 7 parameter study 195 7. . . . . . 219 7. . . . .2 Conversion of Transfer Matrix into Scattering Matrix 294 bibliography 297 . . . 228 7. . . . . . . . .9 Orifice Angle . . . . . . .7 Orifice Cross-Section Shape . . . . .4 Simultaneous Grazing & Bias Flow . . . . . . . . . . . . . 204 7.17 Strouhal Number . . . .10 Double-Skin Configuration . . . . . 195 7. . . 284 A.5 Porosity . . . . . . . . . . . . . . 283 A. . . . . . . . . . . .13 Perforation Placement . . . . . .1 Equations of Ideal Gas Properties . 264 7. 231 7. . . . . . . . . . . . . . . 249 7. .2 Ideal Gas vs. . . . . 255 7. . . . 236 7. . . . . . . . . 269 7. . . . . . . . . . . . . . . . . . . 253 7. .11 Cavity Volume . . . . . . . . . . . . . . . . 285 B some mathematics 293 B. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Orifice Edge Geometry . . . . . . . . . . . . 241 7. .15 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 Rayleigh Conductivity and Impedance . . . . . . . . . . . . . . . . . . . 213 7. . . . . . . . 222 7. . . . . . . . . . . . .12 Partitioned Cavity . .1 Sound Pressure Level . . . . . Real Gas . . . . . . .14 Perforation Pattern . . . . . . . . . . . . . . . . .2 Bias Flow . . . . . . . . . . . . . .3 Tables of Real Gas Properties . . . .

.

mn Cut-on frequency of mode m:n .LIST OF SYMBOLS latin symbols A Amplitude A Cross-section area Aopen Liner total open area Arel Relative error a Attenuation rate B̂ Complex stagnation enthalpy amplitude B0 Fluctuating stagnation enthalpy C Acoustic compliance C Circumference Cc Contraction coefficient Cd Discharge coefficient Cr Resistance coefficient Cv Velocity coefficient c Speed of sound cp Specific heat capacity at constant pressure cph Phase velocity cv Specific heat capacity at constant volume D Duct diameter D Energy dissipation coefficient Dc Cavity diameter Dh Hydraulic diameter d Orifice diameter dc Cavity depth e Exponential function err Fit error err Average fit error over all microphones f Darcy friction factor f Frequency fc.

i Vibrational relaxation frequency of constituent i Gxx One-sided auto-spectral density Gxy One-sided cross-spectral density He Helmholtz number Hm Hankel function of order m h Humidity as molar concentration of water vapor Im{z} Imaginary part of complex number z Im Modified Bessel function of the first kind of order m i Imaginary unit Jm Bessel function of the first kind of order m jmn Eigenvalue of Jm Km Modified Bessel function of the second kind of order m KR Rayleigh conductivity k Wave number k± Convective wave number k0 Free field wave number kr.mn Radial wave number of mode m:n ks Stokes wave number k0s Effective Stokes wave number kx.mn Axial wave number of mode m:n L Length L Perforation length Lc Cavity length Leff Effective perforation length l Orifice length l0 Length end correction leff Effective orifice length M Acoustic inertance M Mach number Mb Bias flow Mach number Mb Mean orifice bias flow Mach number Mg Grazing flow Mach number m Circumferential mode order ṁ Mass flow rate .xiv list of symbols fr Resonance frequency fr.

list of symbols xv ṁb Bias flow mass flow rate ṁg Grazing flow mass flow rate n Radial mode order P Acoustic power P Acoustic energy flux P Perimeter p Pressure p0 Mean pressure p∗ Critical pressure p0 Acoustic pressure p̂ Complex pressure amplitude pc Pressure at the critical point of water p̂mn Complex amplitude of mode m:n pref Reference pressure pws Saturation water vapor pressure Pr Prandtl number Prt Turbulence Prandtl number Q Resonance parameter Q0 Heat release fluctuation Qmn Hard-wall Eigenvalue of Ym q Acoustic volume velocity q̂ Complex volume velocity amplitude R Acoustic resistance R Duct radius R Energy reflection coefficient R Specific gas constant R Radius Rc Cavity radius Re{z} Real part of complex number z Re Reynolds number Rec Critical Reynolds number for pipe flow RH Relative humidity r Orifice radius r Radial coordinate r Specific acoustic resistance .

see Tab. A.xvi list of symbols r Amplitude reflection coefficient re Amplitude end-reflection coefficient rh Hydraulic radius S Scattering matrix Sxy Scattering matrix element Sh Shear number St Strouhal number s Entropy s Spacing between microphones s Spacing between orifices s0 Entropy fluctuations sθ Spacing between orifices in circumferential direction sx Spacing between orifices in axial direction T Energy transmission coefficient T0 Fluctuating temperature T Temperature T Transfer matrix Ta Constant. A.1 Tb Constant. see Tab.1 Tc Temperature at the critical point of water Td Temperature at dew point Tref Reference temperature Ts Sutherland constant Txy Transfer matrix element t Time t Wall thickness t Amplitude transmission coefficient TL Transmission loss U Mean flow velocity Ub Bias flow velocity Ub Mean orifice bias flow velocity Ug Grazing flow velocity V Cavity volume v Velocity v∗ Friction velocity .

list of symbols xvii v0 Mean velocity v0 Acoustic particle velocity v̂ Complex velocity amplitude v Acoustic mass velocity v̂ Complex mass velocity amplitude w Window function X Acoustic reactance Xi Mole fraction of constituent i x Specific acoustic reactance x Axial coordinate Y Wall shear layer admittance Ym Bessel function of the second kind of order m y Transverse coordinate Z Acoustic impedance based on the mass velocity Z Acoustic impedance Z0 Characteristic acoustic impedance of a lumped element Zrot Rotational collision number z Specific acoustic impedance z0 Characteristic acoustic impedance greek symbols α Orifice pitch angle α Absorption coefficient α Attenuation coefficient α± Convective attenuation coefficient α Average convective attenuation coefficient β Coefficient of thermal expansion χ Normalized specific acoustic reactance χ Thermal diffusivity δ Imaginary part of the Rayleigh conductivity δ Boundary layer thickness δτ Thickness of the viscous sublayer of turbulent flow δν Thickness of the viscous acoustic boundary layer δχ Thickness of the thermal acoustic boundary layer δij Kronecker delta .

xviii list of symbols ∆cph Change of phase velocity ∆P Bias flow pressure drop ∆Pt Bias flow total pressure drop in a double-skin configuration η Compliance η Hub-to-tip ratio in an annular duct Γ Propagation constant γxy Coherence γ Heat capacity ratio γ Real part of the Rayleigh conductivity κ Kármán constant κ Thermal conductivity λ Ratio of bias flow to grazing flow λ Second viscosity coefficient λ Wavelength µ Dynamic viscosity µB Bulk viscosity (also called volume viscosity) ν Kinematic viscosity ν0 Effective kinematic viscosity including thermal conductivity losses π Mathematical constant Φ Acoustic losses ϕ Phase angle φ Viscous dissipation rate θ Circumferential coordinate θ Normalized specific acoustic resistance θi Characteristic vibrational temperature of constituent i ρ Density ρ0 Fluctuating density ρ0 Mean density σ Perforation porosity τij Viscous stress tensor ω Angular frequency ψ Fok function ζ Normalized specific acoustic impedance .

und Raumfahrt DUCT-C Duct Acoustic Test Rig . e.Circular Cross-Section EDM Eldredge and Dowling Method HAT Hot Acoustic Test Rig ISA International Standard Atmosphere ICAO International Civil Aviation Organization ISO International Organization for Standardization SDOF Single-Degree-of-Freedom SNR Signal-to-noise ratio SPL Sound Pressure Level TMM Transfer Matrix Method . viscothermal. against flow direction subscripts 1 Section 1 2 Section 2 a Excitation A ac regarding the acoustic mode b Excitation B cl regarding classical. i. in flow direction − in negative x-direction. losses within the fluid cr regarding classical and rotational relaxation losses within the fluid ent regarding the entropy mode fluid regarding losses within the fluid rot regarding rotational relaxation losses within the fluid total regarding total losses turb regarding viscothermal and turbulence losses at the wall vib regarding vibrational relaxation losses within the fluid vor regarding the vorticity mode wall regarding viscothermal losses at the wall acronyms DDOF Double-Degree-of-Freedom (also: 2DOF) DLR Deutsches Zentrum für Luft. list of symbols xix superscripts + in positive x-direction.

.

2013 [231]) The energy industry relies heavily on gas turbines for electric- ity production.8% 6 115 TW h 22 126 TW h Figure 1.7% Natural gas 12.3% Nuclear Nuclear 3. 1 INTRODUCTION Gas turbines convert the chemical energy inherent in a gaseous or liquid fuel into mechanical energy.9% 4. (Key World Energy Statistics © OECD/IEA. Depending on their applica- tion they are designed to deliver shaft power or thrust.6% 21.3% 41.5% 38. . 1 Mtoe = 11630 GWh [231]. Fig- 1 Million tonnes of oil equivalent.2% Oil Natural gas Oil 24. respectively.1: 1973 and 2011 fuel shares of electricity generation. 1973 2011 Hydro Other Other 0.0% 15.8% 4.6% Coal/ peat Hydro Coal/ peat 21. for exam- ple to generate electricity in a power plant or as aircraft propul- sion. The global electricity consumption has more than tripled from 439 Mtoe1 in 1973 to 1582 Mtoe in 2011 [231].3% 11.

As of December 2013 there are a total of 44836 jet engines in service on active commercial aircraft in operation with airlines [159]. 4 Compared to traditional reciprocating engines of the same power rating. the power supply can be balanced flexibly at peak times or when there is a temporary shortage of wind or solar power. fossil fuels (coal. The gas turbine is essential to support this tran- sition. with predictions of 31 % rise in passenger demand by 2017 compared to 2012 [230]. gas turbines are one of the most widely-used power generating machines with industry leading efficiencies of around 40 % in simple cycle and up to 60 % in combined cycle2 operation [331. oil.2 introduction ure 1. With a collective share of 68 % in 2011. and reliable operation have quickly made the gas turbine the engine of choice throughout the aviation industry. Many countries are working on the Energiewende3 . for example. compact size4 . . While gas turbines can run on a variety of fuels. The first flight of a gas turbine powered aircraft took place on 27th of August 1939 in Rostock. 439]. the German term Energiewende has been adopted here. Due to its short start-up time. 2 A combined cycle gas turbine (CCGT) power plant reuses the hot exhaust gas of the gas turbine to additionally drive a steam turbine. Germany [80]. natural gas is the most common for power generation. The contribution of natural gas has nearly doubled from 1973 to 2011.1 illustrates the fuel shares involved in the electricity gener- ation. Gas turbines will remain a dominant technology for power generation and business analysts predict a stable growth for the gas turbine industry [163]. 3 Due to the lack of an appropriate equivalent in the English language (see the discussion in [499]). Light weight4 . This number is expected to grow quickly. and natural gas) remain the largest supplier of primary energy for electricity generation. compared to other means of power generation. that is in- creasing the use of renewable energy as an alternative to fossil and nuclear fuels. Today. indicating the growing dominance of gas turbines in that field.

The Rolls-Royce Trent 900 is one of the most powerful jet engines in operation. 2012. (Courtesy of Rolls-Royce plc © Rolls-Royce plc. The increased environmental awareness has made the reduction of pollutant emissions from the combustion process one of the key challenges for modern gas turbines [292]. a modern high-bypass turbofan engine powering the Airbus A380. introduction 3 Figure 1. The global5 aircraft en- gine emission standards are set by the Committee on Aviation En- vironmental Protection (CAEP) of the International Civil Aviation Organization (ICAO) and are published in the current edition of 5 The emissions of power plant gas turbines are regulated regionally.2.) A modern high-bypass turbofan engine is shown in Figure 1.2: Rolls-Royce Trent 900. .95 m and a length of 5.478 m [151]. so that a wide diversity exists. An overview is given in [306]. delivering a maximum thrust of 334–374 kN with a fan diameter of 2.

The lean combustion increases the air-fuel-ratio beyond the stoichiometric mixture9 . High amplitude instabilities can result in a flame blow-off or in the immediate and fatal damage of components. This problem is addressed by modern combustion concepts.4 introduction Annex 16. is taking part in the combustion. As a result. The instabilities lead to excessive wearing or even cracking of exposed components within or adjacent to the combustor. the reduction of NOx is a major concern throughout the gas turbine industry. which operate in the lean regime. e. the flame temperature is lowered and NOx production is reduced. respectively. so that the levels have been reduced continuously. 7 These values are relative to the capabilities of a typical new aircraft in the year 2000. in line with the optimization for low fuel consumption. which adversely affects the efficiency and the CO2 production. The amount of CO2 in the exhaust gases is directly related to the combustion effi- ciency [292]. The prime factor in reducing NOx is to lower the flame temperature8 . the implementation of this concept is often prevented by its tendency to promote combustion instabilities. po- 6 This collectively includes nitrogen monoxide NO and nitrogen dioxide NO2 . Volume 2. more air. Thus. 8 Two additional factors are a uniform temperature distribution and a short residence time. relative to the fuel. 9 A stoichiometric mixture contains sufficient oxygen for a complete combus- tion of the available fuel. However. including both aviation and power. Particularly rewarding is the lean pre-mixed/pre-vaporized com- bustion (LPP) [292]. . Combustion instabilities are pressure pulsations resulting from a thermoacous- tic feedback between the heat release of the flame and acoustic pressure oscillations. i. The Advisory Council for Aviation Research and Innovation in Europe (ACARE) has formulated future goals for the emissions of carbon dioxide CO2 and nitrogen oxide6 NOx in their Vision 2020 [4] and Flightpath 2050 [5] reports. It should be aimed for a reduction in CO2 of 50 % and 75 % and NOx of 80 % and 90 % until 2020 and 20507 .

This feedback loop in itself does not necessarily lead to instability. and domestic or indus- trial furnaces [375]. instability does occur when [376]11 Z p0 (t) Q0 (t) dt > Φ. g. Commonly. . thermoacoustic instabilities occur at one dominant frequency. so that Φ = 0. e. p0 is the acoustic pressure fluctuation. afterburners of turbojet engines [67. 190. Q0 is the heat release fluctuation. Lieuwen and Yang [305] give an overview of the recent situation and mitigation strategies in the gas turbine industry. 11 The original derivation by Putnam and Dennis [376] neglects the acoustic losses. combustion instabilities are not limited to gas turbines.1) T where T is the period of one oscillation. and Φ describes the 10 Named after John William Strutt. The process leading up to an instability is not yet fully under- stood. Lord Rayleigh (1842-1919. The gen- eral mechanism can be described as follows: The oscillating heat release of the flame produces sound. 390]). the combustion becomes unstable only when the heat release fluctua- tion and the acoustic pressure oscillations are in phase ([388. Thus. The expression including the acoustic losses can be found in [238. The maintenance costs that can be directly associated with damages due to com- bustion instabilities exceed $1 billion annually [305]. liquid and solid propellant rocket engines [102. This necessary condition is referred to as Rayleigh criterion10 . 102]. which is reflected at the com- bustor boundary. 497]. feeding back to the heat release oscillation. ramjet engines [102. In mathematical terms. 440]. English physi- cist). Low amplitude instabilities require an increased downtime of gas turbines for inspections and repairs if necessary. for example. 506]. introduction 5 tentially releasing detached pieces into the turbine downstream. but are a widespread problem in combustion systems. it is not possible to predict the occurrence of insta- bility or avoid it in the first place. which is mainly dependent on the combustor geometry and the operating condition. Furthermore. (1.

Traditionally. Increasing the acoustic losses of the system (passive control). a broadband damp- ing characteristic would be preferred. Vol. Active control measures mostly act on the modulation of the fuel flow. p. Fuel-Oil Journal. . at low fre- quencies the resonators grow to a considerable size and are cer- tainly difficult to integrate into the engine. drill two holes!” One particular feature of the holes in a combustor liner is. Equation (1. Putnam [375] humorously quotes an anonymous author12 : “To stop pulsation. so that the industry is relying heavily on passive control concepts to suppress com- bustion instabilities. drill one hole [. For example.1) suggests two possibil- ities to control the instabilities: 1. Furthermore.6 introduction acoustic losses of the system. 284. 85. Adjusting the phase relationship for a destructive interac- tion between the heat release fluctuation and the acoustic pressure (active control). An alternative approach is to use the acoustic absorption capa- bilities of the perforated combustor liner [131]. Generally. 1940 . Bechert [38] has shown that the absorption mechanism is dominated by the transfer of acoustic energy into the shedding 12 Anonymous: “To reduce pulsations”. 2. 16. that they are always purged with a cooling flow. several barriers to implementation in full-scale gas turbine combustors still exist [98. so that several resonators of different sizes would be required to cover various operating conditions. 189. if that doesn’t work. Even though it has been successfully demonstrated in lab- oratory setups. the damping abilities of these resonators are limited to a rather small frequency range. It is a well known fact. 333]. Unfortunately. 395]. This bias flow through the orifices has been found to have a substantial effect on the absorption characteristics of the liner. At the same time the increasing weight prevents their installation in an aero-engine. that means adding Helmholtz resonators or quarter-wave resonators to the combustor system [42. . 137. that perforations or orifices in general can be applied to damp acoustic pulsations. 18. ].

On the other hand. While an acoustically well designed combustor liner could de- cide over the stability of the combustion. One obvious complexity is due to the multitude of pa- rameters that are involved. Commonly. 1. Through- out the industry overview given in [305]. so a generalization of conclusions can only be assumed. While a considerable amount of research is available. it is only Dowling and Stow [131] who address the acoustic properties of the combustor liner in their modeling. In order to distinguish such a configuration from a perforated liner without flow. these studies are focused on one parameter. there exist plenty of studies aimed at the combustor application and many impor- tant conclusions could be drawn. the industry is clearly hesitant to rely on the existing models. while it is a key parameter for the combustor application.1 motivation 7 vorticity when a bias flow is present. which additionally might be limited in its range. most ge- ometries or operating conditions are very far from what is encoun- tered in a gas turbine combustor. However. under which circumstances this potential can be retrieved. 1. it still seems to be an open question. The amount of literature that deals with acoustic properties of orifices or perforations in general is overwhelming. In particular. the presence of the bias flow is often not considered. that the bias flow liner is able to provide a broadband damping and that the actual components are already available in a combustor.1 motivation The high potential of bias flow liners as dampers in a gas tur- bine combustor is well-known. it is commonly referred to as bias flow liner. so that all the parameters considered here have been addressed in the literature in some way. its acoustic properties are rarely considered in the design of a new combustor. The main advantages over Helmholtz and quarter-wave resonators are. . However.

a com- prehensive assessment of the existing models can be provided. The various modeling approaches that are available to predict the acoustic performance of bias flow liners are illustrated in Chapter 4. which is needed to understand the acoustic phenomena occurring within a duct. Chapter 6 discloses all the details about the realization and the analysis of the measurements. so that possible improvements can be suggested if necessary.8 introduction It seems to be necessary to take a step back and look at the broader picture. and flow parameters are defined with respect to the experiments performed here and put into perspective with the definitions found in the literature. acoustic. the available ex- perimental data varies immensely in setup.2 outline Chapter 2 begins with a detailed overview of the liner setup and the operating conditions within a gas turbine combustor. This includes a detailed description . thermodynamic. and describing quantities. these effects become more significant at elevated pressure and temperature. The second part focuses on the experimental parameter study. is required for the evaluation of the information. As will be shown later. so that an independent set of data. experimental method. The rel- evant geometric. concluding the review part of the thesis. The goal of this holistic approach is to identify and determine the influence of the significant parameters. so that its definition is complemented by a review of the available research involving bias flow liners in Chapter 3. Chapter 5 collects the essential theoretical background. Furthermore. The scattered fragments of information need to be collected and put into perspective. 1. that a sound wave experiences when propagating through a hard- walled duct. based on one foundation. Special attention is put on the various loss mechanisms. However. The parameter in focus is the bias flow.

The results of the study are presented in Chapter 7. When applicable. The influence of each parameter is discussed separately. the models are compared to the experimental results.2 outline 9 of the features and properties of the Duct Acoustic Test Rig and the Hot Acoustic Test Rig. regarding the current findings as well as previous results from other studies. which have both provided their services in the parameter study. 1. .

.

2. The amount of air taking part in the combus- tion process. the layout of the holes in a combustor wall has been determined by combustion and cooling requirements and not by acoustic demands. yields only 40 % . Introducing additional air to the combustion process (secondary air holes in Figure 2. that is the primary air injected through the swirler and the secondary air injected through the walls. cooling down the hot gas before it enters the turbine (dilution air holes in Figure 2. and 3.1 illustrates the setup of a conven- tional combustion chamber. Figure 2. Traditionally. The parameters that are relevant for the acoustic performance of a perforated liner are collected and discussed individually. cooling of the combustor wall (corrugated joint in Figure 2. Due to the essential nature of the bias flow in this work. the openings in the liner serve the following purposes: 1.1). 2.1).1). The discussion includes a brief review of the presence and definition of each parameter in the literature.1 combustor liner Until today. 2 BIAS FLOW LINERS This chapter takes a detailed look at the setup of typical gas tur- bine combustors and their operating conditions. the literature review regarding the bias flow effect receives its own dedicated chapter (see Chapter 3). not limited to combustor liners but for orifices in general.

The walls are perforated by a large number of small holes. The remaining 60 % are required for dilution (20 %) and wall cooling (40 %) [408].12 bias flow liners Figure 2. where a shallow . As a consequence.1: Setup of a conventional combustion chamber.2 and 2. Ac- cording to Lefebvre and Ballal [292]. The arrangements of two modern combustors. courtesy of Rolls-Royce plc. (From [408]. The flow distribution has drastically changed in modern com- bustors. Both examples employ angled effusion cooling of the walls. are shown in Figures 2. the wall cooling air could be reduced by half to 20 % for combustors that are now in service [292].) of the total airflow. Lean combustion demands more air to take part in the combustion process. this is the most promising advancement in cooling methods regarding its potential for fur- ther significant reductions in cooling air requirements.3. designed for low emissions and high efficiency. With novel materials being available and opti- mized cooling techniques. a reduction of cooling air became necessary.

and with a relatively low flow velocity within the cham- ber. 253]. Aero-engines are designed to deliver much higher pressure . i.64-8.1 combustor liner 13 Figure 2.82 mm [88]. 183]. The ori- entation of the cooling orifices might be additionally skewed in the circumferential direction [92. designed for low emissions and high efficiency. the combustor operates at high pressure. courtesy of General Electric Company. e.5-1.2: The GE twin annular premixing swirler (TAPS) combustor with effusion cooling. The cooling efficiency can be further increased by using shaped holes.5 mm [351]. high tem- perature. Generally. Typical orifice inclination angles are between 20° – 60°.) angle in combustor mean flow direction provides two advantages for the cooling [292]: A larger surface area within the hole for increased heat removal and the establishment of a cooling film along the surface of the wall. 2. (From [157]. A typical wall thickness is in the range of 0. holes with an enlarged exit area where the velocity is reduced and the lateral spreading of the cooling air is improved [178. The hole diameters range between 0.

14 bias flow liners Figure 2.3: View into the annular combustor of the Rolls-Royce E3E Core 3/2 technology demonstrator. . the pressure ratio of the Rolls-Royce Trent 900 shown in Figure 1. Stationary gas turbines typically operate at pressure ratios between 10 and 25. optimized for lean burn combustion and NOx reduction. (Courtesy of Rolls-Royce plc. Current engine configurations achieve an overall pressure ratio1 between 30 and 52. Currently. A characteristic temperature of a gas turbine is the turbine inlet temperature.2 is 39. the flame temperature within the com- 1 The overall pressure ratio is defined as p3 /p1 . However.) ratios than stationary gas turbines. temperatures of nearly 1900 K can be achieved [331]. while typical values are between 1300-1700 K [80]. where p1 is the pressure at the inlet and p3 the pressure delivered to the combustor [408]. that is the temperature of the flow leaving the com- bustor and entering the turbine. For example. While a high turbine inlet temper- ature is desirable from an efficiency point of view. the maximum temperature is limited by the material properties of the turbine blades and the applied cooling.

1. when the flow velocity is slower than the flame speed. an additional classifi- . Due to the latter restriction the mean Mach number in the combustor is fairly low and commonly around 0. when the flame starts propagating upstream into the supply pas- sages. Combustion instabilities are a tonal phenomenon. The cooling air provided by the compressor is typically between 500-800 K [408]. grouped into four categories based on their physical origin. The actual frequency is depending mainly on the com- bustor geometry and the operating condition.1 gives an overview of these parameters. e. 2. so that they are observed as a narrow peak in a frequency spectrum of the com- bustor. 333]. The upper limit of the flow veloc- ity is given by the blowoff condition.05 [305]. Some of these parameters are fixed by the operating condition and are not available for modification. the flow velocity is restricted by the flash- back and blowoff limits. thus the frequency is not fixed but might change during operation. i. Several types of instabilities do exist. 305] for an overview and clas- sification).1 Parameter Overview The setup within a gas turbine combustor reveals a multitude of parameters that might be relevant for the acoustic performance of the liner. Table 2. The velocity of the bias flow through the perforated wall is de- termined by the pressure drop across the wall. the flow velocity exceeds the flame speed in any real combustor. In order to stabilize the flame. Typical operating conditions correspond to a 3 % pressure drop [417]. with lean burn tempera- tures usually below 2000 K [292]. Thermoacoustic instabilities in gas turbine combustors are typically observed at frequencies in the range of 100-1000 Hz [272. depending on vari- ous coupling mechanisms (see [301. In pre-mixed flames. where the flame cannot be stabilized and is convected downstream by the flow [304].1 combustor liner 15 buster can be as high as 2200-2600 K [80]. Flashback occurs at low flow velocities. 2. Thus.

A similar setup was used in [16. the circular shape and the grazing sound incidence. 2.1: Overview of parameters relevant to the acoustic performance of gas turbine combustor liners.3. The dependent parameters cannot be used as a design tool to improve the acoustic performance. the geometry of a combustion chamber is rather complex. for example.16 bias flow liners Table 2. 501]. .2 and 2. the geometric features can be abstracted and simplified. their im- pact on the performance is of great importance. e. geometry parameters thermodynamic parameters Orifice Geometry* Pressure Perforation Geometry* Temperature Cavity Geometry* acoustic parameters flow parameters Frequency Bias Flow* Amplitude Grazing Flow * independent parameter cation into dependent and independent parameters makes sense. Putting the focus on the perforated liner. This configuration is generally referred to as a cylin- drical perforated liner and will be used in the parameter study in Chapter 7. is illustrated in Figure 2. 279. The independent parameters are indicated by a star in Table 2. Nonetheless. 294.4. 310. The independent parameters can be adjusted within certain restrictions to optimize the damping. 17. They are determined by the operational requirements of the gas turbine combustor. 142. i. 198. The independent parameters are avail- able for modification to improve the acoustic performance.1.2 geometry parameters As seen in Figures 2. A simplified geometry that resembles the characteristic features.

1 Orifice Geometry The orifice geometry is composed of three main features: the ori- fice cross-section shape. 336. crown [181]. rectangle [89.5 gives an overview of the orifice shapes that have been studied in the literature re- garding their acoustic properties: circle. as indicated in Figure 2. ellipse [336. 169]. 181]. Figure 2. 89. 129.5j. e. i. 2. The orifice cross-section shape is given by a cut through a plane normal to the direction of the orifice. A single rectangular orifice with a high aspect ratio. eye2 [7. a long and thin slit. cross [89. and the orifice profile.2. and trapezoid [213]. 169. star [7. 389]. square [7. 282].2 geometry parameters 17 r θ x Figure 2. 2. is often used for its two-dimensional characteristics 2 The eye-shape is obtained by sliding a perforation consisting of circular ori- fices over another identical perforation. 169. 181]. tri- angle [7. 169].4: Simplified geometry of a combustor liner for acoustic studies: a cylindrical perforated liner. the orifice edge. 385]. . 225. oblong [410].

so that a circular shape is assumed if not stated otherwise. . 138. respectively. [10.5: Overview of orifice cross-section shapes. 332. The hydraulic diameter relates the cross-section area to the perime- ter Dh = 4A/P. 478]. The majority of the literature treats circular orifices only. In fluid dynamics the hydraulic diameter defines the diameter of an equivalent circular geometry for a non-circular shape. Introducing sharp corners and breaking up straight edges.5a) is the standard orifice that serves as a status quo in all3 the references given above. the orifice perimeter P. For a circular orifice the area and perimeter are given by A = πr2 and P = 2πr. 469. 204. The circular orifice (Figure 2. 3 Except Howe [213]. who compares the trapezoid to a square. 267. 203. 468. i. the different cross-section shapes can be compared by the orifice cross- section area A and the length of the orifice edge. e. where r is the orifice radius. 168.18 bias flow liners 2r (a) circle (b) ellipse (c) square (d) rectangle (e) oblong (f) triangle l1 l2 (g) cross (h) star (i) crown (j) eye (k) trapezoid Figure 2. that have been stud- ied in the literature regarding their acoustic properties.

Several orifice edge shapes. Looking at a cut through the wall material reveals the orifice profile shape.2 geometry parameters 19 (a) square (b) round (c) bevel (d) wedge Figure 2. as demon- strated in Figure 2. 141. inclined [16. and wedge [269. Figure 2. 55. 142. 286. 332. inclined orifices are very com- mon for cooling the combustor wall. and sharp. shown in Figure 2.6: Overview of orifice edge shapes. 480]. 2. The orifice inclination angle α increases the orifice length at constant wall thickness.02. 478]. The di- mensions are specified in ISO 5167-2:2003 [234]. How- ever. 478]. In this case. the orifice length l is identical to the wall thickness. (2. that have been studied in the literature regarding their acoustic properties.7b.7 compiles some geometries that have been treated in the literature: straight. a sharp orifice is a straight orifice with square edges and l/d < 0. 284. that are found in the literature. is a standardized geom- etry employed in flow measurements with orifice plates. for structural reasons the wall thickness is often larger than l. so that the downstream edge needs to be beveled to maintain .6: square. As shown in the previous section. The orifice length is then given by l = t/ sin α. Generally. conical [286]. 17.1) The sharp orifice. bevel [360. 282. 412]. the straight orifice is the most common. A characteristic dimensionless quantity is the orifice aspect ratio l/d. 284. Generally.7d. straight orifice is defined by its di- ameter d = 2r and the wall thickness t. round [258. are illustrated in Figure 2. The square edge is the most common geometry and treated in most studies. A circular.

8 illustrates different perforation patterns. an orifice unit area sx × sθ can be assigned to each orifice (in the case of the square . Any curvature effects are neglected for now. the arrangement of a multitude of orifices in a liner adds some more degrees of freedom and complexity.7: Overview of orifice profile shapes. Based on the perforation spacing. forming a square grid.20 bias flow liners l 2r 2r t t Ug ⇒ α (a) straight (b) inclined ⇓Ub t t l α (c) conical (d) sharp Figure 2.2. so that the liner is considered to be flat. Shown is the plan view of a segment of a cylindrical liner with the coordinates x in axial direction and θ in circumferential direction. that have been studied in the literature regarding their acoustic properties. The most simple pattern is the square perforation pattern pre- sented in Figure 2.8a. Figure 2.2 Perforation Geometry Owing to their simplicity. the sharp edge characteristics. While many parameters can be investigated with a single orifice. The orifices are arranged in straight rows in axial and circumferential direction. According to the ISO standard. The perforation spacing s is identical in both directions. single orifices are often employed in sci- entific studies. 2. the bevel angle needs to be within 30° – 60°.

Here.3). .8. This is indicated by the dashed grid lines in Figure 2. For a circular orifice in a square perforation pattern that yields σ = πr2 /s2 .8: Overview of perforation patterns that have been studied in the literature regarding their acoustic properties. 2. pattern that is s2 ). For a uniform perforation pattern the porosity σ is de- fined as the ratio of the open-area of one orifice to its unit area4 σ = A/(sx sθ ).2. For a specified surface area the same 4 In the literature this ratio is often called open-area-ratio.2 geometry parameters 21 Lc Leff L θ θ s sθ s x sx x (a) square (b) rectangular sx0 Leff θ θ α sθ sx x x (c) staggered (d) nonuniform Figure 2. the term open- area-ratio is reserved for the ratio of the open areas of two liners in a double- skin configuration (see Section 2. This is illustrated by the filled black orifice and the gray area around it.

This can happen easily when designing a cylindrical liner of a cer- tain porosity. The axial spacing has to be adapted accordingly to obtain the desired porosity. It is as well possible. producing a ‘negative’ margin. The selection of circumferential spacing is not con- tinuous. the cavity length Lc is defined by the dimensions of the cav- ity behind the liner. The rectangular pattern is defined by the perforation aspect ratio sx /sθ . In other cases the perforation only covers a fraction of the available length.8c is the most wide- spread in technical applications. Then. Finally. 289]. that the ef- fective perforation length becomes longer than the cavity length. the more mass flow is needed to achieve a certain bias flow velocity. the perforation placement could be shifted along the axial coordinate. In a theoretically constructed setup the effec- tive perforation length and the cavity length would most probably be identical (as they are in Figure 2.8b is similar to the square pattern described above. the effec- tive perforation length is shorter than the cavity length. 288. The rectangular perforation pattern in Figure 2. The default position is at the center with equal margins on both sides. The only difference is that the ori- fice spacing is not identical in axial and circumferential direction. The maximum axial distance between two orifices defines the perforation length L. which is the open area of all orifices combined. The staggered perforation pattern in Figure 2. An important parameter regarding the efficiency of a bias flow liner is the total open area. The influence of the aspect ratio on the acoustic performance was studied in [17.22 bias flow liners porosity can be realized by applying many small orifices or fewer large ones.8a. but dependent on the circular pitch resulting from the number of orifices around the circumference. An effective perforation length Leff of the perfo- ration can be specified according to the unit areas of all orifices. resulting in some margins on both sides. The structural strength of the ma- . resulting in large mar- gins of hard wall. In Figure 2. The larger the open area.8b).

By doing so. In the example. that a potential interaction of two adjacent orifices in grazing flow direction is re- duced as their distance is doubled (the orifice in-between is moved sideways). The term nonuniform is used for patterns where the porosity is not constant.1°. Figure 2. cir- cumferential direction. the sheet material retains the most strength while offering the largest possible open area. the ‘lo- cal’ porosity increases towards the center of the liner. This value might seem odd at first. 259. In order to obtain a 60° stagger angle the axial distance between the rows has to be reduced. The pattern presented here has a stagger angle of 53. 2. either in axial direction (as shown). The influence of nonuniform patterns is studied in [116. but it can be easily obtained by just turning every second row of the square pattern about half the orifice circumferential spacing. but it provides a comparative value between uniform and nonuniform patterns.2 geometry parameters 23 terial is enhanced compared to a rectangular or square pattern. as in both cases sθ = sx . the pattern is not completely random and shows some sort of regularity. or both. Common perforation stagger angles α are 60° (sometimes called triangular pattern) and 45° (sometimes called diagonal pattern). This gives an overall or average porosity. Another effect of the circumferential stagger is. 282] with very dif- ferent configurations.8d.8d demonstrates an example of a nonuniform perfora- tion pattern. . so that the porosity would be increased in relation to the square pat- tern. the porosity remains constant compared to the square pattern. as indicated in Fig- ure 2. Thus. At 60° the distances between any neighboring orifices are equal sθ = sx0 . For nonuni- form patterns the porosity is given by the ratio of the total open- area to the area defined by the active length. Still. The significance of this value might be questionable.

the inner arc length (sθ = πR α/180◦ for α in degree) will be used as the sound is incident from that side.9: Effect of the curvature on the definition of the circumferential perforation spacing. Shown is the top half of a cut through the axisymmetric cavity. For a cylindrical liner sθ is given as the arc length between the centers of two neighboring orifices. except the liner. The difference is very small for the geometries studied here.10a illustrates a standard single-skin configuration. 2.24 bias flow liners sθ0 sθ α R Figure 2. Figure 2. Figure 2. . Curvature Any effects of curvature on the geometric specifications above were neglected. double- skin.9 illustrates the curvature effect. and partitioned configuration.3 Cavity Geometry The volume behind a cylindrical liner forms an annular cavity (see Figure 2. All boundaries of the cavity. Figure 2. but might become more substantial for smaller duct radii and/or larger wall thickness. Here.4).10 gives examples of a single-skin. However. the curvature requires a more detailed look at the specifications of the circumferential perforation spac- ing. Due to the curvature it makes a difference if the inner arc length sθ or the outer arc length sθ0 (or the center arc length) is used. The liner has the same radius as the duct and it forms the inner wall of the cavity.2. are acoustically hard.

The annular cavity volume can be calculated from V = π(R2c − (R + t)2 )Lc . 429. when a lower pressure drop is desired for the in- ner liner5 . A variation of the cavity geometry can be obtained by parti- tioning the cavity into several smaller cavities. 481]. 417. The second liner is introduced to match the required pressure drop in a combustor. 310. According to their functions.10: Overview of common variations in cavity geometry. Such a partitioned cavity is illustrated in Figure 2. 5 Such a configuration can already be found in a combustor with an impinge- ment cooling setup. 2. . The cavity length Lc corresponds to the axial dimension of the cav- ity. Partition walls might be necessary for structural reasons. 282. 241. 434. the inner liner is called damping liner and the outer liner metering liner. The double-skin configuration in Figure 2. Variations of the cavity volume have been presented for plane liner configurations with normal sound inci- dence in [219.2 geometry parameters 25 Lc Rc R2 R R1 Lc1 Lc2 Lc3 (a) single-skin (b) double-skin (c) partitioned Figure 2. 417]. The liner radius R and the cavity radius Rc are indicated in the drawing. 423.10c by introducing two solid par- tition walls in axial direction. The velocities through the orifices of the two liners are related via the open-area-ratio between the two liners. for example.10b introduces a sec- ond liner in-between the existing liner and the cavity wall. for example. Double-skin configurations are used in [142.

484]. involving perforated liners [381]. Appendix a. 381. Only a few studies exist where the temperature was controlled.6).1 Temperature The temperature within the combustor is around 2000 K and the cooling flow is provided at temperatures between 500-800 K. 2.3 thermodynamic parameters A combustor operates at extreme pressure and temperature condi- tions. the tempera- ture is much higher. 174. 382]. the temperature of the grazing flow and the bias flow is around 288 K. 294. 379. 462]. Many fluid properties change considerably when these two parameters are changing. 251. None of the references above include a bias flow. 481. Mea- surements of porous materials at reduced temperature (172 K) are presented in [9]. measurements are presented from the Hot Acoustic Test Rig (see Section 6. . The mean temperature of the grazing flow entering the lined section will serve as a reference tempera- ture. 403. a single orifice with cavity [144]. 166. The bias flow is provided at a con- stant temperature of 288 K. or porous materials [93. Here. In that case.3.26 bias flow liners 2. but it is fixed at that level so that the influence of varying temperature on the absorption cannot be determined. i. e. 370. g. Some results are reported for tests including combustion [55.3 gives an overview of the behavior of some properties of air with variation of pressure and temperature. Most laboratory test rigs for liner measurements operate at ambient temperature. 355. 346. e. The only configuration where temperature effects have been stud- ied including a bias flow is a duct termination issuing a hot jet [104. 310. 357. It provides an acoustically defined environ- ment where the temperature of the grazing flow can be adjusted between ambient and 823 K.

swept-sine [87]. so that the influence of the static pressure on the absorption can be determined. Commonly. Closely related to the frequency. The Hot Acoustic Test Rig allows to increase the static pressure in the duct from ambient up to 1100 kPa (see Section 6.2 Pressure The combustor operates at very high static pressure levels.4. 162. 481]. single-sine [52. 2. most studies are limited to plane waves. is the spatial structure of the sound field. [140. 6 That means. How- ever.4 acoustic parameters 2. the frequency characteristic of the liner is one of its most important features. 247]. Even the measurements that involve combustion are typically at atmospheric conditions [55. the performance of a liner is measured over a range of frequencies. the exact frequency where the instability occurs cannot be predicted and it changes with the operating condition of the com- bustor. this is not accounted for when testing perforated liners.4 acoustic parameters 27 2. In order to obtain such a performance spectrum. Com- monly. multi-sine[82. Due to the low frequency (< 1 kHz) nature of the combustion instabilities. 310. or more precisely the wave length. g. 2.6). Thus. the acoustic field quantities are a function of the axial coordinate only (see Section 5. Some theoretical studies are available that describe the interaction of higher order modes with an acoustic liner. or broadband [90].3. 393]. e.1 Frequency A combustion instability is a discrete frequency phenomena. g. . it is often the plane wave mode6 that is dominant [305].3). e. Therefore. various test signals can be applied. 282].

72.2 Amplitude / Sound Pressure Level The influence of the amplitude on the acoustic properties of an orifice has been the subject of many studies. while it becomes relevant in the nonlinear regime. 228. [26. the amplitude of the incident wave [8. 441. 474. e. the peak amplitude of the standing wave field in the hard-wall duct section [327]. 225. 327. 58. It is generally assumed that the physical quantity relevant to the nonlinear behavior is the particle velocity in the orifice.28 bias flow liners 2. that the output to a sum of inputs is equal to the sum of the outputs produced by each input individually: f(x1 + x2 ) = f(x1) + f(x2 ) [47]. Now. the pressure amplitude at the liner surface [41. a linear or a nonlinear behavior is observed. 105. i. or the amplitude in the cavity behind the perforation [227. 415. 473. 123. However. e. 441]. Depending on the amplitude. 61. so that the amplitude is often given in terms of sound pressure level (SPL). Authors refer to: the particle velocity amplitude in the orifice [227. 7 Additive means. The definition of a linear system is given by Bendat and Piersol [47]: The response characteristics are additive7 and homogeneous8 . . 243.4. Unfortunately. 243]. in most cases it would take a great effort to measure the particle velocity in the orifice. 161. 227. the amplitude in the loudspeaker mounting [7]. At high amplitudes the behavior of the orifice is not homogeneous anymore. 361. that the output produced by a constant times the input is equal to the constant times the output produced by the input alone: f(cx) = cf(x) [47]. the response depends on the excitation amplitude. the knowledge of the exact amplitude is rather unimportant in the linear regime. 142. 8 Homogeneous means. 199. As a conclusion. the three most common ap- proaches are discussed and evaluated for their comparability and practicality. 425]. 505]. 481]. 158. g. the definition of the amplitude is not quite con- sistent in the literature. often no clear definition is given. 247]. Actually. the amplitude at a fixed reference location in the hard-walled duct section in front of the liner [82.

flow condition. 161. The amplitude at the liner surface is dependent on the reflection co- efficient of the liner. Normally10 . Furthermore. It is either mea- sured directly with a microphone installed at the liner surface [158. when there is no liner installed and the duct is terminated anechoically. 11 When disregarding any losses within the hard-walled duct. i. the incident wave amplitude cannot be measured directly. while it requires a microphone within (or at least very close to) the liner surface. 10 The incident wave amplitude can be measured directly when there is no reflected wave. so that a wave decomposition is necessary. However. so that typically several iterations (pre- liminary measurements) are necessary to set the desired SPL. the incident wave amplitude is independent of the liner properties. The wave decompo- sition method enables to extrapolate the sound field from micro- phones placed along the hard-walled duct onto the liner surface. 105. 425]. so that for keeping a constant SPL the loud- speaker output needs to be adjusted when modifying any param- eters that change the liner properties. 2. Again. 415. e. frequency. the wave decomposition is not performed in real time with the measurements. so that for low frequencies the amplitude value can be considered identical to the amplitude at the liner surface. The direct measurement has the advantage of being very straightforward and fast. e. Another approach considers the SPL of the incident wave only [8.g. .4 acoustic parameters 29 In a setup with normal (perpendicular) sound incidence it is common to specify the SPL at the liner surface. 9 The reference location in [161] and [243] is chosen to be very close to the liner surface. The desired output of the source needs to be determined once for each frequency and SPL and can then be applied to different liner configurations and flow settings. or determined from a wave decomposition based on microphone measurements in the hard-walled duct section [41]. The advantage over the previous approach is that the incident wave amplitude is independent of the axial position within the hard-walled duct11 . geometry. this requires an iterative measurement procedure. 243]9 .

so that the amplitude |p0 | at the reference position xref is 125 dB for both configurations. 199. in a setup with grazing sound incidence the liner extends in axial direction.11: Illustration of the sound field at a nominal amplitude of 125 dB measured at position xref for a liner configuration with a small and large reflection coefficient R. Figure 2. the amplitude at the liner location at x = 0 would be very dif- ferent when choosing another axial position as reference or when the reflection coefficient changes.30 bias flow liners Microphone 3 4 5 6 Liner 125 small R large R SPL in dB 115 |p0 | 105 |p̂+ | xref 0 Axial Coordinate Figure 2. However.11 demonstrates that the results can be quite misleading. so that the amplitude changes along the liner (due to the presence of a standing wave . 248]. However. The loudspeaker output is adjusted. Plotted is the sound pressure level over the axial coordinate obtained from a wave decomposi- tion in the hard-walled duct section upstream of the liner for a configuration with a small and a large reflection coefficient R. The amplitude at the liner surface is independent of the test rig and thus serves well as a reference quantity when comparing different measurements. The most straightforward approach defines the SPL at a fixed reference location in the hard-walled duct section in front of the liner [82. Plotted is the amplitude of the sound field including the reflected wave |p0 | and the amplitude of the incident wave alone |p̂+ |. However. 142.

The incident wave amplitude should be the preferred quantity in a grazing incidence setup.4 acoustic parameters 31 content in the sound field and due to the absorption of the liner). Figure 2. Analyzing the data in the frequency domain disregards these events which are a product of the super- position. However. In this case.12 illustrates the superposition of two frequency components to one multi-sine signal in the time domain. Using multi-sine signals12 can reduce the measurement time dramatically. the multi-sine signal shows events of varying amplitudes. Multi-Sine Signals 2 1 p0 in Pa 0 -1 -2 2 6 10 Time in ms Figure 2. when specifying the amplitude of a multi-sine signal the 12 A multi-sine signal is synthesized by combining multiple single-sine signals. it is difficult to choose the ’correct’ reference plane. That means. when the amplitude of an event is high enough. so this approach is not applicable in a grazing incidence setup. However. 2. .12: Superposition of two single-sine signals ( 330 Hz and 800 Hz) to one multi-sine ( 330 + 800 Hz) signal in the time do- main. While the single-tone sig- nals have a constant peak amplitude. the orifice ´sees’ these events and might behave nonlinear. multi-sine signals should be used with special care when studying high amplitude effects.

281]. However. 24. Most of the references given above study the influence of the flow boundary layer on the acoustic behavior. 81. 106. 209. 290. i. This is in contrast to gas turbine combustors where Mg = 0. in order to en- sure that one acts in the linear domain the overall SPL should be consulted. g.5 flow parameters The flow paths in a combustor liner were discussed in Section 2. 215. Another characteristic of multi-sine signals in the nonlinear do- main is that the different frequency components might influence each other [68. 264–266. 244. 322. 56. 426.32 bias flow liners overall SPL is the more appropriate quantity. 153–155.05 is a common value. e. 396.5-0. 145.7 are typical. 83. 384. 268. 122. For a single-sine signal the amplitude of the tone and the overall SPL are identical. The flow in a combustion chamber as well as in the . 177. 480]. where grazing flow Mach numbers of 0. 406. [364] conclude that the mean flow velocity is the adequate parameter when the flow is turbulent and fully developed. the significance of the friction velocity over the mean flow velocity. 175. and not the SPL of each tonal component separately. 467.5. Peat et al. 478. 364.1. referred to as grazing flow. and the cooling flow through the orifices of the liner. 312. 341. [21.1 Grazing Flow The effect of a grazing flow on the acoustic properties of orifices and liners has been studied in many publications. In other words. referred to as bias flow. This interest in the grazing flow effect is mostly motivated by the application of liners in aero engine inlets and bypass ducts. 329. So the typical grazing flow Mach number is an order of magnitude lower. The general motion of fluid at the liner is a combination of a flow grazing the liner surface tangentially on the inside of the combus- tor. 274. e. 2. 2.

p.5 flow parameters 33 Ug r x Ucl Figure 2. . .35 f where f is the Darcy friction factor15 . 346] Ucl Ug ≈ √ . so that only the mean grazing flow Mach number is considered. assuming turbulent pipe flow. 496]. 15 Named after Henry Darcy (1803-1858). 108 . the mean velocity can be determined from velocity profile measurements or.2) assumes that the boundary layer profile is described by the logarithmic law [496]. 13 The grazing flow Reynolds number is larger than the critical Reynolds num- ber Rec = 4000 [496].316 4 Re [65.2) 1 + 1. The relationship between mean and center-line velocity is illustrated in Figure 2. Otherwise. 14 Equation (2. 2. In turbulent flow the center-line velocity Ucl is related to the mean velocity via14 [496. French engineer.5 and 6. test rigs used here (see Sections 6. 2.6) is fully turbulent13 . if the mass flow rate ṁ is known. it can be approximated by f = 0. from the measurement of the center-line velocity only. (2. For Reynolds numbers in√the range Re = 4000 .13: Illustration of the relationship between mean grazing flow velocity Ug and maximum velocity at the center-line Ucl . .13.5.2 Bias Flow The bias flow in a combustor is driven by a steady pressure differ- ence across the liner. The mean grazing flow velocity can be computed from the con- tinuity equation Ug = ṁ/(ρA). The higher pressure is applied to the cavity.

this study exclusively treats bias flow directed into the combustor or test duct. A steady bias flow can be defined by its mass flow rate. or the velocity through the orifices. (2. 276. The effect of a periodic oscillating. when a grazing flow is present the results are substantially different [332]. unsteady bias flow is studied by Heuwinkel et al. [200] and Lahiri et al. the pressure drop across the wall.34 bias flow liners so that the fluid discharges from the cavity. A certain operating condition yields a fixed pressure drop. such a concept cannot be applied to a combustion chamber. [280. . However. 281] and also. e. is an operational quantity of a gas turbine combustor. Therefore. 241. However. that is in the opposite direction as described above. Otherwise the intended air distribution might change. into the combustor. Some authors have studied the effect of neg- ative bias flow [6.3) the pressure drop refers to the total pressure drop across both liners. The pressure drop across the liner can be measured with a differential pressure meter via static pressure taps on both sides of the liner. 87. g.3) pduct In a double-skin configuration (see Section 2. 26. in the cavity and in the duct. The air flow in the combustor is regulated by the relation of the pressure drops of the different components. The bias flow pressure drop. As these three quantities provide different information and enable different conclusions it is an advantage to have all three available at the same time. will not be included here. it is very important to match the pressure drop when replacing a liner with a new de- sign. e. The pressure drop is then given in relation to the absolute pressure in the duct as pcavity − pduct ∆P = × 100 % . 332]16 . that is the pressure difference across the liner. Tonon et al. The discharging hot gas would compromise the integrity of the liner and other components down- stream.2. Thus. i. as this 16 Jing and Sun [241] and von Barthel [26] found that blowing and suction have the same effect. through the orifices. [478] report that the acous- tic resistance appears to be a factor four lower for a grazing-bias outflow compared to a grazing-bias inflow case.

A typ- ical pressure drop across a combustor wall is about 3 % [417]. Thus. 17. Most lab- oratory experiments use a mass flow controller to adjust the bias flow. In Figure 2. Howe [210] argues that the vorticity convection velocity is the mean velocity in the plane of the orifice. The bias flow velocity is the quantity that is related to the absorp- tion of sound. a closer look at the steady flow field in the vicinity of the orifice is required. The streamlines cannot follow the sharp contour of the geometry. Italian physicist and mathematician. resulting in a contraction of the emerging jet. e.14a the fluid discharges through a thin orifice with square edges. usually referred to as mean orifice velocity. Figure 2. The difference between these two velocities can be as large as a factor of 2. a liner achieving the same damping performance at a lower mass flow rate is more efficient. 219] use the velocity based on the orifice area. i. should be compared. but estimated from the pressure difference across the liner or the mass flow rate through the liner. 210. 55. Some authors [8. Typically. 214.5 flow parameters 35 is the relevant quantity for the operation of the combustor. so that the total mass flow rate. its value is available in most cases. is ac- credited with the observation of this phenomenon. For a better understanding of these differences. 142. They form a smooth path. The efficiency is evaluated for the liner as a whole. . The mean orifice velocity can be calculated when the mass flow rate and the open area of the liner is known. 2. instead of the mass flow rate per orifice. The bias flow mass flow rate can be a measure of the efficiency of the liner. The cross- section area of the jet is reduced to a value that is smaller than the actual area of the orifice17 .14 illustrates the steady flow through an orifice for three significant geometries. The location of maximum contraction 17 Evangelista Torricelli (1608-1647). The literature offers two views on a definition of the bias flow velocity. The second definition is based on the jet veloc- ity [41. the velocity is not measured directly. 282].

18 The term vena contracta was introduced by Sir Isaac Newton (1642-1727.14: Illustration of the bias flow through orifices of three signifi- cant geometries. typical liner geometries are often within this transition region. In the transition region. natural philosopher.02.36 bias flow liners vena contracta 1 1 1 2 2 2 (a) thin orifice (b) thick orifice (c) round edge Figure 2. but the friction losses increase. The friction losses are similar to the thin orifice and are small for both geometries. The emerging jet is of the same area as the orifice. The area of the resulting jet is identical with the orifice area. En- glish physicist. sharp orifice the vena contracta is located approximately half a diameter downstream of the leading edge of the orifice [76. The reattachment of the flow occurs for orifice aspect ratios beyond l/d > 2 [119. mathematician.14b the flow separates at the sharp inlet edge and then reattaches to the wall before it discharges through the outlet. so that the flow field is subject to wide variations [320]. alchemist. Unfortunately. and theologian) [353]. 353]. is named vena contracta18 (latin for “contracted vein”). and the pressure in the jet is equal to the surrounding pressure. . the velocity is at max- imum. the flow might or might not reattach. say 0. astronomer.14c guides the flow along the smoothly curved inlet of the orifice. ISO 5167-2:2003 [234] defines a thin orifice as l/d < 0. Here. For a circu- lar. For the thick orifice shown in Figure 2.02 < l/d < 2. The round edge orifice in Figure 2. demonstration the jet contraction and friction losses. 302]. the streamlines of the jet are parallel.

99 [76.4) Uideal The velocity coefficients for sharp or round orifices are similar and range from 0. 303]. e.5) The contraction coefficient Cc relates the area of the jet Ajet to the cross-section area of the orifice A. the friction losses are often expressed by the resis- tance coefficient Cr .98 is given [25. The term should not be confused with ideal gas. Rayleigh [386] derives a theoretical limit of 0.g. which is related to Cv by Cr = 1/C2v . Typically. where Ajet corresponds to the vena contracta. . they have zero vis- cosity and thermal conductivity [338]. Measurements with a circular. The area of the jet corresponds to location 2 in Figure 2. 2. (2.611. where a typical value is given by Cv = 0. so that Cv is often neglected for these geome- tries.5 6 Cc 6 1. The friction losses increase considerably in a thick orifice (l/d > 1). 456]. (2. Kirchhoff [262] gives an analytical expression for the contraction of a jet through a circular orifice in a thin and infinitely extending wall as Cc = π/(π + 2) ≈ 0. sharp orifice find a 19 Ideal fluids are free from all dissipative phenomena. a value of Cv = 0. accounting for the friction losses within the orifice U Cv = . In engineering.95 to 0. 91.5 flow parameters 37 Hydraulic coefficients of the orifice The friction losses and the jet contraction can be accounted for by introducing the hydraulic coefficients of the orifice [25. The velocity coefficient Cv relates the theoretical velocity of an ideal fluid19 to the actual velocity of a viscous fluid.8 [25]. (2.14 and the contraction coefficient is given by Ajet Cc = .6) A An appreciable contraction is only observed for the thin orifice. 456].

sharp orifices in a straight pipe there exists a large amount of empirical formulas [221] and tabulated values [76. 478]. 407.98 0. However.00 0. in many cases Cv is close to unity. Typically.38 bias flow liners Table 2. However.14b) 1. 3-60]).00 0. 124. Nominal values of the hydraulic coefficients for geometries cor- responding to Figure 2. Indeed. (2. and maybe grazing flow.80 Round edge (see Fig.62 is used [25].67 [76]. 185. There is data available for more realistic flow conditions and geometries [84. 2. so that Cd = Cc .80 0. 254. It considers the jet contraction as well as the friction losses and is given by the product of the velocity and contraction coefficients ṁ Cd = = Cc · Cv . p.98 0. Orifice geometry Cc Cv Cd Thin orifice (see Fig. 184.2: Nominal values of the hydraulic coefficients for three signifi- cant orifice geometries corresponding to Figure 2.62 0. one should keep in mind that this is just an approximation and that the influence of Cv increases with thicker orifices. . but in most cases it is more reliable to determine the hydraulic coefficients experimentally.14 (from [25. 2.98 contraction coefficient in the range from 0. the discharge coefficient and the contraction coefficient are used interchangeably in literature20 . The discharge coefficient Cd is the ratio of the actual to the theoret- ical flow rate through the orifice.61 to 0.61 Thick orifice (see Fig. 125.2. the flow conditions at a liner can include non- uniform inflow conditions. Cc = 0.14a) 0. For circular. orifice interaction. so that the standard formulas are not applicable. 442. 20 This was already noted by Ahuja and Gaeta [7].14c) 1.14 are summarized in Table 2. 443].7) ṁtheoretical Often. 2.

Then the bias flow velocity Ub is given by 2 r Ub = Cv (p1 − p2 ). (2. The pressure at location 2 refers to the pressure within the duct. so that the velocity on the inlet side is assumed to be zero (the flow is driven by the pressure difference only). For thin ori- fices Cv is close to unity. (2. see Equation (2. .8) 2 2 where gravity is neglected. 2.11) 21 Air can be considered incompressible for M < 0.9) ρ The friction losses can be accounted for by the velocity coefficient Cv .8).ideal = (p1 − p2 ). the location of point 1 is chosen to be far upstream.14.14c). Viscosity is neglected in Equation (2.14b and 2. Equation (2.3 [33]. so that the location of point 2 is at the vena con- tracta (Figure 2. Swiss mathematician and physi- cist). For convenience. the pres- sure and velocity of any two points along a streamline are related via the Bernoulli equation22 (Bernoulli [49]. or [33.14a) or the orifice outlet (Figure 2. 283]) 1 1 p1 + ρU21 = p2 + ρU22 .4). so that it is often neglected.8) can be applied to the orifices presented in Figure 2.5 flow parameters 39 Determination of the bias flow velocity from the pressure drop For the steady flow of an ideal. so that U2 is the velocity of the jet of an ideal fluid 2 r U2 = Ub. Determination of the bias flow velocity from the mass flow rate The theoretical mass flow rate of an ideal fluid through an orifice is given by the continuity equation: ṁtheoretical = ρAUb. (2. incompressible21 fluid. (2. 22 Named after Daniel Bernoulli (1700-1782.ideal .10) ρ The bias flow velocity corresponds to the jet velocity.

The actual mass flow rate includes the jet contraction via the con- traction coefficient Cc . This approach assumes that the mass flow is divided up evenly through all orifices. the mass flow through the orifices might vary due to a non-uniform flow distribution on the inlet side and manufacturing differences of the orifices.10). where n is the number of orifices. (2. so that Ub corresponds to the mean velocity in the plane of the orifice. This is indicated by the bar over the index b.ideal represents the jet velocity of an ideal fluid.9). In that case. so that ṁ = ρ · Cc A · Cv Ub. i. In a liner with many orifices ṁ is the total mass flow rate and the corresponding area is nA. In practice. as well as friction losses with the velocity coefficient Cv .13) ρCc A Ub corresponds to the jet velocity including viscosity effects. Unfortunately. Cc is mostly not known a priori and can only be approximated. Measurement of the discharge coefficient The discharge coefficient can be determined when the pressure difference across the liner and the mass flow rate through the . Then. cor- responding to Equation (2.13) yields the mean velocity of all orifices of the liner.ideal . e. Equa- tion (2. the pressure drop and Equation (2.40 bias flow liners A is the orifice cross-section area. it is not taking the jet contraction into account. (2.ideal is given by ṁ Ub = .12) Here. Ub.10) should be used to determine Ub . The bias flow velocity Ub = Cv Ub. corre- sponding to Equation (2.

where it is assumed that Cv = 1. Here.13). so that Cd ≈ Cc .10) and (2. while both quanti- ties are recorded. (2.10) and (2. 2. the velocity is calculated from the pres- sure difference with Equation (2. The bias flow velocity is given in dimensionless form as the bias flow Mach number Ub Mb = (2.5 flow parameters 41 liner are both available. together with (2. the bias flow velocity refers to the velocity of the jet as defined in Equations (2. Then.7) yields ṁ Cd = .15) c Furthermore. A certain mass flow rate or pressure difference is set for the measurement.13). Combining Equations (2. Cd is calculated from the mass flow rate and the pressure difference with Equation (2.13).14) 2 q ρA ρ (p1 − p2 ) For sufficiently thin orifices it can be assumed that Cv is close to unity. .10).

.

respectively. who eliminated resonances by deliberately placing orifice plates into the exhaust ducts of combustion engines and reciprocating engines. the real and imaginary part of the acoustic impedance.2. 3 L I T E R AT U R E O N B I A S F L O W L I N E R S Historically. while a resonance with a low Q has a higher bandwidth. 3.1 bias flow as a concept of impedance control The first study of a more fundamental nature was published in 1950 by McAuliffe [319]. the damping effect of a bias flow orifice was observed in 1916. Other au- 1 The resistance and the reactance are. He demonstrated the dependency of the orifice impedance on a steady bias flow. In fact. 2 The Q factor. he created a constriction by partially closing a throttle valve. When applied to a Helmholtz resonator. A high Q factor represents a sharp peak. the bias flow provokes a shift of the resonance frequency to higher values and a decrease of the Q factor2 of the resonance. short for quality factor. presenting the results of impedance tube measurements. Increasing the velocity revealed a substantial drop of the orifice reactance and a linear increase of the orifice resistance1 . when Borth [73] reports on the successful suppression of resonances in the ducts of a piston blower with the help of a bias flow orifice. His observation was confirmed by Maier and Lutz [311] and Lutz [309]. . which will be introduced in Section 4. describes the bandwidth of a resonance. respectively.

1]). 484]. ] to control the air flow in the space behind the liner and through the perforations thereof in such a manner so as to com- pensate for variations in the absorption of the liner with variations in pressure level” (US2941356 [62. 170. and a shift in resonant frequency. An early patent describes. 326. . p. 227. thus reducing the need for expensive trial-and-error testing. This helps to overcome the inaccuracies of the avail- able models. . this is often not reflected in the models. p. 491] use this approach with varying em- pirical constants. 459. The effect of the bias flow is commonly modeled by a resistance that depends linearly on the bias flow velocity. 367.” In the meantime. 62. Several authors [118. 3] summarize the general effects of the flow: “Alteration in absorp- tion. 300. While an effect on the reactance has also been reported. with broadening of the bandwidth characteristics. In conclusion of these studies. 368. [326] studied a duct termination. [325] employed screens of porous materials. Utvik et al. . and Mechel et al. technological advancements have produced further applications of the bias flow concept. 78. Mechel et al. The impact of the bias flow on the impedance characteristics is very similar to the behavior observed at high amplitude excita- . namely the suppres- sion of combustion instabilities in rocket engines [20. “[. which can either increase or decrease the damping effective- ness of the liner configuration. . the theoretical setup of Ingard [226] consisted of a perforated screen. the experi- ments of Barthel [26] were performed with a Helmholtz resonator with normal sound incidence. Dean and Tester [118] see the greatest benefit in the ability to easily tune the impedance of the liner when it is installed in an engine. [483.44 literature on bias flow liners thors reproduced these findings with a wide variety of different setups: The same setup was used by Westervelt [491]. 171. 180. Feder and Dean [158] and Dean and Tester [118] propose the bias flow concept as a method of impedance control to turbofan inlet liners. ] a combus- tion chamber with a perforated absorption liner having a regula- tor [.

The impedance model of Bauer [34] is often referred to when perforations with bias flow are involved. so that the relevant quantity in both cases is the velocity in the orifice. e. 57]. 373]. 276. 72. After a long break. 143]. 228. How- ever. In their experi- ments they observed a substantial attenuation at low frequencies for pure tone sound propagating through a nozzle along a tur- bulent jet. but not an explicit bias flow term. 61. g.6. The bias flow term is integrated in the nonlinear impedance term by using the sum of the acoustic particle velocity and bias flow velocity as pro- posed by Premo [373].2 sound absorption due to vorticity shedding 45 tion [26. 123. [14. The work regarding the bias flow involves measurements of various liner geometries as well as impedance modeling [52–54. 474. 162. 225. 473. Dean and Tester [118] include the bias flow in their impedance model by simply replac- ing the acoustic particle velocity v0 with the bias flow velocity Ub .7. [39. 361. 243. The Bauer model is presented in Section 4. while the acoustic signal was introduced into the . 58. 3. 227. 505]. 54] is based on the state of the art models used in the industry with the addition of a bias flow resistance term. 327. as Dean and Tester [118] demonstrated. it seems common practice to replace the acoustic velocity in the nonlinear term with the bias flow velocity. 423. 3. Actu- ally. this can only be correct if Ub  |v0 |. 473].2 sound absorption due to vorticity shedding In the late 1970s Bechert et al. Indeed. 18. The conical nozzle issuing the jet was located in an anechoic room. Bauer’s model includes only a v0 term describing the nonlin- earity at high amplitudes. 105. the bias flow concept was revived in a com- prehensive study initiated by the Boeing Company [56. The impedance model will be referred to as Betts model and is discussed in more detail in Section 4. 441. the high amplitude produces an oscillating flow through the orifice [228. However. The resulting impedance model proposed by Betts [52. 161. 40] could provide some insight into the interaction of sound with a turbulent jet.

has been a long de- sired characteristic of an acoustic absorber. 267]: “We believe that this luxuriance of interaction phenom- ena at low Mach numbers is all to be explained by coupling between sound and fluid flow. low frequency sound absorption.” 5 Named after Martin Wilhelm Kutta (1867-1944. e. Bechert [37.8 or 2000 Hz for the geometry in the experiments. in flow models where viscosity is not ex- plicitly included. It was verified that the sound power is neither shifted to other frequencies nor that it contributes to the broadband jet noise amplification. acoustic energy is converted into energy of fluctuating vorticity. 4 The existence of such a phenomenon was already suspected by Gordon and Smith [180. . 38] il- lustrates the physical mechanisms of the absorption phenomenon and proposes a model to account for the absorption. i. at the sharp edges of the vent. He could experi- 3 That is below He = 0. broadband. A quanti- tative explanation has not been attempted.46 literature on bias flow liners duct upstream of the nozzle. i. In his theory. measurements at dif- ferent sound power levels revealed that the absorption effect is independent of the sound power level. German mathematician) [275]. Bechert’s observation and his associated theory are regarded as a breakthrough in understanding the physical phenomena of the interaction of flow and sound at an orifice. it has a linear behav- ior. While the sound power is conserved without flow. These results reveal the existence of a sound absorption pheno- menon which has not been considered previously. particularly vortices. The transmitted sound power from within the duct was compared to the sound power radiated into the far field at various jet velocities between 0 6 M 6 0. The Kutta condition requires that all velocities remain finite in the vicinity of the edge. Shortly after his experimental observations. p. emphasize that the characteristics of the new phenomenon. Also. which is shed from the nozzle edge and is dissipated into heat fur- ther downstream4 . This is implemented in his model by applying a Kutta condition5 at the edge to enable the vorticity shedding. From [338]: The Kutta condition is applied to a sharp trailing edge to simulate the effects of viscosity.7. with flow it is atten- uated at low frequencies3 . e. Bechert et al.

Howe [208] analytically derives a model based on Powell’s vor- tex sound theory [207. . Jing and Sun [241] introduce a thickness term in Howe’s model by physical reasoning. 7 A recent mathematical discussion on the effective compliance for various orifice geometries is given by Laurens et al. 142.2. 6 The concept of acoustic compliance is introduced in Section 4. 3. it will be referred to as Jing model (see Section 4. He analyti- cally derives an expression of the Rayleigh conductivity for an ori- fice with bias flow and demonstrates the application of his model to a perforated plate with bias flow.4). 219. [285]. 214] presents a model for the unsteady flow through a circular orifice.3. 371] applied to a low Mach number nozzle flow. The model itself is discussed in detail in Section 4. 489].1. Howe [210. Based on the work of Leppington [297] they develop a smooth boundary condition for the perforation in terms of an effective compliance6. 430. 42. 317. The predictions of Howe’s model agree well with the experimental data.7 . when flow discharges through a nozzle. 391. 417. which does not exist in practice. 131. Hughes and Dowling [219] and Dowling and Hughes [129] ap- ply Howe’s approach to a perforated plate with a solid back wall. Inspired by Bechert’s experiments. His approach confirms that the main mechanism of sound absorption.2 sound absorption due to vorticity shedding 47 mentally demonstrate the existence of a new mechanism of sound absorption and was able to give an explanation of the responsible physics. 282. The resulting expression of the Rayleigh con- ductivity is commonly referred to as modified Howe model. 318. stems from the transfer of acoustic energy into vorticity. One of the shortcomings of Howe’s model is the assumption of an infinitesimal thin wall. respectively. Howe’s model has become a quasi-standard when modeling bias flow orifices or perforations [16. The perforation consists of a uniform array of cir- cular orifices or slits. Fol- lowing the notation here. They show theoretically and experimentally that it is possible to absorb all the incident sound energy when a solid back wall is provided.

Eldredge and Dowling [142] apply Howe’s model to a cylin- drical geometry. 452– 455] and used by Rolls-Royce plc [282]. In contrast to the other impedance models mentioned in Section 3. and is then modified to include the bias flow effect. an an- alytical solution can not be obtained. 107] to de- rive a simplified formula for the Rayleigh conductivity of a circu- lar bias flow orifice in an infinitesimally thin wall and in a wall of finite thickness. mass reactance. so that the equations are solved with the boundary element method. Howe’s model [210]. Luong et al. A more comprehensive approach is followed by Bellucci et al. viscosity.48 literature on bias flow liners Later. The Eldredge and Dowling method solves the acoustic equations in the lined sec- tion of a duct. [308] use the Cummings equation [105. nonlinearity. This model will be referred to as Bellucci model and is described in Section 4. 131. However. included in a Low-Order Thermo-Acoustic Network model (LOTAN) developed by Stow and Dowling [130.8. as well as the later adaptations by Jing and Sun [241] and Luong et al.5).2.9. This approach is. orifice interaction. which are coupled to the sound propagation in the cavity via the compliance of the perforation. This extension yields an improved agreement to their measurements. focus exclusively on the bias flow and do not include predictions of an orifice without bias flow. g. the bias flow resistance in this model is based on Howe’s formulation. [308]. for example. This approach is often referred to as simplified Howe model and is labeled Luong model in the notation used here (see Section 4. Jing and Sun [242] include a finite wall thickness and the contraction of the jet in Howe’s equations.1. compared to the modified Howe model. They present an impedance model which takes into account all the usual effects when there is no flow. . [42]. as described in Section 4. e. end correction.

19] have published a series of papers while building up their expertise regarding bias flow liners. they are able to compare the energy found in the unsteady flow field to the lost acoustic energy and find a surprisingly good agreement. and perforation aspect ratio. In a later paper [419]. An- dreini et al. In [417].3 recent developments Based on publications within the last five years. 18. which is representative of a realistic combustor situation. the Howe model (see Section 4. 17. and the Bauer model . these are a 1D network tool based on [142]. Three different models for the perforation are applied. The same setup is used to perform Particle Image Velocime- try (PIV) measurements. the Jing model (see Section 4. Andreini et al.3). orifice angle. Rupp et al. there are several groups doing active research regarding bias flow orifices. [16] compare numerical tools for the evaluation of bias flow liners. e. loughborough university / rolls-royce (uk) Rupp and Carrotte [415] present an experimental study regarding high amplitude effects at a single orifice with and without bias flow. i. The main parameter in the study is the dis- tance between the liners in a double-skin configuration. The airflow through the fuel injector produces a complex flow field at the surface of the liner. university of florence / avio (italy) Andreini et al. but without combustion.4). In particular. 3. and large eddy simulations with the OpenFOAM toolbox. porosity.3 recent developments 49 3. They are able to find an optimum distance with the help of a simple model. [16. bias flow velocity. including frequency. present an experimental study with a model combustor geometry. the FEM solver COMSOL. The results are processed with a Proper Orthogonal Decomposition (POD) technique to identify the co- herent structures of the periodic velocity field. [17] present an experi- mental study regarding various parameters. A full annular geometry is analyzed in [18].

pure bias outflow. . They designed two perfo- rated plates that are tested in an impedance tube as well as in an atmospheric combustion test rig. e. . In [504] they derive an impedance model from a modi- fication of the Cummings equation [105. Scarpato et al. In [70] they make a detailed experimental study of the transition between the dominance of the nonlinear high amplitude effect and the bias flow effect. kth royal institute of technology (sweden) Bodén and Zhou [70] and Zhou and Bodén [504] study the combi- nation of bias flow and high amplitude acoustic excitation. [332] and Tonon et al.6). ecole centrale paris / cnrs (france) Tran et al. 429. the peak absorp- tion frequency can be modified by changing the cavity depth. . [478] present a comprehensive treatment of the grazing and bias flow interaction. low in- flow.50 literature on bias flow liners (see Section 4. 430] present a low Strouhal number analysis and find that the optimal bias flow ve- locity is controlled by the porosity only. eindhoven university of technology (netherlands) Moers et al. They present an ana- lytical model for the inflow regime and present impedance mea- surements for various orifice geometries. i. [481] investigates the use of perforated plates backed by a cavity in a combustion chamber. Scarpato et al. [427] conduct Large Eddy Simulations of an orifice with bias flow at low and high sound levels. different from the steady flow discharge coefficient. pure grazing flow. They discuss the different steady flow regimes. [19] study the cooling effectiveness of the perforation geometries that have been investigated for their acoustic properties in [17]. Then. which were introduced by Baumeister and Rice [35] and Rogers and Hersh [406]. . 107]. They include a new model for the discharge coefficient. which introduces a separate discharge coefficient for the periodic acoustic flow. . In their most recent paper Andreini et al. [428.

More results are announced to be presented in future publications. Haufe et al. are applied to a bias flow liner in [201]. Particle Image Velocimetry (PIV) and Laser Doppler Anemometry (LDA). A third optical technique is used in a cooperation with TU Berlin and TU Dresden. [219]). [55] and Schmidt et al.3 recent developments 51 They achieve a larger absorption bandwidth in the low Strouhal number regime and claim. 192] perform measurements applying Doppler Global Ve- locimetry with Frequency Modulation (FM-DGV). and radius to thickness ratio. The acoustic absorp- tion is measured for various perforation geometries and bias flow velocities at ambient conditions with loudspeaker excitation. In combustion tests. 318] present results from an ongoing numerical study. cambridge university (uk) Bhayaraju et al. Both techniques deliver similar results and are able to resolve the acoustic flow structures in the vicinity of the orifice. orientation. [191. shape. Schulz et al. [434] conduct acoustic mea- surements in a rectangular model combustor. The results reveal a correlation . g. university of dayton (usa) Mazdeh and Kashani [317. they evaluate the influence of the different per- foration geometries on the flame. So far they have presented impedance results while varying the thickness to radius ratio of the orifices. dlr german aerospace center (germany) Two optical measurement techniques. This work is a joint effort with ONERA who performed the LDA measurements. that a low Strouhal number design is generally superior compared to previous efforts looking at the Helmholtz regime (e. 3. [436] investigate the energy transfer with a spectral analysis ap- proach based on FM-DGV data. They are using Large Eddy Simulation (LES) and are working on including geometry parameters like the hole size.

The nature of the work in the references can be mainly theoretical. promises a dramatic reduction of the required mass flow rate while achieving a similar damping performance. and numerical simulations. Heuwinkel et al. All details can be found in the following chapters. respectively. the last columns list some common parameters that are addressed in each publication: frequency f. The table includes not only the perforated liner setup. Helm- holtz resonator. 282] is the basis of this thesis. . wall thickness t. bias flow Mach number Mb . grazing flow Mach number Mg .1. This new concept. A ‘c’ in the experimental column indicates that combustion tests have been performed. [280. A more comprehensive overview is given in Table 3. duct termination. 279. and S (simulation) columns. sudden area expansion in a duct. the Zero Mass Flow Liner. porosity σ. [200] and Lahiri et al. 281] study the effect of an oscillating unsteady bias flow.52 literature on bias flow liners between the absorbed energy and the production of additional spectral turbulence fluctuation components. E (experimental). The unsteady bias flow is gen- erated by a high amplitude acoustic excitation within the cavity of the liner. cavity volume V.4 chronological overview Due to the large amount of publications addressing the bias flow effect. orifice angle α. indicated by the cross in the T (theoret- ical). double- degree-of-freedom (DDOF) liner. orifice diameter d. experimental studies. the previous sections only present a limited selection. The comprehensive experimental parameter study with cylin- drical bias flow liners presented in [277. sound pressure level p0 . but various setups where the bias flow effect can be observed: single orifice. perforated plate. For a quick reference. 3. and tempera- ture T . single-degree-of-freedom (SDOF) liner.

Mb Mg Westervelt (1951) [491] x orifice f Mb σt Barthel (1958) [26] x Helmholtz res. Author T E S Setup Parameters Borth (1916) [73] x orifice Mb Maier (1934) [311] x orifice Mb Lutz (1934) [309] x x orifice Mb McAuliffe (1950) [319] x orifice Mb d McAuliffe (1950) [319] x Helmholtz res. 422] x duct termination f Mb T Sullivan (1982) [461] x perforate f Mb Cummings (1983) [109] x x duct termination f p0 Mb σ continued on the next page . 412] x orifice f Mb Bechert (1979) [37. f p0 Mb Ingard (1959) [226] x perforated plate f Mb Gordon (1965) [180] x orifice Mb σ Mechel (1965) [326] x x duct termination f Mb d Mechel (1965) [325] x orifice f Mb d Utvik (1965) [483. and S indicate if the nature of the work is mainly theoretical. 38] x x duct termination f Mb Howe (1979) [208] x duct termination f Mb Howe (1979) [210] x orifice f Mb d Nilsson (1981) [354] x duct expansion f Mb Rienstra (1981) [400] x duct termination f Mb Salikuddin (1981) [421.4 chronological overview 53 Table 3. 412] x duct expansion f Mb Feder (1969) [158] x perforated plate f p0 Mb Mg σ Garrison (1969) [170. 3. f Mb Alfredson (1971) [11] x x duct expansion f Mb Oberg (1971) [357] c perforated liner Mb σ Cummings (1975) [103] x duct expansion Mb d Dean (1975) [118] x x DDOF liner f Mb σ Munt (1977) [346.1: Chronological overview of publications concerned with the bias flow effect. 399. E. re- spectively. or numerical simulations. A ‘c’ in the experimental column indicates that combustion tests have been performed. 347] x duct termination f Mb Bechert (1977) [39. 224] x duct termination f Mb Richter (1978) [397. 171] x perforated liner f p0 Mb σ Tonon (1970) [479] x Helmholtz res. The columns T. 40] x x duct termination f Mb Imelmann (1978) [223. experimental. 484] x c perforated liner f Mb Mg σ Ingard (1967) [227] x orifice p0 Mb Ronneberger (1967) [409.

134] x duct expansion f Mb Wendoloski (1998) [489] x orifice f Mb d Ahuja (1999) [6. 43] xx perforated liner f p0 Mb σt Bielak (2002) [56] x DDOF liner f p0 Mb Mg T Dupère (2002) [136. 54. 425] x perforated liner f p0 Mb σt V Fukumoto (1991) [168] x perforated plate f Mb Dowling (1992) [129] x perforated liner f Mb Keller (1995) [258] xx Helmholtz res. 412] x orifice f Mb Pallek (1984) [360. f Mb Sun (2002) [463] x orifice f Mb Mg σ t Sun (2002) [463] x perforated plate f Mb Mg σ t Eldredge (2003) [142] xx cylindrical liner f Mb Mg Forster (2003) [164] x perforated plate f Mb σ Forster (2003) [164] x cylindrical liner f Mb Mg d Eldredge (2004) [140] x cylindrical liner f Mb Rademaker (2004) [378] x 3DOF liner f p0 Mb Mg Boij (2005) [71] xx duct expansion Mb Luong (2005) [308] x orifice f Mb dt Heuwinkel (2006) [197. 137] xx Helmholtz res. 412] x duct expansion f Mb Peat (1988) [363] x duct expansion f Mb Hughes (1990) [219] x perforated liner f Mb σ V Salikuddin (1990) [424.1 – continued from the previous page Author T E S Setup Parameters Whiffen (1983) [494] x duct termination Mb Pallek (1984) [360. f Mb Dupere (1998) [133. 198] x cylindrical liner f Mb Mg σ continued on the next page . 162] xx perforated liner f p0 Mb σt Jing (2000) [242] xx perforated plate f Mb σt Betts (2001) [53] xx perforated plate f Mb σ Dupere (2001) [135] x duct expansion f Mb Durrieu (2001) [138] xx orifice f Mb σ Durrieu (2001) [138] xx perforated plate f Mb σ Hofmans (2001) [138. 412] x perforated plate f Mb Pallek (1984) [360. 203] xx orifice f Mb Follett (2001) [161] xx DDOF liner f Mb Bellucci (2002) [41.54 literature on bias flow liners Table 3. 87] x DDOF liner f Mb Mg Jing (1999) [241] x perforated liner f Mb σt V Kwan (1999) [276] x SDOF liner f p0 Mb Mg Premo (1999) [373] xx SDOF liner f p0 Mb σ Zhao (1999) [502] x perforated liner f p0 Mb σ Ahuja (2000) [8] x orifice f Mb Betts (2000) [52.

3. 501] x cylindrical liner f p0 Mb Bhayaraju (2010) [55] c perforated liner f Mb σ α Heuwinkel (2010) [201] x cylindrical liner f Mb Mg Lahiri (2010) [202. 282] x cylindrical liner f Mb Mg σ t V Lei (2010) [294] c perforated liner Mb σ T Rupp (2010) [415.1 – continued from the previous page Author T E S Setup Parameters Macquisten (2006) [310] c cylindrical liner Mb Mg σ T Efraimsson (2007) [139] x orifice f Mb Eldredge (2007) [141] x perforated plate f Mb Mg α Lee (2007) [291] xx perforated plate f Mb σ Leung (2007) [299] x orifice f Mb d Dasse (2008) [112. 430] x perforated liner f Mb σ V Zhong (2012) [503] x cylindrical liner f Mb Mg Andreini (2013) [18] x perforated liner f Mb T Schulz (2013) [436] x perforated liner f Mb Mg Tonon (2013) [478] x slit f Mb Mg α Zhou (2013) [504] x orifice f Mb t Rupp (2014) [420] x Helmholtz res. 417] x perforated liner f Mb V Scarpato (2011) [427] x perforated plate p0 Mb Andreini (2012) [17] x cylindrical liner f Mb σ α Bodén (2012) [70] x orifice f p0 Mb Jayatunga (2012) [240] x perforated liner f Mb σt Jörg (2012) [252] x cylindrical liner f Mb V Lahiri (2012) [279] x cylindrical liner f Mb Mg VT Mazdeh (2012) [318] x orifice f Mb dt Moers (2012) [332] x slit f Mb Mg α Scarpato (2012) [428. 419] x orifice p0 Mb t Schmidt (2010) [434] x perforated liner f Mb Mg σ α V Andreini (2011) [16] x cylindrical liner f Mb Mg σ α Mazdeh (2011) [317] x orifice f Mb Rupp (2011) [416. 418] x orifice p0 Mb Tran (2009) [481] c perforated plate f p0 Mb VT Zhao (2009) [500. 188] x perforated plate f Mb Heuwinkel (2008) [199] x cylindrical liner f p0 Mb Mg Mendez (2009) [328] x perforated plate f Mb Rupp (2009) [415. 328] x perforated plate f Mb Gullaud (2008) [187. f p0 Mb t .4 chronological overview 55 Table 3.

.

the pressure difference across the orifice Re{(p̂1 − p̂2 ) eiωt } produces the fluctuating volume velocity q0 = Re{q̂ eiωt } through the orifice. He introduced the concept of acoustic conductivity as an analogy to Ohm’s law1 in electricity. When the flow is regarded 1 Named after Georg Simon Ohm (1789-1854. Then.1 rayleigh conductivity The first attempt at a theoretical description of the acoustic prop- erties of an orifice was presented by Rayleigh [385. 4 MODELING OF BIAS FLOW LINERS 4. so that p01 = Re{p̂1 eiωt } for the fluc- tuating pressure above and p02 = Re{p̂2 eiωt } for the fluctuating pressure below the orifice. . It is commonly referred to as Rayleigh conductivity2 . the acoustic conductivity of an orifice is given by the ratio of the vol- ume flow through the orifice to the driving pressure difference. While the electric conductivity of a circuit element is described by the ratio of the electric current to the potential difference. 390] when he was studying Helmholtz resonators. English physi- cist). Let the fluctuating pressure on both sides of an orifice be a harmonic function of time. German physicist and mathe- matician) [358]. 2 Named after John William Strutt. Lord Rayleigh (1842-1919.

Ch. Ch. (4.3].3) leff The effective length is longer than the physical length of the orifice l. 5. (4. Morfey [336]. as incompressible and the orifice is acoustically compact. and Howe [214.58 modeling of bias flow liners 2r l0 p̂1 t leff p̂2 v̂ Figure 4. The acoustic volume velocity is de- fined as q̂ = Av̂. then the Rayleigh conductivity is defined as [385. In an ideal fluid the Rayleigh conductivity is determined by the orifice geometry alone. 390] A KR = .4]. The additional length l0 accounts for the additional mass of fluid that takes part in the oscillatory motion outside of the orifice 3 In addition to Rayleigh’s original derivation detailed discussions are given by Stewart and Lindsay [447. (4. 390] KR = 2 r . that is λ  r. 2. For a circular orifice in a wall of infinitesi- mal thickness Rayleigh found that [385.1) p̂1 − p̂2 where ρ is the mean density. For an orifice in a wall of finite thick- ness the Rayleigh conductivity can be expressed as the ratio of the area of the orifice and an effective length [385.1. with the orifice area A = πr2 and the acoustic particle velocity v̂. .1: Illustration of the quantities determining the Rayleigh con- ductivity of an orifice. This is illustrated in Figure 4.2) where r is the orifice radius. 390]3 q̂ KR = iωρ .

5 This configuration is often referred to as flanged pipe.849 r. 100.4) to be in the range 0. 260. This is illustrated in Figure 4. the impedance concept was introduced by Heaviside [193] for treating alternating currents of electricity. The exact value of the end correction has been an ongoing topic of discussion. 4. In a further analysis Rayleigh [385.2) and the end correction can be determined from Equation (4.1.85 r .82 r as the appropriate value for an orifice in an infinite wall. Rayleigh derives an upper and lower limit of the conductivity4 [385. [286]. 4. . but allows a more comprehensive description of a dy- namic system.2 acoustic impedance The concept of acoustic impedance is similar6 to the Rayleigh con- ductivity. how the Rayleigh conductivity is converted into an impedance. In continuation. 390] π r2 π r2 16 < KR < (4. Originally. 390] suggests l0 ≈ 0. The effective length is defined as the sum of the physical length of the orifice and the end correction for both ends: leff = l + 2 l0 . 6 It is shown in Appendix b. (4. 339] 8 l0 = r ≈ 0.2 acoustic impedance 59 and is commonly referred to as end correction. 225.785 r < l0 < 0.5) 3π While the end correction can be neglected for l  r it contributes substantially when l → 0. in contrast to an un- flanged pipe.4) l + 3π r l + π2 r For l = 0 the upper limit coincides with Equation (4.1. 4 A recent mathematical discussion of the limits is given by Laurens et al. A generally accepted value for one end of an orifice in a wall5 is [72.

It describes the opposition of an element to transmit a current when a voltage is applied. (4. z = r + ix . Then. . (4. and ζ = θ + iχ . (4. i. 7 The notation is for an eiωt time dependency. for example a perforated wall.9) where the real part is referred to as resistance and the imaginary part as reactance. For a negative exponent the imaginary part in Equation (4. the impedance is represented by a complex num- ber7 Z = R + iX . e. For plane waves z0 = ρc. (4.6) q̂ Another form of impedance is given by the specific impedance p̂ z= .9) changes sign.7) v̂ The specific impedance is related to the acoustic impedance by Z = z/A and is used to describe the acoustic properties of a medium or material. the normalized specific impedance ζ = z/z0 . the acoustic impedance is defined as the complex ratio of the acoustic pressure to the acoustic volume velocity (Webster [487] or [369]) p̂ Z= . The specific impedance of a fluid is a characteristic quantity of the medium only and is therefore referred to as char- acteristic impedance z0 . A convenient dimensionless quantity is obtained by the ratio of specific and characteristic impedance.8) Mathematically. v̂ is the normal component of the acoustic particle velocity directed into the surface.60 modeling of bias flow liners The electric impedance is defined by the ratio of voltage to current and is measured in Ohm. Accordingly.

(4. (4. Internal Impedance First. 327]. and the compliance C. In a mechanical system the resistance represents a linear. while the reactive elements store energy. g. mass- less.10) The three parameters R. massless spring [156]. 4. Helm- holtz resonator. etc. viscous damper. the process of building up an impedance model of a per- forate is demonstrated.11) σ F(k0s r) . the impedance within an orifice is considered.2 acoustic impedance 61 4.2. and C are commonly referred to as lumped parameters and an element that complies with the acous- tic compact assumption is a lumped element. The acoustic impedance can be written as the sum of these three quantities [260] Z = R + i [ωM − 1/(ωC)] . are much smaller than the relevant acoustic wavelength. its characteristic behavior can be described by three parameters: the resistance R. The normalized specific impedance for a unit length l of an infinite tube filled with a viscous fluid is given by (Crandall [100] or [327]) 1 ikl ζ= . Generally. and the compliance describes the reciprocal of the stiffness of a lumped. the resistance de- scribes the energy dissipation. linear. M. orifice.1 Impedance Modeling of Perforations When the dimensions of an acoustic element.. the inertance corresponds to a lumped mass. without any end effects. Similar. e. the inertance M. Now. either in the form of potential energy in a spring-like ele- ment or kinetic energy in a mass-like element [338]. but more detailed derivation can be found in [271.

A] derives 2 J1 (x) F(x) = 1 − . Depending on the shear number Sh = r ω/ν they yield [100. The Bessel functions with a complex argument produces a com- plex result. so that ks = −iω/ν0 . e. [100. two approximations do exist. 327]: 8νl 1   k ζ≈ +i l+ l for Sh < 2 (4. e.62 modeling of bias flow liners The division by the porosity σ converts the impedance of a single orifice into the impedance of an array of orifices in a perforation8 . 9 Named after Friedrich Wilhelm Bessel (1784-1846.4. this was not done traditionally.11). which are more p physically revealing. and Kirchhoff [261]. Sometimes the heat conduction is 0 neglected entirely. e. Based on the theories given by Stokes [451]. In order to include both effects p an effective kinematic viscosity ν0 = 2. Crandall [100.13) cσr2 σ 3 √ √ 2ωνl 2ωνl   kl ζ≈ +i + for Sh > 10 (4. g. However. 77]. so that the impedance consists of a reactive and re- sistive part. Irish mathematician and physicist.12) x J0 (x) where J0 and J1 are Bessel functions9 of the first kind. (4. [3. Unfortunately. ikl is the inviscid impedance of an infinite tube. It describes the inertia of the oscillating mass of air.1. Helmholtz [195].14) cσr σ cσr 8 Often. more physical insight is not provided by Equation (4. then ν0 is replaced by ν.179ν is used. 327]. 10 Named after Sir George Gabriel Stokes (1819-1903. g. . The effective Stokes wave number10 k0s considers the viscosity as well as ther- mal conductivity losses near a highly conducting wall. They are a standard mathematical tool when dealing with cylindrical geometries. [271]. the impedance is divided also by the discharge coefficient. which are presented in Sec- tion 5. The function F introduces the viscous effects. German mathematician and astronomer). Bessel functions are the solutions to the Bessel differential equation [51]. Appx. However.g.

Besides the obvious introduction of a re- sistance term.11) yields [327. e. that the Stokes wave number external to the orifice considers the viscosity only. Similar to the mass end correc- tion. the additional resistance is included by extending the length of the orifice (now in the resistance term). Resistance End Correction In the idealization of the effective length the orifice is virtually extended. keeping the same cylindrical shape. ks = −iω/ν. that the effective length which needs to be accounted for in the inertance is longer than the actual length of the orifice. [260]. (4. 4. This additional length is defined by the reactive part of the radiation impedance of a circular piston of air in an infinite baffle. Following the discussion of Melling [327]. Applying both end corrections to Equation (4.1. 441] ik 16r/(3π)   l ζ= + .2 acoustic impedance 63 The resistance terms in Equations (4. Sivian finds good agree- ment with experiments when using the same additional length as for the reactance. [260]). g. but he justi- fies the additional resistance with friction losses in the shear layer of the emerging jet. A very similar term is presented by Sivian [441].13) and (4. This is considered by In- gard [225] and he introduces an additional resistive end correction to account for the friction losses on the wall surrounding the ori- fice. Mass End Correction (Radiation Reactance) It was already discussed in Section 4. the viscosity also affects the mass reactance. A typical value considering both ends is 2l0 = 16r/(3π). respectively.and Helmholtz-type losses within a tube (e. p . In reality the air is drawn from the radial direction as well.14) represent the well known Poiseuille.15) σ F(k0s r) F(ks r) Please note. g. e. i. which is increased from its inviscid value. Si- vian’s expression will be used here.

The term above was used by Guess [186] and Bellucci et al.2: Orifice interaction factor ψ plotted over the porosity accord- ing to Ingard (Equation (4.16)) and Fok (Equation (4. a4 = 0 . a1 = −1.17) n=0 a0 = 1 . a7 = 0. a3 = 0. a8 = −0.64 modeling of bias flow liners 1 Interaction Factor ψ 0.03015 .15 0.2 0 0. i.06793 . .33818 . a2 = 0 .02287 .01614 .2 Porosity σ Figure 4.05 0.17)).8 0.4 Fok Function Ingard (1953) 0.1 0. Melling [327] presents a comprehensive discussion on the topic and refers to a solution that was found by Fok [160] 8 X √ n ψ(σ) = an σ . (4.16) 2 The correction factor ψ(σ) reduces the end correction with increas- ing porosity.4092 . with (4. e.6 0. [41]. with decreasing orifice distance. for example. Orifice Interaction Ingard [225] studied the interaction effect of two adjacent orifices and introduces a correction factor for the end correction r σ ψ(σ) = 1 − . a5 = 0. a6 = −0.

For example. Melling defines the Fok function as in Equation (4. describing the nonlinear behavior12 : 1.17). Eq.2. the nonlinearity will affect the reactance as well [143. e. The correct notation was used by Randeberg [383] and the mismatch was also pointed out by Elnady [143]. 479. in Fig. instead of multiplying it as shown in Equation (4. which was neglected by Melling. 327. Applying the correction for orifice interaction to Equation (4. 186]. However. i. 54.17) over the porosity. This is clearly not correct. but divides the end correction by ψ. 271. Melling’s erroneous formula had been reproduced by others. 441.18) will be referred to as Melling model.19) 2c (σCd )2 rms 11 Please note. Various approaches have been followed to include this effect in the impedance formulation [72.2 plots Equations (4. The Fok function predicts a stronger influence of the interaction than Ingard’s approach and will be used here. it seems that Melling uses the Fok function as it would be defined as the reciprocal of Equation (4. (4.18).2 1 − σ2 0 θnl = v .2 acoustic impedance 65 Figure 4. Melling plots 1/ψ. . Melling [327] states that the interaction effects can be neglected for porosities < 4 %. [52. 225.16) and (4. 4. 228. 507]. 6 of [327]. [271] seem to use it correctly). 18]11 ik 16r/(3π)   l ζMelling = + ψ . As shown in Figure 4. For example.17). Unfortunately. 227. (4. the Fok function is smaller than unity and thus needs to be multiplied to achieve the desired effect. 12 Generally. 441]. g. Melling [327] derives a resistance term that depends on the velocity within the orifice. but labels it with ψ. 288. e. the impedance depends on the ampli- tude of the oscillation.15) yields [327. 505. that there is a mismatch in the handling of the Fok function in Melling’s paper [327]. Nonlinear Resistance At high amplitudes the acoustic behavior of the orifice becomes nonlinear [327. 472] (though Kraft et al.18) σ F(Sh 0 ) F(Sh) Equation (4.

20) 2cσCd ρc Including the nonlinear resistance term in Equation (4. (4. . However. 384] or [271]) 1   J1 (2kr) θrad = 1− . The acoustic particle velocity can be determined itera- √ tively from |v0 | = |p0 |/(ρc θ2 + χ2 ) [143].22) can be approximated by [271] (kr)2 θrad ≈ (4. (4.22) or (4.22) σ kr For small kr Equation (4.18) yields the nonlinear Melling model ik 16r/(3π) 1. a first approx- imation of the acoustic particle velocity can be given from the applied sound pressure level SPL by [54] 1 pref 10SPL/20 v0rms = . The difficulty is.21) σ F(Sh0 ) F(Sh) 2c (σCd )2 rms Radiation Resistance Another contribution that was neglected by Melling is given by the resistive part of the radiation impedance13 . similarly to θnl above.2 1 − σ2 0   l ζMelling. p. (4.23) 2σ Equation (4.nl = + ψ + v . that the impedance now depends on the acoustic particle velocity.23) can be simply added to the impedance. So it is not possible to directly calculate the impedance when the acoustic pressure is known. It accounts for the acoustic losses by radiation into the surrounding medium. 13 The reactive part of the radiation impedance is already included by the mass end correction as discussed above.66 modeling of bias flow liners where v0rms is the root-mean-square value of the acoustic veloc- ity within the orifice. The radiation resistance for an array of circular orifices is (Morse and Ingard [340.

some variations can be found in the literature. Especially the importance of the friction velocity. That is the case for the measurements presented here. 290]. e. 194.2. 4. Peat et al. 505].24) σ More sophisticated models are available. . It should be noted.3 Bias Flow Impedance Traditionally. However. rather than the mean flow velocity. However.3 Mg θg = (4. 266. C is a constant that is based on empiricism and 14 The factor of 0. 209. 194]. 459.2. 181. In the first approach the acoustic particle velocity in a nonlinear model is replaced by the bias flow velocity [34. 118. 244. 392]14 0. 186. For a simple linear relationship of resistance and bias flow Mach number given by Sivian [441] this yields C · Mb θb = (4. This replacement is only reasonable if the bias flow ve- locity is assumed to be much greater than the acoustic particle velocity. so that the use of the mean velocity is adequate and the simple model of Equation (4. there are two alternatives to include the influence of a steady bias flow on the impedance of an orifice or perforation. 326. has been studied [175. [364] conclude that the influence of the friction velocity is of no significance when the boundary layer is turbulent and fully developed.2 acoustic impedance 67 4. 4.24) is sufficient. 227.3 is most common. for example Rice [392] suggests 0. 171.24) neglects the reduction of the attached mass by the grazing flow.5.24). [106. g. that Equation (4.25) σ and is very similar to the grazing flow resistance introduced in Equation (4.2 Grazing Flow Impedance The effect of grazing flow on the impedance is often accounted for by a simple contribution to the resistance [34.

15 v0rms )2 + (2 Ub )2 . 4. typically 1.19) yields 1 1 − σ2 θb = |1. Betts [52. In particular.2 v0rms + 2 Ub | (4. Jing and Sun [242] include a finite thickness and allow a contraction of the vortex sheet. [291] determine the impedance of a perforated plate by solv- ing the incompressible Euler equation. 214] analytically derives a formulation of the Rayleigh conductivity for an orifice with a bias flow and includes the ab- 15 This value is often accredited to Bauer [34]. 54] finds the similar expres- sion |1.3 below).2 v0rms + 2 Ub |. Jing and Sun [242] solve Howe’s [210] governing equa- tions with a boundary element method.15 [34]15 . Lee et al. both need to be solved numerically. they consider the interaction effect. However. . Howe himself obtains an analytical solution when assuming infinitesimal thickness and a cylindrical vortex sheet (see Section 4.5. For a steady bias flow Cummings and Eversman [109] derive C = (1 − σ2 C2c )/C2c from the linearized Bernoulli equation.68 modeling of bias flow liners is in the range 1 − 1. See as well the discussion in Section 4. [291]. Bauer himself refer- ences Zinn [505]. Premo [373] suggests a combined velocity term of (1. More advanced approaches are presented by Jing and Sun [242] and Lee et al. where Cc is the contraction co- efficient of the jet. use this term to describe high amplitude effects. which requires a numerical solution of the equations. so that they do not provide a simple parametric description of the impedance.26) 2c (σCd )2 Both approaches are mostly based on empiricism and lack the influence of the bias flow on the reactance. However. not a steady bias flow. Adding this to Equation (4. The second approach combines the effects of the acoustic par- ticle velocity and the steady bias flow velocity in one resistance term.6. Both. p For example.3 howe rayleigh conductivity model Howe [210. Zinn [505] and Bauer [34].

The extreme values of the parameter study here (T = 823. The convection velocity of the vortex rings is taken to be the mean velocity within the orifice. He treats a single. The wall is infinitely thin: t = 0 After a lengthy derivation Howe arrives at [210. On the inlet side the flow field resembles that of a potential sink at the center of the orifice and on the outlet side that of an axisymmetric jet flow with a potential core and shear layer. low pressure.15 K. 19 Here. . The radius of the vortex rings coincides with the orifice ra- dius and remains constant when traveling downstream. The Mach number of the bias flow is low. 18 The highest limit is given at high temperature. Mb < 0. 214]19 KR = 2r(γ + iδ) (4.5 mm) restrict the applicability of Howe’s model to Mb  0.3. The assumptions can be summarized: 1. so that the wavelength is much larger than the orifice radius (the orifice is acoustically compact)16 : λ  r 2.3 howe rayleigh conductivity model 69 sorption caused by vorticity shedding. Viscosity is neglected except at the rims of the orifice. 4. forming a cylindrical vortex sheet. and small radius.0003. Higher temperatures and smaller radii in- creases the limit further. The frequency of the sound is low.27) 16 For the largest orifice dimension treated in the parameter study here (r = 1. r = 0. p = 101. where the acoustic perturbations trigger the periodical shedding of vortex rings. so that the fluid can be considered incompressible17 : Mb  1 3. The Reynolds number of the bias flow is high. this yields a frequency limit of f  272360 Hz at standard tem- perature and pressure conditions. circular orifice in a wall of infinitesimal thickness.25 mm). so that viscos- ity is only considered at the rims of the orifice18 : Ub r/ν  1 4.325 kPa. 17 Typically. the sign is adjusted to the eiωt convention.

which is.4 0.28) St 2 I1 (St) e π − iK1 (St) cosh(St)  −St where St is the Strouhal number. while δ is a resistance term responsible for the acoustic absorption. the vorticity shedding has a negligible influence and δ reverts to its value in the absence of the flow.8 0. e. Equation (4. The maximum of the imaginary part is found at a Strouhal number just beyond unity. Figure 4. respec- tively.70 modeling of bias flow liners 1 γ Rayleigh Conductivity/2r 0. assumed identical to the mean velocity within the orifice Ub . with 2 I1 (St) e π −St + iK (St) sinh(St) 1 γ + iδ = 1 + . The Strouhal number is based on the vorticity convection velocity. . γ represents the inertia of the orifice. (4.3 plots the real and imaginary part of Equation 4.6 0. At very high Strouhal numbers δ → 0.3: Real part γ and imaginary part δ of the normalized Rayleigh conductivity plotted over the Strouhal number.28 over the Strouhal number.2 δ 0 0 2 4 6 8 10 Strouhal Number Figure 4. and Im and Km are the modi- fied Bessel functions of order m of first and second kind. according to Howe [210].28) is a function of the Strouhal number only. i. and the orifice radius St = ω r/Ub .

142. (4. 318. Their argumentation is expressed in terms of impedance. This assumption does not hold for any practical application. A modified Howe model that includes a finite thickness is proposed by Jing and Sun [241] (see below).1. The δ term represents the acoustic resistance due to the bias flow. 417. 282.31) 2 (γ2 + δ2 ) 2 (γ2 + δ2 ) where viscosity effects have been neglected. 131. 391. 42. 489]. 430. (4. The relation between the normalized specific impe- dance ζ and the Rayleigh conductivity is given by20 ikA ζ= .29) KR For a circular orifice A = πr2 with bias flow KR = 2r(γ + iδ) and after splitting the real and imaginary parts by multiplication with (γ − iδ)/(γ − iδ) this results in kπrδ kπrγ ζ= 2 2 +i .4 jing model (modified howe model) Jing and Sun [241] propose a way to include the wall thickness in Howe’s model. the γ term describes the end correction when a bias flow is present.4 jing model (modified howe model) 71 Howe’s model has become a quasi-standard describing the bias flow effect at an orifice [16. 4. However. Howe’s model assumes an infinitely thin wall. . Most references listed above resort to the modified version to include the thickness. 4. (4. 219. The thickness 20 This relation is derived in Appendix b. and the t term adds the mass inertance of the fluid within the orifice. 317.30) 2 (γ + δ ) 2 (γ2 + δ2 ) The thickness adds a mass inertance to the system so that the impedance for a circular orifice of finite thickness and with bias flow yields [241]   kπrδ kπrγ ζJing = +i + kt .

54 4 1.13-2. the modified Howe model agrees well with the experimental data. it exhibits large scatter which makes a comparison dif- ficult. . while the measurements of Eldredge and Andreini include a cylindrical liner. Table 4. Another comparison is presented by Macquisten et al.32) has been compared to experi- mental data by several authors.005-0.13 0. Jing and Scarpato study a perforated screen with normal sound incidence.2-3 3 2. [282] is presented in Chapter 7.3 d mm 2-5 0.02 n/a σ % 1. so that these parameters are not listed in the ta- ble. The data from Lahiri et al. (4. The conductivities need to be added reciprocally 1/KR = 1/K1 + 1/K2 . 430] f Hz 300-1400 100-700 200-1500 100-2000 SPL dB 90-125 90-120 n/a 100 Mb – 0.045 0. While the data is a valuable contribution to demonstrate the behavior at combustion conditions.0. [310] for measurements in a combustion test rig. so that the Rayleigh conductivity for a circular orifice of finite thickness with bias flow yields 1 2t −1   KR = 2r + . Jing Eldredge Andreini Scarpato [241.057 0.75 0.72 modeling of bias flow liners Table 4.6-1 t mm 0.32) γ + iδ πr The outcome of Equation (4.1 gives an overview of the parameter ranges from these measurements.5 1 α deg 90 45 30-90 90 term was simply added to the impedance.1: Parameter range of experimental data that has been compared to the modified Howe model.005-0.07 0. 242] [142] [17] [428.004-0.03-0.045 Mg – n/a 0-0. Generally.8 0.16 1-15.

18.33).4. 143.3)) and Ub is the mean velocity in the plane of the ori- fice. The principal effect of nonlinearity is expressed in a small reduction of the steady bias flow velocity. Including a finite thickness they arrive at ! A ωleff /Ub KR = (4.33) is a good approximation for the linear and nonlinear regime. They modify the Cummings equation using the assumption that the steady pressure drop across the orifice is much larger than the acoustic pressure. He compiles a resistance expression from 21 For an infinitesimal thin wall leff represents the end correction only. the bias flow term was included to describe nonlinear effects at high amplitudes. This is demonstrated in Figure 4. when flow reversal does not oc- cur. However. in particular a reversal of the flow within the orifice against bias flow direction is excluded.6 bauer impedance model Bauer [34] presents an impedance model that includes the effect of bias22 and grazing flow. state that Equation (4. [12. 4. where both results are compared. Lu- ong et al. Luong et al. From comparison with experiments. in that case leff can be replaced by πr/2 in Equation (4. 107] to de- rive a simplified formula for the Rayleigh conductivity of a circu- lar bias flow orifice. derive the Rayleigh conductivity for an infinitesimal thin wall21 . Cummings [107] recom- mends a contraction coefficient of Cc ≈ 0. 363] . [308] use the Cummings equation [105. 4.33) leff ωleff /Ub + Ci2 c where A/leff is the Rayleigh conductivity without flow (see Equa- tion (4. Luong et al. cf. [308] find that the agreement between Howe’s linear theory [210] and their approach is best for this particular choice of Cc .5 luong model (simplified howe model) 73 4. it has become common practice to use this term for the steady bias flow and negelect nonlinear effects. 14.5 luong model (simplified howe model) Luong et al. 22 Originally.75.

25 in the reactance term was found to match well when a grazing flow is present.3 Mg 1.74 modeling of bias flow liners 1 Rayleigh Conductivity/2r Howe (1979) 0. and an expression for the reactance proposed by Rice [394] and Dean [117] based on experimental data.25 is used even without grazing flow. σ The factor of 0.8 γ Luong et al.34) k (t + 0.16 as result. However. 117]. (2005) 0. Bauer [34] does not give any guidance on the actual value without grazing flow.25 d) +i . Zinn actu- ally gives 1.4: Comparison of the Rayleigh conductivity derived by Howe [210] and Luong et al. The bias flow term is originally derived by Zinn [505].75. Commonly .2 δ 0 0 2 4 6 8 10 Strouhal Number Figure 4. an empirical bias flow term given by Zinn [505]. the viscous term of Ingard [225].61. The factor 1. an empirical grazing flow term from Dean [117].4 0.1406 where Cd = 0.15 results from 4/(3πC2d ) = 1. The resulting normalized specific impedance yields √ 8µρω 0. [308] for an infinitesimal thin wall and Cc = 0. The differences in the second digit after the decimal point are probably due to rounding errors.6 0. so that 0. but seems to be somewhat higher without grazing flow [34.15 Mb   t ζBauer = 1+ + + ρcσ d σ σ (4.

as was done by others [14. 54] is based on Equa- tion (4. 143. As a result half the Poiseuille viscous term is added to the Helmholtz approximation. so that it does not need to be divided by Cd as was done in [14.7 betts impedance model 75 1. However. However. 18.15 is used. dividing by the square of the porosity produces unreasonable results.19). Bauer [34] divides the bias flow term by σ2 instead of σ as in Equation (4. e. i. the dependency on the discharge coefficient is already included in this value. 143.34). More importantly. Equation (4. 4. Betts modifies the non- linear term to include the bias flow as shown in Equation (4. Then. he does not use the exact solution of Equation (4. so that the division is done with σ only. 290]. 4. so that23. 18. 290].24 √ 4νl 2ωνl 1 − σ2 .7 betts impedance model The impedance model proposed by Betts [52.18) with the extension for nonlinear effects proposed by Melling [327]. but a combination of the low and high shear number approxima- tions.18).26). The approximations are combined in order to obtain an intu- itive impedance formulation without Bessel functions that ’works’ for all frequencies.

.

eff .

.

ζBetts = + + .

Mb .

. 54] and as well Kraft et al. However.35). not divided by ψ (please see footnote 11 on page 65). 54]. cσCd r2 cσCd r σCd √ (4. σCd cσCd r 3π ψ 23 While Melling [327] still includes the thermal conductivity losses inside the orifice with an effective viscosity (see discussion on page 62). Betts [52. The error is kept for consistency with [52. If this modification was made intentionally or unconsciously remains unclear. Anyhow. The end cor- rection needs to be multiplied by ψ. For consistency with [52. 24 Please note that the use of ψ is not correct in Equation (4. the differences are small for the porosities considered here. 54] the regular viscosity is used here as well.35) 2ωνl 16 r   kl +i + + . Melling’s notation might be mis- leading as he refers to the regular viscosity as µ0 and the effective viscosity as µ. [271] only consider the regular viscosity when they present the approximations.

2. However.24 − 1.8 bellucci impedance model The impedance model of Bellucci et al. 4.2: Parameter range of experimental data from Betts [52. [41. f Hz 1000 − 3000 SPL dB 120 − 140 Mb – 0 − 0. The parameters of the study are listed in Table 4.5 d mm 0. The impedance model and the measured data show a generally good agreement for these parameters. they do not consider the thermal conductivity losses. 54] that has been compared to the Betts impedance model.51 − 1.9 − 16.76 modeling of bias flow liners Table 4. He compares the model to impedance measurements of a perforated plate with normal sound incidence.11). 43] is based on Cran- dall’s impedance model given in Equation (4.48 t mm 0.0175 σ % 0.02 dc mm 272 where Meff b is the effective bias flow Mach number considering the acoustically induced flow as well as the steady bias flow [54] √ ! 0 10SPL/20 2 1 p Meff b = ref + Mb (4. so that the nor- .36) 2cσCd ρc Betts calls this model the Perforate Bias Flow Intermediate Fre- quency (PBFIF) model.

41) 0. 401] (4.42) 6.0/St2 ) + 1 l0b = from [242. It should be Γ = [. (4. the bias flow l0b . (4.12).38) here. i. [41]. so 25 Please note that there is a spelling mistake in Eq. 366.39) The individual terms. con- sisting of contributions from the radiation reactance l0rad .8 bellucci impedance model 77 malized specific impedance of a perforation without end correc- tions and without flow yields25 1 ikl ζ= . are given by: (0.3(6. Including both ends this yields l0Bellucci = l0rad · l0int · l0b · l0nl .37) for large shear numbers √   √  1 2 2  ζ = iklΓ with Γ = 1+ −i . ]−1 (see [327]). Equation (4. compiled from various publications.8216 r 1 + from [356] (4.77 He)2 −1   l0rad = 2 · 0. 19 of reference [41]. is correct. . However.38) σ Sh Sh Several terms are added to include the effect of the bias flow and the end corrections.37) σ F(ks r) where the function F is defined in Equation (4. e.43) The references are the ones given by Bellucci et al. They introduce an end correction length. and nonlinear effects of high amplitudes l0nl .3/St0.6 ac from [366] (4. . (4. Bellucci et al. 4. .0/St2 + 1 l0nl = 1 − 0.77 He l0int = 1 − σ/2 p from [225] (4.40) 1 + 0. use the common approximation of Equation (4. 18 of reference [41]. The Helm- holtz and Strouhal numbers are based on the orifice radius. the approximate expression for Γ given in Eq. the ori- fice interaction l0int .

49)  4 1 − 3π In Equation 4. so that the additional radius term needs to be dropped.46)       3π    1/3   0.28).45) |x| if |x| > 1 The term ξ in Equation (4.44) cσ |v̂| where G(x) is given by Keller and Zauner [258] as   √  2 x · arcsin(x) + 1−x2 (2 + x2 )  3 if |x| 6 1 G(x) = π (4. (4.44) to (4.49. Stac 6 (Stac )qs (4.61/C6d . ξ0 refers to Equations 4. and Stac = ωr/|v̂|. Stac > (Stac )qs (4. 29 of reference [41].78 modeling of bias flow liners that He = kr. See Equation 4. The coefficients for l0b and l0nl have been fitted to their experimental data. γ and δ are.48.50)  ζBellucci = G cσ |v̂| σ 26 Please note that there is a spelling mistake in Eq. Adding Equations (4. . the real and imaginary part of Howe’s Rayleigh conductivity. [41] regarding the bias flow effect and the nonlinearity due to high amplitudes yields   ξ Ub θBellucci = G |v̂| .47.5 Stac if Ub = 0. The limit of the quasi-steady assumption for the acoustic Strouhal number is given by (Stac )qs = 0. St = ωr/Ub . (4. de- fined in Equation (4.48)   2 γ2 + δ2      4     ξ 0 (1 − G) + ξ b G −   3π if 0 < Ub < |v̂| (4. depending on Stac . respectively. The radius is already contained in the Strouhal number.47)   4   ξ= π δ St   if Ub > |v̂| (4.39) and (4. The resistance term of Bellucci et al.37) yields ik   ξ Ub |v̂| + lΓ + l0Bellucci . and ξb refers to Equation 4.48.44) is evaluated depending on the ratio of bias flow velocity to acoustic particle velocity26 :   1/C2d if Ub = 0.46 or 4.

[41] that has been compared to the Bellucci impedance model. Bellucci et al. These type of liners are commonly called liners with extended reaction or non- locally-reacting liners. f Hz 50 − 600 SPL dB n/a Mb – 0 − 0.3.9 application to a cylindrical geometry 79 Table 4.023 σ % 1. How- ever. while Bellucci et al. 4. [41] also included the cavity reactance.3: Parameter range of experimental data from Bellucci et al.50) describes the impedance of the perforation only.9 application to a cylindrical geometry It is straightforward to calculate the plane wave reflection and ab- sorption coefficients of a perforated liner with normal sound inci- dence when its impedance is given (see [260].51) 2σCd ρc Bellucci et al. [41] present a comparison of the model with exper- imental data from measurements of the reflection coefficient of a perforated plate with normal sound incidence. Here. e. for example).8 t mm 1. [41] determine the acoustic particle velocity |v̂| iteratively with a Newton-Raphson method.5 − 43 Equation (4. the sound prop- agation within the cavity needs to be considered.03 − 2. [54] √ ! 1 p0ref 10SPL/20 2 |v̂| = . The impedance model and the measured data show a good agreement for these parameters. 4.31 d mm 4 − 13. with grazing sound incidence. the particle ve- locity is approximated as suggested by Betts et al. The parameters of the study are listed in Table 4. i. when the same liner with an unpartitioned cavity is mounted as a sidewall. (4. That is in contrast to a locally-reacting sur- .

linear dynamic systems.1 Transfer Matrix Method The transfer matrix method is a convenient approach for model- ing one-dimensional. where the acoustic particle velocity at one point depends on the acoustic pressure at only the same point [369].9. For a non- locally-reacting liner the acoustic field inside the duct and the acoustic field inside the cavity need to be modeled. 452. Two methods are presented that follow this approach for a cylindrical geometry.52) q̂1 q̂2 T21 T22 The individual terms are [343]: . 455]. It is especially popular in muffler modeling [343] and as well widely used for the model- ing of thermoacoustic effects in gas turbine combustors [44. while they are coupled via the impedance boundary of the perforated liner in-between. commonly the acoustic pressure p̂ and the acoustic volume veloc- ity q̂. 4. 362. A transfer matrix T relates the input and output state variables.80 modeling of bias flow liners face. it has become a standard tool in acoustics (Igarashi and Toyama [222] or [323. For a two-port system [343]: T11 T12       =T 2 T= p̂1 p̂ with (4. With its origins in electrical circuit theory. 343]).

.

.

.

p̂1 .

.

p̂1 .

.

q̂1 .

.

q̂1

T11 =

T12 =

T21 =

T22 =

p̂2 q̂2 =0 q̂2 p̂2 =0 p̂2 q̂2 =0 q̂2 p̂2 =0

The transfer matrix of a network of elements is obtained by multi-
plication of the individual transfer matrices of all the elements of
the network. For example, the transfer matrix T of a network of n
elements, each represented by a transfer matrix Tn , is given by

T = T1 · T2 · . . . · Tn . (4.53)

where x = 0.54) T21 = (i/Z0 ) sin(kL) e−iMkL T22 = cos(kL) e−iMkL with the convective wave number k = k0 /(1 − M2 ). Z0 is the char- acteristic acoustic impedance of a duct element.5. Helm- holtz resonator [323. 343] T11 = cos(kL) e−iMkL T12 = iZ0 sin(kL) e−iMkL (4. e. (4. Uniform Tube The transfer matrix method is demonstrated with a simple uni- form tube. where x = L.9 application to a cylindrical geometry 81 p̂1 . to position 2. The uniform tube setup is illustrated in Figure 4. 4. q̂1 p̂2 . 342]. The literature provides transfer matrices for many common duct elements.55) q̂ Av̂ A .5: Transfer matrix representation of a uniform tube. g. uniform tube. sudden area change. along the uniform tube with constant flow Mach number M is given by the elements [323. q̂2 M 1 D 2 x=0 L x=L Figure 4. For plane waves p̂ p̂ ρc Z0 = = = . orifice. The transfer matrix from position 1.

or in muffler terminology called concentric-tube resonator. the equations are coupled by the connection of the duct and the cav- ity through the perforated wall. The sound field in the duct and in the annular cavity are described by their respective differential equation. was presented by Sullivan and Crocker [460] and Sullivan [458.57) . The transfer matrix is given by T = A(x = 0) A−1 (x = L). Unfortunately. (4. The set of coupled differential equations for the setup illus- trated in Figure 4.82 modeling of bias flow liners Concentric-Tube Resonator A description of the transfer matrix for a cylindrical perforated liner. A condition which is unlikely in most applications. The state of the art approach is described in [323. [345] and Peat [363]. where Peat claims that his solution is more stable numerically. 343] and will be summarized here. A decoupling approach that yields a closed form so- lution of the coupled differential equations was presented by Ja- yaraman and Yam [239].6a can be described by the 4 × 4 transfer matrix relation p̂1 (0) p̂1 (L) T11 T12 T13 T14       p̂2 (0) p̂2 (L) T21 T22 T23 T24  q̂1 (0) = T q̂1 (L) with T = T31 T32 T33 T34  . so that the solution is much more compact. However. This approach overcomes the necessity of forming small control volumes. but it requires relatively high computing effort and might lead to numerical instabilities at high porosities. re- spectively. 459]. (4. These shortcom- ing have been solved in the formulations by Munjal et al.56)       q̂2 (0) q̂2 (L) T41 T42 T43 T44 where the indexes 1 and 2 refer to the inner and outer tube. According to Peat [363] this approach is very flexible in the rep- resentation of different geometries. the decoupling requires for the grazing flow Mach number within the duct and the cavity to be equal. Sullivan used a segmentation ap- proach by dividing the geometry into many small control volumes.

9 application to a cylindrical geometry 83 x=0 x=L p̂2 (x).59)  1 0 0 0  0 1 0 0 where  2 ka + k20 k2a  2 iM1 iM1 ka − k20   α1 = − .n eβn x (4.58) A3. A2. The elements of matrix A(x) are A1. 1 − M21 k0 1 − M21 1 − M21 k0  2 kb − k20  2 kb − k20  2 ka − k20 iM2    α4 = − . ψ and β are respectively the eigenmatrix and eigenvector of the matrix −α1 −α3 −α2 −α4   −α5 −α7 −α6 −α8   . 1 − M21 1 − M22 k0 1 − M22  2 iM2 kb + k20 k2b  α7 = . (4. 2.n = ψ4. eβ n x ψ2.n eβn x . 4.n eβn x . 3. α2 = . coupled via a perforation. α6 = − . q̂1 (x) D2 2 L2 D1 D2 p̂2 (x). 4. and A4.n = − . α5 = . α3 = .n = − .n = ψ3. q̂2 (x) (a) Two concentric tubes (b) Concentric-tube resonator.60) 1 − M22 k0 1 − M22 . Figure 4. ik0 + M1 βn ik0 + M2 βn with n = 1.6: Transfer matrix representation of a concentric-tube resonator. α 8 = (4. q̂2 (x) M L1 1 L D1 p̂1 (x).

61). k2b = k20 − . The equations above include the convective effect of a grazing flow. X1 = −i tan (k0 L1 ) .d indexing scheme has been used. there is no grazing flow within the cavity. The acoustic be- havior of the perforation is represented by the impedance term ζ in Equation (4.61). .  where T12 .6b. The remaining quantities are given by A1 = X1 T22 − T42 /F . The setup of a concentric- tube resonator. T32 . as illustrated in Figure 4. Tb = T14 + B1 A2  Z0 (4.56) and (4.c. so that it can be represented by the 2 × 2 transfer matrix27 Ta Tb       p̂1 (0) p̂2 (L) =T with T = (4. Commonly. B1 = X1 T24 − T44 /F .62) v̂1 (0) v̂2 (L) Tc Td with Ta = T12 + A1 A2 . X2 = i tan (k0 L2 ) . 27 To avoid confusion between the elements of the matrices in Equations (4. (4.b.62).56). B2 = T31 + X2 T33 . All elements of the transfer matrix in Equation (4.61) c D1 ζ (D22 − D21 ) ζ and ζ is the normalized specific impedance of the perforation.  The transfer matrix in Equation (4. so that M2 = 0.   A2 = T11 + X2 T13 .84 modeling of bias flow liners with ω 4ik0 4ik0 D1 k0 = . T14 . k2a = k20 − .62) describes the relation of acoustic pressure and velocity when a sound wave is traveling from position 1 to 2 in Figure 4. and F = T41 + X2 T43 − X1 T21 + X2 T23 .56) can be cal- culated from Equations (4.6b. and T34 are elements of the matrix in Equa- tion (4. an a.58) to (4. introduces hard-wall boundary conditions at both ends of the cavity. and Td = T34 + B1 B2 .63) Tc = T32 + A1 B2 /Z0 .

For example. or insertion loss. level difference.9 application to a cylindrical geometry 85 Acoustic Performance Criteria Traditionally. the performance of a muffler is evaluated by its transmission loss. the transmission loss28 can be calculated directly from the transfer matrix elements as [323] T Zn 1 + M1 . 4.

.

T11 + Z12n + T21 Z1 + T22 Zn1 .

.

Z r .

.

! TL = 20 lg .

.64) Z1 1 + Mn . (4.

2 .

.

Here. This will be indicated. where the Transfer Matrix Method is used in combination with the Bauer impedance model. For a plane wave. 214] B0 = p0 /ρ + vv0 . short TMM.2 Eldredge & Dowling Method Eldredge and Dowling [142] develop equations for the stagnation enthalpy29 and the acoustic particle velocity in the lined section 28 The transmission loss describes the difference between the incident power and transmitted power in dB. the dissipation coefficient can be calculated from the reflection and transmission coefficients given by the scattering matrix according to Section 6. for example. The acoustic properties of the liner are included in Equation (4.2. the fluctuating stagnation enthalpy yields B̂± = p̂ ρ (1 ± M). This is demonstrated in Appendix b. where the acoustic pressure and velocity are related via the impedance ρc = p̂/v̂. Then. by TMM:Bauer. the acoustic performance is evaluated with the dissipa- tion coefficient.4.9. where the index 1 refers to the first and the index n to the last ele- ment of the network and the terminations are assumed anechoic. the transfer matrix is converted into a scattering matrix. 4.61) and can be represented by an impedance model of choice. Therefore. . 29 The fluctuating stagnation enthalpy is defined by [142. that is TL = 10 lg (Pi /Pt ). This method will be referred to as Transfer Matrix Method.

Open ex- terior. and the acoustic particle velocity through the perforation v̂1 specified below. . the equations for the lined duct are given by dψ+ ikLeff 1 C1 Leff =− ψ+ + v̂1 and (4.66)   2 where Ug is the grazing flow velocity and v̂x is the complex am- plitude of the particle velocity in axial direction within the duct. and 3. For convenience.86 modeling of bias flow liners of a duct by applying a mass and momentum balance to a de- fined control volume. 2. the effective perforation length30 Leff . (4.65)   2 − 1 ψ = (1 − Ug ) B̂ − (1 + Ug ) v̂x . Please note. so that the stag- nation enthalpy and acoustic particle velocity need to match at the respective boundaries.68) dx 1 − Ug (x) 2 A with the two boundary conditions − ψ+ (1)re eik −L +L ψ+ (0) = 1 and ψ− (1)e−ik eff eff = 0 . (4. Then. (4.67) dx 1 + Ug (x) 2 A dψ− ikLeff 1 C1 Leff = ψ− − v̂1 .69) the liner circumference C1 . Annular cavity enclosed by a rigid wall. that the grazing flow velocity increases with x over the length of the liner C1 Leff Ug = Mg + Mb x .2).70) A Eldredge and Dowling consider three configurations: 1. (4. 30 The effective perforation length yields best agreement with the experimental data (see the discussion in Section 4. Annular cavity enclosed by a second liner. Only the last two are of impor- tance here. and (4.10. the duct cross-section area A. they introduce the char- acteristic quantities 1 ψ+ = (1 + Ug ) B̂ + (1 − Ug ) v̂x . An analytic solution does exist for the hard- walled duct sections upstream and downstream.

(4. for example in Matlab with the function bvp4c.73). (4. This can be done accordingly for the case with a second liner with equations (4. (4. (4. and v̂x.69) and (4. i.1 C1 Leff = −ikLeff B̂1 − v̂1 . (4.75) dx Ac Ac with η2 v̂2 = B̂2 (x) − B̂1 (x) .67).1 .72) dx Ac v̂x. (4.1 (1) = 0 .68). When the surrounding enclosure is suffi- ciently large. (4.75). e.1 and (4.72) form a closed system for the four unknowns ψ+ .1 (0) = 0 and v̂x.68). ψ− . The liner velocity v̂b is given in Equation (4. (4. the four differential equations (4. B̂2 = 0.71) dx dv̂x. the stagnation enthalpy fluctuation external to the second liner can be assumed negligible.1 is the axial particle velocity in the cavity with the rigid wall boundary conditions v̂x.71) and (4.71) and (4. . The boundary value problem is solved using the shooting method. B̂1 .9 application to a cylindrical geometry 87 The equations within the annular cavity enclosed by a rigid wall yield dB̂1 = −ikLeff v̂x.76)  ikL For the rigid wall case.72) is modified to dv̂x.74)  ikLeff When a second liner is present Equation (4. (4. 4.74) and the four boundary conditions by Equations (4.73) The particle velocity through the liner is given by η1 v̂1 = B̂1 (x) − ψ+ (x) − ψ− (x) .67).1 C1 Leff C2 Leff = −ikLeff B̂1 − v̂1 + v̂2 .

the compliance delivers a homogeneous boundary condition for the perforation by dividing the single orifice value by the porosity. It is defined as 1 πr2 1 l = + . scaled by the energy incident upon the lined section.1. Here. thus . (4. when needed. Eldredge and Dowling [142] define an absorption coefficient as the net energy absorbed by the liner. that is Equation (4. following the ar- gumentation given by Jing and Sun [241].74). the compliance includes a thickness term. The acoustic properties of the bias flow orifice are given by the Howe Rayleigh conductivity model. e.4.27). As was done with the impedance in Section 4.2.88 modeling of bias flow liners The damping liner is represented by the compliance31 η1 in Equation (4. see Section 4.77) η1 σL2eff KR σLeff The compliance of the second liner η2 is given accordingly. i.

+ .

2 .

− .

2 .

B̂ .

+ .

B̂ .

α = 1 − .

2+ .

2 .

1− .

78) .2 . (4.

B̂1 .

+ .

B̂2 .

It should not be mixed up with the acoustic compliance that was introduced in Equa- tion (4. as shown in Section 6. the duct end is assumed to be anechoic.10). Then. Equation (4. with and without flow. . the absorption coefficient and the average dissipation coefficient are identical and can be compared directly. 31 This term was introduced by Hughes and Dowling [219] to describe the homogeneous boundary condition of an acoustically ’soft’ wall.3.79) 1 + |re |2 |ψ+ (1)|2 Eldredge and Dowling [142] give several expressions to calculate the end reflection coefficient re . so that re = 0. (4.78) can be expressed in terms of ψ+ and ψ− as 2 2 1 + |re |2 − 1 |ψ+ (1)| − |ψ− (0)|  α= . where the index 1 and 2 refers to the upstream and downstream hard-walled duct section adjacent to the liner.4. Here.

4.10 some comparisons 89

This approach will be referred to as Eldredge and Dowling
Method, short EDM, and EDM:Jing when including the descrip-
tion of the liner according to Equation (4.77).

4.10 some comparisons

4.10.1 Impedance Models

The first comparison is done for a configuration without any flow.
The linear acoustic behavior of a perforation without flow is fairly
well understood, so that the models should predict a similar be-
havior. The Howe model and its derivatives are only valid when
a bias flow is present, so that they are excluded from this compar-
ison.
Figure 4.7 plots a) the resistance, b) the reactance, and c) the
dissipation over the frequency. The liner geometry corresponds to
configuration dc006 of the parameter study. The specifications are
given in Table 7.1. Even without flow the impedance prediction of
the models varies slightly. The effect of these variations on the ab-
sorption are revealed in Figure 4.7c. The theoretical dissipation is
calculated with the transfer matrix method and compared to the
experimental results. Surprisingly, the oldest model, i. e. the linear
Melling model, shows the best agreement with the experiments.
The other models do not include some effects that are considered
in the Melling model, i. e. thermal conductivity losses within the
orifice are neglected or approximations of the Bessel functions are
used, for example. The prediction quality for the no flow case
would have benefited from including these effects. The peak fre-
quency predicted by the Bauer model is significantly higher than
the other models. This results from the empirical correction factor
in the reactance, which is strictly valid only when a grazing flow
is present (see discussion in Section 4.6).
The impedance of the same setup as above, but including a bias
flow of Mb = 0.03 is plotted in Figure 4.8. The Melling model does

90 modeling of bias flow liners

0.8
Normalized Reactance Normalized Resistance
a)
0.6

0.4

0.2

0
b)
4

2

0
c)
Average Dissipation

0.4

0.2

0
0 300 600 900 1200
Frequency in Hz

Figure 4.7: Comparison of the impedance and the dissipation predicted
by various models plotted over the frequency for dc006 without flow: a)
normalized specific resitance, b) normalized specific reactance, c) aver-
age dissipation coefficient. Legend: Melling; Bauer; Betts;
Bellucci; Experiment.

4.10 some comparisons 91

Normalized Reactance Normalized Resistance
6 a)

5

4

3
b)
3

2

1

0
0 300 600 900 1200
Frequency in Hz

Figure 4.8: Comparison of the impedance models plotted over the fre-
quency for dc006 with Mb = 0.03. Jing; Bauer; Betts;
Bellucci.

not consider bias flow, so that it would remain unchanged from
Figure 4.7 and has been omitted here. Instead, the Jing model is
plotted.
All models predict a significant increase of the resistance due
to the bias flow. The resistance terms of Jing and Bellucci are both
based on the Howe model (with an addition contribution from
viscous losses in the Bellucci model), so that they give similar re-
sults. The resistance predicted by Betts and Bauer is much lower.
The Betts and Bauer models neglect any influence of the bias flow
on the reactance, so that the reactance is identical to the config-

92 modeling of bias flow liners

Normalized Reactance Normalized Resistance
a)
30

20

10

0
b)
3

2.5

2
0 0.05 0.1 0.15
Bias Flow Mach Number

Figure 4.9: Comparison of the impedance models plotted over the bias
flow Mach number for dc006 with f = 1000 Hz. Jing; Bauer;
Betts; Bellucci.

uration without flow in Figure 4.7. The reactance of Bellucci is
significantly reduced compared to the no flow configuration.
Figure 4.9 plots the impedance over the bias flow Mach number
for a fixed frequency. All models predict a mainly linear increase
of the resistance with the bias flow Mach number. As discussed
above, the reactance is constant for the Betts and Bauer models.
On the other hand, Jing and Bellucci predict a decreasing reac-
tance, when increasing the bias flow. In that case, the reactance is
rapidly reduced at low bias flow Mach numbers and approaching
a constant value at high bias flow Mach numbers. The effect is
much stronger in the Bellucci model. The more complex models

4.10 some comparisons 93

EDM:Jing

Average Dissipation
0.6 TMM:Jing
3
0.4

0.2 2 1

0
300 600 900 1200 1500 1800
Frequency in Hz

Figure 4.10: Comparison of the Eldredge and Dowling method and the
transfer matrix method for three configurations: 1. dc006 with Mb =
0.015; 2. dc007 with Mb = 0.029; 3. dc008 with Mb = 0.049.

of Jing and Sun [242] and Lee et al. [291], which require a numer-
ical solution and are not plotted here, predict a similar effect (not
shown here).

4.10.2 Transfer Matrix Method vs. Eldredge & Dowling Method

Theoretically, the Eldredge and Dowling Method (EDM) and the
Transfer Matrix Method (TMM) should give identical results when
the same model for the perforation is used. The remaining differ-
ences are due to the two methods. This will be demonstrated for
three configurations, varying liner geometry as well as bias flow
Mach number, applying the Jing model. The liner geometries cor-
respond to configurations dc006, dc007, and dc008 of the param-
eter study. The specifications can be found in Table 7.1.
A first comparison is given in Figure 4.10. Both methods predict
similar results, but they are not identical. While both methods pre-
dict a similar level of dissipation for setup 3, TMM yields slightly
higher dissipation levels for setups 1 and 2. However, the pre-
dicted peak frequencies agree for setups 1 and 2, while they are

94 modeling of bias flow liners

EDM:Jing

Average Dissipation
0.6 TMM:Jing
3
0.4

0.2 1
2

0
300 600 900 1200 1500 1800
Frequency in Hz

Figure 4.11: Comparison of EDM using the effective perforation length
and TMM using the perforation length plus additional hard-wall sec-
tions for three configurations: 1. dc006 with Mb = 0.015; 2. dc007 with
Mb = 0.029; 3. dc008 with Mb = 0.049.

slightly different for setup 3. The origin of these differences is not
obvious and no general trend can be observed (e. g. one method
always predicts higher values than the other).
It was found that the axial length is a crucial parameter. It is not
entirely clear which length should be used here, that is either the
perforation length L, the effective perforation length Leff , or the
cavity length Lc as defined in Figure 2.8. Additionally, TMM can
account for hard-walled sections at either end of the perforation,
see Figure 4.6b. For the theoretical comparison of the two methods
in Figure 4.10, the effective perforation length was used in both
methods and any additional hard-wall sections were neglected in
TMM.
When comparing the models to measurement data, it will be
made use of the feature of TMM which takes the hard-walled
sections into account. In that case, the perforation length is used
and the length of the hard-walled sections is treated separately.
However, due to the lack of this feature in EDM, it will continue
to use the effective perforation length. The comparison between

4.10 some comparisons 95

EDM:Jing

Average Dissipation
0.6 EDM:Luong 3

0.4

0.2 2 1

0
300 600 900 1200 1500 1800
Frequency in Hz

Figure 4.12: Comparison of the Howe model and the Luong model for
three configurations: 1. dc006 with Mb = 0.015; 2. dc007 with Mb =
0.029; 3. dc008 with Mb = 0.049.

the two methods looks slightly different when they are based on
different characteristic lengths. This is shown in Figure 4.11. Var-
ious length combinations have been applied to the models and
were compared to the experimental data. There was no clear best
match of any one length. Using the effective perforation length
with EDM and the perforation length plus additional hard-wall
sections with TMM yielded the best overall agreement. Thus, this
setup is applied throughout this work.

4.10.3 Howe vs. Luong

Figure 4.12 compares the results when using the Jing model or
the Luong model to represent the perforation. Both cases are cal-
culated with the Eldredge and Dowling method, setting Cc = 0.75
for EDM:Luong. The Luong model for zero thickness is used with
the thickness term applied by EDM. The three configurations are
the same as in the previous section.
The results from the two models are very similar. However,
EDM:Luong predicts a slightly lower peak frequency in all three

Due to their similar results.96 modeling of bias flow liners cases. . only EDM:Jing will be com- pared with the experimental data in Chapter 7.

59]) D ∂ = + v∇ . that its macroscopic behavior is the same as if it were perfectly continuous in structure [33]. . (5. It relates the velocity of the fluid to its density and is valid without any restric- tions. An alternative formulation is given by replacing the local time derivative ∂/∂t with the substantial derivative (Stokes [449]. 2 The continuum hypothesis disregards the discrete molecular nature of a fluid and assumes.2) Dt ∂t 1 This and the following equations are given in Gibbs notation (Gibbs and Wil- son [173]. The Nabla operator ∇ is used as a convenient mathemati- cal notation for either the gradient grad f = ∇f. The conservation of mass in the absence of any sources is described by the partial differential equa- tion1 (Euler [148]. and energy. except that the fluid and its properties must be continuous2 . the divergence div f = ∇ · f. or rotation (or curl) rot f = ∇ × f [77]. 369]) ∂ρ + ∇ · (ρv) = 0 . or [59]). momentum. (5. or [33. or [33.1) ∂t It is commonly known as the continuity equation. 5 DUCT ACOUSTICS 5.1 acoustic wave equation Fundamental Equations of Fluid Dynamics The motion of fluid is governed by the conservation laws for mass.

It follows from Equation (5. Swiss mathematician and physi- cist. 369]) Dv ρ = −∇p .3a) Dt The conservation of momentum for an ideal fluid is given by (Eu- ler [148] or [33. An adiabatic process is isentropic provided it is reversible.3) is commonly referred to as Eu- ler equations6 . thus it reveals the convective effect of the mean flow. which for an ideal gas4 is given by (Clapeyron [97]. ideal fluid. 3 Named after Leonhard Euler (1707-1783).98 duct acoustics The substantial time derivative follows the motion of the fluid.” 6 Even though only the first two originate from Euler. (5. 369]   ∂p Ds = c2 with = 0. This equation is also called ideal gas law. The acoustic motion is considered to be isentropic5 . .3b) Dt where gravity is neglected. Air of normal atmospheric temperature and pressure can be considered to behave ideal. They describe the motion of a compressible. This equation is also called equation of motion. while this special case for an ideal fluid is commonly known as the Euler equation3 . or [33]) p = ρRT . It relates the pressure to the veloc- ity of the fluid. The conservation of energy is expressed by the equation of state.3c) ∂ρ s Dt This system of three Equations (5.1) that [33] Dρ + ρ (∇ · v) = 0 . (5. (5. so that the thermodynamic relation between the density and the pressure can be given by [33. 4 An ideal gas is an idealization where the molecular interaction is neglected. 5 Morfey [338]: “In an isentropic change of thermodynamic state the entropy remains constant.

(5.6) in a few simple steps: Substitute the fluctuating density ρ0 in Equation (5.6a)  Dt Dv0 ρ0 + ∇p0 = 0 . taking the time derivative of Equation (5.3). and density) that are used in the previous section can be written as the sum of their mean and fluctuating (acoustic) values: p = p0 + p0 v = v0 + v0 (5. 5.6a) with (5.6c) Acoustic Wave Equation The acoustic wave equation can be obtained from Equation (5. so that the approximate acoustic equations are given by [176. (5. (5.6b) yields the acoustic wave equation 1 D2 p0 ∇2 p0 − = 0. This assumption allows to disregard higher order terms in Equations (5.6b) Dt p0 = c2 ρ0 .4) ρ = ρ0 + ρ0 A linearization of Equations (5. Then. and (5.7) c2 Dt2 .6a) and replacing its second term with Equation (5.5) p0 v0 ρ0 which is fulfilled in most applications. 369] Dρ0 + ρ0 ∇ · v0 = 0 . velocity.1 acoustic wave equation 99 Equations of Linear Acoustics The field quantities (pressure. (5.3) is possible by assuming that the acoustic quantities are much smaller than the mean quantities p0 v0 ρ0 1 1  1.6c).

independent of position. or [369]) 1 ∂2 p0 ∇2 p0 − = 0. . French mathematician).e. i.9) c2 ∂t2 7 The Laplace operator ∇2 . is named after Pierre- Simon de Laplace (1749–1827.7) results in the commonly used form of the convected wave equation (Howe [207. It prescribes the sum- mation of the second partial derivatives of a function f with respect to a given coordinate system.2).” 9 Morfey [338]: “Linear acoustics is limited to small amplitude signals or os- cillations.” 10 In uniform flow the magnitude and direction of the velocity is spatially uni- form. 150]. can be applied to stationary or moving medium.8)) reduce to the well known form of the classical acoustic wave equa- tion (d’Alembert [111] and Euler [149. In general: ∇2 f = ∇ · (∇f) = div(grad f) [77] 8 Morfey [338]: “A homogeneous medium is a medium whose properties are spatially uniform.8) c ∂t In stationary medium. see Equation (5. so that the relation between any two oscillatory quantities is inde- pendent of amplitude. (5. (5. v0 = 0. This general notation includ- ing the material derivative. 343]) 2 1  2 0 ∂ ∇ p − 2 + v0 ∇ p0 = 0 . the material derivative can be re- placed by the local time derivative and Equation (5. 214]. Linear sound waves propagate independently of one another. The assumptions that have been introduced are: • homogeneous medium8 • ideal fluid • linear acoustics9 • isentropic relation between pressure and density • uniform flow10 or stationary medium Expanding the material derivative in Equation (5.7) and ((5. or [340.100 duct acoustics where ∇2 is the Laplace operator7 . or sometimes denoted as ∆. without interaction.

(5. [206.g. e. density. and potential [86]. 352. However. for example. the radial coordinate r. t) = p̂+ mn e −ikx. or Morse and Ingard [340]. since all coefficients are constant they apply to all acous- tic variables. (5.mn r) imθ (5. 12 The original derivation of the modal character of a sound field goes back to Duhamel [132] and Rayleigh [387].10) ∂x2 r ∂r ∂r r2 ∂θ2 so that the three-dimensional wave equation in cylindrical coordi- nates yields ∂2 p0 1 ∂ ∂p0 1 ∂2 p0 1 D2 p0   + r + − = 0. pressure. g. g. 5. r. The Laplace operator in cylindrical coordinates is given by (e. t) = fx (x) · fr (r) · fθ (θ) · ft (t) and reduces the partial differential equation into a set of ordinary differential equations (Bernoulli [50] or [77]). 11 The separation of variables introduces a substitution in the form p0 (x.mn r) + Qmn Ym (kr.7) can be expressed in cylindrical coordi- nates with the axial coordinate x. 5. [77]) ∂2 p0 1 ∂ ∂p0 1 ∂2 p0   2 0 ∇ p = + r + .11) ∂x2 r ∂r ∂r r2 ∂θ2 c2 Dt2 This is a linear second-order partial differential equation which can be solved by the separation of variables11 . Morfey [335].12) ×e × eiωt . θ.mn x p̂mn (x. θ. r. Detailed discussion are given by Tyler and Sofrin [482]. 444])12 ikx. and the circumferential coordinate θ. .2 three-dimensional waves 101 The wave equations above are expressed for the acoustic pressure.2 three-dimensional waves For the description of the wave propagation in a circular duct the wave equation (5. A general solution for a circular duct is given by (e. velocity.mn x + p̂− mn e × Jm (kr.

and Qmn is the n-th Eigenvalue of Ym . . t) = p̂mn (x. 2. the radial pattern is defined by Bessel functions. 15 Morfey [338]: “A mode is a spatial pattern of vibration. (5. whose shape remains invariant as the vibration propagates spatially. } and n ∈ {0. This special case is called fundamental mode or plane wave and will be discussed separately in Section 5. . 0.mn + k2r. Jm is the Bessel function of the first kind. . Equation (5. For m = 0 and n = 0 there are no nodal lines in transverse direction. The ac- tual sound pressure of a time harmonic wave is given by the real part of the superposition of an infinite number of modes15 XX p0 (x. −1. −2. Ym is the Bessel function of the second kind. In a cylindrical duct.12) represents the so called modal solution. }. 1. θ. German mathematician). (5.12) describe the spatial (axial. respectively. so that the solution for a time harmonic 13 Named after Heinrich Martin Weber (1842-1913. . r. Each mode is characterized by its circumferential mode order m and its radial mode order n. . and circumferential) and temporal shape of the mode. without any central hub.” . . often called Weber13 or Neu- mann14 function. . r.102 duct acoustics where the axial and radial wave numbers are related by  ω 2 = k2x.14) m n where m ∈ {. radial. While the spatial patterns in axial and circumferential direction are sinusoidal (as well as the temporal development). commonly only called Bessel function.12) becomes zero. 14 Named after Carl Gottfried Neumann (1832-1925. . German mathematician). 1. Qmn in Equa- tion (5. so that the sound field is one-dimensional with variations in x-direction only.13) c The four rows in Equation (5. All other modes are referred to as higher order modes. respectively.mn . The integers m and n describe the number of nodal lines in circumferential and radial direction. 2. θ.3. . t) .

337. g. (5.1. . Abramowitz and Stegun [3].1.6b) follows the boundary condition for the 0 acoustic pressure: ∂p ∂r = 0 at r = R. (5.3176 5. (From [3. From Equation (5.15) with the axial wave number  s   2 k  jmn  k± x.mn x J mn m (kr. 411]) m:n 0:0 1:0 2:0 0:1 3:0 4:0 1:1 jmn 0 1. The mathematical description of the acoustic pressure in a cylin- drical duct is demonstrated graphically in Figure 5.17) with the radial wave number jmn kr. Some values of jmn are given in Table 5. (5.mn = −M ± 1 − (1 − M2 ) .16) 1 − M2 kR The hard-wall boundary condition16 requires that the derivative of Jm vanishes at the wall. so that for r = R J0m (kr.1: Eigenvalues of the hard-wall solution for a cylindrical duct. e.2 three-dimensional waves 103 Table 5. 444])   −ik+ − imθ iωt p̂mn = p̂+ mn e x. The spatial distribution of the acoustic pressure is plotted for the four modal 16 The hard-wall boundary condition demands that the radial velocity vanishes at the wall.mn R) = 0 .8412 3. 343. More extensive tables can be found in mathematical references. g.3314 wave propagating in a cylindrical duct with hard walls and uni- form mean flow yields (e.18) R jmn is the n-th root of J0m or the n-th Eigenvalue of Jm for the hard-wall boundary condition. sorted in ascending order of jmn .mn x + p̂− e−ikx. p. [315.2012 5. (5. 5.8317 4.0542 3.mn r) e e .mn = . 330.

1 Cut-On Frequency The axial wave number k± x. 1:0. Equation (5.104 duct acoustics 0:0 0:1 p0 1:0 1:1 Figure 5.16) allows for two fundamentally dif- ferent propagation characteristics depending on the evaluation of the square root in the equation:  2 1. Another characteristic feature of modes of higher circumferential order m > 0 is that their transversal pattern is rotating around the x-axis. the spinning modes produce the spiral pattern of the acoustic pressure at the duct surface. and 1:1 and the resulting sound field p0 at a single frequency. 5. and 1:1 and the resulting sound field p0 = p00:0 + p01:0 + p00:1 + p01:1 . A behavior which has coined the term spinning mode [482]. While the images are snapshots of the instantaneous acoustic pres- sure at t = 0. components (m:n) 0:0.mn defines the propagation of a mode in x-direction. 1 − (1 − M2 ) jkR mn > 0 → k± x.mn ∈ R . 0:1.2. 0:1. The nodal lines of the circum- ferential and radial patterns are visible in the duct cross-section.1: Illustration of the acoustic pressure in a cylindrical duct for four modal components (m:n) 0:0. 1:0.

the speed of sound. so that jmn √ fc.19) 2πR The cut-on frequency is dependent on the duct radius.mn is a real quantity.mn = c 1 − M2 .2 Evanescent Modes A mode that is excited below its cut-on frequency is not able to propagate. i. the mode is not able to propagate. M. The Eigenvalue jmn of the fundamental mode 0:0 is zero and it follows that the cut-on frequency is also zero.mn ∈ C In the first case k±x. Only a limited number of higher order modes are able to propagate at a given frequency. the fundamental mode is able to propagate at all frequencies. These modes are referred to as evanescent modes or 17 The ’first’ higher order mode refers to the higher order mode with the lowest cut-on frequency. . the mode starts propagating. At this cut-on frequency the term under the square root in Equation (5. i.2 three-dimensional waves 105  2 2.mn is complex and its imaginary part serves as an attenuation coefficient.1).2. so that the remaining non- propagating modes can be neglected (in most cases). (5. The amplitude of the mode decays expo- nentially with axial distance from the source. The cut-on behavior of the higher order modes introduces a ma- jor advantage. e. In the sec- ond case k± x. 1 − (1 − M2 ) jmn kR <0 → k± x. The mode propagates along the duct unattenuated (considering ideal fluid). the mean flow Mach number. e. c = constant) the numerical order of the cut-on frequencies is given by the numerical order of the values of jmn (as they are given in Table 5. i. For a fixed setup (R. A characteristic frequency can be defined as the frequency where a mode becomes ’cut-on’. 5. and the Eigenvalue of the associated mode. e. The cut-on frequency of the first17 higher order mode defines the transition between one-dimensional and three-dimensional sound propaga- tion.16) vanishes. 5.

Their amplitude decays exponentially with axial distance from their source. so that the evanescent mode attenuation coefficient is given by18 . The decay is determined by the imag- inary part of the axial wave number k± x.mn in Equation (5.16).106 duct acoustics cut-off modes.

 .

2π q.

αmn = .

Im k± .

f2 − f2 .

20) . . (5.

.

=
x,mn 2
c (1 − M ) c,mn

The unit of the attenuation coefficient is Np/m, i. e. Neper19 per
meter. It can be converted to the attenuation rate amn in dB/m by
applying

amn = 20 · lg (eαmn ) ≈ 8.69 αmn . (5.21)

The attenuation decreases when increasing the frequency towards
the cut-on frequency of a mode.
At a certain frequency only a limited number of modes are able
to propagate and it is reasonable to neglect all evanescent modes.
However, evanescent modes might become important if the fre-
quency of interest is close to the cut-on frequency and the location
of interest is in the vicinity of the source, e. g. a microphone which
is located near a loudspeaker (see Section 6.5.4).
Figure 5.2 concludes the discussion of cut-on frequency and
evanescent modes by demonstrating the phenomena visually. The
acoustic pressure of the plane wave p00:0 , the first higher order
mode p01:0 , and the resulting sound field p0 = p00:0 + p01:0 is plotted
for three distinct frequencies:

1. The frequency is below the cut-on frequency,

2. the frequency is equal to the cut-on frequency, and

3. the frequency is above the cut-on frequency.
18 This notation is according to Tyler and Sofrin [482], where it was given for a
stationary fluid.
19 Similar to the decibel, the Neper is a logarithmic scaled ratio of field or power
quantities. While the Neper uses the natural logarithm ln, the decibel uses
the decadic logarithm lg.

5.2 three-dimensional waves 107

f < fc,1:0 f = fc,1:0 f > fc,1:0

0:0 0:0 0:0

1:0 1:0 1:0

p0 p0 p0

Figure 5.2: Illustration of the cut-on phenomena of higher order modes
and the influence on the resulting sound field for mode 1:0.

The source of the circumferential mode is located at x = 0, that
is the visible cross-section. The amplitude of the circumferential
mode is chosen, so that the maximum acoustic pressure at the
duct wall corresponds to the maximum acoustic pressure of the
plane wave.
In the first case, the circumferential mode is excited at a frequency
below its cut-on frequency. The mode is not able to propagate
and becomes insignificant within a short axial distance. The re-
sulting sound field, shown at the bottom, reflects the influence of
the higher order mode close to its source, while it blends into the
one-dimensional plane wave further along the duct. In the middle
column the frequency is equal to the cut-on frequency, so that the
circumferential mode is able to propagate along the duct. The re-
sulting sound field is three-dimensional throughout the pictured

108 duct acoustics

Table 5.2: Eigenvalues of the hard-wall solution for an annular duct ac-
cording to Equations (5.22) and (5.23) with η = 0.59 and sorted in as-
cending order of jmn .

m:n 0:0 1:0 2:0 3:0 4:0 5:0 6:0 0:1 1:1
jmn 0 1.269 2.529 3.772 4.990 6.179 7.338 7.772 7.887
Qmn 0 -0.320 -0.298 -0.239 -0.178 -0.126 -0.084 0.935 -0.745

duct segment and beyond. Even though the circumferential mode
starts spinning at its cut-on frequency, the spinning motion can be
observed much better at higher frequencies (as shown in the right
column).

5.2.3 Resonances in an Annular Cavity

Starting from Equation (5.12), a resonance condition for a hard-
walled annular cavity can be derived. The annular cavity is con-
fined by two concentric cylinders with axial boundaries in both
directions. The solution in radial direction is now dependent on
the radius-ratio of the inner and outer wall, the so called hub-to-
tip ratio η = Rhub /R. jmn and Qmn are determined by solving the
equations (e. g. [255])
0 0 0 0
Jm (jmn )Y m (ηjmn ) − Jm (ηjmn )Y m (jmn ) = 0 (5.22)
J 0 (jmn )
Qmn = − m0 (5.23)
Y m (jmn )

Equation (5.22) needs to be solved numerically. Some values of
jmn and Qmn for a hub-to-tip ratio η = 0.59 are listed in Table 5.2.
The radial wave number is still defined by Equation (5.18), with
the appropriate values of jmn for the annular geometry.
The hard-wall boundary conditions in axial direction require
that p̂+
mn = p̂mn at x = 0 and that sin(kx,mn Lc ) = 0 at x = Lc ,

where Lc is the length of the cavity. The latter one is fulfilled for

5.3 plane waves 109

any integer multiples of π, so that the axial wave number for the
annular cavity is given by

πl
kx = with l ∈ {1, 2, . . . } . (5.24)
Lc
Inserting Equations (5.18) and (5.24) into (5.13) yields the reso-
nance condition20
s 
πl 2 jmn 2
 
c
flmn = + . (5.25)
2π Lc R

Equation (5.25) defines the resonance frequencies in a hard-walled
annular cavity for resonances in axial, radial, and circumferential
directions.

5.3 plane waves

At low21 frequencies the sound field in a duct is one-dimensional.
The acoustic field quantities of a wave traveling in x-direction vary
with time and x, but are constant in any plane normal to the di-
rection of wave propagation, i.e. in the duct cross-section. Such
one-dimensional waves are called plane waves. For plane waves
the acoustic wave equation (5.7) reduces to (e. g. [369])

∂2 p0 1 D2 p0
2
− 2 = 0. (5.26)
∂x c Dt2
A general solution regarding stationary medium (v0 = 0) is given
by (d’Alembert [111] and Euler [149], or [343, 369])
 x  x
p0 (x, t) = f1 t − + f2 t + , (5.27)
c c

20 This equation is equally valid in a cylindrical geometry, with the appropriate
value for jmn (e. g. [284]).
21 The frequency limit is given by the cut-on frequency of the first higher order
mode, as discussed in Section 5.2.1.

110 duct acoustics

where f1 and f2 are arbitrary functions, with the only limitation
that they have continuous derivatives of first and second order.
The equation describes the superposition of two plane waves trav-
eling in opposite direction with the propagation speed c. The
acoustic fluctuations are assumed to be sinusoidal, so that f1 and
f2 can be approximated by a Fourier series22 . In complex expo-
nential notation the Fourier series is expressed as (Fourier [165]
or [369])
X
f(t) = p̂ω eiωt . (5.28)
ω

The coefficient p̂ = A eiϕ is a complex quantity with amplitude
A = |p̂| and phase ϕ = arg p̂, commonly called complex pressure
amplitude. Applying (5.28) to (5.27) yields (e. g. [48])
X

iω(t− xc ) − iω(t+ xc )
p0 (x, t) = Re p̂+
ω e + p̂ ω e . (5.29)
ω

where p̂+ω is the complex pressure amplitude of the wave traveling
in positive x-direction and p̂− ω for the wave in opposite direction.
The use of the complex exponential notation is convenient mathe-
matically. However, the relevant physical quantity is only the real
part. This fact is indicated above by taking the real part Re{z} of
the expression. It is common practice to omit this explicit nota-
tion, while still taking the real part. This reduced notation will be
adopted here.
The acoustic pressure is given by the sum over all frequency com-
ponents. Typically, the solution is given for one frequency compo-
nent or a single frequency wave only, so that the summation of
the frequencies can be dropped.
Applying these simplifications and introducing the wave number
k = ω/c yields the familiar notation (e. g. [260, 343])

p0 (x, t) = p̂+ e−ikx + p̂− eikx eiωt . (5.30)


22 Named after Jean Baptiste Joseph Fourier (1768-1830, French mathematician
and physicist).

5.4 attenuation of sound 111

Equation (5.30) describes the temporal development and spatial
distribution of the acoustic pressure of a one-dimensional, single
frequency sound wave in a stationary medium. The behavior is
illustrated in Figure 5.3 for three distinct cases depending on the
values of p̂± :

1. Traveling wave: |p̂− | = 0

2. Standing wave: |p̂+ | = |p̂− |

3. Mixed wave: |p̂+ | 6= |p̂− |

Introducing a uniform mean flow in x-direction affects the propa-
gation speed of the wave. In flow direction the propagation speed
c is increased by the mean flow velocity, while it is decreased
against flow direction, i.e. c ± v0 respectively. The convective ef-
fect of the mean flow is included in the wave number
ω/c
k± = . (5.31)
1±M
where the superscript ± denotes in and against flow direction,
respectively. Equation (5.30) changes to (e. g. [340, 343])
 
p0 (x, t) = p̂+ e−ik x + p̂− eik x eiωt .
+ −
(5.32)

5.4 attenuation of sound

Until now all dissipative effects have been neglected. In many ap-
plications the dissipative effects can be considered small and ne-
glecting them reproduces the physical behavior adequately. How-
ever, certain conditions require their inclusion to describe the phy-
sics accurately.
The absorption of sound results in an attenuation of the wave
amplitude with propagation distance and a change in propagation
speed, i. e. phase velocity. Both can be expressed by adjusting the
wave number. Disregarding all absorption phenomena the wave

112 duct acoustics

Acoustic Pressure a) Traveling Wave

b) Standing Wave
Acoustic Pressure

c) Mixed Wave
Acoustic Pressure

t1 t2 t3 x1 x2 x3
Time Axial Coordinate
p0 at x1 /t1 , p0 at x2 /t2 , p0 at x3 /t3

Figure 5.3: Illustration of the temporal development (on the left) and
spatial distribution (on the right) of the acoustic pressure for three char-
acteristic plane wave sound fields: a) Traveling Wave, b) Standing Wave,
c) Mixed Wave.

35)  k0 cph where α = k0 Re{Γ }. 475. 498]) α c p 0 = p̂+ e−k0 Γ x + p̂− ek0 Γ x eiωt with Γ = +i . g. Confusingly. Some authors prefer the formulation with the propagation con- stant Γ instead of the wave number. . Defined in ISO 80000-8:2007 [235] the propagation constant is related to the wave number by Γ = ik.4 attenuation of sound 113 number is real and given by k0 = ω/c (in a stationary medium). Without any absorption α = 0. When cph is a function of the frequency. (5. The phase velocity reflects a possible change in propagation speed. The attenuation rate a in units dB/m can be computed according to Equation (5. cph = ω/Im{Γ }. Yet another notation can be found in [126–128]23 c α p 0 = p̂+ e−ik0 Γ x + p̂− eik0 Γ x eiωt with Γ = − i . i.21). [359.30) remains unchanged ω p0 (x.36)  cph k0 23 Unlike here. there are some slightly different notations in use. e.34)  cph where α = Re{Γ }. the propagation is dispersive. another common notation seems to be (e. while Equation (5. and k = k0 Γ /i. 492. 5. The attenuation coefficient accounts for the amplitude attenuation.33)  cph where α = −Im{k} is the attenuation coefficient in Np/m and cph = ω/Re{k} is the phase velocity in m/s. Disregarding any absorption cph = c and the propagation is non- dispersive. so that (e. 508]) ω p 0 = p̂+ e−Γ x + p̂− eΓ x eiωt with Γ = α + i . t) = p̂+ e−ikx + p̂− eikx eiωt with k = − iα . the references include the sign of the exponent ik0 Γ in the prop- agation constant Γ . and k = Γ /i. (5. cph = c/Im{Γ }. However. (5. [257. waves of different frequencies propagate at different speeds. g. (5. Including absorption the wave number becomes a complex quan- tity. 405.

Navier-Stokes-Fourier Model The equation of continuity 5. and k = k0 Γ . 25 That is in particular gravity. The bulk viscos- ity will be introduced in Section 5.4. and is repeated here for convenience ∂ρ + ∇ · (ρv) = 0 . (5.g. (5. and assuming Newtonian fluid27 it is given by (Navier [348] and Stokes [449]. or [495]) neglects the bulk viscosity µB = λ + 32 µ = 0. Kirchhoff [261] included the thermal conductivity losses and presented a theoretical description of the sound propagation including viscosity and thermal conductivity. and theologian). or [33. Inspired by the experiments of Kundt [273]. French engineer and physicist) and Sir George Gabriel Stokes (1819-1903. His result is still the state of the art used today. 283. the effect of viscosity on sound propagation has been studied by Stokes [451] and von Helmholtz [195]. the wave number notation as in Equation (5. cph = c/Re{Γ }.3 27 Named after Sir Isaac Newton (1642-1727. applying the Stokes hypothesis26 . .37b) Dt 24 Named after Claude Louis Marie Henri Navier (1785-1836.3a is generally valid.4. Here.114 duct acoustics where α = −k0 Im{Γ }.1 Losses at the Wall Historically. e. 26 The Stokes hypothesis (Stokes [449].37a) ∂t The equation of motion for a viscous fluid is given by the Navier- Stokes equation24 . 495]. where λ is the second viscosity coefficient. ideal fluid or otherwise. Irish mathematician. politician. air and water. When neglecting all body forces25 . English physicist and mathemati- cian). These differ- ences should be kept in mind when browsing the literature. 5. physicist.33) will be used. the viscous stresses are linearly related to the rate of strain [33. In Newtonian fluids. 495]) Dv ρ = −∇p + ∇ · τij .

37c de- scribes the motion of fluid in a compressible. or unit tensor. The fluctuating pressure of the acoustic mode yields 1 ∂2 p0ac ∇2 p0ac = . (5. entropy. 5. German mathematician).4 attenuation of sound 115 with the viscous.38a) c2 ∂t2 28 Named after Leopold Kronecker (1823-1891. and Entropy Modes After applying the typical acoustic linearization (corresponding to Equation (5. German physicist) and Jean Baptiste Joseph Fourier (1768-1830. ∂xj ∂xi 3 ∂xk where δij is the Kronecker delta28 . or in [369]. Acoustic.37 can be obtained by assuming that the sound field is a superposition of vorticity. . stress tensor ∂vi ∂vj 2 ∂vk   τij = µ + − δij . is such that for i = j → δij = 1 and for i 6== j → δij = 0 [77]. 283. 495]) Ds ρT = ∇ · (κ ∇T ) + φ . ∂xj This system of partial differential equations 5. and acoustic modes.37a to 5. and heat conducting fluid. viscous. or deviatoric. 29 Named after Gustav Robert Kirchhoff (1824-1887.37c) Dt where the viscous dissipation rate is specified as ∂vi φ = τij . The energy equation is given by the Kirchhoff-Fourier equation29 (Kirchhoff [261] and Fourier [165]. (5. It is commonly referred to as Navier-Stokes- Fourier model.5)) a solution to Equations 5. Wave equations for the three compo- nent modes are derived by Cremer [101] and Kovasznay [270]. The Kronecker delta. or [33. Vorticity. French mathematician and physi- cist).

The fluctuating velocity of the vorticity mode is expressed as ρ ∂v0vor ∇2 v0vor = .38c) κ ∂t The subscript at the field variables indicates the associated mode. but only by superposition with the vorticity and entropy mode.9). [369])     0 Tβ 0 0 0 T Tac ≈ pac .38b) µ ∂t Finally. is given by T 0 = Tac 0 0 + Tvor 0 + Tent . The second one requires that the heat capacity of the wall material is much larger than the heat capacity of the fluid. (5. ‘vor’ for the vorticity mode. The first bound- ary condition is generally valid for any combination of fluid and solid wall material. and ‘ent’ for the entropy mode. the entropy mode is described by ρcp ∂s0ent ∇2 s0ent = . ‘ac’ for the acoustic mode. (5.38a) is identical with (5.37.116 duct acoustics All dissipative effects within the fluid are neglected. However. respectively.40) ρcp 0 cp 0 ent 30 The discussion applies to the acoustic particle velocity in a similar manner (as shown in [369]). The fluctuating temperature field. Tent ≈ s0 .39) The temperature fluctuations are related to the field quantities used in Equations 5. . (5. Tvor = 0 . a condition which is generally met when the fluid in the duct is air. so that Equa- tion (5. these boundary conditions can not be satis- fied by the acoustic mode alone. This will be illustrated for the acoustic temperature fluctua- tions30 .38 by (e. In a viscous and heat conducting fluid the boundary conditions at the stationary wall are vanishing velocity (no slip) and con- stant temperature (isothermal wall). (5. which is a solution to Equations 5. g.

but tend to infinity as f → 0. These characteristic lengths can be regarded as a boundary- layer thickness. The idea is that dissipative effects only need to be considered in a thin layer very close to the wall.” . The vorticity mode does not contribute to the temper- ature fluctuations. The acoustic boundary layer. eventually filling the entire cross-section of the pipe. cannot grow very much because the fluid never flows in one direction for more than half a period. on the other hand. 5.4 attenuation of sound 117 where the subscript ‘0’ indicates that the approximation is of ze- roth order. as well as the dis- tance from the wall y. They are thin at high frequencies. They are related via the Prandtl number δν /δχ = Pr.41) ωρ ω ωρcp ω for the vorticity wave and the entropy wave. The thickness of the viscous and the thermal acoustic boundary- layers are of the same order of magnitude. Assuming the acoustic mode to be a lossless plane wave.√ respectively. Schlichting and Gersten [432] give detailed discussions on all aspects of the subject. The thickness of the steady flow boundary layer grows with the distance. Tent which the secondary waves become insignificant are 2µ 2ν 2κ 2χ r r r r δν = = and δχ = = (5. The numerical values at 1 kHz and standard 31 The boundary-layer concept was introduced by Prandtl [372] in 1904 and is considered one of the most important advances in fluid dynamics [15]. Further- ent 0 must vanish for y → ∞. The boundary condition is satisfied if Tac 0 + T 0 = 0 for y = 0. with the latter one be- ing slightly thicker. The characteristic lengths after more. where the excess temperature is a function of the ax- ial coordinate x and the time t only. 32 Blackstock [64]: “Notice that the acoustic boundary-layer is qualitatively dif- ferent from the ordinary viscous boundary-layer that develops in a pipe when the flow through the pipe is steady (unidirectional). The fluid dynamic boundary-layer concept31 was adopted to an acoustic boundary-layer by Cremer [101]32 . for a cylindrical geometry y = R − r. The entropy mode introduces a secondary wave which is dependent on time. while the remaining fluid can be regarded as ideal where the acoustic wave is not affected.

4 over a complete oscillation cycle in steps of π/4. respectively. (5.43) The temperature profiles of the entropy mode alone and of the su- perposition according to Equations (5. Equation (5. At the distance given by the boundary-layer thickness δχ the magnitude of the combined wave |T 0 | is within 86 % of |Tac 0 |. are plotted in Figure 5.37 in or- der to quantify the absorption of sound.081 mm.5 mm).43).42) and (5. .068 mm and δχ = 0. the superposition shows that the temperature fluctuations vanish at the wall and fade into the acoustic mode away from the wall. (5. The conversion can be done √ with the help of the relation i = (1 + i)/ 2 [101].42) The temperature field close to the wall resulting from the super- position of the acoustic mode with the entropy mode is then given by [214]33   T 0 = Re Tac 0 1 − e−(1+i)y/δχ eiωt .38c) together with (5. which is gen- erally small compared to the wavelength (at the same conditions λ = 340. With some mathematical 33 The notation in the reference√is slightly different. Please also note the e−iωt time-dependency in the reference.40) yields an entropy wave expressed by the fluctuating temperature of the form [101. 340.118 duct acoustics conditions are δν = 0. Kirchhoff’s Solution Kirchhoff [261] was seeking a solution of Equations 5. 369] 0 0 −(1+i)y/δχ iωt Tent = Re −Tac e e . As required.

5. b) Fluctuating temperature field T 0 = Tac0 + T 0 . . a) Temperature fluctuations of the entropy mode.4: Illustration of the entropy mode and the resulting fluctuating temperature field at a wall. The temperature ent profiles follow a complete oscillation period in π/4 steps.4 attenuation of sound 119 a) Distance from the Wall δχ 0 0 −Tent 0 =0 Tent 0 +Tent b) Distance from the Wall δχ 0 −T 0 T0 = 0 +T 0 Figure 5.

or [475. p p a1 = He 2 He and x1 and x2 are the small and large roots of He2 4 γ He2 1 4 He2      γ 1+x 1+i 2 + +i +i x2 = 0 . where p He = kR is the Helmholtz number (or reduced frequency). Sh = R ω/ν is the shear number (square root of the Stokes num- ber). radial velocity is zero at tube axis. 475]):  1 Sh2 − 2 1 1 J1 (a1 )   2 2 iΓ Γ − i 2 − He x1 x2 J0 (a1 ) γ He2 1   1 J1 (a2 ) + 2 −i (Γ 2 − x1 ) 2 (5. 2. Sh 3 Pr Pr Sh2 γ 3 Sh2 The notation that is used here was introduced by Tijdeman [475]. . so that end-effects can be neglected.44) Pr Sh x1 J0 (a2 ) γ He 2 1   1 J1 (a3 ) − 2 −i (Γ 2 − x2 ) 2 = 0. 34 The propagation constant is defined as in Equation (5. and J0 and J1 are Bessel functions of the first kind of zeroth and first order. 35 The wave length and the tube radius must be large in comparison with the mean free path. respectively.44) is subject to the following assumptions (Kirchhoff [261]. 3. Pr Sh x2 J0 (a3 ) with Sh2 r Γ2 − i .120 duct acoustics effort he derived a complicated transcendental equation for the propagation constant34 Γ (Kirchhoff [261] or [390. linear acoustics (small amplitude. 492]): 1. Equation (5. a2 = He Γ 2 − x1 . a3 = He Γ 2 − x2 . semi-infinite tube.35). 4. For air of normal atmospheric temperature and pressure the limits are f < 108 and R > 10−4 mm [475]. Pr = cp µ/κ is the Prandtl number. Homogeneous medium35 . sinusoidal perturbations).

5.45) 2 Sh Pr The notation is again that of Tijdeman [475] with the quantities specified above and the propagation constant is defined as in 36 For non-cylindrical geometries Davies [113] suggest to replace the radius with the hydraulic radius rh = 2 A/P. In his original paper Kirchhoff [261] presents an approx- imate solution by restricting himself to a 10.4 attenuation of sound 121 5.δχ  R [369]. which are an approximation to first order. Ultimately. temperature boundary condition). 39 A survey of models for various tube dimensions is given in [475]. stationary medium. 448]. 475]) (1 + i) γ−1   Γ = i+ √ 1+ √ . cylindrical geometry36 . 448. 37 A discussion including the effect on higher order modes can be found in [36. (5. 79]. plane waves37 . it fails an analytic solution. 492]. so that the wide tube case is limited to δν . 8. More detailed (and more complicated) treatments are given in [108. and 9. 7. 390. and thus has served as a starting point for several further studies [257. velocity vanishes at the wall (no-slip boundary condition). Unfortunately. . the influence of the wall affects only a small layer of fluid which is an insignificant fraction of the whole of the con- tents. isothermal walls (const. 261. 6. that is the same assumption that Cremer [101] treated rigorously about 80 years later. or [369. wide tube39 . The wide tube assumption yields the following expression for the propagation constant (Kirchhoff [261].37). 38 It is an exact solution of the linearized Navier-Stokes-Fourier model (lin- earization of Equations 5.44) is given by the fact that it is exact38 . The importance of Equation (5. That means.

This corresponds to a change of 0. As a numeric example. suggest to neglect the change of phase velocity. 42 See as well the discussion by Peters et al.ν and αwall. and wave number this yields ω 1 γ−1 1 νω γ − 1 χω   r r αwall = √ 1+ √ = + (5.35).47) The first term in Equation (5. the wall attenuation coefficient at 1 kHz and ISA conditions40 is αwall = 0.χ 1 γ − 1 −1    cph. 508] to derive a corrected attenuation coefficient42 1 γ−1 1 γ−1 1 γ−1      ω αwall = √ 1+ √ + 2 1+ √ − γ . The effect on the phase velocity is rather small. so that the wave number yields k = ω/c − iαwall . The attenuation √ is proportional to the square root of the fre- quency αwall ∝ f and also dependent on the duct radius and the properties of the fluid.41 At 1 kHz and standard condi- tions the phase velocity is reduced from 340.46b) 2 Sh Pr ω ω kwall = − iαwall = + (1 − i)αwall (5. g.96 m/s. .23 dB/m. corresponding to an attenuation rate a = 0.46c) cph.0266 Np/m. c 2 Sh Pr Sh Pr 2 Pr (5. Ronneberger [411] used the low reduced frequency solution of [475. 115].14 %. phase velocity. Written in terms of attenuation coefficient.45 m/s to 339.47) is identical to the attenuation co- efficient found by Kirchhoff and the second term adds a constant 40 The International Standard Atmosphere [220] defines the temperature and pressure at sea level as 288.122 duct acoustics Equation (5. e.46a) c 2 Sh Pr Rc 2 Rc 2 | {z } | {z } αwall.ν αwall. [113.325 kPa. 41 Some authors.15 K and 101.wall = c 1 + √ 1+ √ (5. respectively. respectively.wall c where αwall.χ are the separate attenuation coefficients due to viscosity and thermal conductivity. [366].

127] presents an asymptotic solution of the con- vective acoustic equations for large shear numbers Sh  1 follow- ing the simplifications given by Zwikker and Kosten [508]. His result is different from Equation (5. Now. but without any further explanation. He presents a quasi-laminar theory that. the convective effect of a mean flow will be discussed.31). 5.49) c 1 ± ΓM 1 ± ΓM 1 ± ΓM where Γ is the propagation constant that corresponds to the Kirch- hoff solution. as is discussed in [114].50) 2Sh Pr ω The earliest and probably most rigorous derivation is given by Ronneberger [411. Davies et al. 43 Please note that Equation (3) in the reference is erroneous. Its relative contribution to the total losses is larger at low frequencies and small duct radii. Without any dissipative effects the convective wave number is given by Equation (5.36) (1 − i) γ−1   c Γ = 1+ √ 1+ √ = 1 + (1 − i) αwall . where the mean flow introduces a dis- tinction between the wave number in and against flow direction. Convective Effect of a Mean Flow The assumptions listed above require a stationary medium. . A more profound approach including the additional absorption due to a turbulent flow boundary-layer is treated in the succeeding section. [115]43 suggest that the appropriate wave numbers including viscothermal losses at the wall are given by ω/c α k± wall.4 attenuation of sound 123 value which is independent of frequency.48) and reads ω Γ ω/c α k± wall. but given in the notation of Equation (5. 413].Davies = + (1 − i) wall (5.Dokumaci = = + (1 − i) wall (5. (5. where the first term becomes smaller.48) 1±M 1±M Dokumaci [126. Following this definition.

In the end. it increases against the flow. Here. the small differences in the result of these two models (Dokumaci vs. Comparison to experimental data has shown [366] that Ron- neberger’s model includes the physical effects most accurately.wall = cph. The approach of Davies shows the smallest effect.5b. To first order and with 1 − 0.χ are the attenuation coefficients due to vis- cosity and thermal conductivity losses at the wall given in Equa- tion (5. Ron- neberger) can be neglected in most practical applications. The influence of convection on the phase velocity is visualized in Figure 5.46a).wall − c (1 ± Re {Γ } M). While the change of phase velocity in a stationary medium is small already. ∆c± ph. 205]. the convection introduces an even smaller variation around that value. The values of Dokumaci’s and Ronneberger’s model are very close.wall = cph.wall shows only the additional contri- bution that is introduced by the losses. but Ronneberger predicts a slightly stronger influence of the flow. browsing in recent literature it seems that Dokumaci’s approach is the most accepted [13. (5. The corresponding attenuation coefficients α± wall = −Im{k± } are plotted in Figure 5.wall − c (1 ± M) and for Dokumaci ∆c± ± ph.χ kwall.Ronneberger = + (1 − i) + . respectively. describes the damping of acoustic waves quite accurately. While the attenuation is reduced in flow direction. Ronneberger’s model predicts the opposite and a generally smaller effect.51) 1±M (1 ± M)3 (1 ± M) where αwall. so that for Davies and Ronneberger the change is given by ∆c± ± ph. 147.ν αwall. compared to the stationary case αwall . The limit that the acoustic boundary-layer thickness needs to be . The results of Davies’ and Dokumaci’s models are identi- cal.124 duct acoustics according to Peters et al. when the acoustic boundary-layer thick- ness is small compared to the thickness of the viscous sublayer of the flow. They express a reduction of phase velocity in flow direction and an increase against the flow. [366].18 M 2 ≈ 1 (for small Mach numbers) Ronneberger results at [411]   ± ω/c αwall.5. However.ν and αwall.

wall ∆cph.Dokumaci -6 1 10 102 103 104 Frequency in Hz Figure 5. .wall ∆c± -4 ∆c− ph.Davies -5 ∆c+ ph.Dokumaci in m/s -2 ∆c− ph.035 m at ISA conditions.Ronneberger ∆c+ ph. 5.Davies -3 ∆c+ ph.1 and R = 0.Ronneberger ph.5: Convective effect of a mean flow on the wall attenuation coefficient for M = 0.4 attenuation of sound 125 10−1 a) α− wall in Np/m Ronneberger α− Dokumaci 10−2 α− Davies α± αwall α+ Davies α+ Dokumaci α+ 10−3 Ronneberger 0 b) -1 ∆c− ph.

263. This fact has been studied experimentally (e.53) βω2 √ iωχ iω + χ FA Prt .4. 212. 366. some extensions have been proposed in [263. 490]). then the additional losses due to the interaction with the turbulent flow boundary- layer need to be considered. except for low frequencies at high static pressures. 229. 366. 490]. This is discussed in the next section.126 duct acoustics smaller than the thickness of the viscous sublayer of the turbulent flow δν . Al- lam and Åbom [13] presented a comparison of the available mod- els and concluded that the model by Howe [212. 413. Howe [212.52) 1 ± M (1 ± M) Dh where Y is the wall shear layer admittance iπ " ω/c 2 √ s r ! e− 4 iων iω  Y= 3/2 ν FA 2 2 .2 Losses due to Turbulent Flow The absorption of sound in a duct is increased by the presence of a turbulent flow boundary-layer. (5. g. When δν . [211. including the attenuation due to turbulent flow: ω/c 2iρc k± turb = + Y. δχ > δτ . [229. a phe- nomenon which occurs when the acoustic boundary-layer thick- ness is larger than the viscous sublayer of the turbulent flow. The additional absorption is due to the transfer of acoustic energy to the turbulent stresses. 214] is the most complete model developed so far44 . δν ρω 1±M κ v∗ ν s s !# (5. 5. which are not included here. δν . 413]) and theoretically (e. δχ < δτ holds for all three models and is generally ful- filled for the parameters studied here. 214] derives the following expression for the wave number. . g. cp κ2 v2∗ χ 44 Since then.

The thickness of the viscous acoustic boundary layer δν is calcu- lated from [212. 432] U(y) 1  v∗ y  = ln +C. v∗ is the friction velocity.5 1 + . so that βc2 /cp = γ − 1 [212] 47 Schlichting and Gersten [432] state that Prt = 0. which will be used here 48 Named after Hermann Hankel (1839-1873. Its empirical value is κ = 0. and h i (1) (1) i H1 (a) cos(b) − H0 (a) sin(b) FA (a.55) v∗ κ ν where U(y) is the flow velocity at a specified distance from the wall y. Also known as Bessel functions of the third kind. when setting y = R and C = 2 in the above equation. 214] gives a value of Prt = 0. 432]. κ is the Kármán constant45 (please note that the same symbol was used to repre- sent thermal conductivity before). β is the coef- ficient of expansion at constant pressure46 . German mathematician). U(y) can be replaced by the mean flow velocity U.54) H0 (a) cos(b) + H1 (a) sin(b) (1) (2) with the Hankel functions48 Hm and Hm . 432]. The Kármán constant is a universal constant for turbulent boundary layers.7. δν is the thickness of the viscous acoustic boundary layer. κ is again the Kármán constant. and C is an empirical con- stant. where the adjustment of C accounts for the dif- ference between the mean flow velocity and the maximum flow velocity at the centerline (y = R) of the duct [212.41 [214. . 214] δν v∗ σ(ω/ω∗ )3   = 6. (5. 5.87. aerospace engineer. 432].4 attenuation of sound 127 Dh = 4A/P is the hydraulic diameter of the duct (with the duct cross-section area A and the duct perimeter P). The friction velocity v∗ can be computed from the empirical pipe flow formula [212. Prt is the turbu- lence Prandtl number47 . Hungarian-American math- ematician. (5.56) ν 1 + (ω/ω∗ )3 45 Named after Theodore von Kármán (1881-1963. 46 For an ideal gas β = 1/T . b) = (1) (1) (5. cp is the specific heat capacity at constant pressure per unit of mass. usually specified as C = 5 [212. while Howe [212. and physicist) [256].

Ronneberger αwall α+ wall.035 m at ISA conditions in comparison to the stationary and Ronneberger’s convective attenuation coefficient: a) attenuation coefficient.Ronneberger −4 ∆c+ ph ∆c− ph −6 1 10 102 103 104 Frequency in Hz Figure 5.wall.128 duct acoustics 10−1 a) α− turb αturb α+ turb α in Np/m 10−2 α− wall. b) change of phase velocity.6: Losses at the duct wall including the effect of a turbulent flow boundary-layer for M = 0.Ronneberger 10−3 0 b) ∆c± ph. .1 and R = 0.turb −2 ∆cph in m/s ∆c± ph.

7 for the average attenuation coefficient and the Mach numbers 0. 214]. At low frequencies. 5. The wall attenuation coefficient for a stationary medium αwall is given as reference and the average turbulence attenuation coefficient αturb = (α+ − turb + αturb )/2 is introduced. The turbulence attenuation coefficient is plotted in Figure 5. 214] ω∗ ν/v2∗ ≈ 0. The same behavior is ob- served for the phase velocity in Figure 5.2 α in Np/m M = 0. where the critical frequency ω∗ can be roughly estimated by [212.1 10−2 M = 0.05. The additional absorption due to the interaction with the turbulent boundary- layer occurs at low frequencies only. which yields better agreement with experiments [212] than the commonly used value of 7.6b. when ω → 0. The dependency on the Mach number is illustrated in Figure 5.035 m.7 yields best agreement with exper- imental results [212.7: Average attenuation coefficient due to losses within the tur- bulent flow boundary-layer at various Mach numbers and R = 0. . where it is compared to Ronneberger’s convective attenuation co- efficient.5. At high frequencies. This value increases at higher frequencies to account for viscous diffusion and for the reduced efficiency of turbulent diffusion. α± turb becomes identical with α± wall.56) estimates the thickness of the viscous acoustic bound- ary layer with δν v∗ /ν = 6.05 αwall αturb 10−3 1 10 102 103 104 Frequency in Hz Figure 5.Ronneberger .01.6a.4 attenuation of sound 129 10−1 M = 0. and σ = 1. Equa- tion (5.

the losses within the fluid become significant at high frequencies (be- yond the audible range) or when sound travels long distances. Bass [28] discussed the validity of the available models under high temperature conditions and came to the conclusion that they are accurate within 10 % for low water vapor concentrations (less than 10 %). [29. [30].4. The loss mechanisms within the fluid can be grouped into classi- cal absorption mechanisms. it will be shown in Section 6. While this generally holds for air at normal atmospheric temperature and pressure.2.5 that the losses within the fluid might become important at elevated pressure and temperature. referred to as molecular relaxation losses. This section will be limited to a collection of the necessary for- mulas. 31. Furthermore. so that higher frequencies are affected. 32]. for example. In duct acoustics. A detailed historical review of the subject is presented by Delany [120] and an in-depth discussion of the physics is given by Bass et al. Such conditions are met when dealing with waves propagating through the atmosphere.1. Today’s state of the art models that describe the sound absorption within the fluid have been estab- lished in a series of publications by Evans et al. and 0. the losses due to the turbulent flow in- crease at higher Mach numbers.6. Generally. [152] and Bass et al. resulting from the fundamental trans- port properties of a fluid. . and the losses based on the movement of the molecules.3 Losses Within the Fluid At normal atmospheric temperature and pressure conditions. these ef- fects are commonly neglected as they are two orders of magnitude smaller than the losses at the duct walls. the subject is treated extensively in the book of Pierce [369].130 duct acoustics 0. which ultimately led to the international stan- dard ISO 9613-1:1993 [236]. 5.

49 The effect of diffusion is known to be small. . mainly because the two ma- jor constituents of air. [314]). the classical absorption is also referred to as viscothermal absorption. for example nitrogen and oxygen (the main constituents of air). However. or [30. so that the classical attenuation coefficient due to viscothermal absorption within the fluid is given as (Stokes [451] and Kirchhoff [261].3 % to the total classical absorption at audio frequencies. thermal conductivity.57) 2ρc3 3 2ρc3 cp 2ρc3 3 Pr | {z } | {z } viscosity thermal conductivity The classical attenuation coefficient is proportional to the square of the frequency: αcl ∝ f2 . 5.4 attenuation of sound 131 Viscothermal Losses The classical absorption is associated with the transfer of kinetic energy of molecules into heat. The effect of viscosity on sound propagation has been treated by Stokes [449. so that only the first two are considered here. (from [120]) 50 The effect of radiation is very small (Stokes [450]). this process typically contributes only 0. 451]. There- fore. The responsible fluid properties are viscosity. Both effects are additive. the degrees of freedom within the molecules generate additional energy losses. and radiation. (5. except possibly at very low pressures or very high tem- peratures (Markham et al. In polyatomic gases. [152]). so that the classical absorption underestimates the attenuation. and the dependency on thermal conductivity has been introduced by Kirchhoff [261]. are of such similar molecular weights. as Rocard [404] and Herzfeld and Litovitz [196] have shown. Radiation absorption plays little role in gases. diffusion. It accurately describes the sound at- tenuation in a monoatomic gas. 369]) ω2 4 ω2 (γ − 1)κ ω2 µ 4 γ − 1   αcl = µ + = + . nitrogen and oxygen. the contribution of diffusion49 and radiation50 to the total losses is very small. Radiation losses are significant only at very low frequencies (Evans et al.

51 In the order of 108 Hz [120]. The attenuation coefficient due to rotational relaxation is given as [30] ω2 γ(γ − 1)R αrot = µ Zrot .58) and (5.6 µ [29. The losses are maximized if the associated relaxation frequency corresponds to the acoustic frequency. 182.8 T . Each relaxation process has its own characteristic relaxation frequency. (5. the absorption is proportional to f2 (a behavior already found for the classical absorption).59) 2ρc3 The bulk viscosity is related to the dynamic viscosity by µB = 0.6 e−16.25 cp where R is the specific gas constant and the rotational collision number is given by −1/3 Zrot = 61.59) give nearly identical results at normal atmospheric temperature and pressure. The time that is necessary for energy to be transferred between the states of motion is called relaxation time. The relaxation frequency of rotational relaxation is much higher than any frequency of interest51 .58) 2ρc3 1. While Equations (5. For frequencies below the relaxation frequency. 369] ω2 αrot = µB . 369]. . How- ever. (5. above the relaxation frequency the absorption approaches a constant value. so that the associated attenuation is proportional to f2 over the whole audible range and beyond [120. A simplified approach accounts for the rotational relaxation losses with a bulk viscosity µB [29.132 duct acoustics Molecular Relaxation Losses The molecular relaxation losses consider the transfer of energy from various states of molecular motion into rotational and vi- brational movements. 369].

i = (αλ)max. 236] 2f f/fr.62) 35 T where Xi is the mole fraction.781.i = αvib. pref T ref T0 p and T are the ambient pressure in kPa and the ambient tem- perature in K. while for air only the two main constituents. while pref = 101.04 · 10 h .15 K are atmospheric reference values.0 K. and pref 0. (5. so that Equation (5. θN = 3352.O + αvib.59) underestimates the attenuation at higher tempera- tures.i αvib.O = 24 + 4. The resulting attenuation coefficient is the sum over all species.i = Xi e−θi /T . For oxygen O and nitrogen N [32.325 kPa and T ref = 293.i is the vibrational relaxation frequency of constituent i.02 + h   p fr. h is the molar concen- tration of water vapor in % (fraction of air molecules that are wa- . 236. 414]: θO = 2239. 4 0. and fr.N (5.i .61) c 1 + (f/fr. 5. are considered: X αvib = αvib. nitrogen N and oxygen O.58) will be used. (5. The relaxation frequencies of the vibrational movements can be much lower than the one for rotation and must be treated for each species separately.170 −1 . θi is the characteristic vibrational temperature.391 + h  1  "  1 #! T −2 T −3  p fr.4 attenuation of sound 133 Equation (5.209.1 K. XO = 0.N = 9 + 280 h exp −4. respectively.60) i The attenuation coefficient due to vibrational relaxation for one particular constituent is given by [30.i )2 with the maximum attenuation over a distance of one wavelength given in Np/m  2 2π θi (αλ)max. XN = 0.

096 K a2 = −1.65) The classical and rotational relaxation losses are functions of pres- sure. Combined Losses Within the Fluid The losses within the fluid are additive52 .134 duct acoustics ter).84408259 a5 = −15.064 × 106 Pa a3 = −11. .64) with a1 = −7. 486]   pws Tc ln a1 ϑ + a2 ϑ1.85951783 a4 = −22. [30] states that this is valid for frequencies up to 10 MHz. (5. temperature and frequency only. It can be calculated from the relative humidity RH (also in %) via the saturation water vapor pressure pws [236. temperature. the state of the art formula is given by [485. as well as pressure.O + αvib.N (5.80122502 ϑ = 1 − T /Tc The saturation water vapor pressure is a function of temperature and is defined for temperatures up to the critical point of water at Tc .7866497 a6 = −1. 369] pws h = RH .5  = pc T (5. and frequency. Often they are combined in one expression αcr = αcl + αrot . so that they are given by the sum αfluid = αcl + αrot + αvib. 52 Bass et al. where the properties of the gas and liquid phase converge.63) p While ISO 9613-1:1993 [236] presents a simplified calculation for pws .5 + a3 ϑ3 + a4 ϑ3. The vibrational relaxation at- tenuation depends on the particular atmospheric constituent and the mole fraction of water vapor.9618719 pc = 22.5 + a5 ϑ4 + a6 ϑ7.6807411 Tc = 647.

53 Please note that the references use a e−iωt time dependency.max.66) c c 1 + i f/fr.8b.018 m/s. Around the relax- ation frequency of each constituent. The wave number regarding the attenuation within the fluid is given by [369. 5. The corresponding increments of oxygen and nitrogen at 288 K are cph.4 attenuation of sound 135 The characteristics of the attenuation within air at normal condi- tions is illustrated in Figure 5.O = 0. The effect of the losses within the fluid on the phase velocity is even smaller than the wall influence (at the specified radius).65) are plotted. where the different contributions according to Equation (5. This is illustrated in Figure 5.O and nitrogen fr.max. while it remains unchanged and identical to c by the classi- cal and rotational relaxation losses alone [414]. The combined losses within the fluid (the sum of all contributions) is given by the solid black line. it should be noted that the attenuation at the wall reduces the phase velocity. the total losses within the fluid are given by αcr alone. At high frequencies. the phase velocity increases by cph. while it is increased due to the losses within the fluid. respectively.N = 0.i = c/π(αλ)max.8a.N are indicated with arrows.174 × 10−3 dB/m. (5.104 m/s and cph.i i The phase velocity is only affected by the vibrational relaxation losses. depending on the relative humidity.i .9 shows the dependency of the attenuation on the rel- ative humidity. Further- more. The attenuation rate of the total losses at 1 kHz is afluid = 8.i kfluid = − iαcl − iαrot − (αλ)max. The relaxation frequen- cies of oxygen fr. . 414]53 ω 2f X i f/fr. Figure 5.max.i .

05 fr.O αfluid in Np/m 10−2 10−3 fr. The relaxation frequencies are fr.N 0 101 102 103 104 105 106 Frequency in Hz Figure 5.136 duct acoustics 100 a) fr.N = 102 Hz.O = 6691 Hz and fr.fluid in m/s 0.N ∆cph.N 10−5 cl α t ro α 10−6 b) fr.O ∆cph.O 0. .O 10−1 αvib.8: Losses within air and its different contributions at ISA condi- tions and RH = 20 % (h = 0.1 ∆cph.337 %): a) attenuation coefficient. b) change of phase velocity.N 10−4 αvib.

.9: Influence of humidity on the combined losses within air at ISA conditions. 5.4 attenuation of sound 137 100 RH = 0 % 10−1 RH = 20 % RH = 100 % αfluid in Np/m 10−2 10−3 10−4 10−5 cr α 10−6 1 10 102 103 104 105 106 Frequency in Hz Figure 5.

.

Two acoustic measurements are performed. e. is commonly referred to as two-source scattering matrix method [1]. the lined duct section is regarded as a black box and only the input and output quantities are considered. [1] . For the evaluation of its acoustic performance. sometimes the term four-pole is used instead of two-port. 138. The reflection and transmission coeffi- cients of a two-port can be determined from the incoming and out- going wave amplitudes based on two linearly independent mea- surements. most of these publications are hard-to-come-by reports. i. The same methodology. two independent physical quantities. 198. the lined duct section is represented by an acoustical two-port2 . 397–399. one with acoustic excitation from a loudspeaker located beyond 1 Unfortunately. the- ses. 360. e. which was developed further by Ronneberger and his students [74. Based on its origin from electrical network theory. 83. The liner is inserted in between two hard-walled duct sections. The state at the input and output is fully defined by two state variables. This method was used for liner or orifice characterization in [17. 399]. within which the acoustic pressure is measured at a minimum of two axial locations. 6 E X P E R I M E N TA L M E T H O D & A N A LY S I S The general methodology of the measurements is based on an approach proposed by Ronneberger [411]. physical system with one input and one output port. in principle. for example. 146. 412]1 . i. or conference papers. 2 An acoustical two-port is a linear.

1 spectral analysis The time series data recorded by the microphones is transformed into the frequency domain following the Welch method3 [488]. In short. . In order to reduce the leakage that is produced by calculating the Fourier transformation of non-periodic4 signals. 6. 4 For example. where the length does not correspond to multiples of the sine period. when compared to not using window functions. the non-periodicity of a harmonic sine signal is introduced by taking time-series data of finite length. Welch [488] proposes to apply a window function to each segment. so that more averages must be taken to obtain the same statistical accuracy of the estimated spec- trum. The con- cept of the Bartlett method involves the averaging over several sig- nal segments in order to reduce the variance of the resulting spec- trum estimation. However. windowed signal segments. The overlapping re-uses the data attenuated by the window function and provides more seg- 3 Named after Peter D.1. Welch [488] overcomes this by allowing the segments to overlap.1 Welch Method The Welch method is an extension and more generalized form of the Bartlett method [27] for spectral density estimation. Welch. Additionally. 6. This would require a longer measurement time.140 experimental method & analysis the microphones in the first duct section and another one with a second loudspeaker located beyond the microphones in the sec- ond duct section. that is the calculation of an averaged spectrum from over- lapped. using this procedure attenuates the signal at the begin- ning and end of each segment. Such a measurement pair is repeated for each frequency of interest. turbulent noise is rejected by using a coherence function method as proposed by Chung [95].

Here. For a discrete time series x(n) with N samples the Hann window function is defined as [47] 1h  2πn i w(n) = 1 − cos with n = 0. . keeping the overall length of the signal the same. 1. The discrete Fourier trans- 5 Named after Julius Ferdinand von Hann (1839-1921.8 2 Time in s Figure 6. Top: A two second time series signal is split into several segments of one second length with 50 % overlap. the term Hanning was introduced by Blackman and Tukey [63] for the process of applying the Hann window function. Originally.6 1. (6. .6 0. This procedure is illustrated in Figure 6.1. 2.1: Illustration of the Welch method. the microphone signals are split into segments of one sec- ond with 50 % overlap and the Hann5 window function is used for the modification of the data.8 1 1. . Bottom: Each segment is modified by applying the Hann window function.1 spectral analysis 141 1 0 p0 -1 1 2 3 1 0 p0 -1 0 0.1) 2 N The modified segments are transformed into the frequency do- main by Fourier transformation [165]. Austrian meteorologist). . It is often referred to as Hanning window.4 1. .2 1. 6. N − 1.4 0. ments to the averaging process.2 0.

.4)   1 X S    S N [X∗s (k) · Ys (k)] for k = 0. (6. The one-sided auto-spectral density. . 6 The complex conjugate of the complex number z = a + ib is given by invert- ing the sign of the imaginary part. e. . 2. . . . Using the Hann p window function X(k) can be corrected by multiplication with 8/3. or short cross spectrum. g. of two signals x(n) and y(n) is given by [47]    2 X ∗ S   [Xs (k) · Ys (k)] for k = 1.2) n=0 The use of a window function introduces some losses to the mag- nitude of the Fourier transformed signal. . or short auto spectrum. In acoustics. (N/2) − 1  S N s=1 Gxx = (6. g.3)   1 X S   2  S N |Xs (k)| for k = 0. (N/2) s=1 where S is the number of segments. N − 1. z∗ = a − ib (e. is then given by averaging over all segments s and normalizing with the number of samples [47]    2 X S   |Xs (k)|2 for k = 1.142 experimental method & analysis formation X(k) of a discrete time signal x(n) with N samples is given by (e. . 2. 2. . . . the spec- trum is commonly scaled to the root-mean-square value by divi- √ sion with 2. (N/2) s=1 where X∗ is the complex conjugate6 of X. 1. [77]). . [47]) X N−1 X(k) = x(n) · e−i2πkn/N with k = 0. (N/2) − 1  S N s=1 Gxy = (6. . Accordingly the one-sided cross-spectral density. i.

The coherence between any two signals x and y is calcu- lated from the auto-spectral and cross-spectral densities [47] . Chung [95] presents a method to reject the turbulent flow noise using three signals measured at different lo- cations. 6.1 spectral analysis 143 6. while the sound pressure is completely coherent. The basic premise is that the flow noise at these three positions is uncorrelated.1.2 Rejection of Flow Noise The pressure fluctuations measured with the microphones are a superposition of the sound pressure and flow induced turbulent pressure fluctuations.

2 γ2xy = .

Gxy .

(6. / Gxx · Gyy .5) .

In that case index 1 refers to the target microphone. and index 3 to the speaker input signal. (6. The result of Equation (6. Here.6)  The indexes 1. 2. and 3 refer to the three signals.6) is a real valued auto spectrum with unit Pa2 . phase reference. the signal of two microphones and the input signal of the loudspeaker are used.  The ‘noise-free’ auto-spectral density G0xx yields [95] G011 = G11 γ12 · γ31 /γ23 . However. index 2 to an arbitrary reference microphone. In- serting Equation (6.2 below) relies on the phase information of each microphone. but fixed for all microphones. the plane wave decomposition (see Section 6.6) and adding the phase of the cross correlation between the target signal and the loudspeaker signal yields |G12 | · .5) into (6. The phase information is added by using the loudspeaker input signal as an arbitrary.

G13 .

i arg(G13 ) .

.

0 G11 = .

.

G23 .

.

f) = G011 (f). .7) The complex pressure amplitude at thep position of the target mi- crophone x1 is then given by p̂ (x1 . e . (6.

which is repeated here in terms of complex amplitudes + p̂− eik +x −x p̂(x) = p̂+ e−ik . this is expressed by Equation (5. .2. xm . 6. . Mathematically. From now on. as illustrated in Figure 6.2: Schematic illustration of the plane wave decomposition with wall flush microphones. . p̂(xm ) x p̂+ p̂− Figure 6. each frequency component will be treated separately and only the tonal compo- nents are of interest. . .2 plane wave decomposition The spectral analysis provides a complex pressure amplitude for each frequency and microphone.9) . x2 . + p̂− eik +x −x p̂(xm ) = p̂+ e−ik m m . Assuming plane waves.32).. This yields the following system of equations: + p̂− eik +x −x p̂(x1 ) = p̂+ e−ik 1 1 + p̂− eik +x −x p̂(x2 ) = p̂+ e−ik 2 2 (6. (6. . . the sound pressure measured by a mi- crophone is a superposition of two waves traveling in opposite direction.144 experimental method & analysis p̂(x1 ) p̂(x2 ) .8) The plane wave decomposition identifies the complex amplitudes p̂+ and p̂− from the knowledge of p̂(x) at several axial positions x1 .

Weyl (1885-1955. for which Uh U = I. and Sir Roger Penrose (born 1931.10) can be obtained by solving [110]7 x = A+ b (6. English mathematical physicist). It can be calculated from the singular value decompo- sition9 of matrix A. French mathe- matician) [249.. American mathematician). More details can be found in [110] or [179]. for example. . 250]. . Schmidt (1876-1959. Then.  e−ikxm eikxm p̂(xm ) | {z } | {z } | {z } A x b The size of the matrices is defined by the number of unknowns n and the number of available data m. for example. the size of x is [2 × 1].   . A solution to Equation (6.10)      .. the pseudoin- verse is a unique. so that n = 2. Italian mathematician) [45].. C. square matrix. E. generalized inverse that exists for matrices of any size.J. where I is the identity matrix. J. German mathematician) [433]. More details can be found in [110] or [179].2 plane wave decomposition 145 This system of linear equations can be written in matrix form as eikx1  −ikx1    e p̂(x1 )  e−ikx2 eikx2   +   p̂(x2 )  p̂  . 8 The Moore-Penrose pseudoinverse is named after Eliakim Hastings Moore (1862-1932. and m equals the number of microphones at distinct axial positions. Here. [110] . 9 The singular value decomposition originates from the works of E. German mathematician and theoretical physicist) [493]. The singular value decomposition is the fac- torization of the matrix A into the product of a unitary10 matrix 7 The reference gives an overview of other solution methods as well. English mathematician) [464– 466]. the size of A is [m × 2]. Sometimes it is called general- ized inverse or just pseudoinverse. the unknowns are p̂+ and p̂− . 6. Jordan (1838-1921. While an inverse matrix only exists for square matrices. [334]. and H. 10 A unitary matrix U is a complex-valued.  × p̂− =  .  (6.11) where A+ is the Moore-Penrose pseudoinverse8 of matrix A. Sylvester (1814-1897. The real-valued equivalent to the unitary matrix is the orthogonal matrix. Beltrami (1835-1899. and the size of b is [m × 1]. [365]. Stewart [446] gives an historical overview from where the original references are taken.

where s is the spacing between the microphones and n = {0.g. The two-microphone method is established as a standardized mea- surement method defined in ISO 10534-2:1998 [232] and ASTM E 1050-12 [22]. (6. by Johnston & Schmidt [245]. or sometimes called conjugate transpose. 435. 1. The sensitivity to errors has been studied by several authors. where Vt denotes the transpose and V∗ denotes the matrix with complex conjugated entries. so that A = UΣVh . For n = 1 this corresponds to a spacing of half a wave length s = λ/2 . 218.10) exists. 69. can be handled. . it can be distinguished between two cases. This so- lution method is very universal.12) 11 The Hermitian transpose. a diagonal matrix Σ. and another unitary matrix Vh . depending on the number of microphones: 1. Then. in slightly varying forms. Seybert & Ros [437]. This means. the system in Equation (6. as it can be used with matrices of any size and even badly conditioned matrices.10) has one unique solution. 2. Generally. in particular [2. where Vh is the Hermitian transpose11 of V [110]. 2. This is commonly referred to as the two-microphone method. With only two linearly independent equations. The number of equations/microphones is greater than the number of unknowns. e. singular or near-singular matrices. . that the solution becomes singular when ks = nπ. It was initially proposed. This is commonly referred to as the multi-microphone method. A solution to Equation (6. One restriction of the two-microphone method is. and Chung & Blaser [96]. [110] . if at least two equations are linearly independent. 438]. is de- fined as Vh = (V∗ )t .146 experimental method & analysis U. microphones need to be installed at least at two distinct axial positions. the pseudoinverse is given by A+ = VΣ+ Uh [110]. }. . The number of equations/microphones and the number of unknowns are equal.

so that Åbom and Bodén [2] suggest to limit the two-microphone method to the frequency range ks 0. Brandes [74] investigates the influence of different microphone distributions on the error in the scattering coefficients. However. The microphones are distributed exponentially with a factor of 1. 411. 237. 12 The method of least-squares minimizes the sum of squared residuals [60]. respectively).1π < < 0. He demonstrates. The effect of the number and spacing of the microphones was studied by Jones [246] and Jang and Ih [237]. The multi-microphone method yields an overdetermined sys- tem of equations.5 and 6. 246]. An exponential distribution. fulfills this requirement best.72 (HAT).2 plane wave decomposition 147 The errors increase rapidly in the vicinity of these singularities.11) is a best fit in the least-squares12 sense. that the density of the microphones should be higher towards the test object. . Both conclude that an equidistant microphone spacing delivers the best results.62 (DUCT-C) and 1.6. (6. He requires accurate measurements over a large frequency range (20-4000 Hz). 246. where the spacing increases with a factor of 1. The solution given by Equation (6. the accuracy of the results can be improved by using more than two micro- phones [94. 412]. the problem with a singular solution at a certain microphone spacing remains. 6. This principle was introduced by Legendre [293] and Gauss [172].72. This reduces errors when extrapolating the result of the wave decomposition from the microphone section to the test ob- ject.13) 1 − M2 which includes the influence of mean flow in the duct. Experimental methods that make use of a least-squares fitting procedure over 3+ microphone posi- tions are presented in [167. Generally. This approach has been followed at the DUCT-C and HAT facilities (see Section 6.8π . The influence of the singu- lar solution is reduced to a smaller frequency range when using more microphones.

148 experimental method & analysis 101 2 Mics.3 compares various microphone setups. some noise of the input signal is compensated in the least-squares fit and the SNR of the result is improved compared to the input signal. 200 mm 100 101 5 Mics. DUCT-C setup 100 0 300 600 900 1200 Frequency in Hz Figure 6. The relative er- ror Arel describes the influence of errors in the input signal on the result of the wave decomposition. 471]. 100 mm Relative Error Arel 2 Mics. indicated by the gray background. 100 mm 3 Mics. The noise of the input signal is reduced when the relative error is below unity. 100 mm Relative Error Arel 5 Mics. The results can be inter- preted as follows: When Arel < 1. 200 mm 3 Mics. The SNR remains un- changed when Arel = 1 and it is amplified when Arel > 1.3: Evaluation of different microphone setups. The procedure of calculating the relative error is explained in [470. Figure 6. equidistant. As . relative to the signal-to-noise ratio (SNR) of the input signal.

At the same time the error at low frequencies is increasing. Thus. The solution to Equa- . e.3 compares microphone setups with two and three microphones. 531 mm compared to 400 mm. At the long distance. Both results are very similar. This singularity moves to 1700 Hz.3 compares a five micro- phone equidistant distribution to the exponential distribution of the DUCT-C. or rather when the wave length is long. The plot at the bottom of Figure 6. The amount of noise and the distance between the microphones determines the frequency un- til which the wave decomposition is able to resolve a small phase difference.12) is fulfilled at 850 Hz. for the frequency range here the dissipation coefficient measured with either microphone distributions would probably be identical. while the DUCT-C setup will not. respectively. However. Adding a third mi- crophone generally improves the result dramatically. The low frequency error is mainly dependent on the maximum distance between two microphones. yields only a very limited fre- quency range where Arel is just below unity. p̂ is measured at five axial po- sitions xm in a hard-walled duct section. is that the error of the equidistant setup will of course peak at 1700 Hz. The phase difference between two microphones becomes smaller at low frequencies. Three micro- phones should really be the minimum configuration. Using two microphones. i. resulting in large errors when using either two or three micro- phones. the results at f < 400 Hz is nearly identical for two micro- phones spaced at 200 mm or three microphones at 100 mm each. A typical result of a plane wave decomposition from DUCT-C data is presented in Figure 6. 6. especially for measurements with flow. The top plot in Figure 6. when using the tighter spacing. At low frequencies the DUCT-C setup is slightly better than the equidistant setup due to its longer maximum distance between the microphones. What is not visible here. spaced equidistantly by 100 mm and 200 mm.4. a good microphone distribution should achieve Arel 6 1 for the relevant frequency range.2 plane wave decomposition 149 a consequence. condition (6. just outside the fre- quency range.

Then. tion (6. p̂. phone and average value.2 -π 0 0 x1 x2 x3 x4 x5 1 2 3 4 5err Axial Position Microphone (a) Complex pressure amplitudes: (b) Fit error of each micro- Experiment.4 0.6 errφ in deg Phase 0. The de- viation regarding magnitude and phase for each microphone be- tween the fitted p̂ and the experimental data yields the fit error . p̂− .4 0 0.6 errmag in % Magnitude 4 0.150 experimental method & analysis 6 0. p̂+ and p̂− respectively. the sound field can be reconstructed with Equation (6.4: Typical result of a plane wave decomposition from DUCT-C data at 714 Hz.2 2 0 π 0.8). p̂+ .10) provides the amplitudes of the waves propagating in positive and negative x-direction. Figure 6.

|p̂exp (xm )| − |p̂fit (xm )| .

.

.

errmag (xm ) = .

.

.

14) |p̂exp (xm )| . · 100 % and (6.

.

.

.

arg p̂ (x ) − arg (p̂ (x )) .

180◦ exp m fit m .

errφ (xm ) = .

.

· . (6.15) .

arg p̂exp (xm )  .

.

π .

16) M m=1 where M is the total number of microphones contributing to the wave decomposition. When the acoustic excitation is introduced by speaker A. Now. The acoustic properties of the two-port are characterized by the reflection and transmission coefficients. as indicated by the bold arrow in Figure 6. we want to determine the four scattering coeffi- cients from the measured wave amplitudes.5a. The four waves are related by the scattering coefficients of the liner and the end-reflections of the duct. from microphone measurements via a plane wave decomposition as discussed in the previous sec- tion.3 reflection and transmission coefficients 151 respectively. which are commonly referred to as scat- tering coefficients. p̂+ and p̂− respectively.3 reflection and transmission coefficients As discussed in the beginning of this chapter. The average fit error of the wave decomposition is given by 1 X M err = err(xm ) . the scattering . 6. They are obtained. (6. The three remaining wave amplitudes can be ex- pressed in terms of the incoming wave amplitude. 6. the lined duct sec- tion can be represented by an acoustical two-port. sepa- rately for both sections 1 and 2.5. it pro- duces the incoming wave p̂+ 1a . This is illustrated in Figure 6. The plane wave sound field in the two hard-walled duct sec- tions on both sides of the liner is fully defined by the complex pressure amplitudes of the waves traveling in positive and nega- tive x-direction.

r .18b) . the scattering coefficients r . and t .152 experimental method & analysis Speaker A p̂+ 1a t+ p̂+ 2a r+ r− r+ e p̂− 1a t− p̂− 2a Section 1 − + Section 2 x=0 x=0 (a) Excitation with speaker A Speaker B p̂+ 1b t+ p̂+ 2b r− e r+ r− p̂− 1b t− p̂− 2b Section 1 − + Section 2 x=0 x=0 (b) Excitation with speaker B Figure 6. (6. t .17a) p̂+ 2a = p̂+ + + + + − 1a t + p̂1a t re r . − coefficients. and p̂2 .17a) and (6. (6. which results in two linearly independent equations p̂− + + − − 1a = p̂1a r + p̂2a t and (6.17c) can be inserted into (6.5a yields p̂− + + + + + − 1a = p̂1a r + p̂1a t re t . (6. and the en- + − + − + − dreflection coefficients r+e and re . − p̂2 . Following the ar- rows in Figure 6.5: Relation between the complex pressure amplitudes p̂+ 1 .17c) Equation (6.18a) p̂+ 2a = p̂+ 1a t + + p̂− − 2a r .17b) p̂− 2a = p̂+ + + 1a t re .17b). and (6. and the end-reflection coefficient. p̂1 .

The equivalent ma- trix notation yields [1] p̂1a p̂− +  −   + 1b = S p̂1a p̂1b . as illustrated in Fig- ure 6. and t− . (6. The scattering matrix S is defined as S 11 S12    + − S = S S = rt+ rt− .18) are not depending on the en- dreflections anymore. four linear independent equations are nec- essary.20) 21 22 The scattering coefficients are obtained by solving Equation (6.18 can be solved for the four scattering coefficients r+ . Two additional equations are obtained by a second mea- surement with excitation from speaker B. t+ .5b. so that the two additional equations are given by p̂+ − − + + 2b = p̂2b r + p̂1b t and (6. However.19) or by basic algebraic transformations of Equations 6.18 − − − p̂− 1a p̂2b − p̂1b p̂2a r+ = + −. to determine the four unknown scattering coefficients.19) + − p̂+ 2a p̂2b p̂− 2a p̂2b | {z } | {z } pout pin where pin and pout are matrices containing the complex pressure amplitudes of the incoming and outgoing waves. (6.21a) p̂+ − 1a p̂2b − p̂1b p̂2a p̂+ + + + 2b p̂1a − p̂2a p̂1b r− = + −.21d) p̂+ − 1a p̂2b − p̂1b p̂2a . The equations are derived accordingly to the excitation with speaker A. r− . and (6.18d) The four linear independent Equations 6. respectively.21b) p̂+ − 1a p̂2b − p̂1b p̂2a − + − p̂+ 2a p̂2b − p̂2b p̂2a t+ = + − . (6. 6. (6.3 reflection and transmission coefficients 153 Please note that Equations (6. (6.18c) p̂− 1b = p̂− 2b t − + p̂+ + 1b r . (6.21c) p̂+ − 1a p̂2b − p̂1b p̂2a − + − p̂+ 1a p̂1b − p̂1b p̂1a t− = + −.

A variation of the two-source location method is proposed by Enghardt [146]. the sum of reflection. where both speakers are operated simultaneously. The simultaneous excitation on both sides ensures good signal- to-noise ratios in both duct sections. in a combustion test rig with acoustic excitation by the flame. e. g. In energy terms. while r and t have been the pressure reflection and transmission coefficients. the reflection and transmission coefficients of the liner can be calculated directly from the incoming and out- going wave amplitude of two independent measurements. The independence of the measurements is achieved by applying the so-called two-source location method [1. e.4 dissipation coefficient An acoustical two-port is characterized by its reflection and trans- mission coefficients.22) R and T are the energy reflection and transmission coefficients. (6. respectively. A derived quantity is the energy dissipation coefficient D. Two independent sound fields are produced by keeping the phase of the input signal constant for one speaker. The energy and pressure quantities are . 6. 344]. g. even when the transmission coefficient of the test object is near zero. commonly with the source fixed at one end of the duct and varying the load. transmission. An alternative approach is the two-load method [307. 477]. where the sound is subsequently incident from either side of the liner. While this method is generally inferior to the two-source location method [344]. it could be the only option when the source cannot be moved. 344. and dissipation coefficient is unity R± + T ± + D± = 1 . which is used here to evaluate the acoustic performance of the liner. at the opposite end. while the phase of the second speaker is shifted by 180° between the measurements. the length.154 experimental method & analysis With above equations. as defined above. 476.

4 dissipation coefficient 155 related by the acoustic energy flux P. 6. For plane waves in a duct with cross-section area A and mean flow Mach number M the energy flux yields [66]13 A .

.

2 P± = (1 ± M)2 .

p̂± .

it can be applied to calculate R and T from r and t. the energy dissipation coefficient is defined as (1 ∓ M1 )2 . Accordingly. Combining Equations (6.22) and (6. respectively.23). (6.23) 2ρc This expression relates the acoustic pressure p̂ to the acoustic en- ergy flux P. .

.

± .

.

2 ρ1 c1 A2 (1 ± M2 )2 .

.

± .

.

Equation (6.5.24) reduces to14 (1 ∓ M1 )2 . When the duct cross-section area and the thermodynamic conditions are the same in both sections. (6.24) (1 ± M1 )2 ρ2 c2 A1 (1 ± M1 )2 where the indexes 1 and 2 refer to the quantities in section 1 and section 2 as defined in Figure 6.2   ± D = 1− · r + · t .

.

± .

.

2 (1 ± M2 )2 .

.

± .

.

. In that case D+ = D− = D 13 See as well the derivations in [374.1 Average Dissipation Coefficient Averaging the dissipation coefficients in positive and negative x- direction yields the average dissipation D = D+ + D− /2 .25) can be reduced further. (6.4. if ṁb  ṁg or M ≈ 0 (when including the additional mass flow).25) (1 ± M1 )2 (1 ± M1 )2 6. (6. then the difference can be neglected and Equa- tion (6. 412] or the entry in Morfey’s dictonary [338] under ‘Blokhintsev invariant’. However.26)  The + and − dissipation coefficents are equal when the liner con- figuration is symmetric and Mg = 0. so that always M1 6= M2 . 14 A bias flow liner injects mass flow into the duct in-between the two sections.2   ± D = 1− · r + · t .

T ± = 1. without liner the theoretical result is known and yields R± = 0. 6. For example in a combustor. The measurement and analysis procedure is the same as if a liner was present. the sound wave travels in flow direction away from the flame and is then reflected at the combuster exit traveling against flow direction towards the flame. a comparison of D al- lows the evaluation of the overall performance of the liner. and D± = 0.156 experimental method & analysis and using D is merely a reduction of data. aver- aged over both directions and expressed in percent . The dissipation error is defined as the absolute deviation from this expected value.4. The grazing flow reduces the dissipation in flow direction while increasing it for waves propagating against the flow. Then.2 Dissipation Error The error that is made when measuring the dissipation coefficient can be determined by a reference measurement without liner. where the main source of sound is the flame. An asymmetric con- figuration or a grazing flow yields D+ 6= D− . In order to suppress the acoustic feedback to the flame D+ and D− are equally important. so that the overall performance can be evaluated by their mean value D. However. taking both directions into account.

+.

.

−.

.

D .

+ .

D .

The absorp- tion and dissipation coefficients follow the same principle. 6.5. (6. It is in particular useful when evaluating different setups or changes in the data analysis (see Sections 6.4.78).3 Compared to the Absorption Coefficient Eldredge and Dowling [142] use the absorption coefficient to eval- uate the performance of the liner. see Equation (4. .2 and 6.5.4). ref ref Derr = · 100 %. rep- resenting the difference between incoming and outgoing energy.27) 2 The dissipation error is a measure of the accuracy of the results.

6.4 dissipation coefficient 157

0.8 0.8

Absorption Coefficient
a) b)
Average Dissipation

0.6 0.6

0.4 0.4

0.2 0.2

0 0
300 600 900 1200 300 600 900 1200
Frequency in Hz Frequency in Hz

Figure 6.6: Measurements with and without anechoic terminations, com-
paring the resulting a) average dissipation and b) absorption coefficient.
with anechoic terminations ( swept-sine excitation), without ane-
choic terminations ( swept-sine excitation).

However, the results can be very different, depending on the en-
dreflections of the test duct. Furthermore, the absorption coeffi-
cient does not distinguish between the two directions of wave
propagation. In order to be aware of these differences, the term
dissipation coefficient is used here15 .
All differences are based on the fact that two measurements are
necessary to determine the dissipation coefficient, while only one
measurement is performed to obtain the absorption coefficient.
The combination of the two measurements eliminates the depen-
dency on the end-reflections, as discussed in Section 6.3. The ab-
sorption coefficient is strongly dependent on the end-reflection,
and thus on the position of the liner within the duct.
This is demonstrated in Figure 6.6. The average dissipation and
the absorption coefficient are presented determined from mea-
surements with and without anechoic terminations. The average

15 However, when referring to the physical phenomenon rather than the nu-
meric quantity, the terms dissipation and absorption are used interchange-
ably.

158 experimental method & analysis

dissipation in Figure 6.6a is identical in both cases. It is inde-
pendent of the end-reflections. The absorption coefficient in Fig-
ure 6.6b exhibits large variations due to the end-reflections. Addi-
tional measurements with swept-sine excitation reveal the deter-
ministic structure of the variations. They are reduced, but not re-
moved, when the anechoic termination is installed. After all, the
anechoic termination only reduces the end-reflections and does
not eliminate them completely16 . However, the plot suggests that
the absorption coefficient converges into the average dissipation
when the end-reflection becomes zero.
Without end-reflection and with excitation from speaker A the
reflection and transmission coefficients are simply given by r+ =
1a /p̂1a and t
p̂− + + = p̂+ /p̂+ (see Figure 6.5). Inserting these into
2a 1a
Equation (6.25) and dropping index a for simplicity yields
2 +2
!
+ (1 − M1 )2 · |p̂− 2
1 | + (1 + M2 ) · |p̂2 |
D = 1−

2 . (6.28)
(1 + M1 )2 ·

p̂+
1

From Equation (4.78) follows with B̂− 2 = 0 (because re = 0) and
+

ρ1 = ρ2
2 2 · |p̂+ |2
!
(1 − M1 )2 · |p̂− | + (1 + M )
α = 1− 1

22 2 . (6.29) (1 + M1 )2 · .

p̂+1 .

The result of the Eldredge and Dowling Method. when setting re = 0 in the Eldredge and Dowling Method. 16 The end-reflection coefficients for the DUCT-C are plotted in Figure 6. which is identical to Equation (6. Thus. the dissipation coeffi- cient and the absorption coefficient are equivalent when the end- reflection is zero. which is the result of the measurements and the Transfer Matrix Method. .10. which is the absorption coefficient. can be compared to the dissi- pation coefficient.28).

The test rig consists of two symmetric measurement sections with the liner module in-between and anechoic terminations17 at both ends. the circular cross-section (DUCT-C) is used.5 duct acoustic test rig 159 6. 212 mm in the shorter version and only at 162 mm in the longer version.1 Setup & Instrumentation A schematic setup of the DUCT-C is shown in Figure 6. The microphones are distributed over ten axial positions with a higher density towards the liner.5 duct acoustic test rig The Duct Acoustic Test Rig (DUCT) provides three setups with dif- ferent cross-section geometries of the measurement sections. A total of 12 microphones are mounted flush with the duct wall. A second liner can be inserted concentric to the first one at diameters 162.10. 192.7. . The maximum axial distance between the microphones in each section is 531 mm and the minimum distance between the two innermost microphones is 60 mm. The two microphones are mounted opposite each other. In order to reduce the interaction between the grazing flow and the opening it is covered by a per- forate within the duct. Each section is made of an 1200 mm long aluminum tube with an inner diameter of 140 mm. The inner diameter of the cavity of both modules is 240 mm. Two microphones are installed at the two outermost positions close to the loudspeakers. They are connected to the duct via a conical horn.5. The di- ameter of the liners is identical to the duct diameter. 18 The influence of evanescent modes is discussed in Section 6.4. 6. canceling out evanescent modes generated by the loud- speaker18 . so that lin- ers with a length of 60 mm and 280 mm can be studied.5. Here. 6. Two loudspeakers are installed at the circumference of the duct at x = ±1041 mm. Two liner modules of different lengths are available. There are four inlets for the bias flow at the outer cir- 17 The effectiveness of the anechoic termination is demonstrated in Figure 6.

Section 1 Liner Section 2 Speaker Microphonesp Microphonesp Speaker Anechoic Anechoic 3 4 56 Bias Flow 7 Termination A 2 8 9 10 11 B Termination experimental method & analysis +−x +x Grazing Flow 140 0 0 200 731 mm 1041 mm 1 12 Figure 6. All dimensions in mm.7: Schematic illustration of the Duct Acoustic Test Rig with circular cross-section (DUCT-C) with speak- ers A and B and microphones 1-12. 160 .

8 mm. 1173 Hz signal 3: 816.19) for speed of sound at ISA conditions and the maximum grazing flow Mach number. 918. the controller is excluded and the flow rate is set manually by adjusting the line pressure. 969. 306. The maximum mean grazing flow Mach number is 0. The grazing flow Reynolds number of all velocities studied here is larger than the critical Reynolds number Rec = 4000 (e. distributed with an even pitch of 90◦ . 765. 357. 255. 1275. 867 Hz signal 2: 408. 1224. 663. 1377 Hz The frequencies are distributed between the signals. the inlets are at the upstream end of the cavity with tubes extending into the cavity and directing the bias flow in axial direction. The signal has been 19 Defined by Equation (5. All three signals together cover a frequency range from 204 to 1377 Hz in steps of 51 Hz. their frequencies are below19 1413 Hz. This was con- firmed by velocity profile measurements [197]. 1071. In the short module they are centered axially and discharge into the cavity in radial direction through an opening of diameter 17. . 714. so that no multiples of a frequency are contained in one signal. no mass flow information is available in this approach. 1122. g. All tonal components are in the plane wave range. 612. 561. the pressure difference across the liner is measured with a MKS Baratron differential pressure meter. At the same time. 1020. so that the flow can be considered fully turbulent. Three signals have been synthesized containing eight tones each: signal 1: 204. How- ever. e. In the rare case when higher mass flow rates are needed. 1326. 459.5 duct acoustic test rig 161 cumference. i. 510. The temperature of the flow is measured in section 2 of the test duct with a J-type thermocouple connected directly to an Agilent 34970A data acquisition / data logger switch unit. The bias flow is controlled with a Bronkhorst ELFlow mass flow controller with a maximum flow rate of 100 kg/h. In the longer version.13. The acoustic test signal is a multi-tone sine signal. 6. [495]).

the output signal of the function generator is recorded at the same settings as a reference signal for the analysis. relative to one reference microphone. The signals are fed through a KME SPA240E stereo amplifier. micro- phones installed around the circumference of the duct at the same axial position should measure the same values for amplitude and phase. so that the relative difference between the microphones can be determined. the acoustic field quantities are constant in any cross- section of the duct and vary only in x-direction. 6. type 40BP condenser microphones in combination with type 26AC pre-amplifiers.2 Microphone Calibration The analysis relies on the magnitude and phase relations between the microphones. which powers two Monacor KU- 516 speakers.5. The microphones used for the acoustic measurements are 1/4" G. Their signals are recorded over a time of 10 s with a 16 track OROS OR36 data acquisition system with a sampling frequency of 8192 Hz. i.162 experimental method & analysis adjusted in a way that the amplitude of each tonal component inside the duct is about 102 dB.A.R. An in-duct calibration method is applied that provides accurate information about the relative magnitude and phase over the whole frequency range. cor- rection values are obtained for a whole range of frequencies. The test signal is generated by an Agilent 33220A function generator. e. The result of such a calibration measurement is a pair of correction values for magnitude and phase for each microphone. Several measurements are required when the total number of microphones is larger than the available positions around the cir- . so that any frequency dependent behavior is accounted for. Additionally. The cor- rection values are then applied in the data analysis. Therefore.S. Typically. The method is based on the fact that the plane waves sound field in the duct is one-dimen- sional.

Microphone 2 cumference.5° 300 600 900 1200 Frequency in Hz Figure 6. Measurement: Microphone 1. Generally.04 1.01 1. In that case it is important. shows some bumps and a non-linear behavior towards the upper frequency limit. The behavior of the wall-flush micro- phones is expected to be smooth and linear over the frequency.99 1. the differences between the Mi- crophones are very small.0° 0. however. as it is unphysical for the microphone response to jump like that between frequencies.8 shows the result of an in-duct calibration measure- ment for two exemplary microphones.00 0. Microphone 2. Figure 6.0° -1. The sound field can no longer be regarded as one-dimensional when the frequency approaches the cut-on frequency of the first higher order mode.0° Phase -0. Linear Fit: Microphone 1.8: Results of an in-duct calibration measurement for two ex- emplary microphones.02 1. The plot. that the same reference microphone is used for all calibration measurements. 6.5° 0. . The non-linear behavior of magnitude and phase for frequencies beyond 1200 Hz can be at- tributed to evanescent modes. The bumps are certainly mea- surement errors.5 duct acoustic test rig 163 1. considering the very detailed scale of the y-axis.5° -1.03 Magnitude 1.

The low error for frequencies below 1300 Hz is maintained. .05. Apply- ing the fitted calibration data gives the overall best result.8. Using the measured calibration data reduces the error even more. Here. so that the evanescent mode effect is fully excluded. the data for frequencies beyond 1200 Hz is extrapolated from the fit. except at the upper frequency end.9. the fitted calibration data. Generally it can be said that the error is already very small without applying any calibration data. or no calibration data at all is demon- strated in Figure 6. The difference between applying the measured calibration data. Both problems are overcome by a linear fit through the mea- sured calibration data. the calibration data is tampered by the existence of evanescent modes and cannot be used. Then. as indicated in Figure 6.164 experimental method & analysis 10 fitted calibration data Dissipation Error in % 8 measured calibration data without calibration data 6 4 2 0 300 600 900 1200 Frequency in Hz Figure 6.9: Comparison of the influence of the calibration data on the dissipation error in a reference measurement with Mg = 0. The fit is limited to frequencies below 1200 Hz. while the falsification at higher frequencies is eliminated. This approach is applied to the measurement data of the DUCT-C facility.

so that it will not be considered here. as observed in Figure 6. The admissible limit20 of ISO 5136:2003 is indi- cated in gray and the anechoic termination complies with this limit. and the DUCT-C setup.3 End-Reflections Both ends of the DUCT-C test duct are equipped with anechoic terminations. so that the sound field is consid- ered to be one-dimensional. the anechoic termination. p.4 Influence of Evanescent Modes The analysis is limited to frequencies below the cut-on frequency of the first higher order mode. The setup of the DUCT-C includes a loudspeaker in-between the microphones and the anechoic termination. 15] in a one-sided configuration with a length of 2.5.5.10 compares the measured reflection co- efficients with Mg = 0 at the open duct. This reflection cannot be separated from the end-reflection and results in an increased reflection at frequencies from 600 to 900 Hz. the magnitude of the reflec- tion coefficient at the open end is unity for low frequencies and decreases for higher frequencies.5 duct acoustic test rig 165 6.15 for frequencies > 160 Hz. 6. The sound coming from the active loudspeaker is partly reflected at the opening of the inactive loudspeaker at the other end of the test duct.5 m. This phenomenon showed no implications on the microphone measurements. some local three-dimen- sional higher order mode effects might be encountered. As expected. 6. The loudspeaker is mounted to the circumfer- 20 ISO 5136:2003 [233] requires the maximum value of the pressure reflection coefficient to be below 0. Regardless. . 21 Another three-dimensional effect is the modal scattering at the leading and trailing edges of the liner [402]. Figure 6. that is in particular21 the existence of evanescent modes in the vicinity of the loudspeaker. The anechoic terminations are designed according to ISO 5136:2003 [233. The anechoic termination reduces the reflection over the whole frequency range.10.

10: Magnitude of the reflection coefficient for open end. and DUCT-C setup. anechoic termination. loudspeaker A and microphones 2. .8 0. This is demonstrated for the DUCT-C setup in Figure 6.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 6. second and third microphone.166 experimental method & analysis Downstream End Upstream End 1 0. ence of the duct.7.21). producing a three-dimensional sound field at the point of entry into the duct. The attenuation rate of mode 1:0 is plotted over the frequency for three distances. so that the mode has decayed when it reaches the microphone. At a sufficient distance only the plane wave prevails. and 4 in Figure 6. The lines are the theoretical predictions of the attenuation according to Equation (5. i. In order to avoid evanescent mode effects the distance between the loudspeaker and the microphone needs to be large enough.11. e. The higher order modes are not able to propagate and their amplitudes decay exponentially with the distance from the source. 3. A mode analysis22 based on the data of the eight microphones is per- 22 The mode analysis is conducted with a DLR software tool described in [470]. The distances coincide with the distances from the loudspeaker to the first. Eight microphones have been installed uniformly spaced around the circumference of the duct at each respective axial location.4 0. The symbols are obtained from measurements.6 |re | 0.

6.1:0 0 x = 310 mm Attenuation in dB x = 527 mm −20 x = 684 mm −40 −60 800 1000 1200 1400 Frequency in Hz Figure 6.11: Attenuation rate of the evanescent mode 1:0 for three loud- speaker–microphone distances in the DUCT-C setup. let’s call it double microphone approach.12. is illustrated in Figure 6.12a shows the spatial distribution of the acoustic pressure of the mode 1:0. Symbols: measurement. Lines: theoretical prediction.11 illustrates that the distance between loudspeaker and microphone is crucial for the decay of the mode.mn )| Figure 6. Another approach is to install two microphones opposite each other and taking their average. In order to obtain the attenuation rate the resulting amplitudes have been normalized with the amplitude at the cut-on frequency: |p̂mn (f)|   amn = 20 · lg (6. The cross-section of the duct in Figure 6.12b.30) |p̂mn (fc.5 duct acoustic test rig 167 fc. eliminating the spatial structure of the first higher order mode. Taking the mean value of any two locations that are 180° apart (demonstrated . A micro- phone which moves in θ-direction along the wall observes a sinu- soidal pressure distribution. formed in order to identify the amplitude of mode 1:0. as shown in Figure 6. The theoretical background to this approach.

for 120° and 300°) eliminates the contribution of this particular mode. The smallest distance between loudspeaker and microphone is 527 mm. The acoustic pres- sure is measured at four axial positions in each section. . The acoustic pres- sure is measured at five axial positions in each section. This is the standard configuration with one micro- phone installed at each axial position. This configuration employs the double microphone approach.168 experimental method & analysis °0° θ p0 300° + 120° 0 300° θ 120° − (a) Acoustic pressure distribution (b) Acoustic pressure distribution in a duct cross-section. The acoustic pres- sure is measured at five axial positions in each section. 23 The numbering of the microphones refers to the labels used in Figure 6. 18 microphones Microphones 3-10 are used.7.12: Illustration of the double microphone approach applied to mode 1:0. 12 microphones Microphone 1-12 are used. The effect of the distance and the double microphone approach is demonstrated by studying the following three configurations23 : 10 microphones Microphones 2-11 are used. The smallest distance between loudspeaker and microphone is 310 mm. Figure 6. The shortest distance between loudspeaker and microphone is 310 mm. around the duct circumference.

so that this approach is taken for the sole purpose of demonstrating the influence of the distance.5 duct acoustic test rig 169 30 8 Microphones Dissipation Error in % 10 Microphones 12 Microphones 20 10 0 300 600 900 1200 Frequency in Hz Figure 6. This configuration is referred to as ’8 Microphones’.13 reveals that the effect of evanescent modes is successfully elimi- nated. In order to enlarge the distance in the existing test rig. This is due to the fact that the sound field contains some rele- vant three-dimensional components which are not accounted for. the micro- phones close to the loudspeakers are excluded from the analysis. Looking at the result in Figure 6. Of course the reduction of microphones will create other disad- vantages. The dissipation error for the standard configuration with 10 mi- crophones increases dramatically for frequencies beyond 1200 Hz. based on the total number of microphones that are in use. the microphones are too close to the loudspeaker. reducing the error at frequencies beyond 1200 Hz. In other words. Figure 6. 6. Increasing the distance between loudspeaker and microphones is the obvious solution trying to suppress evanescent mode effects. .13 compares the results of these three configurations.13: Comparison of three reference measurements with differ- ent microphone configurations regarding the suppression of evanescent modes.

. where the theoretical results are known (D± = 0). all micro- phones need to be installed in a double configuration. without imposing any other limitations. The double microphone approach is not limited to evanescent modes. It is obtained from a reference mea- surement without liner.5 Accuracy The dissipation error.4. This approach is applied here. defined in Section 6.2.170 experimental method & analysis The double microphone approach achieves the same result at a much shorter distance between loudspeaker and microphones. This gives the benefit of suppressing evanescent mode effects and thus extending the frequency range. The influence of the circumferential structure of the first higher order mode is successfully suppressed by averaging over two mi- crophones separated by 180°. 6.5. At Mg = 0. Then. The same principle can be applied when the first circum- ferential mode is propagating through the duct. Figure 6. so that the standard setup of the DUCT-C employs 12 microphones. The upper frequency limit would move from the cut-on frequency of the first circumferential mode to the second circumferential mode. the dissipation error is well below 1 % for most fre- quencies. especially at low frequencies. The benefit of such a setup would be an extended frequency range. The setup and anal- ysis has been designed for high precision acoustic measurements. For the DUCT-C geometry that would shift the limit for Mg = 0 from 1425 Hz to 2364 Hz. The grazing flow increases the error slightly.14 plots the typical dissipation error of the DUCT-C at three grazing flow Mach numbers. is a measure of the accuracy of the results.

Temperature and pressure can 24 The HAT is a joint facility of DLR and TU Berlin.05 Mg = 0. 6. while offering a limited accuracy only.6 hot acoustic test rig 171 10 Mg = 0 Dissipation Error in % 8 Mg = 0. So far they have mostly served as a quality check of a given liner design. The main flow can be heated up to 820 K (550 ◦ C) with an electric air heater and the maximum static pressure is 1100 kPa absolute. On the other hand. so that they are very important for the understanding of the basic parameter relations.14: Dissipation error of the DUCT-C determined from reference measurements at various grazing flow Mach numbers. The Hot Acoustic Test Rig24 (HAT) fills the gap between fundamental studies and application tests. 6. Chair of Aero Engines. It enables liner mea- surements at elevated pressure and temperature in an acousti- cally well defined environment. measurements including combustion are extremely demanding and costly.1 6 4 2 0 300 600 900 1200 Frequency in Hz Figure 6.6 hot acoustic test rig Perforated liners are commonly studied at normal atmospheric pressure and temperature conditions. . These tests can deliver re- liable results of high accuracy.

which is connected to the test duct via a cylindrical shaft with an inner diameter of 50. resembling the geometry of the duct. so that the flow velocity within the neck is about 0. The whole test duct is covered by a custom taylored. On the side of the test duct the open- ing is covered with a wiremesh. Its amount is adjusted.6. Air cooled and pressure proof microphone probes are installed at ten axial positions upstream and downstream of the liner. keep- ing thermal losses at a minimum. This low velocity flow prevents the hot air from entering the speaker housing and ensures a negligible influence when discharged into the hot duct flow. 6. For the speakers to operate at elevated and changing static pres- sure it is crucial to allow pressure equalization within the driver.15. two layer insulation.8 mm. It consists of two symmetric measurement sections with the liner module in- between and anechoic terminations at both ends.4 m. They are distributed with an increasing density towards the liner.1 Setup & Instrumentation The layout of the Hot Acoustic Test Rig is based on the success- ful design of the Duct Acoustic Test Rig. but too small to affect the dynamic behavior of the speaker. The speaker housing is flushed with cooling air to ensure a temperature of the speaker below 50 °C at all times. This was accomplished by manually introducing a small gap into the driver’s sealing. The cooling air discharges into the test duct. The test duct is made of a high temperature steel tube with an inner di- ameter of 70 mm and a total length of about 5. Two loudspeakers are installed at the circumference of the duct at x = ±1415 mm.172 experimental method & analysis be adjusted independently and continuously from ambient up to their maximum values.2 m/s. large enough to enable a balancing flow. A schematic illustration of the setup and instrumentation is given in Figure 6. The . They are encased in a pressure proof housing.

temperature probes T1-T7.15: Schematic illustration of the Hot Acoustic Test Rig (HAT) with speakers A and B. All dimensions in mm. 6.6 hot acoustic test rig 173 . and pressure taps P1-P4. Section 1 Liner Section 2 Anechoic Speaker Microphonesp Microphonesp Speaker Anechoic Termination 1 2 3 45 Bias Flow 6 7 8 9 10 Termination A B +−x +x Grazing Flow 70 0 0 200 1115 mm 1415 mm T1 T2 P1 T3 T4 P2 T5 T6 P3 T7 P4 Figure 6. microphone probes 1-10.

maximum axial distance between the microphone probes in each section is 915 mm and the minimum distance between the two in- nermost microphone probes is 85 mm. 1530. are logged at a 0. Three signals have been syn- thesized: 1: 204. 1122 Hz 2: 1224. 1938. 2754 Hz All three signals together cover a frequency range from 204 to 2754 Hz in steps of 102 Hz. 714. 1734. 1428.4 0. while their actual values. The test . 1020. 1836. 408. 2142 Hz 3: 2244. 2448.8 Mach Number Figure 6. as illustrated in Figure 6. The test signal is a multi-tone sine signal.2 0. 306.16: Upper limit of the plane wave frequency range in the HAT. Desired values for the temperature and pressure in the duct. 2346. 612. 510. 1632. 2550. The acoustic measurements are controlled with a separate PC. as well as tem- perature and pressure data.1 Hz frequency. 2040. 1326. The upper frequency limit for plane wave sound propagation is depending on the temperature and grazing flow Mach number. 918.6 0. 816. 2652. as well as the mass flow rates of the cooling and bias flow can be set. depending on Mach number and temperature.174 experimental method & analysis 800 4500 Temperature in K 700 0 400 600 0H z 50 =3 500 fc 00 30 400 00 25 00 300 20 0 0. The operation of the Hot Acoustic Test Rig is controlled with a PC via a LabView interface.16.

The analysis of the acoustic data requires an accurate knowledge of the temperature. Additionally. the output signal of the function generator is recorded at the same settings as a reference signal for the analysis. The temperature of the bias flow is measured in the cavity of the liner (see T7 in Figure 6. as well as a constant temperature along the measurement sections. which powers two Monacor KU-516 speakers. that is upstream and downstream of each loud- speaker and upstream and downstream of the liner (see T1-T6 in Figure 6. as well as the discharge of cooling air through the microphone probes and loudspeaker housings. the amount of bias flow can be substantial.15). The sensors are resistive temperature detectors (RTD) of type PT100 connected in four-wire configuration to a digital multimeter (Agilent AG34972). so that the temperature in the downstream section can be much lower than the temper- .6 hot acoustic test rig 175 signal is generated by an Agilent 33220A function generator. there is a temperature sensor installed in each speaker housing for health monitoring (not shown in Figure 6. while the injection of cooling air through the microphone probes.5°C for the complete temperature range of 280-820 K. The placement of the temperature sensors allows the assessment of temperature changes along the duct. the speaker housing and the liner would cause a sudden change of temperature.15). Additionally. In- sufficient insulation would produce a continuous temperature gra- dient. 6. The acoustic pressure is recorded over a time of 30 s with a 16 track OROS OR36 data acquisition system with a sampling fre- quency of 8192 Hz. The signals are fed through a KME SPA240E stereo amplifier. Their ac- curacy is specified as ±0. However. Temperature measurements The temperature of the main flow is measured at six positions along the duct.15). Several tests have been made and it can be concluded that the temperature gradient due to insufficient insu- lation is negligible.

until now all measurements dealt with double skin configurations . The third pressure tap enables the measure- ment of the pressure drop in a triple skin configuration.15). They are lo- cated upstream and downstream of the liner (see P1 and P2 in Figure 6.) 9116 pres- sure scanners.15). This enables the measurement of the pressure drop over each layer of a double skin configuration. In extreme cases the temperature difference between sec- tion 1 and 2 can be as much as 160 K.15) and at various positions in the liner cavity (see P4 in Figure 6. The pressure in the liner cavity is measured at three different radial positions.15 % FS for the larger ranges. 69. While the aver- age value is used for post-processing. For redun- dancy there are two pressure taps at each location. The actual temperature difference depends on the initial temperature. The change in temperature is unproblematic as long as the temperatures of both sections are known and considered separately in the acoustic analysis. The pressure taps are con- nected to two modules of PSI (Pressure Systems. and 1380 kPa. the measured quantities include the pressure drop of the innermost liner and the total pressure drop across both layers. The pressure in the duct is given as absolute value.5.5 and 5 kPa ranges and ±0. while the pres- sure in the liner cavity is given as the differential pressure refer- ring to the absolute pressure in the duct. In fact. 5. the difference between both values serves as an accuracy indicator.05 % FS for the 2. Inc. The temperature of the upstream section serves as the characteristic temperature of a test point.176 experimental method & analysis ature in the upstream section. However. They have separate sensors for absolute and differ- ential pressure measurements and are divided into four blocks of different pressure ranges: 2. The accuracy is dependent on the full scale (FS) value of the range and is specified as ±0. Static pressure measurements The static pressure is measured using pressure taps. and the ratio of grazing and bias flow. at the nozzle (see P3 in Figure 6.

At the critical pressure the velocity in the nozzle equals the speed of sound M = 1.2 Air Supply The air for the HAT facility is provided by two rotary screw com- pressors that can deliver a total mass flow rate of 0. and a nozzle at the end of the test duct. The air is dried (at a dew point of 276 K). For pressures beyond the criti- cal pressure the velocity remains constant. The liner bias flow. filtered. The air heater can increase the temperature of the flow to any value between ambient and 820 K. The main duct flow is fed through an electrical air heater before it enters the duct. . and the speaker cooling flow are delivered at these conditions.78 kg/s. The resulting flow velocity in the duct is defined by the geometry of the nozzle and can be determined by the laws of gas dynamics. The supply for each consumer can be set individually via mass flow controllers. The flow rate and pressure in the duct is controlled via a pres- sure valve. so that the two outermost pressure taps measured the same pressure.6 hot acoustic test rig 177 only. Raising the pressure in the duct increases the flow velocity until a critical pres- sure is reached. 6. The pressure in the duct can be adjusted continuously between ambient and 1100 kPa absolute pressure. and then delivered into a 2 m3 pressure reservoir at a temperature of 288 K at 1600 kPa of pressure.6. the microphone cooling flow. 6. It is custom made by DLR and has a maximum power rating of 540 kW. a volume flow meter.

32) where Ψ describes the discharge of the fluid. (6.325 kPa. and the pressure in the duct is increased until the critical condition is reached. The mass flow rate through the nozzle is given by [431] √ ṁ = A2 2p1 ρ1 Ψ .1 kPa at 773 K. the pressure downstream of the nozzle is the ambient pressure. so that the pressure upstream is the stagnation pressure and the pressure downstream is the critical pressure and Equation (6.9 kPa at 293 K and p∗1 = 189.178 experimental method & analysis The critical pressure in the duct p∗1 can be calculated from the isentropic relation for the pressure [431]25 γ 2   γ−1 p2 = (6. so that the critical pressure yields p∗1 = 191. The heat capac- ity ratio for air varies slightly within the temperature limits en- countered here (see Table a.31) is adapted accordingly. For duct pressures below p∗ it is given by v "  2   γ+1 # u u γ p2 γ p2 γ ∗ p1 < p1 : Ψ = t − . the nozzle back pressure equals the ambient pressure p2 = 101. the velocity upstream of the nozzle is assumed zero. (6.34) γ+1 γ+1 Applying the continuity equation to Equation (6.10).33) γ−1 p1 p1 while for higher pressures 1 r 2   γ−1 ∗ γ p1 > p1 : Ψ = . Here. Then. Here. p∗1 is only dependent on the heat capacity ratio γ. .35) A1 γ 25 Commonly. . The notation in Equation (6. (6. which remains constant.31) p∗1 γ+1 where the indexes 1 and 2 denote the quantities in the duct and downstream of the nozzle. .32) an expression for the duct Mach number can be found A2 2 r M1 = Ψ. (6.31) is commonly written as p∗2 /p1 = . respectively..

. Please note the log scaling of the pressure axis.17: Relationship between pressure and Mach number for vari- ous nozzle diameters of the Hot Acoustic Test Rig at two temperatures.2 40 ma ssfl ow limit 0.3 m as sfl ow 0.2 40 lim it 0. 6. .4 b) T = 773 K 55 Mach Number 0.4 a) T = 293 K Mach Number 0.3 5 5 0. Lines: theoretical predictions.1 30 20 15 0 0. Symbols: measurement..1 30 20 15 0 1 P*. 3 4 5 6 7 8 9 10 Pressure in 105 Pa Figure 6.6 hot acoustic test rig 179 0.

18. Below the critical pressure the Mach number is a function of the pressure.027 0.180 experimental method & analysis Table 6.106 0. However. Increasing the pressure would also increase the Reynolds number. The limit can be seen in the gap between the experimental values and the mass flow limit in Figure 6.17 for five nozzles with diameters between 15 mm - 55 mm and at temperatures of 293 K and 773 K.17 are lines of constant temperature in Fig- ure 6. The numerical values of the Mach num- bers at critical condition are given in Table 6.35) can be used to describes the Mach number char- acteristics of the Hot Acoustic Test Rig. so that a change of pressure changes the Mach number. The mass flow limit curves in Figure 6.359 Equation (6. so that the flow can be considered fully turbulent. The dependency of the Reynolds number on the temperature is plotted in Figure 6.047 0. The dependency of the maximum Mach number on pressure and temperature is plotted in Figure 6.18.19 for various grazing flow Mach numbers at ambient pressure. Nozzle Diameter in mm 15 20 30 40 55 Mach Number 0. The maximum Mach number for a given nozzle diameter remains constant with temperature. The higher temperature pushes the mass flow limit to higher pressures. the capacity of the air heater is lim- ited. depending on the mass flow rate and the flow speed.1: Available nozzles and associated grazing flow Mach numbers in the Hot Acoustic Test Rig when p1 > p∗1 .17b.35) and the measured values are in good agreement. The Reynolds number of the grazing flow is always larger than the critical Reynolds number (Rec = 4000).189 0.1. This is demonstrated in Figure 6. . The theoretical predictions according to Equation (6.

325 kPa.04 7 0.19: Reynolds number for various Mach numbers plotted over the temperature at p = 101. . Please note the log scaling of the pressure axis.2 =0 0.4 0.1 500 M 400 300 1 2 3 4 5 6 7 8 9 10 Pressure in 105 Pa Figure 6.6 700 0.18: Maximum Mach number depending on pressure and tem- perature.6 hot acoustic test rig 181 800 0.189 105 0. Reynolds Number 0.5 0.027 104 300 400 500 600 700 800 Temperature in K Figure 6.7 Temperature in K 0.106 M = 0.359 0.3 600 . 6.

20: Design of the microphone probes used for acoustic measure- ments at the Hot Acoustic Test Rig. .182 experimental method & analysis cooling air inlet signal output semi-infinite tube cooling air microphone housing sound mounting flange Isometric view Scale: 1:1 measurement location Figure 6.

6.S. In order to overcome this limitation a microphone probe has been developed at DLR. Commercial products are typically based on a piezoelectric sensor. 26AC). 6.A. the design has received a major upgrade to include the ability of measuring in high pres- sure environments of up to 2000 kPa. following the principle of a semi-infinite tube.3 Microphone Probes Accurate acoustic measurements in high temperature flow are very challenging. for acoustic measurements in a laboratory combustion chamber. Recently. 40BP or 40BH) and its pre-amplifier (G. The housing contains a standard quarter-inch condenser microphone (G. The microphone probes have been successfully tested in a combustion facility with respect to their safety and data reproducibility at conditions up to 1350 kPa and 1500 K. After a certain length a microphone is mounted perpendicular to the tube in- side a pressure-proof housing. The overall length of this tube is about one me- ter which is mostly wound-up in a spiral to save space. so that the end of the tube is flush with the wall. Their maximum tolerable temperature is typ- ically around 620 K (in a very few cases the specifications go as high as 920 K). the tube is extended be- yond the location of the microphone. The original design has been used successfully for several years. The tube below the flange is inserted into the wall of the test duct.A. In order to prevent reflections of sound. The probes are mounted to the test duct via a flange.R. g.6 hot acoustic test rig 183 6. The tube extends outwards with a constant diameter.S. The acoustic signal is transmit- ted from the tube to the microphone membrane via a small open- ing. the sensitivity of piezoelectric pressure sensors is much lower compared to a standard condenser micro- phone. However. e.20. This is the measure- ment location where the acoustic pressure is to be recorded.R. The microphone signal is passed on through the wall with a pressure-proof wall bushing that ends in a LEMO connector. The large . The design of the microphone probes is shown in Figure 6.

However. so that the mass flow rate has to be adjusted according to the change in density. i. This corresponds to a mass flow rate of 0. the diameter of the tube. small variation in manufacturing and assem- bly require the determination of an individual transfer function for each probe. so should be their transfer functions. so that the frequency range of interest is free from resonances. The transfer function has to be considered in the analysis. prevent- ing hot gases from the duct to enter the tube and get into contact with the microphone membrane. correcting the sig- nals from the transfer characteristics of the probe. A probe together with a microphone is forming one unit. A small amount of cooling air is fed through the tube. The required value for the cooling flow velocity is held constant at elevated pressures. Each microphone probe has a characteristic transfer function which is determined by the geometric dimensions of the probe. the distance between the measure- ment location and the microphone membrane.054 kg/h at atmospheric conditions. At at- mospheric pressure it was shown that a velocity of 4 m/s in the tube is enough to ensure the safe operation of the microphones. Then. A study regarding the influence of the cooling flow on the acoustic measurement was done with the initial design of the microphone probes (unpublished).184 experimental method & analysis ratio between diameter and length of the tube provides sufficient viscous damping. Three different ways of calibration have been established: . a calibration curve for the whole unit is determined. which is only separated if a component is damaged. While the speci- fied geometry is identical for all probes. the calibration curve includes the transfer function of the probe as well as the behavior of the microphone itself. etc. A disturbance of the signal by the cooling air could only be observed for velocities beyond 30 m/s. e. Instead of deter- mining the transfer function of the probe separately.

Probe 12. is saved in the data acquisition system and is directly applied to the recorded data. 1. Absolute calibration: Calibration with a pistonphon. Probe 16. i. The two other options apply the in-duct calibration method that was described in Section 6. regard- ing the magnitude at one frequency. 6. The absolute calibration defines the deviation of the measured signal to a reference signal supplied by a pistonphon. Probe 08.4 0° -180° ∆f Phase ∆ϕ -360° -540° -720° 500 1000 1500 2000 2500 Frequency in Hz Figure 6.21: Quasi-absolute calibration curves at ambient conditions: Reference. The quasi-absolute calibra- tion employs a wall-flush mounted condenser microphone as ref- . 124 dB at 250 Hz.5.6 hot acoustic test rig 185 1 Magnitude 0.8 0. expressed as a sensitivity in V/Pa. Quasi-absolute calibration: Calibration regarding the mag- nitude and phase over the complete frequency range with respect to a reference condenser microphone.2.6 0. 2. 3. The magnitude correction value. e. Relative calibration: Calibration regarding the magnitude and phase over the complete frequency range with respect to a reference microphone probe.

Probe 08.38 · 10−4 s (the negative sign indicates the delay). say (1020 Hz. The refer- ence magnitude and phase have a constant value of unity and zero. The condenser microphone is calibrated with a piston- phon. The plot shows the frequency dependency of magnitude and phase. -266.0 0. so that their magnitude correction values at 250 Hz are equal to unity.2 Magnitude 1. Probe 16. All three microphone probes have been cali- brated with a pistonphone previously. Assuming c = 343. - 512. Plots of the magnitude and phase correction values for three ex- emplary microphone probes are given in Figure 6. respectively. so that the results can be considered to be quasi-absolute. The continuous slope of the phase reveals that there is no resonance within the frequency range.186 experimental method & analysis 1.15 K) gives .8 0. yields a time delay of ∆t = −6.6 0° Phase -10° -20° 500 1000 1500 2000 2500 Frequency in Hz Figure 6. The slope itself defines the time delay of the acoustic wave traveling from the measurement location to the mi- crophone: ∆t = ∆ϕ/360°/∆f.38 m/s (at 293.22: Relative calibration curves at ambient conditions: Probe 04 (Reference).8°) and (2091 Hz. Picking two random points from the phase curve of probe 16.8°). erence. Probe 12.21.

Figure 6. The influence of temperature and pressure on micro- phone probe 20 is shown in Figures 6. Furthermore. as the microphone probes are flushed with cooling air of constant temperature while the duct temperature increases. The reflections can be avoided by further extending the tube.21. is the same within the probe and the duct. 6. The static pressure on the other hand. respectively. Each of the three methods is important to characterize the be- havior of the microphone probes. The curves look very different to Figure 6. so that the calibration can be done at the ac- tual operating condition of the measurements. The actual temperature within the probes does not change that much. This effect becomes more and more visible while the pressure increases. the pressure dependency can be handled very well. The viscothermal losses within the tube are decreasing considerably at high pres- sure. The advantage of the third method is that it only employs microphone probes. . the proce- dure is based on the in-duct calibration method described before. However. This agrees very well with the actual distance of 220 mm. This is expected. The calibration curves exhibit no dependency on the tempera- ture in the duct. A strong influence on both. so that the assumption of the semi-infinite tube is violated and reflections occur. However. it serves as an easy way of health checking in-between measurements. Again. Doing so reveals the dependency of the microphone probes on temperature and pressure. due to the ability to provide magnitude and phase correction values for each operating condi- tion.23 and 6.22 shows the resulting calibration curves for the same microphone probes from the previous plot. magnitude and phase. The absolute calibration with the pistonphon equalizes the levels of all microphone probes. but they basically contain the same infor- mation. The third calibration method is the relative calibration with re- spect to a chosen reference microphone probe. can be observed. The quasi-absolute calibration serves as a one-time check that the design was successful (no res- onances).1 mm between measurement location and micro- phone.6 hot acoustic test rig 187 a distance of 219.24.

6 0° Phase -10° -20° 500 1000 1500 2000 2500 Frequency in Hz Figure 6. 1000 kPa. 573 K.24: Relative calibration curves of microphone probe 12 at 293 K and 101 kPa. 1. .8 0. 423 K.188 experimental method & analysis 1.0 0. 773 K.8 0. 600 kPa.2 Magnitude 1.0 0.2 Magnitude 1.23: Relative calibration curves of microphone probe 12 at 200 kPa and 293 K. 200 kPa.6 0° Phase -10° -20° 500 1000 1500 2000 2500 Frequency in Hz Figure 6.

64) and results in pws (Td ) = 758.051 %. Its values range from 32.054 Pa.325 kPa).6 hot acoustic test rig 189 it should be noted that applying a new absolute calibration value makes any previous relative calibration invalid (the magnitude curve will have an offset). at least at the beginning of a measurement campaign. The relative humidity in the saturated state is RH = 100 %.096 K and 101. a rel- ative calibration is performed in regular intervals. .63) as h = 0. The temperature in the Hot Acoustic Test Rig is always well above the dew point temperature and the pressure is always well below the saturation pressure at the respective temperature. The relative humidity might be the more well-known quantity.6. The relative calibration at operating condition is crucial for the analysis of the data.15 K and 101. 6.9 % (at 288.325 kPa). the saturation pressure at operating temperature pws (T ). and the op- erating pressure p. so that the molar concentration of water vapor is given by Equation (5.4 Estimation of the Humidity The treatment of the acoustic losses at elevated pressure and tem- perature in the next section requires some knowledge about the humidity.15 K and 1100 kPa) to 0. Therefore. so that the mo- lar concentration of water vapor in the air stays constant and is independent of the operating temperature and pressure. The saturation pressure at the dew point temperature is calculated from Equation (5.051 %.15 K (3 °C) at an op- erating pressure of 1500 kPa. The molar concentration is determined by the operating conditions of the air dryer. The air provided for the Hot Acoustic Test Rig is dried at a dew point temperature Td = 276. while it is 4 % at ISA conditions (288. It can be calculated from Equation (5. The relevant quantity for the acoustic losses is the mo- lar concentration of water vapor h (fraction of air molecules that are water). 6.63) with h = 0.0002 % (at the critical temperature Tc = 647.

36a) . recent measurements [277] revealed an increasing attenuation of sound at elevated pressure and temper- ature.190 experimental method & analysis The given values of relative humidity and molar concentration of water vapor are estimates only. hard-walled duct with flow are discussed in Section 5.15 K (3 °C) and the operating pressure of the air drier of 1500 kPa is a maximum value. 445].051 % can be regarded as a minimum estimation. However. and their dependency on pressure and temperature has been pre- sented by Lahiri et al. The value of h = 0. so that they cannot be neglected any longer. as deviations in the dew point temper- ature are expected rather to higher temperatures than below the specified 276. [278]. the losses due to turbulent flow and the losses within the fluid. which cannot be justified by the viscothermal losses at the wall alone. This is common practice and reproduces the physics accurately [273.4. A rigorous review of the other loss mechanisms. so that they are usually neglected.5 Attenuation at Elevated Pressure and Temperature The various acoustic losses that occur when sound is propagat- ing through a smooth. The uncertainties lie in the ac- tual dew point temperature and variations in the operating pres- sure of the air dryer. then the total losses are given by k± ± ± total = ω/cph.total − iαtotal (6. i. It was found that the molecular relax- ation losses within the fluid become more important when pres- sure and temperature are increased from their ambient values. It is assumed that the losses at the walls and the losses within the fluid are additive and that the effect of convection on the at- tenuation within the fluid can be accounted for by division with 1 ± M. e.6. At ambient conditions the losses within the fluid are generally two orders of magnitude smaller than the viscother- mal losses at the duct wall. 6. 316.

the losses due to rotational and vibrational molecular relaxation.turb are given in Section 5. for example.36c) The total losses include the viscothermal losses within the fluid. The convective in- fluence of the flow on α± total is small.36b) c± ± ph.fluid .total = (c + ∆cph. The attenuation coefficient at elevated pressure and tempera- ture has been determined experimentally with the Hot Acoustic Test Rig. According to Equation (6. and the absorption due to the turbu- lent flow boundary-layer.25.36b). While there is some scatter in the experimental data. the convective effect of the flow on the wall losses. Fig- ure 6.4.turb )(1 ± M) .36b).051 %. However. the viscothermal losses at the duct wall. the total losses are the superposi- tion of αfluid and α± turb . 6. . except at low frequencies where the losses due to the turbulent flow boundary-layer become prominent. Details of the measurements can be found in [278].25 shows a comparison between the experimental data and the theoretical predictions of Equation (6. ± ∆cph. which is a valid assumption. (6. The respective equations for αfluid . so that only the average value αtotal = (α+ total + α− total )/2 is plotted. αturb ± is mostly determined by the viscothermal losses at the duct wall. which are also plotted in Figure 6. α± turb . Setting h = 0. the attenuation in the measured frequency range (indicated in gray) is dominated by the losses within the fluid at both operating conditions.08 % for the model matches the experimental data slightly better than the es- timate h = 0.fluid + ∆cph. This corresponds.6 hot acoustic test rig 191 with the attenuation coefficient and phase velocity given by α± ± total = αfluid /(1 ± M) + αturb and (6. and ∆cph.65 K and 1300 kPa. it matches the theoretical predictions rather well. to a change of operating condition of the air dryer to 280.

984 kPa ui d α fl 10−1 α in Np/m α turb 10−2 αexp αtotal 101 102 103 104 Frequency in Hz Figure 6. M = 0.25: Comparison of the attenuation coefficient determined exper- imentally and from theoretical predictions at two elevated temperature and pressure conditions with h = 0.05. b) 773 K. and R = 0.192 experimental method & analysis a) 773 K.08 %. .035 m: a) 773 K. 196 kPa. 196 kPa 10−1 α in Np/m 10−2 αturb id αexp u α fl αtotal b) 773 K. The frequency range that is covered by the experiments is indicated in gray. 984 kPa.

At normal atmospheric temperature and pressure and without flow. the dissipation error is generally below 3 %. 6. 6.26 for four operating conditions.6. 1100 kPa.325 kPa. 773 K.6 Accuracy The dissipation error is plotted in Figure 6. 1100 kPa.26: Dissipation error as a measure for the accuracy at the four extreme operating conditions: 288 K. 288 K. The error increases at elevated pressure and temperature. . 200 kPa. 773 K. 101.6 hot acoustic test rig 193 20 Dissipation Error in % 15 10 5 0 500 1000 1500 2000 2500 Frequency in Hz Figure 6.

.

The text presents a selection of results of significant im- portance.1. the inci- dent wave amplitude is determined from a wave decomposition and used for the presentation of the data.1 sound pressure level The amplitude of the acoustic oscillations in a gas turbine com- bustor can be very high. The geometric specifications of the liner configurations are listed in Table 7. 7 PA R A M E T E R S T U D Y This chapter presents the results of the experimental parameter study. This section studies the effect of high amplitudes on the dissipation of sound at perforations with and without bias flow.2.4. Each parameter is addressed individually. especially when instabilities occur. . The combination of these two methods eliminates the time consuming adjustment to match a certain SPLvalue. As a result. a combination of the incident wave ampli- tude and the fixed reference location method is applied. always including a bias flow. 7. The loud- speaker output is adjusted to match a nominal SPL at a reference microphone in the hard-walled duct section. the actual SPL values are scattering slightly around the nominal values. The measurement is performed with this nominal SPL value. Various definitions of the measured amplitude are discussed in Section 2. Here. In the analysis.

10c. see Figure 2.75 190 56 1347 hc007 70 2 1.5 90 295 1. d Same as dc017 with partitioned cavity.5 212 60 1161 dc017 212 1 2 90 440 1.1 59.8 59.5 240 60 556 140 1 2.5 43.5 90 2042 6.8 59. Config.5 90 1787 6.5).5 90 295 1. see Figure 7.73 30 724 7. while hcxxx indicates the Hot Acoustic Test Rig (see Section 6.5 90 1787 6.3 45 190 56 1347 hc008 70 2 1.5 240 60 1757 dc007 140 1 2.8 68 240 280 8201 dc024 140 1 2.40b.5 240 60 1757 dc009 140 1 1 90 47 0.1 59.5 192 60 780 dc015 192 1 1 90 85 0.5 240 60 1703 dc011a 140 1 2. .8 130 56 515 hc005 130 1 2 90 217 1.0 66 240 60 1757 dc013c 140 1 2. the first for the damping liner and the second for the metering liner spec- ifications.5 240 60 556 140 1 2.9 190 51 748 hc006 70 2 1. see Figure 2.6b. Double-skin configurations are represented by two consecutive lines.8 59.75 190 56 1347 a Same as dc007 with cross shaped orifices.5 162 60 280 dc014 162 1 1 90 85 0. b Same as dc007 with both edges round.3 51 240 60 941 140 1 2.5g.8 59.0 66 240 60 1757 dc012b 140 1 2.2 66 240 60 1757 dc010 140 3 1 90 286 1.73 60 719 7.6).5 43.8 68 240 280 8201 dc022 140 1 2.196 parameter study Table 7.0 66 240 60 1757 dc008 140 1 2.2 45.1 59.1: Geometric specifications of the studied liner configurations.5 90 295 1.5 90 2042 6.4 42 240 60 1446 140 1 1 90 286 1. c Same as dc007 with nonuniform porosity. D t d α Aopen σ Leff Dc Lc V – mm mm mm deg mm2 % mm mm mm cm3 dc006 140 1 1 90 286 1.5 59.8 68 240 280 8201 70 1 2.73 90 724 7.5 90 1787 6.8 68 240 280 8201 dc023 140 1 2.1 59.1 59.4 54.5 90 2042 6.5 90 1787 6. The configurations labeled with dcxxx are measured in the Duct Acoustic Test Rig (see Section 6.0 66 240 60 1757 140 1 1 90 286 1.5 90 893 7.5 240 280 6896 dc021 140 1 2.5 90 295 1.5 90 2042 6.5 160 280 1164 dc019 160 1 2 90 440 1. see Figure 2.1 59.5 212 60 1161 dc018d 212 1 2 90 440 1.

Configuration dc006 dc008 Sound Pressure Level dB 100-135 100-135 Bias Flow Mach Number – 0. 0. The two liner configurations and four frequencies are chosen due to their different damping behavior.7). described in Chap- ter 6.1.1 sound pressure level 197 The two-source scattering matrix method. dc008: 1020 Hz). The SPL is varied in the range from 100 dB to 135 dB for two bias flow velocities and four frequencies. a medium frequency (816 Hz). The influence of the sound pressure level on the dissipation is studied at configurations dc006 and dc008. 510. The dis- crepancy of the incident wave amplitude between these two mea- surements is generally below 0. and the frequency of maximum dissipation for the respective configuration (dc006: 510 Hz. that is microphone 6 or 7 (referring to the notation in Figure 6.2. a high frequency (1224 Hz). 306. Table 7. The average value is used for the presentation of the results. respectively.1 for specifications). 7. 1224 . Frequency Hz 816.2: Parameter settings for the measurements regarding the influ- ence of the sound pressure level. 1224 1020.036 306. The reference microphone is chosen next to the liner.5 dB.04 0. The only difference between these two liners is the orifice diameter and the resulting porosity (see Table 7. 0. The settings are listed in Table 7. De- pending on the excitation with speaker A or B. The frequencies are chosen by evalu- ating the dissipation curves: a low frequency (306 Hz). requires two consecutive measurements to produce one data point. The dissipation curves of the two configurations and the corresponding bias flow settings are displayed in Figure 7. There is no grazing flow present and temperature and pressure are at normal atmospheric condition. 816.

In the case without bias flow the dissipation increases continu- ously for SPLs beyond 115 dB. The top row. the dissipation value from the measurement at the low- est SPL (around 100 dB). depicts the aver- age dissipation. It is plotted in the bottom row.04. b) dc008 Mb = 0.e. i. when a bias flow is present.2 0 306 510 816 1224 306 816 1020 1224 Frequency in Hz Frequency in Hz Figure 7. shows the behavior without bias flow and the right column. plots b) and d).036. In order to remove this distraction and concentrate on the linear/nonlinear behavior the data has been normalized with the linear dissipation value. As the linear regime is de- fined by being independent of the amplitude it can be concluded that a linear to nonlinear transition occurs at about 115 dB for the . plots a) and b). The different dissipation levels at the various fre- quencies and different bias flow conditions are apparent. Mb = 0. When a bias flow is present the dissipation remains unchanged up to the maximum SPL. plots a) and c). a) dc006 Mb = 0.4 0. This behavior is consistent for all frequencies. plots c) and d).6 0.1: Dissipation curves of the studied configurations. Results The results of the linearity study for configuration dc008 are given in Figure 7.2. The left column. Mb = 0. The normalized dissipation expresses the deviation from the linear dissipation in percent.198 parameter study Average Dissipation a) dc006 b) dc008 0.

1020 Hz.2 0 Normalized Dissipation 2.2: Influence of the sound pressure level on the dissipation of configuration dc008 for the frequencies 306 Hz.5 c) linear nonlinear d) 2 1. 7. .5 1 95 105 115 125 135 95 105 115 125 135 SPL in dB SPL in dB Figure 7.1 sound pressure level 199 Without Bias Flow With Bias Flow 0. 816 Hz.4 0. and 1224 Hz.6 0. In the bottom row the dissipation is normalized with its value at the lowest SPL.8 a) b) Average Dissipation 0.

4 0.5 1 95 105 115 125 135 95 105 115 125 135 SPL in dB SPL in dB Figure 7.3: Influence of the sound pressure level on the dissipation of configuration dc006 for the frequencies 306 Hz.5 c) linear nonlinear d) 2 1.200 parameter study Without Bias Flow With Bias Flow 0. In the bottom row the dissipation is normalized with its value at the lowest SPL.8 a) b) Average Dissipation 0. and 1224 Hz. 510 Hz. .6 0.2 0 Normalized Dissipation 2. 816 Hz.

However. While the general outcome is very similar to the results of configuration dc008.02. Plotted is the average dissi- pation over the bias flow Mach number for a low and high sound pressure level at two frequencies. . This is most obvious at 510 Hz with a maximum at about 120 dB.3b).3a). This coincides with the con- stant dissipation value when a bias flow is present (Figure 7. Without any flow and at low Mach numbers the behavior is dependent on the sound pressure level. these limits are rather vague. When the saturation occurs. 7. a remarkable difference can be observed in the nonlinear regime. Increasing the Mach number eliminates the ampli- tude dependency. its dissipation values without bias flow stay below this limit. This could be one reason why this effect is not observed for configu- ration dc008.035 at 816 Hz. The behavior is linear when both curves agree and nonlinear if they do not. the linear regime extends over the whole range of SPLs tested in this study (up to 135 dB). The saturation occurs for average dissipation values beyond 0.3 for high amplitudes (Figure 7. using an identical arrangement of the plots as in Figure 7.4 shows the transition between linear and nonlinear behavior depending on the bias flow. The results of configuration dc006 are presented in Figure 7.3. Figure 7.1 sound pressure level 201 case without bias flow. At 510 Hz the dissipation values agree for bias flow Mach numbers beyond 0. while the limit is at about 0. The nonlinear increase of dissipation reaches a maximum value at a certain sound pressure level. ultimately resulting in dissipation values be- low that of the linear dissipation. When a bias flow is present. For higher SPLs the dis- sipation is reduced.3.2. Similarly. the dissipation curves for 306 and 816 Hz exhibit a saturation towards high SPLs. it seems that the dissipa- tion curves converge towards a common value of approximately 0.

473. comparing two sound pressure levels 105 dB and 130 dB.3 0.4 0.1 0 0.06 Bias Flow Mach Number Bias Flow Mach Number Figure 7. 105. [481]. 61. This limit seems to be independent of the frequency and the orifice diameter. [199] found a limit of 110 db at 310 Hz for a cylindrical liner configuration. 58. Heuwinkel et al. 1 A reason for this deviation could be that Lee and Kwon [287] use a random broadband signal.02 0. Discussion The dependence of the orifice behavior on the sound pressure level seems to be fundamentally different with or without bias flow. studying the influence of the sound pressure level on the reflection coefficient of two different perforated plates.4: Dissipation of configuration dc006 plotted over bias flow at two frequencies a) 510 Hz and b) 816 Hz. 505]. where the overall SPL should be the significant quantity .5 a) 510 Hz b) 816 Hz Average Dissipation 0. 225. 243. 227. These results are in agreement with previous findings. For exam- ple. The results here suggest a transition from linear to nonlinear behavior when increasing the sound pressure level beyond ≈ 115 dB.04 0. 123. 361. 72.06 0 0. The nonlinear behavior of orifices without bias flow has been demonstrated and discussed in many publications [26.02 0.2 0. 474. 228. Lee and Kwon [287] determine a lower limit at 102 dB1 .202 parameter study 0. 327. 441. the same limit was found by Tran et al.04 0.

They argue that the absolute amount of energy that can be absorbed is fixed. so that the ab- sorption or dissipation coefficient decreases due to the increasing incident energy. Thus. 481]. They only become significant when flow reversal oc- curs. Thus. that the dissipation is indepen- dent of the SPL when a bias flow is present. Luong et al. 419]. so that an instability does not occur. . 420]. This behavior agrees with the findings in [40. that nonlinear effects are of minor importance for bias flow liners. The results of the experimental study by Rupp et al. [415. It is not clear which definition the 102 dB are based on. Below this limit. This is of course cor- rect. 417. 419] exhibits a similar saturation of the absorption at high amplitudes as configuration dc006 in Figure 7. the oscillating acoustic velocity is larger than the steady bias flow velocity. [308] assume that the nonlinear effects are negligible2 . the velocity of (see discussion on page 31). however. the transi- tion between linear and nonlinear regime takes place when flow reversal occurs. results in overheating and failure of the structure. 2 Recently. Furthermore. a successful liner design aims to disrupt the feed- back loop between the heat-release fluctuations and the acoustic pressure fluctuations.3. This changes the absorption characteristics of the liner and. [415. 7. is confirmed by the mea- surements of Rupp et al. flow reversal in a combustor liner leads to hot gas ingestion into the liner cavity [99. However. This limit. Bodén and Zhou [70] showed that the nonlinear effects cannot be neglected entirely below this limit. In that case. the design operating point of a liner should be at much lower amplitudes. that is at very high amplitudes (of course depending on the bias flow velocity). this conclusion might not apply generally. [308]. based on theoretical assumption. According to Luong et al. 199. It could be argued that the pressure ampli- tudes of the instability are in fact very high. in worst case. It can be concluded from the results and the discussion of the literature.1 sound pressure level 203 The experimental results show.

2 bias flow This section discusses the general influence of the bias flow on two representative liner geometries.1 0. Please note that the maximum mass flow rate that is supported by the mass flow controller in the mea- surements is 100 kg/h.5: Measured values of the discharge coefficient (symbols) and average value (solid line).05 0. so that it will be included in the discussion of each parameter in the subsequent sections. However.5.6 0 0. the steady bias flow must be sufficiently high. Their results reflect the typical behavior of the other configurations in this study.25 Bias Flow Mach Number Figure 7. The general influence of the bias flow is demonstrated with configurations dc007 and dc008. so that flow re- versal does not occur.3 lists the characteristic bias flow parameters of the se- lected test points.2. the bias flow is the main parameter of this study. the liner is required to operate in the linear regime. Following these presumptions. 7.204 parameter study 1 dc007 Discharge Coefficient dc008 0. while the other quantities are calculated as described in Section 2. The data is listed in Table 7.3.2 0.8 0. Table 7.15 0. The mass flow rate and the pressure differ- ence are measured. The test points of configuration dc008 that .1. The geometric specifications are given in Table 7.

1 0.54 0.63 0.7 59.069 0.015 0.0 0.217 69.150 0.5 2.54 0.8 0.005 0.127 40.049 15. so that the average value is used.66 0.9 2.8 9.8 – 2.6 0.025 8.52 0.322 0.016 0. dc007 dc008 Mb Ub ṁb ∆Pt Cd Mb Ub ṁb ∆Pt Cd – m/s kg/h % – – m/s kg/h % – 0.3 0.2 6.7 0.069 0.497 0.53 0.8 – 0.051 0.8 11.4 0.8 – 0.5.5 45.6 15.493 – 0.000 0.8 45.5 3.53 0. so that mass flow data is not available here.347 0.67 0.029 9.64 0.3 40.005 0.034 10.1 6.7 0.036 11.9 0.8 0.66 0.8 0.0 0.1 35.058 18.040 12.4 28.0 0.8 0.221 70.9 24.68 0.206 0. .9 54.043 13.67 0.1 4.049 15.016 5. the orifices have the same geometry and dimensions.147 0.197 0.075 24.127 40.009 2. The values are rather constant over the bias flow Mach number.9 10.64 0.3 0.0 0.113 0.0 0.3 0.034 10.099 0.57 0.54 0.073 23.6 0.090 28.065 0.099 0.8 0.971 0.012 0.7 34.63 0.3 19.030 0.892 – required a higher mass flow rate have been measured without the mass flow controller by adjusting the line pressure directly.024 0.65 0.52 0.022 7.9 12.2 bias flow 205 Table 7.033 10.0 0.081 0.52 0.025 8.64 0.67 0.7 13. The result is plotted in Figure 7.009 2.9 66. However.039 0.0 21.051 0.54 0.64 0.014 4. After all.1 0.0 0.52 0.2 7.9 0.4 0.000 – 0.0 0.8 0.0 0.2 76.039 0.1 0.0 0.057 18.987 – 0. 7.090 28.7 50.2 100. the difference between the two configurations is quite surprising.66 0.000 – 0.64 0.040 12.000 0.016 5.029 9.986 0.54 0. just the spacing between them is changed.020 6.3: Measured data of the bias flow parameters of the selected test points.14). The discharge coefficient Cd is calculated from the pressure drop and mass flow rate according to Equation (2.

049. Introducing a bias flow and increasing its velocity shifts the dissipation maximum to higher frequencies. Without any bias flow. config- uration dc007 exhibits a prominent dissipation maximum around 400 Hz.4 0.025. The center frequency of the broadband increase of dissipa- tion is around 1 kHz.127 for two liner configurations a) dc007 and b) dc008. 0.016. Configuration dc008 in Figure 7.090. A maximum dissipation of nearly 60 % is reached when the bias flow Mach number is around 0. 0. Results Figure 7. The Mach number for maximum dissipation is around 0. 0.016 and the op- timum Mach number to achieve a good broadband damping is around 0. The flow through the liner improves the dissipation levels over a broad frequency range.6 0. At the same time.6: Dissipation plotted over the frequency at various bias flow Mach numbers 0.206 parameter study Average Dissipation a) dc007 b) dc008 0.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. and 0. 0. The damping is very poor when there is no bias flow present and there is no obvious peak frequency.049.6b demonstrates different char- acteristics. the maximum level of dissipation is first increased and then re- duced. For .049.6 plots the dissipation coefficient over the frequency for some selected bias flow velocities. while the dissipation becomes more broadband.

When necessary. At high frequencies (1020 Hz and 1326 Hz) the level of dissipation is constant for very low bias flow Mach numbers and then quickly rises to its maximum value at a certain Mach number setting.05 0. except at the low frequency end (< 500 Hz). the data has been smoothed with an appropriate fit. While the previous figures only displayed a selection of the available data. The influence of the bias flow on the dissipation becomes even more clear in Figure 7. 306 Hz. and 1326 Hz for two liner configurations a) dc007 and b) dc008.02-0. 816 Hz. higher velocities the dissipation is mainly reduced. 7.05 0. where it is slightly improved.2 0 0 0. Here.4 0. An interesting feature can be ob- served for configuration dc007. 1020 Hz.7. the dissipation coefficient is plot- ted over the bias flow Mach number for some selected frequencies. the contour plot is based on 416 data values (26 frequencies and 16 bias flow Mach numbers).05) and the curves converge towards a common dissipa- tion value for high velocities.2 bias flow 207 Average Dissipation a) dc007 b) dc008 0.8.1 Bias Flow Mach Number Bias Flow Mach Number Figure 7. 510 Hz.6 0. The dissipation increases for all frequencies when introducing a bias flow. The contents of the two previous figures can be combined to produce a contour plot of the dissipation depending on both fre- quency and bias flow Mach number. The result is shown in Figure 7. A maximum is reached at rather small Mach numbers (0.1 0 0. The limits of the .7: Dissipation plotted over the bias flow Mach number at var- ious frequencies 204 Hz.

the Strouhal number dependency is that ex- . Depend- ing on their porosity they can be divided into two groups: low porosity (≈ 1 %) with characteristics similar to dc007 and high porosity (≈ 7 %) with characteristics similar to dc008. Obviously. The ac- tual value of the Strouhal number will be discussed later in Sec- tion 7. The limits of the colormap are D = 0 (white) and D = Dmax (black).8: Contour plot of the average dissipation depending on fre- quency and bias flow Mach number for two liner configurations a) dc007 and b) dc008.17.9 enlarges a detail of Figure 7. the optimum combination of bias flow ve- locity and frequency is related via the Strouhal number. black and white colormap are D = 0 (white) and D = Dmax (black) and distributed linearly in-between.2 a) dc007 b) dc008 Bias Flow Mach Number Dmax = 0. However. The contour plot gives a con- venient overview of the general characteristics of the two liners. where the bias flow-frequency combinations for optimum damp- ing can easily be determined.1 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. It is shown how the maximum dissipation coincides with a curve of constant Strouhal number. These two examples are representative of all liner configurations that were studied within this work.8a. the two configurations have very different characteristics.208 parameter study 0. Clearly. The star indicates the maximum dissipation.50 Dmax = 0.59 0. as was already discussed with the previous two figures. Figure 7.

i. The differences between the models are largest at the low bias flow Mach number in a).02 0 300 600 900 1200 Frequency in Hz Figure 7. at intermediate bias flow Mach numbers the agreement of the models and the measurements is rather poor. At high Mach numbers.06 Bias Flow Mach Number 0. However. Figure 7. Only EDM:Jing is slightly off in f).10 compares the prediction of the models presented in Chapter 4 to the experimental results of dc007 and dc008 at various bias flow Mach numbers. The solid. Here. in b) and d).04 0.9: Detail of Figure 7.8a. 7. all models yield a similar result. white line represents a curve of constant Strouhal number that includes Dmax . . The general trends of the two geometries and various bias flow Mach numbers is captured by all models. in e) and f). e.2 bias flow 209 0. EDM:Jing and TMM:Bellucci provide the best match with the experiments. which is in excellent agreement with the experimental data. plicit for configuration dc007 only and is not reproduced with configuration dc008.

Mb = 0.049 f) dc008. EDM:Jing.210 parameter study a) dc007. Mb = 0. Mb = 0. TMM:Jing.049 d) dc008.4 0.6 0. TMM:Bellucci.2 0 e) dc008.029 Average Dissipation 0. Mb = 0. Mb = 0.6 0.10: Comparison between experimental data and models for configurations dc007 and dc008 and various bias flow Mach numbers.4 0.6 0.127 Average Dissipation 0. TMM:Betts.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.2 0 c) dc007. Mb = 0.009 b) dc007. . Experiment.016 Average Dissipation 0.4 0.

They derive an optimum ratio of bias flow Mach number to porosity. 219. They found Equation (7. so that they cancel each other out. [310]. This is in contrast to the optimum bias flow Mach numbers revealed by the exper- iments: 0. These findings comply with the results reported in [41. .016-0.1) σ opt 2 2 Applying the equation above to configurations dc007 and dc008. that a bias flow does improve the absorption of a perforated liner compared to the same configuration without bias flow. The two liner configurations presented here demonstrate very different characteristics. 241]. In order to achieve maximum absorption the liner should obey Mb   PL/A = √ .2 bias flow 211 Discussion The results show. (7. especially when the number of orifice rows is low. While the re- sults for dc008 are in the same range.015 for the liner geometry specified in Table 4.1. respectively.46. where for both PL/A = 1.035. 7. configuration dc007 exhibits a typical resonance damping behavior3 . 198. for example.0052 and 0. On the other hand. There exists an optimum bias flow. the incident wave from the duct and the reflected wave from the cavity are 180° out of phase. where the absorption is max- imum. A similar outcome was reported by Macquisten et al. 161. no obvious resonance effect can be observed 3 At the location of the perforation. The bias flow modifies this resonance effect until a sufficiently high bias flow velocity is reached.049 for dc008. 142. beyond which the bias flow effect dominates. the estimation for dc007 is very far off.1) to underestimate the optimum bias flow Mach number. Without bias flow. the optimum bias flow Mach num- bers should be 0.049 for dc007 and 0. Eldredge and Dowling [142] find the optimum Mach num- ber to be 0.

The models seem to dis- agree about the amount that the reactance is reduced by.11. This corre- sponds very well with the maxima observed in Figure 7. . g. [430]. this phenomenon seems to be dependent on the Strouhal number. However. The models seem to provide reasonable predictions (with ex- ceptions). According to Figure 7. The critical quantity during this transition seems to be the reactance. before it leads to the typical broadband absorption. as shown in Figure 4. The transition between both effects is not captured very well. Then. [260]) s c Aopen fr = . when either the resonance effect or the bias flow effect dominates. which would be responsible for shifting the reso- nance frequency. (7. It has been reported very early [319] and more recently [242. primarily around the resonance fre- quency. The resonances that are observed here are of the Helmholtz type.212 parameter study for configuration dc008 without bias flow. 291].2) 2π Vleff which yields 400 Hz for dc007 and 985 Hz for dc008. This behavior can be found in the models of Jing and Bellucci. that the reactance drops considerably when a bias flow is introduced.6. for example. that these two regimes correspond in some way to the high and low Strouhal number regimes observed by Scarpato et al. the development of the reactance with the Strouhal number seems to be essential to describe the behavior in the transition region. however. The dependence of the dissipation on the resonance frequency is fur- ther discussed in Section 7. the bias flow increases the dissipation. as well as the Strouhal number dependency of this process. It is believed. The resonance frequency can be estimated by (e.9.9.

7. This will be discussed in Section 7. without grazing flow. and Mb = 0.1 D+ 0. the grazing flow is studied separately from the bias flow. D+ and D− respectively. However.6 D− D− Dissipation D D 0.11: Demonstration of the influence of a grazing flow depending on the direction of sound propagation and the average dissipation for configuration dc006: a) without grazing flow. This can be observed in Figure 7. Plot- ted are the dissipation coefficients for sound propagating in pos- itive and negative axial direction.3 grazing flow In this section.11a.3 grazing flow 213 a) Mg = 0 D+ b) Mg = 0. Introducing a grazing flow in Figure 7.4. b) with grazing flow. A grazing flow introduces an asymmetry. In Figure 7. . 7. Results The dissipation coefficient of a symmetric liner configuration in stationary fluid is independent of the direction of wave propaga- tion.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. the values are identical.041 in both plots. the operation in a gas turbine combustor includes a grazing and bias flow simultaneously.11b. and their average value D.11 by comparing two measure- ments of configuration dc006 with and without grazing flow.4 0. so that D+ 6= D− .

0. The influence of the grazing flow on the average dissipation is il- lustrated in Figure 7.4. 0. This can be an important feature when the sound wave is mainly com- ing from one direction.13 for two liner configurations a) dc007 and b) dc008 without bias flow. Introducing a grazing flow shifts the frequency of maximum dissipation to higher values.12b experi- ences a general broadband increase of the dissipation with increas- ing grazing flow Mach number. 0.6 0. Configuration dc008 Figure 7. and 0. The average dissipation coefficient is plotted over the fre- quency for various grazing flow Mach numbers.214 parameter study Average Dissipation a) dc007 b) dc008 0.1) and will be used here.12 for the two liner configurations dc007 and dc008.12a dis- plays a dissipation peak at around 400 Hz for configuration dc007 without flow. However.06.1. The peak level reaches an optimum for grazing flow Mach number of 0.06.12: Average dissipation plotted over the frequency at various grazing flow Mach numbers 0.04. in exchange for a more broadband absorption. The maxi- mum value is reduced at higher velocities. The air supply of the DUCT-C test .2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. Figure 7.04-0. the dissipation is increased for sound waves propagating against the flow and reduced when propagating in flow direction.4 0. for a general comparison the average dissipation coefficient is more suitable (see Section 6.

036 0.3 grazing flow 215 Average Dissipation a) dc007 Mg Mb 0. Comparing these results to Figure 7. where maximum dissipation is achieved.016 0.6 0.2 0 b) dc008 Mg Mb Average Dissipation 0. there might exist an even better match for a bias flow veloc- ity that has not been measured. As the curves are chosen from available data.029 0.6.06 0. rig limits the maximum grazing flow Mach number to 0. The best visual match is chosen and plotted as a pair in Figure 7. where the influence of the bias flow is shown.13. . reveals a very similar behavior of the dissi- pation with either grazing or bias flow.13 0. For a better comparison an attempt is made to match the curves of equal dissipation.10 0.034 0.022 0.6 0.4 0.13. 7.2 0 300 600 900 1200 Frequency in Hz Figure 7.13 0.04 0.4 0.13: Visually matched pairs of dissipation curves comparing the effects of either grazing flow or bias flow. This is done as follows: The dissipation curve for one grazing flow veloc- ity is compared to several dissipation curves obtained by varying the bias flow velocity.06 0.10 0.014 0.043 0.

216 parameter study Grazing Flow Mach Number 0.15 0.01 0. This discrepancy can- not be fixed by finding a better matching bias flow velocity. A linear fit. respectively. The solid line is a linear fit through the measurement data and the origin. The over- all agreement is remarkable. However. The legend of Figure 7. The experimen- tal data exhibits a linear relationship between the two quantities.13 presents the matching pairs for both liner configura- tions and various flow conditions. The effect on the dissipation seems to be the same.03 0.02 0. That means the .04 0. It seems to be of no importance if the flow is passing the liner tangentially or if it is penetrating through the orifices of the liner.05 Bias Flow Mach Number Figure 7.14: Grazing flow velocity plotted over bias flow velocity for the matching dissipation curves from Figure 7. The respective grazing and bias flow Mach numbers are listed in the legend of the figure. a systematic difference can be observed for configuration dc007: The frequency of maximum dissipation is slightly higher when applying a bias flow.13 reveals that the grazing flow veloc- ity needs to be significantly higher than the bias flow velocity to achieve a similar dissipation.13. separately for configurations dc007 and dc008. yields a proportionality factor of 3 and 3. Figure 7.13.05 dc007 dc008 0 0 0. Figure 7.6.14 plots the grazing flow Mach number over the bias flow Mach number of the matched pairs as they are listed in the legend of Figure 7.1 0.

1. 349. 329. 321. While the wave normal without flow is assumed to be parallel to the wall (ne- glecting any acoustic boundary layer). It remains unclear if the deviation between the two configurations is a physi- cal reality or if it is a result of the quite vague method of matching the dissipation curves. where δ is the thickness of the flow boundary layer. This effect is discussed in [226. 264. The velocity gradient within the flow boundary layer refracts the direction of propagation of the sound wave. so that the proportionality factor should only be regarded as rough estimate. 2. Thus the ab- sorption per wave length remains constant when a grazing flow is present. 467] and the difference between D+ and D− is ac- credited with two phenomena: 1. it is refracted into the liner when propagating downstream and away from the liner when propagating upstream. 324.3 times higher as the bias flow velocity to achieve a similar absorption. The convective effect of the flow modifies the phase velocity of the wave by a factor 1 + Mg in flow direction and 1 − Mg against. 374. increasing and reducing the wave length respectively. that the absorption is reduced for sound traveling in grazing flow direction while it is increased by a similar amount when propagating in the opposite direction. 274. 322. This would result in an increased absorp- tion for sound waves propagating in flow direction and a reduc- tion against flow. Discussion Evaluating D+ and D− separately reveals.3 grazing flow 217 velocity of the grazing flow needs to be about 3. 7. . while the absorption due to a liner of fixed length is re- duced for waves traveling downstream and increased against flow direction [321]. This phenomenon is pronounced at higher fre- quencies and Tack and Lambert [467] suggest to neglect it when kδ 6 0.

They arrive at ratios of approximately 8 and 5 for the low and high grazing flow velocity. This similarity between the influence of grazing and bias flow on the dissipation was also observed by Feder and Dean [158]. They compared the grazing flow velocity to the bias flow velocity producing the same change of resistance. 22]: “The reason for this similarity is believed to be due to boundary layer fluctuations. p. Their measurements include six perforations with varying poros- ity (10-35 %) and two grazing flow velocities (90 and 150 m/s).61 yields ratios of grazing to bias flow velocities of about 5 and 3. Both terms are identical. Their calculation of the bias flow velocity neglects the orifice discharge coefficient. respectively. In fact. e. except for a factor of 3. This corresponds very well to the experimental finding discussed above. the conversion of acoustic energy into shedding vorticity.3 Mb . the absorption mechanism seems to be the same. The fre- quency of these fluctuations is believed to be low enough so that . which is in good agreement with the values obtained here. The impedance model of Bauer contains a term for the resistance due to the grazing flow and a second term for the resistance due to the bias flow (see Section 4. The effect of the grazing flow on the average dissipation is corresponding to the bias flow effect. Feder and Dean suggest the following explanation of the grazing flow phe- nomenon [158.8 between them. It can be observed in Figure 7.11 that the sec- ond phenomena dominates here. Judging by the similarity of the grazing flow and bias flow ef- fects. i.218 parameter study Both phenomena have a converse effect on increasing and reduc- ing the dissipation. The ratio of grazing to bias flow velocity is virtually independent of the porosity. induced by the grazing flow velocities. but different values are obtained for the two graz- ing flow velocities. similar results are achieved when Mg ≈ 3. Applying a general discharge coefficient of 0.6). causing airflows through the specimens.

However.4 simultaneous grazing & bias flow The effects of grazing flow and bias flow are mostly studied sepa- rately. for example). There seems to be an upper limit of dissipation which cannot be surpassed by the combination of grazing and bias flow.” The results above demonstrate a throughout positive effect of the grazing flow on the acoustic performance of a perforated liner.15b. 7. if any of the two velocities is already at an optimum (for max- imum dissipation).15 shows dissipation curves at two bias flow velocities for varying grazing flow Mach number. in Figure 7. it has been reported. The superposition of grazing and bias flow does not improve the dissipation beyond what can be achieved by applying either one alone. the result is unimpressed by the grazing flow. as was done in the two previous sections. depending on the ori- fice Strouhal number. 377]. when both are present simultaneously. In Figure 7. that at certain combinations of grazing flow velocity and liner geometry. So. 7.4 simultaneous grazing & bias flow 219 the resulting through flows affect the acoustic characteristics of the specimens in a manner similar to steady through flows. .15a. the dissipation increases with the super- position of the grazing flow. However. 216. then the other imposes only a minor or no influence on the result. At the faster bias flow. sound might be generated rather than ab- sorbed (see [23. slightly changing the grazing flow velocity or introducing a bias flow suppressed the phenomena. How- ever. the result might be influenced by the interaction of the two. 75. Results Figure 7. This effect oc- curred during the measurements here for configuration dc008 at Mg = 0. at the low bias flow Mach number. 321.05 and produced a dominant tone at about 1300 Hz.

035 0. it was found that. for liners sub- jected to simulataneous [simultaneous] flow through and past the apertures. 0. higher Mach numbers (Mg > 0. However. Lewis and Garrison [300] get more specific by stating that the grazing flow has no effect 4 As mentioned in a footnote in [171. p. II-3]: “From the results of previous experiments (Reference 2) [that is [313]]. Discussion The results demonstrate that the presence of a grazing flow at a bias flow liner seems to have either a beneficial effect or no effect on the absorption. typical com- bustor Mach number fall into the range studied here. the acoustic resistance should be computed as if only flow through were present.016 b) Mb = 0.035.1) might produce some adverse effects that are not observed here.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.220 parameter study Average Dissipation a) Mb = 0.6 0. In an early experimental study Marino et al. At sub-optimal bias flow Mach numbers the additional grazing flow increases the dissipation. Of course.097 at two bias flow Mach numbers a) Mb = 0.046.15: Average dissipation of configuration dc008 plotted over the frequency for varying grazing flow Mach number 0.” . while it shows no effect when the bias flow Mach number is already optimal.016 and b) Mb = 0. [313]4 come to the conclusion that only the bias flow needs to be accounted for when both bias and grazing flow are present.4 0. and 0.

p. and c3 = 0. with a reference to experimen- tal data of Rice [392]. Eq. was followed by Sun et al. repre- senting one cycle of the acoustic oscillation: zero net flow. . It should be noted. 8 in [463] is not correct. A bias flow simulates the acoustic flow through the orifice captured in a steady state. that the influence of the grazing flow can be neglected when the ratio of bias to grazing flow velocity is larger than Mb /Mg > 0. that the given limit matches very well with the condition Mg ≈ 3. and high inflow. The resulting discharge coefficient is plotted 5 Please note.3). are given by c1 = 2. Rogers and Hersh [406] conducted a detailed study regarding the steady state resistance of a single orifice subjected to grazing flow.e.5 Mb where a similar dissipation is obtained when bias or grazing flows are applied separately (see Section 7. is given in Section 5 of [463].1. They find that the grazing flow imposes a constriction to the flow through the orifice and describe this phenomena with an effective discharge coefficient based on the ratio of bias flow velocity to grazing flow velocity. [313]. Their model computes the grazing and bias flow re- sistance separately and then uses the large one only. low inflow. They define five flow regimes. [463]. The correct version.3. pure bias flow. The empirical constants for the bias-outflow case. e. 7.332. c2 = −0. which is considered here. They propose an effective discharge coefficient5 1 Cd = p . Below this limit the grazing flow dominates the resistance. high outflow. Dean and Tester [118. i. but applied to the acoustic resistance in- stead of the steady state resistance.69 for high values of λ.0566. 23] state.3) c1 + c2 /λ + c3 /λ2 where λ is the ratio of bias flow velocity to grazing flow veloc- ity λ = Mb /Mg . i. low outflow.3) tends to Cd = 0.4 simultaneous grazing & bias flow 221 when the bias flow Mach number is greater than 0. Equation (7.3). (7.1. corresponding to Equation (7. basing their conclusion most likely on the same experiments. A similar approach.

so that the porosity is an ambiguous parameter.3 is a characteristic value. .222 parameter study 1 Discharge Coefficient 0. weather the porosity is sufficient to describe the acoustic characteristics of a perforation or if the orifice area and spacing are significant on their own. in Figure 7. However.8 0.3) [463]. Summarizing the suggestions of the references: The dissipa- tion is largely independent of the grazing flow when roughly Mb /Mg > 0. The grazing flow dominates at smaller velocity ra- tios.2 0 0 0.16: Effective discharge coefficient due to the interaction of bias and grazing flow according to Equation (7. It is unclear.6 0. the ratio of Mb /Mg ≈ 0. 7. which essentially increases the bias flow velocity.8 2 λ = Mb /Mg Figure 7.6 0.16.4 1.2 0.8 1 1. Again.4 0.3.4 0. above which the interaction between bias and grazing flow seems to be negligible. various combinations of orifice area and orifice spacing yield the same porosity.6 1.2 1.5 porosity The porosity is the combination of the two parameters orifice area and orifice spacing. This can be accounted for by applying an effective discharge coefficient to the bias flow velocity. These findings generally agree with the obser- vations made here.

The perforation parameters of these four configurations are sum- marized in Table 7.1. the remaining specifications are given in Table 7.0 6. 3.4.5 22 8.5 2. Informally they are characterized by: 1. medium porosity by few large orifices with loose spacing. Table 7.8 0.5 1 Orifice Spacing mm 8. high porosity by many large orifices with tight spacing.5 porosity 223 (a) dc006 (b) dc007 (c) dc008 (d) dc009 Figure 7. Medium porosity by many small orifices with tight spacing.5 22 . 7. Configuration dc006 dc007 dc008 dc009 Porosity % 1.4: Perforation parameters of the four configurations studied re- garding the influence of porosity.1 1.2 Orifice Number – 364 60 364 60 Orifice Diameter mm 1 2. Four different liner configurations are compared.17: Illustration of the four configurations studied regarding the influence of porosity. and 4. 2. low porosity by few small orifices with loose spacing.

.18 compares the four configurations at various bias flow Mach numbers. Figure 7. For convenience. this trend is only revealed at higher bias flow Mach numbers. or the number of orifices has an impact on the prediction. and the number of orifices. the orifice spacing. however. show a greater variation when comparing dc006 and dc007. it can be observed at a glance that all models capture the general influence of the porosity success- fully. The porosity is clearly the significant parame- ter at higher bias flow Mach numbers.19 reproduces Figure 7. Matching the porosity of two liners yields nearly identical re- sults in the experiments. It is obvious that the models do not depend on the porosity alone. the differences be- tween the two configurations are indicated by filling the area be- tween the respective curves with the associated color. With increasing bias flow Mach number. overruling the orifice area.224 parameter study Results Figure 7. This is in contrast to the experimental results. However. At the same time it becomes obvious. as apart from the porosity the geometric specifications are quite different. but that the orifice diameter. the curves of equal porosity align more and more. This is re- markable.18d and includes the model predictions for all four porosities.20. so that they become identical for Mb > 0. the general agreement between the experiment and the models of these particular configurations is satisfactory at best. the orifice spacing. While the details of each model vanish in the multitude of lines. Especially. Without flow and at low Mach numbers the re- sults of the four configurations are very different. Again. that higher porosities yield an increased absorption. the broadband nature of the damping is not captured correctly.049. these two configurations are plot- ted separately in Figure 7. In order to allow a more detailed comparison. The models.

4 0.5 porosity 225 a) Mb = 0 b) Mb = 0. 6.016 Average Dissipation 0.2 0 c) Mb = 0.219 Average Dissipation 0.18: Average dissipation plotted over the frequency comparing four liner configurations with different porosities 1.6 0. 7.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.6 0.049 Average Dissipation 0.8 %.1 %.4 0.4 0.0 %. 1.059 f) Mb = 0.6 0. and 0.034 d) Mb = 0.2 0 e) Mb = 0.2 % at various bias flow Mach numbers. .

Models: EDM:Jing.3 Average Dissipation 0.19. .049. TMM:Bellucci. TMM:Bellucci.19: Comparison between experimental results and model predictions at Mb = 0.18d. TMM:Betts. TMM:Jing. corresponding to Figure 7. TMM:Betts. Ex- periment: dc006.226 parameter study 0. enlarging the predictions for the two configurations of similar porosity.6 Average Dissipation 0. Experiment: dc006.4 0. Models: EDM:Jing. TMM:Jing. dc008.2 0 300 600 900 1200 Frequency in Hz Figure 7. 0. dc007. dc007.2 0.1 300 600 900 1200 Frequency in Hz Figure 7.20: Detail of Figure 7. dc009.

Indeed. 7. the existing treatment of the interaction does not consider any bias flow. but recently Bendali et al. The single orifice model is then applied to a perforation by constructing a smooth boundary condition. The first assumption. which is reduced due to the loss of shear re- gion when the oscillation through both orifices are in phase. where other interaction effects than the ones considered by Melling might occur. Furthermore. The usual restriction demands that the ori- fice diameter is much smaller than the orifice spacing. This requirement is typically met for the porosities and frequencies relevant in combustion cham- bers. so that interaction might occur. The impedance models of Betts and Bellucci include a correction term for the interaction. A similar effect can be pictured for the resistance. e. i. Betts and Bellucci. [46] showed. so that the effec- tive attached mass is reduced. √ This is a rather vague limit. Melling [327] discusses the interaction effect: The at- tached masses of adjacent orifice are combined. Melling [327] gives σ < 0. The porosity of dc008 is 6. d  s. which translates to σ < 4 %.2. In terms of conductivity that is equivalent to an effective compliance [46. 327]. This simplification seems to give reasonable results. When dealing with impedance this is usually done by dividing the single orifice impedance by the porosity [241.8 %. predict a slightly higher frequency of . i. but it only considers the effect on the reac- tance.5 porosity 227 Discussion It is common practice to model a single orifice only. it seems that the models which take the interaction into ac- count. It is assumed that the orifices are sufficiently far apart so that there is no interaction between them and that the wave length is large in comparison to the orifice spacing. that s < λ/2 would be sufficient. however. 142. 298]. as a limit where orifice interaction can be neglected. e. The latter condition is commonly given by s  λ. 297. might become invalid for some realistic geometries. 219.

measurements with higher porosities would be required. At high bias flow velocities in Figure 7. Results The comparison is presented in Figure 7. but the 1 mm configuration is slightly supe- rior at higher frequencies. . However. in Figure 7. However.21a. The frequency of peak dissipation is moving slightly to higher fre- quencies when the bias flow is increased. The full specifications are available in Table 7. This broadband advantage of the thin liner (dc006) increases at low and medium bias flow velocities.19). Without any bias flow. where the walls are 1 mm and 3 mm thick. Figure 7. However. The frequency shift is somewhat larger for the thinner configuration. 7. this does not improve the agreement with the experi- mental data.21d the results have nearly converged and an appreciable difference is observed towards high frequencies only. respectively. that increasing the porosity requires a higher mass flow rate. When the absorption effect is dominated by the bias flow. the most significant difference is that the frequency of maximum absorption is shifted to lower frequencies when the liner is thicker. so that the interaction effect should be of minor importance. when keeping the bias flow velocity constant.1.6 wall thickness The influence of the wall thickness is studied by comparing con- figurations dc006 and dc010. typical porosities of combustor liners are rather at the lower end of the porosities encountered here.228 parameter study maximum dissipation compared to the Jing model (cf. The peak dissipation is nearly identical for both thicknesses. For a further clarification. one should keep in mind.21. The interaction of the bias flow jets might be less important or might have a different effect on the absorption. the level of dissipation seems to be in some way proportional to the porosity.

027 d) Mb = 0. 7.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.015 Average Dissipation 0.6 0. .6 0.6 wall thickness 229 a) Mb = 0 b) Mb = 0.4 0.2 0 c) Mb = 0.21: Average dissipation plotted over the frequency comparing two liner configurations with different wall thickness 1 mm (dc006) and 3 mm (dc010) at various bias flow Mach numbers.4 0.08 Average Dissipation 0.

TMM:Betts. Experiment: dc006.230 parameter study a) Mb = 0 Average Dissipation 0.22: Average dissipation plotted over the frequency compar- ing two liner configurations with different wall thickness at vari- ous bias flow Mach numbers.4 0. TMM:Jing.2 0 b) Mb = 0. dc010.2 0 300 600 900 1200 Frequency in Hz Figure 7. TMM:Bellucci. Mod- els: a) TMM:Melling/b) EDM:Jing. .027 Average Dissipation 0.4 0.

but exhibit less optimal agreement with the measurements.7 orifice cross-section shape 231 Without bias flow in Figure 7. Both. The predictions follow the trend also at Mb = 0. cf. Both models use an approximation of Equation (4. the agreement is better for the thicker configuration. However. while all models underpredict the dissipation for dc006. Again. A longer orifice increases the mass iner- tance and the viscous losses.5g with the dimensions l1 = 1 mm and l2 = 1 mm.027. 241]. [41]. [41. TMM:Betts and TMM:Bellucci.12). Now. the agreement is better for dc010. which seems to be suited better at this larger thickness. When a bias flow is present the absorption characteristics of the thin wall are clearly superior with a higher peak absorption and covering a wider frequency range. Both effects are reproduced in the exper- imental data presented here and predicted rather accurately by TMM:Melling. The geom- etry of the cross-shaped orifice is illustrated in Figure 2. reproduce the general trend when changing the thickness.7 orifice cross-section shape The standard orifices in this study have a circular cross-section. The other models. the thickness effect is well understood.22a the best agreement with ex- perimental data is achieved by TMM:Melling. 7. This is in agreement with the experimental results of Jing and Sun [241] and Bellucci et al. Without flow. 7. TMM:Betts and TMM:Bellucci. Discussion The wall thickness is a prominent parameter in any of the ori- fice models presented in Chapter 4. yield a better agreement with the experiments for the 3 mm wall thickness. The main effects on the dissipation are the decrease of the peak frequency and the increase of the Q factor. the dissipation characteristics of a circular orifice is com- pared to a cross-shaped orifice with equal open area. The characteristic spec- .

5: Orifice cross-section geometry parameters. . The results from these two fundamentally different geometries agree very well. It can be concluded that the actual cross-section geometry and a varying orifice perimeter is rather unimportant to the dissipation. This results in an increased perimeter length for the cross-shaped orifice and thus a reduced hydraulic diameter.9 12. the open area is kept constant for both con- figurations. The previous work is quickly reviewed here. As already mentioned.5 1.9 5.028.232 parameter study Table 7.5. This is found for config- urations with and without bias flow. Discussion In the literature the influence of the cross-section shape on the acoustic properties of an orifice has only been studied for a sta- tionary medium or including a grazing flow. The given jet contraction coefficient is the mean value taken from the measurements of various bias flow velocities.23 for four different bias flow Mach numbers. Configuration dc007 dc011 Cross-Section Shape – circle cross Orifice Open Area mm2 4.65 0. Only a slight shift of the frequency of maximum dissi- pation can be observed at Mb = 0.69 ifications of these two configurations are given in Table 7.7 Jet Contraction Coefficient – 0.0 Orifice Perimeter mm 7.0 Hydraulic Diameter mm 2. Results The comparison of the two cross-section geometries is presented in Figure 7.

057 Average Dissipation 0.6 0.4 0.2 0 c) Mb = 0. Rayleigh [385. Ingard [225] compares the end-correction of circular and rectan- gular orifices. that the conduc- tivity can be calculated from the area alone. 7.4 0. 390] compares the theoretical conductivity of a circular orifice to that of elliptic orifices with various elongations. He concludes.23: Average dissipation plotted over the frequency comparing two liner configurations with different orifice cross-section geometries circular (dc007) and cross (dc011) at various bias flow Mach num- bers. However.028 d) Mb = 0.014 0. Based on his measurements he argues that the end- .7 orifice cross-section shape 233 Average Dissipation a) Mb = 0 b) Mb = 0.6 0. that the circular orifice is the shape which has a minimum conductivity. he finds that a moderate eccentricity of an ellipse of equal area results only in small change of the conductivity. .2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. so that KR A ≈ const. He states without formal proof.

The cross-section geometries includes a cir- cle. (7.5 yields a slightly different conductivity of 4. He treats a circular. and a crown7 with equal maximum 6 Morfey treats only one side of the orifice. where the longest dimension remains small in comparison with the wavelength.4) pyields KR ≈ 3.9r. narrow openings.85 mm and 5. which includes circular. In conclusion. using the hydraulic diameter to calculate the con- ductivity of the cross-shaped orifice yields KR = 4. Grace et al. various triangles.4) gives more appropriate results for very long. accord- ing to Morfey Equation (7. he studies the influence of the orifice cross-section geometry. However. which agrees 6 with Rayleigh’s result KR = 4 A/π = 4r.61 mm. respec- tively. see Figure 2. Morfey [336] reviews the work concerned with the influence of orifice shape on the sound transmission. elliptic. 7 The crown is an orifice with square cross-section and triangular serrated leading or trailing edge. so this equals twice the value obtained by Rayleigh for both sides. he estimates the conductivity of an orifice of arbitrary shape in terms of perime- ter P and area A with π P KR = . and three differ- ent cross-shaped orifices. Amongst other things. square. square.54 mm. and rectangular cross-sections. [181] numerically calculates the Rayleigh conduc- tivity for a variety of orifice shapes for both one-sided and two- sided grazing flows. Chanaud [89] presents a theoretical study on geometry effects on the resonance frequency of Helmholtz resonators. which matches slightly better to the circular geometry.4) 2 ln (P2 /A) For a circular orifice Equation (7. rectangular.5i. His results suggest that the orifice cross-section shape is not highly signifi- cant. Inserting the dimen- sions of the circular and cross-shaped orifices from Table 7. . all of which have equal area. Instead. confirming Rayleigh’s theoretical assumption.234 parameter study correction scales with the orifice area only.

no data is available that provides such a comparison. star-shaped. Measurements were done at low and high sound pressure levels. and eye-shaped8 with equivalent area. This is confirmed by the experimental results here. 169] at high amplitudes suggest that any effect vanishes when there is a bias flow present. The findings of Ahuja and Gaeta [7. see Figure 2. The cross- section geometries of the orifices are circular. Accounting for the different open areas √ of the orifices. Rayleigh [385. the orifice cross-section shape seems to have a minor effect on the acoustic properties in stationary medium or with grazing flow. 7.5j.5 m/s as limiting value where the transition occurs. 169] refer to this shape as stepped-oval geometry. square. They specify an orifice ve- locity of 0. so that orifices with shapes that are significantly different from circular yield a higher absorption. 390] and Morfey [336] expect a more substantial change of the conductivity for a long narrow slit when compared to a circular orifice. where the triangular shapes produced the largest differences compared to a circular orifice. 169] performed an experimental study on the effect of the orifice shape on its acoustic impedance. Ahuja and Gaeta [7. 8 This is achieved by sliding a perforation consisting of circular orifices over another identical perforation. the conductivities are normalized with respect to A. The results are found not to vary significantly with the orifice shape. triangular. However. at higher amplitudes the resistance appears to be independent of the orifice shape. The length of the perimeter is more than doubled between the circular and star- shaped orifice. Ahuja and Gaeta [7. As a consequence it seems to be sufficient to match the orifice area when dealing with different cross-section geome- tries.7 orifice cross-section shape 235 dimension in streamwise direction. They find that at low incident sound amplitudes the resistance increases with the perimeter. Based on the literature. However. .

6 0. Results The comparison of both geometries is presented in Figure 7.236 parameter study Table 7.6: Flow parameters of the measurements with varying orifice edge geometry. However.64 0.022 6.67 0. Without any bias flow. in Figure 7.043 11.000 0.033 0.1 0.7 0.8 orifice edge geometry This section presents a comparison between orifices with square edges (dc007) and round edges (dc012) as was illustrated in Fig- ure 2.033 11.1.6.034 9. The flow parameters for the two geometries and four bias flow settings presented here are given in Table 7. the mass flow needs to be increased for the rounded geometry.1 0. As discussed in Section 2. the behavior of the flow through the orifice is very sensitive to the edge geome- try.2.044 15.82 0. please keep in mind that a higher mass flow rate needs to be applied for the .24.5.24. dc007 (square edge) dc012 (round edge) Mbnom Mb ṁb Cd Mb ṁb Cd – – kg/h – – kg/h – 0.7 0. the square edge is superior to the round edge.6.6 0. The discharge coef- ficient Cd is closer to unity for the round geometry. corresponding to Figure 7. All remaining geometry parameters are kept constant and are listed in Table 7.000 0.86 7. When a bias flow is present this finding is reversed and the dissipation of the round edge configuration is slightly better than that of the square edge.000 0.044 0. keeping the bias flow Mach number constant.0 – 0.68 0.022 7.87 0.022 0.0 – 0. Thus.24a.

2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. The length of the orifice is effectively reduced by . 7.4 0.4 0. Discussion The results show only minor effect of the orifice edge geometry on the dissipation.2 0 c) Mb = 0.1 with and without bias flow. which adversely affects the efficiency of the liner. The improvements of the round edge over the square edge be- come more explicit when a grazing flow is present. round edge configuration.6 0. This is demon- strated in Figure 7.8 orifice edge geometry 237 Average Dissipation a) Mb = 0 b) Mb = 0.033 d) Mb = 0.6 0.25 for Mg = 0.24: Average dissipation plotted over the frequency comparing two liner configurations with different orifice edge geometries square edge (dc007) and round edge (dc012) at various bias flow Mach num- bers.022 0.044 Average Dissipation 0.

so that the lower absorption without flow could be attributed to reduced viscous losses within the orifice.238 parameter study Average Dissipation a) Mg = 0.01.4 0. Mb = 0 a) Mg = 0. the improvements are rather minimal. Additionally.1 without bias flow: square edge (dc007) and round edge (dc012). the smaller frequency of maximum dissipation in Figure 7. Keller and Zauner [258] observed an increase of absorption by 55 % when rounding both edges of a Helmholtz resonator neck compared to having square edges. a diameter of 9 mm. . rounding both edges.24c and d. and the edge describes an arc of 15 mm radius over an angle of 40° (see Figure 7. The improved absorption was obtained without flow as well as with a bias flow of 4 m/s.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.25: Comparison of the square-edge orifice and the round orifice at Mg = 0. These large improvements by round edges that are reported in the literature are in contrast to the findings here: While round edges are found to be superior to square edges when either bias or graz- ing flow is present. The neck had a length of 100 mm. [284] improved the absorption of Helmholtz resonators and quarter-wave resonators by rounding off the edges.01. could be a result of an increased mass reac- tance due to the enlarged open area at the surface. Similarly. Mb = 0.6 in [258]). no geometric specifications of the setup is given. However.034 0. Laudien et al.6 0.

8 orifice edge geometry 239 Average Dissipation a) Mb based on jet area b) Mb based on orifice area 0. b) Mb based on orifice area.26a compares both configurations by matching the velocity based on the jet area. the dissipation curves for two config- urations with substantially different contraction coefficients are compared.26b.24 or 7.26: Comparison of the square-edge orifice and the rounded ori- fice at Mb ≈ 0. The bias flow velocity here is based on the velocity of the contracted jet.7.26.039: a) Mb based on jet area. Similar re- sults are found at other bias flow velocities and with other con- figurations.26a are in a much better agreement than in Figure 7. Figure 7. These deviation is most probably the results of a different defini- tion of the bias flow velocity.65 and 0. It can be concluded that the jet velocity which includes the jet contraction is the relevant param- .85. In order to determine which definition should be used when comparing bias flow liners. The contraction coefficients for the square-edge orifice and the rounded orifice are approximately 0. This is demonstrated in Figure 7. Keller and Zauner [258] use the ori- fice mean velocity and keep it constant for either square or round edge. that is taking the jet contraction into account.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. The velocity in Fig- ure 7.18. The respective Mach num- bers are listed in Table 7.4 0. cf. The dissipation curves in Figure 7. respectively.26b is based on the orifice area.6 0. Figure 7. 7.

Bett’s model includes the discharge coefficient as a variable parameter. which considers the orifice mean velocity as rele- vant quantity.65 0. the mass flow rate that is discharged through the orifice will be increased.038 0. As a consequence.7: Flow parameters for the comparison of jet velocity and mean orifice velocity.26. cf. In Bauer’s model a constant discharge coefficient of 0. However. 4. corresponding to Figure 7.26b Mb Mb Cd Mb Mb Cd dc007 ( ) 0. The bias flow through a combustor liner is controlled via the pressure difference across the liner.26a Figure 7.058 0. A similar comparison is not provided in the literature. who finds a much better agreement of experimental data with his vorticity wave theory when using the jet velocity instead of the orifice velocity. Figure 7.044 0.040 0.65 dc012 ( ) 0.85 eter for the dissipation.10). Equation (2.240 parameter study Table 7.18 & 4.040 0.026 0.61 is ‘baked’ into the expression for the bias flow. Figs.19]. the dissipa- tion remains constant when changing the orifice edge geometry from square to round at the same pressure difference.038 0. the significance of the jet velocity is supported by Ronneberger [412. All models that are used here take the orifice mean velocity as an input parameter.034 0. The jet contraction is then considered in var- ious ways. This definition has been applied to the experimental data throughout this work. The bias flow contribution in the models of Bellucci and Jing is based on Howe’s model. .85 0. In that case. the jet velocity is independent of the jet contraction and depends only on the pres- sure difference. However.

These ge- ometric features are listed in Table 7.27. so that the amount of dissipated acoustic energy de- pends on the direction of sound propagation relative to the incli- nation angle. 7. the orifices are inclined in positive x-direction. as is expected for a sym- metric configuration without grazing flow.9 orifice angle Inclining the orifices from their 90° angle results in an asymmet- ric behavior.3 4 7.8: Geometry parameters regarding the orifice angle. Looking at inclined orifices in a grazing flow environment (the right col- umn in Figure 7. Decreasing the incli- nation angle increases the dissipation of sound propagating in positive x-direction. as specified in Equation (2. Configuration hc006 hc007 hc008 Orifice Angle deg 90 60 30 Wall Thickness mm 2 2 2 Orifice Length mm 2 2. Keeping the wall thickness constant increases the orifice length for angles other than 90°.1). It was shown before (see Figure 7. The dissipation curves for sound propagating in positive and negative direction are com- pared for liners of different inclination angles. while it decreases the dissipation of sound propagating in the opposite direction.27) reveals that both effects counteract each other . Results The influence of the inclination angle on the directional perfor- mance of the liner is shown in Figure 7. Here.11) that a grazing flow has the opposite effect. so that while reducing the inclination angle the direction of the bias flow injection aligns more and more with the grazing flow. The curves for the 90° angle (no inclination) are identical.9 orifice angle 241 Table 7.8.

Mg = 0 d) 60°. bottom row: 30°) on the dissipation at two grazing flow velocities (left column: Mg = 0.2 e) 30°. D− . Mg = 0. Legend: D+ .242 parameter study 0.4 0. D .2 c) 60°. right column: Mg = 0.6 Dissipation 0. middle row: 60°.1) with a constant bias flow Mach number of 0.6 Dissipation 0.1 0 0.27: Influence of the orifice angle (top row: 90°.2 a) 90°. Mg = 0.1 0 500 1500 2500 500 1500 2500 Frequency in Hz Frequency in Hz Figure 7.4 0. Mg = 0.6 Dissipation 0.1 0 0. Mg = 0 f) 30°.087. Mg = 0 b) 90°.4 0.

2 0 c) Mb = 0. The overall performance is represented by the average dissipation. From a design point of view. The average dissipation is already included in Figure 7.4 0.4 0. is essential.6 0. that is considering both directions of wave propagation. but the three configurations with varying angle are finally compared in Figure 7.1. see Section 6.28. when the dissipa- tion in the other direction is of minor importance. and cancel each other out when the angle and the grazing flow is chosen properly (here around 60° and Mg = 0. 60°. the orifice angle can be used to improve the damping in positive x-direction.6 0. for example in an exhaust duct.27. For a combustor the overall damping perfor- mance.4. 7.087 d) Mb = 0. .1).2 0 500 1500 2500 500 1500 2500 Frequency in Hz Frequency in Hz Figure 7.172 Average Dissipation 0.28: Average dissipation plotted over the frequency comparing three liner configurations with different orifice angles 90°. 30° at various bias flow Mach numbers.044 0.9 orifice angle 243 Average Dissipation a) Mb = 0 b) Mb = 0.

As shown in Fig- ure 7. Figure 7. their performance without bias flow is rather poor. the orifice angle seems to be of no importance when looking at the average dissipation or D+ and D− separately (not shown here). there is hardly any influence of the inclination angle. EDM:Jing.2 a) hc006.087: a) hc006. Figure 7. α = 30° 0 500 1000 1500 2000 2500 Frequency in Hz Figure 7. At medium bias flow velocities.28c. α = 90°. TMM:Betts. TMM:Jing.28d.28a.28b and 7. α = 90° 0 Average Dissipation 0. Also. The liners are designed for bias flow operation.29 plots the model predictions for hc006 with 90° in- clination angles and hc008 with 30°. At high bias flow velocities.244 parameter study Average Dissipation 0.6 0. The inclination angle is in- .6 0. Figure 7. b) hc008. Experiment. smaller angles achieve a slightly better broadband dissipation.4 0.2 b) hc008.29: Comparison of the experimental data and the model predic- tion for two inclination angles at Mb = 0.4 0. TMM:Bellucci. α = 30°.

However.75 t/ sin α (7.27. (6) of [17] only includes the wall thickness and not the orifice length. Section 7. the experimental results show the opposite effect. As the angle is included by adjusting the orifice length. Figure 2.9 orifice angle 245 cluded in the models by simply using the orifice length (Equa- tion (2.7b). e. a detailed inter- pretation of the model results is difficult. instead of the wall thickness (cf. Discussion Eldredge and Dowling [142] include the inclination angle in their analysis by simply using the orifice length. Unfortunately. This is taken into account by an empirical correction factor of 0. the area is slightly larger. 7. The behavior of the mod- els is rather similar. cf. i. [141] and compared to data obtained from a large- eddy simulation of an inclined bias flow orifice. 9 Eq. The result is a reduction of the broadband absorption with increas- ing orifice length/decreasing angle. They are comparing the results to EDM:Jing and yield good agreement9 . their mea- surements include a grazing flow which might counteract on the effect of the inclination angle as shown in Figure 7.1)) instead of the wall thickness.5) Andreini et al.6. the predicted effect is equivalent to increasing the wall thickness and keeping the angle constant. the agreement be- tween models and experiment is poor at frequencies > 1500 Hz. However. so that the corrected ori- fice length for an inclined orifice is given by l = 0. They found that the agreement can be further improved by considering that the ac- tual openings of the orifice are elliptical and not circular when it is inclined. this must be a spelling mistake as the predictions for the two con- . Thus. already for the standard 90° configuration. The same approach was used by El- dredge et al. while the 90° orientation is slightly superior at low bias flow velocities.75. [17] report that any effect of the inclination angle disappears with increasing bias flow. However.

Equation (7. The effect of the orifice length alone (without inclination) was discussed in Section 7. it seems to have the opposite effect on the dissipation. This good agreement is quite surprising as the bounds in [285] are derived for stationary fluid and the simulation in [141] includes a bias and grazing flow. who applied an empiri- cal correction (cf.5)). 10 This was already suggested by Eldredge et al.246 parameter study Here. so that 60° in [141] is equivalent to 30° here. They find that the empiri- cal correction of Eldredge et al. [141]. it was determined for a 30° configuration12 . Anyhow. which is avail- able here as well. The data of Andreini et al. [285] derive upper and lower bounds of the Rayleigh conductivity of an inclined orifice with circular cross-section and elliptical openings11 . 12 Please note. Laurens et al. While the inclination angle also increases the length. the angle of 30° doubles the length compared to the 90° configuration. Applying a correction factor figurations produce different results when the inclination angle is the only modified parameter. in fact it is close to the mean value of the upper and lower bounds. Laurens et al. 142. This hints to an additional effect that has not been accounted for in the models. As suggested in the literature. [141]. This can be done by applying the correction factor given by Eldredge et al. [17] is limited to frequencies below 1500 Hz. . [285] present a mathematical treatment of the Rayleigh conductivity for inclined orifices in the absence of flow. They show that the effect of the orifice angle is not entirely cap- tured by the intuitive extension of the length. How- ever. [141] falls within these bounds. but this is the rele- vant configuration for gas turbine combustors. as described by Equation (2. the elliptical open- ing should be considered as well. The correction factor is presumed to be dependent on the inclination angle. that the angle in [141] is measured from the wall normal. 141. 328]. but that the increased area of the elliptical openings needs to be consid- ered as well10 .1) and applied previously [17.6. the largest effect is found to be at higher frequencies. 11 They extend their treatment to other geometries as well.

The peaks associated with the metering liner are eliminated when applying a damping liner bias flow of Mb = 0. Results The experimental results in Figure 7. 7. but will not surpass the 90° result. 7. This would yield the desired effect for a bias flow setup.10 double-skin configuration 247 of 0. At the given open-area- ratio.30 demonstrate this effect for two double-skin configurations.6 has shown that the influence of the wall thickness dis- appears at high bias flow velocities. as it is in a combustor. that corresponds to a metering liner bias flow Mach number . the maxima at the higher frequencies correspond to the damping liner. so that each max- imum can be related to the one or the other.75 for the elliptical opening will produce a result that lies in- between the results for 30° and 90° without the correction factor. Section 7. it could be proposed to keep the rele- vant length constant to the wall thickness and apply a correction for the increased opening area only. As a consequence the effect will be weakened.014. The frequencies of these two maxima correspond to the Helmholtz resonances of the damping and metering liners and their associated cavities. Without any bias flow the dissipation curve ex- hibits two maxima (one maximum in Figure 7. The bias flow velocity through the damping liner is then determined by the open-area-ratio between damping and metering liner. but not be turned around. The velocity through the metering liner is typ- ically much higher.3. When transferring this find- ing to the inclined orifice.10 double-skin configuration A double-skin configuration allows to reduce the bias flow veloc- ity of the damping liner when the total pressure drop is fixed. The open-area-ratio of both con- figurations is 0.30a is expected just beyond the plotted frequency range). Here.

replacing the second liner with a hard wall. where a second liner can be included.6 0.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.014: with second liner. Generally.30 for two cases: 1. Experiments: Mb = 0. due to the small open-area-ratio the differences between sec- ond liner and hard wall predictions vanish when increasing the velocity further (not shown here). The result for dc015 is not as clear. While the inclusion of the second liner gives a better match with experiments at low frequencies. neglecting the second liner’s influence. the predictions of EDM:Jing benefit from the inclusion of the second liner.30: Influence of a double-skin arrangement on the dissipation with and without bias flow. i.014.4 0. How- ever. it slightly underestimates the peak level.05. Pre- diction of EDM:Jing for Mb = 0. . The result suggests that the influence of the meter- ing liner can be neglected. In- cluding a second liner in the model.248 parameter study Average Dissipation a) dc014 b) dc015 0. Mb = 0. of nearly 0. Predictions of EDM:Jing are also plotted in Figure 7. so that the actual geome- try is reproduced. e. with hard wall. and 2. The prediction of dc014 is improved when accounting for the second liner. This conclusion can be verified with the Eldredge and Dowl- ing method.

31. allowing for a sig- nificant effect of the second liner. In a different setup the second liner could be implemented to improve the absorption. while the max- imum associated with the outer volume disappears at high bias flow velocities. 7. The three configurations are illustrated in Figure 7. Discussion Jayatunga et al. However. [240] develop an impedance model for a plane double-skin configuration. In that case the open-area-ratio between the two perforations should be close to unity. nor a compa- rable measurement are presented.31: Illustration of the configurations studied regarding the in- fluence of the cavity volume. Configuration dc006 is the base configuration with- out metering liner. 7.11 cavity volume 249 (a) dc006 (b) dc014 (c) dc015 Figure 7. similar to a locally-reacting double- degree-of freedom liner. yielding a of 1757 cm3 . Modeling a double-skin liner with bias flow demonstrates the same behavior as above: Two absorption maxima are dominant at low bias flow velocities. The metering liner in . no details of the geometry.11 cavity volume The cavity volume is modified by using double-skin arrangements with various dimensions.

13 The peak of dc014 is expected just beyond the plotted frequency range. The double- skin arrangements of dc014 and dc015 are accounted for in the modeling. configurations dc014 and dc015 reduces the volume to 280 cm3 and 780 cm3 .4 0.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7. Prediction by EDM:Jing: dc006.250 parameter study Average Dissipation a) Mb = 0. The geometric specifications are given in Table 7. the model consistently yields a lower dissipation than the measurements. At higher bias flow velocities. Results The comparison of the three different volumina is plotted in Fig- ure 7.1.015 b) Mb = 0. The influence of the cavity volume is reproduced well by the model. dc015 (780 cm3 ). The damping liner is the same in all configurations.32: Influence of the cavity volume at two bias flow Mach numbers. dc014 (280 cm3 ). dc015. The peak dissipation is shifted to higher frequencies when the volume is reduced13 . The peak level of dissipation re- mains unchanged.32.053. Experiment: dc006 (1757 cm3 ).32 also plots the predictions of EDM:Jing. At Mb = 0. respectively. dc014.053 0.32b. Figure 7. Figure 7.6 0. the dissipation increases with the frequency up to the resonance fre- quency of the respective configuration and remains at the maxi- mum level for higher frequencies. .

This behav- ior can not only be observed for the three configurations in Fig- 14 The second liner in configurations dc014 and dc015 invalidates the hard-wall assumption. Figure 7. the frequency of maximum dissipa- tion is related to a resonance frequency of the system. The reduction of the reactance seems to be a generally accepted fact.32a. this behav- ior is underestimated by the Jing model as well as the numerical model presented in [242]. Assuming the reactance becomes zero at sufficiently high bias flow velocities. The impedance measurements of Jing and Sun [242] demonstrate the behavior of the reactance.2). the resonance frequency is directly dependent on the cavity volume. In- terestingly enough.9 do not exactly match with the dissipation maxima in Figure 7. this is typi- cally a Helmholtz resonance. While some models do not in- clude any influence on the reactance at all.2. so that the theoretical values in Table 7. In that case. In fact. 7. but the literature does not provide a collective opin- ion on the more specific behavior.30). other models predict very different strengths of the phenomena (see Figure 4.9). this would yield a resonance frequency of infinity. However.32b no resonance is visible. The bias flow shifts the resonance frequency to higher values14 . the dissipation increases with the frequency until it reaches its maxium at the theoretical no flow Helmholtz resonance and remains constant at higher frequencies. Wester- velt [491] found that the reactance ultimately becomes zero when a bias flow is present. . The theoretical Helmholtz resonance frequencies for the three configurations are included in Table 7.32a. and in fact it is close to zero at a certain bias flow setting. When all dimensions are much smaller than the wave length.11 cavity volume 251 Discussion As discussed in Section 7. so that the measured resonance frequencies are slightly higher than the theoretical ones already without flow (cf. This is due to the decrease of the reactance by the bias flow (see discussion in Sec- tion 7. at Mb = 0. This relationship is an obvious conclusion from Figure 7.9.053 in Figure 7.

In agreement with Figure 7. 3 in [219]). so that the absorption maximum moves away from unity (cf. 16 Please note the different volume definition compared to Equation (7. so that Equation (7. Hughes and Dowling [219] introduce a resonance parameter15 dc Q = (k s cos θ)2 . For normal sound incidence. (7.252 parameter study Table 7.7) can be readily applied to a cylindrical configuration. Fig. Applying Equation (7.9. . Obviously. (7. which was introduced on page 43.6) to a grazing incidence configuration yields zero (as cos 90◦ = 0). and dc009 in Figure 7. dc008. that the bias flow alters the resonance frequency. but as well for configurations dc007.18d.6) 2r where θ is the incidence angle of the sound wave.6) corresponds to the squared ra- tio of frequency to Helmholtz resonance frequency of a resonator with a neck length of zero. They use it as a non-dimensional fre- quency parameter for the absorption of a thin liner with normal sound incidence.6). Equation (7.32b. The associated Helmholtz resonance frequencies are listed in Table 7. dc006 dc007 dc008 dc014 dc015 hc006 q Aopen fr = c 2π Vleff 512 400 985 1283 768 679 ure 7.7) Aopen fr 15 The resonance parameter Q should not be confused with the Q factor. they were operating in the resonance controlled regime.9: Theoretical Helmholtz resonance frequency in Hz for a hard- walled cavity with Mb = 0. Neglecting the incidence angle for wave lengths much larger than any liner dimension and account- ing for the finite thickness of the liner yields16  2 2 Vleff f Q=k = .32a they found. They show that the peak dissipation is obtained when the resonance parameter equals unity.

The associated resonance fre- quencies are listed in Table 7. Mb = 0. 7. Mb = 0. Mb = 0.33 plots the dissipation coefficient of various liner ge- ometries over the resonance parameter of Equation (7.9.34. Mb = 0.33: Average dissipation of various liner geometries plotted over the resonance parameter Equation (7. Mb = 0. so that the dissipation is dominated by the bias flow effect. hc006. The Helmholtz resonance frequency of the liner seems to be the lower limit of the broadband absorption of the liner. Only configuration hc006 exhibits a slight decrease of the dissipation levels for Q > 5. Mb = 0. The bias flow Mach number for each liner was chosen. The details of the partitioned cavity are illustrated Figure 2. 7.12 partitioned cavity One design choice can be to divide the liner cavity into several smaller cavities.6 0.7).12 partitioned cavity 253 Average Dissipation 0. The axial . dc006.120. Figure 7.049.053.7). The absorption increases until Q = 1 (indicated by the gray background) and is largely constant for higher values of the resonance parameter. dc015.4 0. dc014.10c.049.051.2 0 0 2 4 6 8 10 2 Resonance Parameter Q = (f/fr ) Figure 7. This is demonstrated in Figure 7. The as- sociate resonance frequencies are listed in Table 7.9. when it is operating in the bias flow regime. dc007. dc008.049.

which slightly reduces the total volume compared to the unpartitioned configuration. The open-area-ratio between metering and damping liner is kept constant for each partition. so that their damping liner bias flow velocity should be approximately the same. It can be observed that the res- onance frequencies of the two configurations are slightly differ- ent. yielding an average value .34: Illustration of the configurations studied regarding the in- fluence of dividing the cavity into several partitions. lengths of the three cavities are Lc1 = 7. In configuration dc018 the ‘acoustic’ volume is re- duced by inserting the partition walls. The theoretical Helmholtz resonance frequency of dc017 is at fr = 1211 Hz. containing one. The remain- ing specifications are given in Table 7.75 mm. and four orifice rows respec- tively. two.25 mm.1. At the low bias flow Mach num- ber the absorption is significantly increased by the partitions.35. Results The acoustic performance of the unpartitioned and the partitioned cavity is compared in Figure 7. The two parti- tion walls are 3 mm thick each. and Lc3 = 32.254 parameter study (a) dc017 (b) dc018 Figure 7. It can be assumed that the resonance effect dominates and that it seems to be amplified for dc018. Lc2 = 14 mm.

079 are virtually identical.35: Average dissipation plotted over the frequency compar- ing two liner configurations with a single large cavity (dc017) and a partitioned cavity (dc018) at two bias flow Mach numbers.13 perforation placement The position of the liner within a combuster is very important. the partitions could be tuned to respond to different resonances. To broaden the frequency range at low bias flow velocities. It can be as- sumed that the bias flow effect dominates the absorption charac- teristics. that the resonance frequency was largely kept constant. the sound field describes a distinctive spa- . This difference agrees well with the experimental data.079 0. In that case. This effect does in- crease the absorption for a small frequency range around the reso- nance. Es- pecially at resonance. of fr = 1299 Hz.6 0.016 b) Mb = 0. the partitions seem to have no influence.4 0. 7. In that case. the dissipation behaves as if the Q factor of the resonance was increased.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.13 perforation placement 255 Average Dissipation a) Mb = 0. The results at Mb = 0. 7. Discussion The cavity was partitioned in a way.

The position of the liner relative to the external sound field can be accounted for in the Eldredge and Dowling method.256 parameter study (a) dc021 (b) dc022 (c) dc023 (d) dc024 Figure 7. that the liner is most effective when it is located at a maximum of the pressure fluctuation [501]. the position of the liner does not characterize the performance of the liner itself. However. within the liner cavity. the perforation placement can be used as a design parameter for the liner itself. i. It has been shown. tial pattern of maximum and minimum pressure fluctuations. In contrast to the position of the liner. is studied. This means. the perforation placement relative to the internal sound field of the liner. The perforation placement can only be important. e. Here. when a spa- tial structure of the sound field does exist.36: Parameter study perforation placement configurations. but rather reflects the properties of the combustor or test rig. when look- .

2 0 300 600 900 1200 Frequency in Hz Figure 7. The perforation is either moved axially (cf. The config- urations studied here are illustrated in Figure 7. dc022. and dc024).36.38. that is a λ/2-resonance with fr = 612 Hz and a λ-resonance with fr = 1224 Hz. Results The results of the four configurations are compared in Figure 7. ing at the placement in axial direction. dc022 and dc023) or split up into several segments of equal local poros- ity (cf.6 0.049. The dissipation drops to significant lower values . The total amount of perforation is constant in all four configurations. The behavior is strongly dependent on the ax- ial resonances within the cavity. With 280 mm the cavity is nearly five times longer than the previous config- urations. 7.1.8 Average Dissipation 0. that the axial dimension of the cavity should be larger than half a wave length.4 0. The geometric specifications are listed in Table 7. dc023. dc022.37 where Mb = 0. Two axial resonances lie within the frequency range.13 perforation placement 257 0. and dc024. The magnitude of the ax- ial pressure distribution for these two resonances is illustrated in Figure 7. dc021.049.37: Average dissipation plotted over the frequency comparing four liner configuration with different perforation placement at Mb = 0. dc021.

5 1 0 0. Discussion The configurations in the previous sections meet the requirement that the wave length of the frequencies of interest is much longer . the perforation of config- uration dc022 is located at a pressure node of the λ/2-resonance and the dissipation curve exhibits only a minor effect at that fre- quency. Especially the resonance effect is reproduced quite accurately. The design with alternating hard wall and perforated sections seems to improve the sound absorption significantly.38: Magnitude of the acoustic pressure at resonance. The dissipation coefficient surpasses values beyond 0. For example. at these resonance frequencies. Configuration dc024 yields much higher dissipation levels com- pared to the other setups.5 1 x/Lc x/Lc (a) λ = Lc (b) λ/2 = Lc Figure 7. This effect can be reduced when placing the perforation at a pressure node (indicated by the white background in Figure 7. The transfer matrix method allows to include hard-wall sections at both ends of the perforation. Figure 7.38. The models predict a very similar behav- ior and the general agreement is good. so that configurations dc022 and dc023 can be modeled without any modifications.8 at frequencies around 800 Hz.258 parameter study 1 1 |p0 | |p0 | 0 0 0 0.39 compares the model predictions with the experimental data for these two configurations.

Experiment. [116.13 perforation placement 259 Average Dissipation 0. This transition was already observed by Davis et al.. than the dimensions of the cavity.2 a) dc022 0 Average Dissipation 0.4 0. In that case the resonances of the system are of the Helmholtz type (see Section 7. TMM:Bellucci. the resonance becomes a length-controlled phenomenon instead of a volume controlled one [. the axial resonances become dominant. . 43]: “If the resonant chamber is itself long.11) and the resonance frequency is dependent on the combination of cavity volume and orifice dimensions..049.]”.4 0.6 0.6 0. 7. TMM:Jing. When the axial length of the cav- ity is increased beyond the wave length restriction.39: Average dissipation plotted over the frequency com- paring four liner configuration with different perforation place- ment at Mb = 0. TMM:Betts.2 b) dc023 0 300 600 900 1200 Frequency in Hz Figure 7. p. in fact when Lc > λ/2.

Due to multiple reflections within the length of the liner portions of the wave pass the liner several times.2. When axial resonances do occur. the transition from hard-wall to impedance wall and vice versa. 7.2. [116] found that the perfora- tion should be located a quarter wave length away from the hard wall. In that case. the reflection from the hard-wall is 180° out of phase with the incoming wave at the perforation. The same conclusion was drawn from the experimental data above. Davis et al.14 perforation pattern The geometric characteristics of different perforation patterns are defined in Section 2.38). its ab- sorption is generally higher compared to the other configurations. i. Figure 7. It can be assumed that portions of the sound wave are reflected at each change of the wall boundary condition. yielding an increased attenuation. . e. 16].260 parameter study (a) uniform (dc007) (b) nonuniform (dc013) Figure 7. p. then the placement of the per- foration becomes crucial.40: Parameter study perforation pattern configurations. The measurements compare a uniform. This condition is meet when the perforation is located at the pressure nodes of the respective reso- nance (cf. While configuration dc024 with alternating hard-wall and per- forated sections exhibits the same resonance phenomena. so that they can- cel each other out [116.

The circumferential structure of the perforation pattern resembles the spatial pattern . staggered perforation to a nonuniform pattern with circumferen- tial variation of the porosity. 7.14 perforation pattern 261 Average Dissipation a) Mb = 0 b) Mb = 0.41: Average dissipation plotted over the frequency comparing two liner configurations with different circumferential perforation dis- tribution uniform (dc007) and nonuniform (dc013) at two bias flow Mach numbers. This peak is associated with the cut-on frequency of the first circumferential mode in the annular cavity.4 0.6 %. The remaining geometric specifications are given in Table 7. i. except for an additional peak at 714 Hz produced by the nonuniform perfora- tion.6 0. Without any bias flow in Figure 7. resulting in an overall porosity of 1 %. dc007 has a uniform porosity of 1 %. e. The local porosity of the nonuniform con- figuration (dc013) changes around the circumference from 3 % to 0.1.41 for two bias flow Mach numbers. The layout becomes more clear in Figure 7.40. Results The comparison of the acoustic performance of both setups is plot- ted in Figure 7.043 0.41a the results are very similar. where both configurations are shown. the ratio of total open area to total hard wall is identical to dc007.2 0 300 600 900 1200 300 600 900 1200 Frequency in Hz Frequency in Hz Figure 7.

This frequency is shifted to slightly higher values due to the perforated wall at the inner radius (cf.8. the dif- ference would not be reproduced by any of the models.45. The absorption of both geome- tries is rather similar. stretched in axial direction and compressed in circumfer- ential direction. so that it is excited. The change in perforation aspect ratio did not show any influence on the dissipation. When thinking in terms of porosity and diameter (which are both kept constant).4 %. 289] vary the perforation aspect ratio from a nearly square configuration with sx /sθ =0. the reso- nance effects are largely suppressed and the broadband damping due to the bias flow dominates.41b. [288. e. This devi- ation between square and rectangular pattern becomes larger at higher frequencies. resulting in sx /sθ =2.96 to a rectangular setup.93.82 %. Lee et al. [281]). Without any flow they observe an increase of the transmission loss for the rectangular arrangement. [17] present experimental results with two rect- angular perforation patterns. Andreini et al. Discussion The only experimental data that is found in the literature regard- ing a variation of the perforation pattern is concerned with the perforation aspect ratio. the abandonment of a square perfora- tion pattern for a rectangular pattern as illustrated in Figure 2. though two weak maxima are still visible for configuration dc013. The theoretical cut-on frequency for a hard-walled annulus of the same geometry is at 577 Hz. as shown in Figure 7. The requirement for the ori- .24 to 1. varying the perforation aspect ratio from sx /sθ =1.262 parameter study of the mode. When a bias flow is present. This difference is somewhat surprising. The porosity was kept constant at 8. Both measurements are done with si- multaneous grazing and bias flow and retain a porosity of 1. The actual orifice spacing seems to be important. i.

For a square perforation pattern this limit is associated with a certain porosity σ 6 4 % (see Section 7.9 mm in axial and circumferential direction. The increased orifice interaction could be one possible explana- tion for the effect observed by Lee et al.6 mm and 7. while both bias and grazing flow is present in [17]. Both values fall short of the required limit given above. one major difference between the two studies is. On the other hand. the porosity cannot be used as a sole indica- tor for the interaction anymore. [288. [288.5). The ratio of minimum spacing to orifice diameter is reduced from smin /d = 3 for the square pattern to 2 for the rectangular one. The spacing between the orifices has to be evaluated in both directions separately. Due to the distortion of the orifice grid. [288.7 and 5. this approach works only when the . Naturally. 289] are performed without any flow. This adds another degree of freedom. This is achieved by resembling the spatial structure of a cavity mode with the pattern of the perforation. 289] and their con- tradictory results to Andreini et al. even when the porosity is kept constant. This is larger than the ratios in Lee et al. that the experiments of Lee et al. The spacing of the nearly square pattern is 7.4. 7. The closer proximity of the orifices in circumferential direction could promote the interaction.9.9 mm in circumferential direction. for example for the simultaneous absorption of multiple tonal components. [17]. The spacing to diameter ratios of Andreini et al. respectively. [17] are smin /d = 4. 289] and above the spacing require- ment. Translating the porosity requirement into a minimum spacing requirement yields smin /d > 4.14 perforation pattern 263 fices to act independently without interaction is given by d  s. The experiments here have shown that the perforation pattern can be designed to promote certain cavity resonances. The rectangu- lar pattern is streched to 12 mm in axial direction and compressed to 4.

15 temperature The temperature in a combustion chamber is very high and little is known on the effect of high temperature on the acoustic perfor- mance of a bias flow liner. while the bias flow remains constant at approximately 288 K. except TMM:Betts which continuously under- estimates the data. but a shift .1. where the data is slightly more scattered. that is without or only slow bias flow. The temperature of the grazing flow is increased. comparing them to the predictions of the models. The frequency where the dissipa- tion is at maximum remains unchanged. The liner setup hc005 is a double-skin configuration (see Table 7.42 are reproduced in Figure 7. 7. The speed of sound used in the calculation of the bias flow Mach number is based on the temperature of the cold bias flow. assumed to be constant at 288 K. The operating conditions of a combus- tor liner were discussed in Section 2. Results Figure 7. yielding a bias flow Mach number of 0. For frequencies below 1800 Hz it can be observed that the dissipation decreases with in- creasing temperature. It is not only that the tem- perature is very high.43. The latter setup is reproduced here. the models seems to predict a con- stant level of dissipation for both temperature setting. Generally. The results for the two extreme temperature of Figure 7.1 for specifications).42 compares the experimental results when increasing the grazing flow temperature. The agreement between models is good for frequen- cies up to 1800 Hz. The total pressure drop across both liners is at 3 % of the duct pressure. but there is also a significant difference in the temperatures of the bias flow and the grazing flow. This trend is not so clear at higher frequen- cies.264 parameter study resonance effect is dominant.04 at the damping liner.

42: Influence of the temperature on the dissipation of configu- ration hc005 at 600 kPa and Mb = 0.6 0. TMM:Bellucci. 773 K. Average Dissipation 0.2 0 500 1000 1500 2000 2500 Frequency in Hz Figure 7. TMM:Betts.2 0 500 1000 1500 2000 2500 Frequency in Hz Figure 7.4 0.4 0. 573 K.15 temperature 265 Average Dissipation 0. . Experiment: 288 K. 773 K.04. 7. Models: EDM:Jing. 423 K. 288 K.04.43: Comparison between experiment and models at differ- ent temperatures for configuration hc005 at 600 kPa and Mb = 0.6 0. TMM:Jing.

This is a very different behavior from what is observed by the experiments. which they placed in an oven. The tempera- ture is increased up to 573 K and measurements are carried out at 120 dB (linear regime) as well as at 140 dB (nonlinear regime). that treat the influence of tem- perature on the acoustic properties of a liner. The quantities that determine the tempera- ture dependence are the flow resistance17 .266 parameter study of the dissipation maximum to higher frequencies with increas- ing temperature. that is responsible for the dependence of the flow resistivity. Elnady et al. While the reactance agrees well with the measured data. an impedance model which was not adapted for high tempera- ture use. Their findings are summarized here. The reactance is reduced with higher temperatures at both sound pressure levels. He finds that the Delany-Bazley model18 is capable of providing useful predictions. [144] measure the acoustic properties of a single orifice and cavity. In the linear regime the resistance is increasing with the tempera- ture and its maximum is shifted to higher frequencies. and the speed of sound. the density of the fluid. Discussion Very few publications are available. Christie [93] studies the acoustic absorption of porous materials at high temperatures. the resistance is overpredicted at higher frequencies. the general trend when increasing the 17 It is most probably the change in viscosity of the fluid. At 140 dB only the resonance frequency changes. However. The results are compared to a standard19 impedance model. while the maximum level remains constant with temperature. 18 The empirical Delany-Bazley model [121] describes the acoustic properties of porous materials at room temperature. so that it is not related to the tempera- ture influence. when the empirical con- stants are adjusted to the high temperature condition. . 19 That means. This mismatch occurs already for ambient temperature.

while the resistance of the linear and the porous liner is significantly increas- ing with the temperature. [380] present a series of impedance measure- ments of perforated. It was found that the resistance of the perfo- rated liner was nearly independent of the temperature. However. and porous liners at varying tem- peratures up to 773 K and including a grazing flow with Mach numbers up to 0. The viscosity increases with the temperature (see Table a. The mass reactance describes the inertia of the oscillating mass and is regarded to be proportional to the wave number (see Equa- tion (4. so that the resistance grows accordingly. their results show consistently. Sun et al. the dominant absorption effect is due to vorticity shedding. 7.15 temperature 267 temperature is captured by the impedance model. They conclude that the temperature effect is sufficiently described by adjusting the properties of air (i. While this cannot be observed in the 20 Linear liners are (nearly) independent of sound pressure level and grazing flow. and speed of sound) accordingly. e. [462] confirm these observations regarding the im- pedance in a comprehensive experimental and theoretical study with metallic porous materials. so that the influence of viscosity is of minor importance. viscosity. . Commonly. that the sound absorption is reduced when the tem- perature is increased.18). Raising the temperature increases the speed of sound (see Table a. All configurations exhibit a reduction of the mass reactance with higher temperatures.4). linear20 .4. the acoustic performance of a perforated liner de- pends mainly on the resonance effect and only to some extent on the viscous dissipation within the orifices. this is achieved by applying a wire mesh layer on top or behind the perforated facesheet.6). The results can be interpreted as follows. Without flow and in the linear regime. so that the wave number and the reactance are reduced. for example). The absorption mecha- nism of linear or porous liners is mainly based on viscous dissipa- tion. density. Furthermore. at high sound pressure levels or in combination with either bias or grazing flow. Rademaker et al.

However. where. [380] it will be mentioned here for com- pleteness. i. e. Nayfeh and Sun [350] discusses the influence of a transverse temperature gradient on the absorption of a liner. but constantly at around 288 K..268 parameter study data of Rademaker et al. This approach with different bias and grazing flow temperatures re- flects the conditions at a combustor liner in a gas turbine. the bias flow is not heated. which demonstrates the dependency of the absorption of a bias flow liner on the temperature.] cool- ing the duct walls leads to channeling of the sound towards the walls [. however. the viscosity effect increases the reactance with the temperature..14)).3). Their theoretical experiment yields higher attenuation when the temperature of the wall is decreased. [462] for a porous liner.. The effect can be compared to the refraction effect of a sheared grazing flow (see the discussion in Section 7. the temperature within the cavity and the orifices of the liner is rather constant near its ambient value. The main effect seems to be a slight reduction of the absorption with increasing temperature.. The temperature dependency observed in Figure 7. with the difference that it is independent of the direction of sound propaga- tion.13) and (4. Here.]”. This agrees with the results of Sun et al. the resonance frequency is not shifted to higher frequen- cies as reported in the literature. The viscosity effect is commonly small compared to the wave number effect. that the reactance is as well affected by the viscosity (see Equations (4. The results here present for the first time experimental data. but remains unchanged. Nayfeh and Sun [350] come to the conclusion that “[. Additionally. both temperatures are at a much higher level. A key difference between the references and the experiment here is of course the bias flow. which might be the reason for the rather small influence of the temperature. .42 might be of an entirely different nature. which contradicts the findings here. but it counteracts the results. while the given temper- ature corresponds to the mean grazing flow temperature.

2 0 500 1000 1500 2000 2500 Frequency in Hz Figure 7. [55. g. The temperature is kept at ambient.6 Average Dissipation 0. 1000 kPa.04 at all operating conditions.04. 481]). 7.16 pressure 269 0. theoretical studies. Even most combustion tests are performed at atmospheric pressure conditions (e. 7. or simula- tions. Results The results of the acoustic measurements at different static pres- sures are compared in Figure 7.1 for specifica- tions). The Hot Acoustic Test Rig allows for a controlled increase of the static pressure up to 1100 kPa.44: Influence of the pressure on the dissipation of configuration hc005 at 288 K and Mb = 0. so that the influence of the pressure on the absorption of a bias flow liner can be studied. 200 kPa. Tests were performed at am- .4 0. 101 kPa. The total pressure drop across both liners is at 3 % of the duct pressure. 600 kPa. The liner configuration is again hc005 in a double-skin arrangement (see Table 7. so that the mass flow rate of the bias flow needs to be adjusted for the varying duct pressure.44.16 pressure The high static pressure in a combustion chamber is commonly not accounted for in experiments. Doing so results in a constant bias flow Mach number of 0. 310.

The dis- sipation coefficient is largely unimpressed by the varying static pressure. 200 kPa. The predictions within the pressure range of the experiments are indistinguish- able. Ahuja et al. The absorption coef- ficient remains constant while the flow resistance changes signifi- cantly. the absorption is expected to change with the pressure. e.270 parameter study bient pressure (101 kPa). As a conclusion.43. Giese et al. The plot would reproduce the results at ambient conditions in Figure 7. 600 kPa. However. e. The same result is predicted by the models. and 1000 kPa. It seems that the static pressure does not change the acoustic properties of the liner. the experimental results with the bias flow liner paint a very clear picture: The absorption due to the bias flow effect is independent of the operating pressure. . that the absorption of the bias flow liner is not at all depending on the static pressure (i. Judging from the references. it is insensitive to changes in the fluid density. so that they are not shown here. It was observed that the absorption of porous materials is decreasing when the pressure is lowered. mainly). Discussion The experimental results demonstrate. [174] plot the maximum reach- able absorption coefficient and the corresponding flow resistance of a porous ceramic tile at various pressures. when the flow resistance remains constant. the density of the fluid. i. the acoustic properties of porous materials are indeed depending on the pressure. In a theoretical study. [9] report about acoustic measurements of porous materials at sub-atmospheric pressures.

210. the ’optimum’ Strouhal number alone does not grant an optimum dissipation. It is often- times used to describe the dependency of the acoustic properties of a bias flow liner on these quantities (e. Commonly. Here. Each row is a different liner configuration with varying orifice radii and lengths. Czech physicist) [457]. 7. However.17 strouhal number The Strouhal number21 is a dimensionless quantity relating the frequency. 242]). a characteristic length. 219.45. the Strouhal number is based on the orifice radius. the maximum dissipation oc- curs in a smaller range of Strouhal numbers around unity. while it is based on the orifice length St = kl/Mb in the right column. The levels of dissipation measured at Strouhal num- bers of unity for varying frequencies and bias flow velocities show a large diversity. The Strouhal number where the dis- sipation is at maximum changes for the different configurations. as shown in the right column. In that case. it seems that the Strouhal number can be an indicator for maximum dissipation. This variation is much smaller when the Strouhal number is based on the orifice length. Results The experimental results of various frequencies and bias flow ve- locities is plotted over the Strouhal number in Figure 7. and the flow velocity. In the left column the Strouhal number is based on the orifice radius St = kr/Mb .17 strouhal number 271 7. 21 Named after Vincenc Strouhal (1850-1922. g. consis- tently for all configurations. [70. as plotted in the left column. Mb is the mean bias flow Mach number in the orifice. This is indicated by the gray background. .

2 0 dc010. r = 0.6 0. l = 1 mm 0.2 0 dc008.2 0 dc007.45: Average dissipation plotted over the Strouhal number based on the orifice diameter (on the left) and the orifice length (on the right).25 mm dc007.5 mm dc010. l = 3 mm 0. l = 1 mm 0.5 mm dc006.25 mm dc008.6 0. . r = 1.4 0.2 0 0 1 2 0 1 2 St = kr/Mb St = kl/Mb Figure 7. r = 0.6 0.4 0. r = 1.4 0. The gray background indicates the variation of the Strouhal number of maximum dissipation for the different configurations.6 0.272 parameter study dc006. l = 1 mm 0.4 Average Dissipation 0.

. This confirms Howe’s approach of using the Strouhal number as the only parameter in his model of the Rayleigh conductivity. the dis- sipation coefficient does not have a unique relationship to the Strouhal number. maximum absorption can be achieved for Strouhal numbers of unity. 7. However.17 strouhal number 273 Discussion Bodén and Zhou [70] successfully collapse their experimental data when expressing it in terms of Rayleigh conductivity and plot- ting it over the inverse Strouhal number. Unfortunately. when the Strouhal number is based on the orifice length rather than the orifice radius.

.

In the first case. Two scenarios can be distinguished: The bias flow acting on the resonance effect and the pure bias flow effect. acoustic. The bandwidth of the absorption is in- creased while its level is typically reduced. the bias flow shifts the frequency of maximum dis- sipation to higher values. An extensive amount of available research has been collected and reviewed. 8 CONCLUDING REMARKS This thesis provides a comprehensive experimental parameter stu- dy of bias flow liners regarding their geometric. True broadband ab- sorption is obtained when the bias flow effect dominates. and flow parameters. compared to the acoustic performance without flow. that is when flow reversal does not occur. The consistent database of experimental results allowed to put the existing information into perspective. A combustor liner is required to operate in the linear regime to prevent hot gas inges- tion. . thermodynamic. The significant features of each parameter can be summarized as follows: sound pressure level The behavior of the liner is linear when the bias flow velocity is larger than the acoustic velocity. bias flow The bias flow was able to improve the absorption of all tested configuration.

at certain combinations of graz- ing flow velocities and liner geometries sound might be produced rather than absorbed. in particular concerning the bandwidth. where a similar dissipation is obtained when bias or grazing flows are applied separately. simultaneous grazing & bias flow The influence of the grazing flow can be neglected when the ratio of bias to grazing flow velocity is larger than Mb /Mg > 0. In order to achieve similar results the mean grazing flow Mach number must be a factor of ≈ 3.276 concluding remarks grazing flow The effect of the grazing flow on the acoustic properties of a perforated liner is identical to the bias flow effect. when the . wall thickness The wall thickness seems to be of minor sig- nificance at high bias flow velocities. orifice cross-section shape The acoustic performance is largely independent of the orifice cross-section shape. A thin configuration is su- perior.5 higher than the Mach number of the bias flow jets.3. This corresponds to the condition Mg ≈ 3. and number of orifices are unimportant. Below this limit the influence of the grazing flow can be accounted for by an effec- tive discharge coefficient. Due to the convective effect of the grazing flow. The individual parameters orifice area. which essentially increases the bias flow velocity. the absorption of sound waves traveling against flow direc- tion is increased while it is reduced in the opposite direction. at low bias flow velocities. The situation is more complex without bias flow and in the transition region between resonance regime and bias flow regime.5 Mb . orifice spacing. porosity When the bias flow effect dominates the magnitude of the dissipation is directly related to the level of porosity. However.

The experimental data reveals that it is indeed the jet velocity and not the orifice mean velocity. below which the absorption falls off from its broadband maximum level. cavity volume In the resonance dominated regime. while it is reduced in the other direction. In the bias flow regime the no flow Helmholtz res- onance frequency provides a cut-off frequency. At lower velocities more shallow angles result in a slightly superior bandwidth. double-skin configuration A double-skin configuration allows for a reduction of the bias flow velocity through the damp- ing liner when a certain pressure drop needs to be met. Introducing a bias flow shifts the dissipation maximum to higher frequencies. The in- fluence of the second liner on the acoustic performance becomes insignificant when the bias flow velocity increases. orifice angle The average dissipation is independent of the orifice angle at high bias flow velocities. the orifice edge geometry exhibits only a minor influence on the absorption. Complying with this rule. an effect is expected for long and narrow openings. the fre- quency of maximum dissipation is associated with the Hemholtz resonance. The observed behavior might be the result of the effectively reduced length and the in- creased open area at the surface when both edges are rounded. . The ab- sorption of sound waves traveling in direction of the bias flow injection is increased. orifice edge geometry The velocity of the bias flow jet is very sensitive to the orifice edge geometry. concluding remarks 277 open area remains constant. which needs to be constant when comparing different se- tups. which is inversely proportional to the cavity volume. However. For a typical installation that counteracts the convective effect of a grazing flow.

In that case. a Strouhal number of unity seems to be a neccessary condition only. the partitions show no influ- ence when the bias flow velocity is high. However. as experienced in a combustor. temperature Increasing the temperature of the grazing flow results in a reduction of the absorption. However. when the Strouhal number is based on the orifice length rather than the orifice radius. but is not sufficient that maximum absorption will be achieved. ampli- fies the resonance effect. perforation placement The placement of the perforation relative to the liner cavity becomes important when the cavity is longer than half an acoustic wave length. pressure The pressure seems to have no influence on the ab- sorption. so that this result reflects the influence of the temperature difference between grazing and bias flow. However. The experiments show that a configuration with alternating areas of hard-wall and perforation yields an improved absorption. However. strouhal number Maximum dissipation can be obtained at a Strouhal number of unity. in this setup the bias flow is not heated.278 concluding remarks partitioned cavity Dividing the cavity into several smaller cavities while keeping their Helmholtz resonance constant. perforation pattern The absorption seems largely inde- pendent of the perforation pattern. it should be preferred to place the perforation only at the pressure nodes of the axial resonance in the cavity. . a nonuniform cir- cumferential perforation distribution can be used to promote a circumferential resonance.

is not clear. it seems to correlate with the porosity of the liner. The Helmholtz resonance frequency is determined by the combination of the porosity. if a resonance is provoked or suppressed. Here.33). The upper frequency limit of the broadband absorption is pushed to higher values while the bias flow velocity is increased.8 %). the acoustic performance of the liner is very poor without bias flow. i. The level of the broadband absorption is depend- . It provides a cut-off fre- quency. the bias flow suppresses the resonance effect leading to a broadband absorption. below which the absorption falls off from its broadband maximum level (see Figure 7. at the penalty of lowering the absorption level. The bias flow continuously increases the absorption until an optimum is reached. concluding remarks 279 Many of the parameters are becoming insignificant when the bias flow effect dominates. The resonance effect occurs at low porosities (here around 1 %) and is absent at high porosities (here > 6. Ultimately. the wall thickness. while it reduces the level of the broadband absorption. Bias flow velocities suc- ceeding this optimum increase the broadband damping towards higher frequencies. the porosity. typically at a much lower level then the maximum observed due to the resonance effect. The domination of the bias flow effect on the acoustic damping performance is mainly determined by three parameters: the bias flow velocity. The condition that decides which behavior will occur. and the cavity volume. however. e. where the acoustic per- formance surpasses the best achievement of a resonance type liner in absorption level and frequency range. then the bias flow shifts the absorption maximum to higher frequencies and broadens its frequency range. depending on the resonance characteristics of the liner-cavity system. and the no flow Helmholtz resonance frequency of the liner-cavity system (assuming all dimensions are much smaller than the wave length). If a reso- nance occurs without flow. There seems to be a fundamentally dif- ferent behavior of bias flow liners at low velocities. In the case without resonance.

by Jing and Bellucci. The best agreement is achieved where the bias flow effect is described by Howe’s theory. EDM provides the ability to in- clude a second liner in a double-skin configuration. Recent numerical models by Jing and Sun [242] and Lee et al. e. Both methods have some unique features readily available. the Jing model is restricted to the presence of a bias flow. all models tend to under- estimate the dissipation. Higher porosities yield an improved absorption when keeping the velocity constant. While the Bellucci model has been extended to function as well when there is no bias flow. while EDM pre- dicts slightly higher dissipation values. However. i. none seems to capture it quite right. considering the different approaches taken.280 concluding remarks ing on the porosity and bias flow velocity. TMM allows for the simple integration of hard-wall extensions on both sides of the perforation. It is quite surprising that the deviation between the models themselves is rather small. This transition seems to be mainly governed by the reduc- tion of the reactance. Models All models capture the general trends of the experiments fairly well. While there is a wide variety in the models of the behavior of the reactance. One common problem of the models is the transition between the resonance dominated damping to the purely bias flow related regime. The agreement between the models and the experimental data is generally satisfactory. in the case where the perforation does not cover the entire length of the cavity. . [291] predict a much stronger effect on the reactance. The Eldredge and Dowling method (EDM) and the transfer ma- trix method (TMM) give largely similar results. which often reproduces the experimental results a bit better.

[436] point into this direction. They all use optical measurement techniques to capture the physical effect right at the orifice. An enhanced phys- ical understanding of the underlying mechanisms is needed to improve the prediction quality of the models. Recent studies by Rupp and Carrotte [415] and Heuwinkel et al. [201] and the ongo- ing work of Schulz et al. concluding remarks 281 Outlook The models and conclusions that have been drawn here and else- where are often based purely on empiricism. . The results of further efforts in this direction are highly anticipated.

.

9) Pr = cp µ/κ (A. A P R O P E RT I E S O F A I R a.1 equations of ideal gas properties Air at ambient conditions is commonly regarded to behave like an ideal gas.4) T + Ts Tref 3/2 Tref + Ta e−Tb /Tref  T κ = κref (A.6) γ−1 cv = R/(γ − 1) (A.7) ν = µ/ρ (A.11) with reference quantities and constants listed in Table a.10) z0 = ρc (A.1. [414]) p ρ= (A.8) χ = κ/(ρcp ) (A.5) T + Ta e−Tb /T Tref γ cp = R (A. so that the properties are approximated by (e.1) RT p c = γRT (A. g.3) cv  3/2 T + Ts T µ = µref ref (A.4 (A.2) cp γ= = 1. .

the properties of air are calculated based on the real gas equations throughout the entire work. the ideal and real calculations agree very well at ambient conditions. Pressure pref = 1.284 properties of air Table a. low temperature. the first phenomenon might be of importance. Therefore. 2 Batchelor [33] gives a temperature of 3000 K for oxygen and 6000 K for nitro- gen. However. Dynamic Viscosity µref = 1. real gas The ideal gas assumptions reproduce the behavior of air at ambi- ent conditions very accurately.05287 J kg−1 K−1 Sutherland Constant Ts = 110. the prop- erties of air are calculated based on the real gas equations given by Lemmon et al.2 presents a comparison at ISA conditions and at the extreme condition of the Hot Acoustic Test Rig HAT.6 K a.4 K Empirical Constant Ta = 245. Ref. Motivated by this rather large discrepancy.4 K Empirical Constant Tb = 27. Table a.1: Some reference quantities and constants. 1 That can either be at high pressure. [296] and Lemmon and Jacobsen [295] and com- pared to the ideal gas results.5343×10−2 W m−1 K−1 Specific Gas Constant R = 287. .15 K Ref.2 ideal gas vs. Compiled from ICAO 7488:1993 [220] and [414].01325×10 Pa −5 Ref. Temperature Tref = 288. As expected. While the second ef- fect is far beyond the temperature range encountered in this work. However.7894×10−5 N s m−2 Ref. at the elevated pressure and temperature condition their differences can become as large as 12 %. or both. Thermal Conductivity κref = 2. real gas effects become more prominent when the density of the molecules is increased1 or at very high temperatures2 when polyatomic molecules might dissociate into their constituent atoms [33].

31 % ν 10−5 m2 /s 1. 101.5499 0.4045 0. 2000 kPa.6884 3.4660 0. 101.36 % 0.6759 -1.4029 3.15 K. • HAT extreme conditions: 823.0682 0.1763 7.325 kPa 823.8595 -0.4022 0.3527 -3.15 % 1.5314 2.6366 -0.2252 0.38 % 3.1686 4.7962 0.15 K.02 % 4.1741 12.7894 1.3 tables of real gas properties 285 Table a.00 % 7. 288.0568 2.49 % z0 102 kg/s/m2 4.62 % cp 103 J/kg/K 1.1429 -10.15 % 1.6553 4.15 K.70 % κ 10−2 W/m/K 2. 1100 kPa) calculated according to ideal gas and real gas relations and their difference in percent.07 % 26.4607 1.21 % γ 100 – 1.74 % a. a. Each table lists a variety of pressure and temperature combi- nations.1986 12.8147 3.317 -1.1763 0.1714 0. 101.7515 5.15 K.19 % 6.0047 1.0880 -0. and • exemplary combustor conditions: 2000 K.1763 8. Some distinguished conditions are (all indicated in bold): • ISA conditions (sea level): 288.3 tables of real gas properties The following tables for dry air are based on the real gas equa- tions given by Lemmon et al. [296] and Lemmon and Jacobsen [295].2596 1.15 K.0062 0.2: Various properties of dry air at ISA conditions (at sea level: 288.40 % c 102 m/s 3.15 K.72 % 5.8911 5.4000 1.8227 3.7923 0.1018 7.2903 7. 1100 kPa.21 % Pr 10−1 – 7.775 26.05 % 5.4000 1.54 % χ 10−5 m2 /s 2.1057 9.325 kPa.14 % cv 102 J/kg/K 7. 1100 kPa ideal real ∆ ideal real ∆ ρ 100 kg/m3 1.325 kPa) and the extreme conditions of the Hot Acoustic Test Rig (823. .55 % 1.33 % µ 10−5 N s/m2 1.2250 1.0047 1.

15 332.23 387.15 0.80 477.02 866.15 0.085 323.065 773.6578 12.0438 1.15 349.1562 6.2758 1.9457 1.95 363.37 565.66 Temperature T in K 293.24 333.77 350.00 551.15 411.58 476.38 344.15 0.2252 2.8457 2.2092 1.91 344.8338 1.617 423.68 411.9315 8.45 360.777 288.4503 0.04 387.91 481.4230 0.00 0.286 properties of air Table a.76 388.22 553.4563 0.61 2000.29 349.9456 823.98 388.673 23.1764 0.4715 10.135 13.59 566.15 1.3: Density ρ in kg m−3 .6366 8.51 863.48 331.15 476.79 361.1492 1.3779 7.15 343.5532 7.15 1.353 Temperature T in K 293.0547 6. Pressure p in kPa 100 101.325 200 600 1000 1100 2000 273.89 412.113 25.59 569.17 361.83 573.1885 1.15 0.26 864.19 288.97 549.54 340.81 415.4194 7.9119 3.44 351.79 323.27 823.870 21.39 .15 387.17 349.9360 8.6454 4.15 1.353 24.82 332.17 349.2991 6.7384 1.15 1.4038 2000.09 551.3227 10.37 341.49 342.45 340.38 343.48 343.30 350.1439 11.00 863.2145 3.9045 11.04 387.0779 1.48 331.81 423.1645 2.48 567.38 343.8229 0.54 331.5334 4.6075 0.15 1.57 476.2165 4.4888 4.378 573.823 14.95 390.5965 9.519 12.4: Speed of sound c in m s−1 .66 864.45 360. Pressure p in kPa 100 101.9333 0.15 331.93 341.50 373.67 478.790 11.97 548.2928 2.254 18.2043 2.119 23.15 0.922 13.51 863.8664 5.6382 6.0296 16.15 360.50 345.38 565.6971 4.87 865.1741 0.2700 12.9003 2.15 565.73 478.4707 Table a.73 413.595 373.0922 2.325 200 600 1000 1100 2000 273.68 411.59 361.15 340.45 340.19 550.05 773.3482 1.59 413.6771 12.15 548.74 303.4286 0.37 567.6155 1.914 303.2109 9.

2 3988.00 150.068 68.69 777.3 2631.04 2415.8 4296.537 36.129 24.689 18.084 38.15 289.34 579.359 18.8 2000.081 68.544 288. Pressure p in kPa 100 101.7 Temperature T in K 293.511 303.204 2000.0 4441.0 4035.7 4786.117 18.2 7275.909 21.218 17.92 2478.980 773.071 68.49 2547.8 423.28 406.8 373.15 388.1 2898.296 17.1 323.819 29.703 18.15 18.70 2171.777 19.30 1435.5 7849.4 5803.705 19.5 1653.7 4519.15 408.15 401.890 29.564 36.07 1738.854 29.4 8268.15 17.15 361.15 36.027 24.92 428.22 366.961 17.15 239.35 152.7 2473.205 18.0 303.144 423.10 1503.023 22.6 573.325 200 600 1000 1100 2000 273.115 38.658 823.084 38.8 3007.916 373.7 288.6: Dynamic viscosity µ in 10−6 kg m−1 s−1 .986 323.50 293.4 3396.343 18.14 242.531 36.15 17.0 3736.15 29.0 8588.976 18.5: Characteristic acoustic impedance z0 = ρc in kg s−1 m−2 .635 19.249 573.2 8121.53 816.3 3903.094 68.15 247.15 422.027 24.591 36.117 .49 494.141 24.649 19.8 4105.60 803.038 24.26 677.45 1483.21 250.15 24.6 6810.4 4259.761 18.7 773.959 22.15 411.898 29.00 68.083 24.2 3188.838 18.15 21.54 393.597 36.147 38.635 19.010 22.6 4142.272 Temperature T in K 293.100 18.77 343. Pressure p in kPa 100 101.140 38.15 38.15 19.090 38.325 200 600 1000 1100 2000 273.3 4689.01 722.1 2392.52 846.3 tables of real gas properties 287 Table a.896 21.5 4560.092 68.220 18.762 19.3 2720.15 338.811 29.0 Table a.34 300.206 18.280 18.31 478.689 18.3 3624.72 2035.233 17.3 823.14 823.15 18. a.218 17.70 902.530 36.811 29.68 417.896 21.12 413.962 17.0 8344.8 4949.380 17.068 68.363 17.037 18.48 2337.822 18.78 2456.

495 44.4806 573.462 23. Pressure p in kPa 100 101.906 26.726 773.9237 2.262 16.15 28.1128 3.977 9.360 24.001 35.879 32.83 195.4144 4.15 16.694 2000.94 385.660 7.325 200 600 1000 1100 2000 273.8315 1.360 24.9387 2.060 40.15 25.357 28.390 28.117 7.609 4.668 25.1895 423.52 114.169 35.00 114.7173 1.746 31.873 25.496 13.620 31.442 573.2530 1.042 15.040 26.433 44.9366 4.5588 1.182 26.6621 2.696 373.949 27.626 Table a.696 24.541 24.15 49.4810 1.567 303.498 25.15 58.082 28.215 17.53 114.134 423.5461 2000.854 14.15 18.595 58.8224 323.226 39.0979 823.15 25.51 65.816 14.3567 0.121 80.908 56.0448 1.649 26.49 114.7: Kinematic viscosity ν = µ/ρ in 10−6 m2 s−1 .3609 2.4916 1.15 55.15 29.433 24.319 6.49 114.1479 1.736 25.499 25.585 58.045 9.15 35. Pressure p in kPa 100 101.531 25.6661 0.9223 373.7498 2.739 3.418 44.644 31.318 15.852 31.15 14.198 28.118 288.490 58.582 13.618 26.00 390.620 31.2059 4.8836 2.7503 Temperature T in K 293.022 35.395 24.851 55.56 .3540 1.5385 1.4864 0.083 28.110 35.2273 4.0454 8.778 26.417 44.038 88.4849 773.282 323.795 55.896 55.6735 1.4298 2.914 26.813 25.288 properties of air Table a.049 8.8: Thermal conductivity κ in 10−3 W m−1 K−1 .3995 0.618 26.15 81.15 26.4908 2.553 8.2315 0.862 45.000 35.501 58.542 58.201 35.491 58.112 28.074 48.15 15.15 44.556 8.232 28.15 13.575 44.616 19.795 55.1515 7.325 200 600 1000 1100 2000 273.6341 1.15 23.155 11.559 44.014 823.6806 288.874 25.51 114.851 26.806 55.49 114.15 24.224 35.15 90.15 31.7741 303.1349 2.207 Temperature T in K 293.218 26.

3517 1.3837 3.4111 1.325 200 600 1000 1100 2000 273.2981 1.3516 1.4017 1.2894 2.91 56.733 9.3521 1.1406 323.706 11.3088 2000.3593 823.5922 2.3566 1.391 Table a.1881 3.981 18.84 518.4080 1.0192 6.9534 1.4136 1.15 23.4184 1.4223 1.516 12.2981 1.677 47.718 18.97 87.3943 1.469 3.15 1.15 1.23 123.4167 1.429 6.3855 773.3579 1.3795 1.3809 1.2981 1.4010 1.2981 .15 25.4030 1.3581 1.2981 1.4788 2.2981 1.682 10.15 20.854 25.007 22. a.3428 1.350 5.15 1.4021 1.4100 1.325 200 600 1000 1100 2000 273.3792 1.4136 1.3980 1.3526 1.4020 1.807 3.4015 1.680 7.8809 1.15 21.4280 373.8083 2.4019 1.15 1.8910 1.15 69.3989 1.4445 288.15 1.15 1.4016 1.59 62.4203 1.9331 288.4027 1.7084 0.39 111.4184 1.3979 1.3825 1.15 1.922 4.4801 3.987 11.638 20.15 18.4193 1.898 26.4040 1.4369 303.5822 3.10: Heat capacity ratio γ = cp /cv .0803 1.912 12.4170 423.3568 1.3822 1.15 1.0698 303.632 21.4022 1.3979 1.496 3.807 41.4007 1.4049 1. Pressure p in kPa 100 101.033 34.9740 4.3538 2000.567 11.0999 573.8232 2.15 113. Pressure p in kPa 100 101.15 1.4151 1.261 20.948 69.4387 Temperature T in K 293.380 10.4007 1.1468 1.348 10.0350 Temperature T in K 293.00 1.907 6.1499 1.3516 1.3951 1.2866 373.956 20.4038 1.936 11.5245 773.4082 573.5777 2.2981 1.4213 1.3 tables of real gas properties 289 Table a.7135 823.9: Thermal diffusivity χ = κ/(ρcp ) in 10−6 m2 s−1 .4027 1.15 41.4337 323.3499 3.3566 1.4021 1.496 33.4032 1.3792 1.6771 423.3943 1.3527 1.3011 2.15 33.747 5.058 16.15 1.00 525.4117 1.4079 1.726 52.4070 1.4245 1.15 125.0814 1.0456 1.3574 1.96 262.

33 963.42 721. Pressure p in kPa 100 101.15 1011.0 1023.42 719.22 730.9 1023.2 323.44 963.22 373.15 817.65 719.9 1020.9 1025.8 1051.68 730.1 1038.2 1008.19 722.85 806.9 773.95 758.53 .39 722.4 1008.15 1006.07 817.7 Temperature T in K 293.9 1044.9 1093.5 1011.7 1092.9 1007.7 1006.7 1007.15 723.29 718.90 718.03 817.5 1250.69 Temperature T in K 293.11 717.11: Specific heat capacity at constant pressure cp in J kg−1 K−1 .63 717.13 720.8 1029.6 1250.2 1019.00 1250.15 805.76 2000.8 373.7 1009.9 1032.23 823.30 722.41 719.15 1007.5 1048.15 1045.6 1250.65 729.00 963.01 726.12: Specific heat capacity at constant volume cv in J kg−1 K−1 .9 1005.3 1030.46 720.9 1015.40 720.67 303.96 720.7 1008.0 1105.28 724.2 1006.84 718.80 573.7 1039.15 1092.65 729.38 817.4 303.41 817.79 731.11 720.4 1006.4 1018.9 1022.95 288.7 1107.5 1023.1 1048.02 758.55 723.11 717.325 200 600 1000 1100 2000 273.5 1094.77 730.64 805.55 723.33 963.9 823.9 1023.95 757.63 717.48 805.64 722.7 1024.3 1104.55 758.53 719.38 963.73 323.2 1094.15 1006.325 200 600 1000 1100 2000 273.15 1017.4 1250.31 423.5 1250.03 720.81 805.34 963.1 2000.86 725.03 817.15 717.42 718.0 1014.3 1104.22 773.3 288.4 1020.3 1014.2 423.4 1095.290 properties of air Table a.7 Table a.0 1014.43 723.15 719.43 805.88 720. Pressure p in kPa 100 101.4 1250.42 963.4 1045.15 1006.15 729.15 718.7 1092.22 817.5 1012.4 1016.15 1005.5 573.4 1105.15 1104.62 720.3 1019.7 1021.29 758.83 717.09 719.29 718.15 757.43 805.6 1022.15 717.23 721.1 1020.62 759.7 1047.15 717.4 1045.2 1036.0 1014.4 1017.69 724.6 1105.6 1020.

13: Prandtl number Pr = cp µ/κ.7194 0.7076 0.7112 0.7035 0.7016 0.15 0.7033 0.7190 0.7068 0.7159 0.7175 0.7156 0.7208 0.7110 0.7166 0.15 0.7249 Temperature T in K 293.7004 0.7435 0.7164 0.7003 0.7009 0.7052 0.7435 0.15 0. a.7107 0.7026 0.7088 0.6984 0.7097 0.325 200 600 1000 1100 2000 273.15 0.15 0.7294 288.7199 0.7206 2000.7168 373.7154 0.7435 0.15 0.7079 0.7046 0.7236 303.7164 0. Pressure p in kPa 100 101.7436 0.7048 0.7163 0.7436 0.7198 0.7120 0.7029 0.00 0.7110 0.7160 0.7172 823.7017 0.15 0.6984 0.7435 0.7068 0.7437 .6988 0.7052 0.7016 0.7105 0.15 0.7190 0.7081 0.7093 423.7051 773.7045 0.7004 0.7081 0.7020 0.7138 0.7155 0.3 tables of real gas properties 291 Table a.7051 573.7123 0.7191 0.7090 0.15 0.7154 0.7145 0.7211 323.7088 0.15 0.7199 0.7132 0.7018 0.

.

B S O M E M AT H E M AT I C S b.4) v0 so that it is related to the Rayleigh conductivity by iωρA z= (B.6) ρc KR .2) results in v0 KR = iωρA (B.1) p̂1 − p̂2 with q̂ = v0 · A (B.1 rayleigh conductivity and impedance The Rayleigh conductivity q̂ KR = iωρ (B.5) KR or normalized with ρc z ikA = (B.3) p̂1 − p̂2 The specific acoustic impedance is defined as p̂1 − p̂2 z= (B.

(B.12) z0 1 1 ikx v̂2 = (p̂+ e−ikx − p̂− 2 e ). (B. 23].13) z0 2 For convenience one can write ikx ikx P1+ = p̂+ 1 e −ikx . and (B.8) and (B. P1− = p̂− 1 e . Inserting Equations (B.14) z0 P1+ − P1− = (P2+ + P2− )z0 T21 +(P2+ − P2− )T22 .10) ikx p̂2 = p̂+ 2 e −ikx + p̂− 2 e .11) 1 ikx v̂1 = (p̂+ e−ikx − p̂− 1 e ).7) v̂1 T21 T22 v̂2 Equation (B. (B. and (B.9) The acoustic pressure and the acoustic particle velocity at posi- tions 1 and 2 can be given in terms of incoming and outgoing waves as ikx p̂1 = p̂+ 1 e −ikx + p̂− 1 e .294 some mathematics b. and P2− = p̂− 2 e .13) into Equations (B.2 conversion of transfer matrix into scattering matrix This section follows closely the derivation given in [217.7) can be written as p̂1 = p̂2 T11 + v̂2 T12 (B.8) v̂1 = p̂2 T21 + v̂2 T22 (B. P2+ = p̂+ 2 e −ikx .9) with the simplification above yields T12 P1+ + P1− = (P2+ + P2− )T11 +(P2+ − P2− ) .15) . (B.10) to (B. pp. The transfer matrix T relates the acoustic pressure and particle velocity at the input of a system to its output: T T12      p̂1 p̂2 = 11 (B.

18) and (B. S12 = .16) P1+ − P1− = Y + P2+ + Y − P2− . (B.17) which can be rearranged to X+ − Y + + X+ Y − − Y + X− − P1− = P + P2 . z0 then Equations (B. X + Y+ X + Y+ .15) yield P1+ + P1− = X+ P2+ + X− P2− . and (B.14) and (B.19) X + Y+ 1 X+ + Y + 2 Written in matrix form.19) result in the common expression of the scattering matrix p̂− S S12 p̂+      1 = 11 1 (B. X+ + Y + X+ + Y + 2 X− + Y − S21 = + . and (B. Equations (B. (B.b.20) p̂+ 2 S21 S22 p̂− 2 with the elements X+ − Y + X+ Y − − Y + X− S11 = .2 conversion of transfer matrix into scattering matrix 295 Introducing the substitutions T12 X± = T11 ± and Y ± = z0 T21 ± T22 . and S22 =− + .18) X+ + Y + 1 X+ + Y − 2 X− + Y − − P2+ = + P + − P .

.

) . K. J. 2000. P. Stegun. 1991. K. Agostino. D’Agostino.) [7] Ahuja. J. Gaeta. 1970. Mechanical Systems and Signal Processing.: “Sound absorption of a 2DOF res- onant liner with negative bias flow”. 35.) [9] Ahuja.) [2] Åbom. Contractor Report NASA-CR-2000-210633. J.. (Cited on pages 28. 2000. H. R.) [3] Abramowitz.BIBLIOGRAPHY [1] Åbom. NASA.) [5] ACARE: “Flightpath 2050 . Contractor Report NASA-CR-2000- 210637. 1988. Advisory Council for Aviation Research and Innovation in Europe.: “Error analysis of two-microphone measurements in ducts with flow”. 153. Gaeta.K. Advi- sory Council for Aviation Research and Innovation in Europe. Gaeta.: “Active control of liner impedance by varying perforate orifice geometry”. National Bureau of Standards..) [8] Ahuja. Gaeta. and 154.. (Cited on pages 139.D.. and 235. K. and 54.J. Hsu. (Cited on pages 146 and 147. R. K.) [4] ACARE: “European aeronautics: A vision for 2020”. M. R.. NASA. Journal of the Acoustical Society of America.: “Acoustic properties and durabilitiy of liner materials at non-standard atmospheric conditions”. Report. 28.: Handbook of mathematical functions. (Cited on pages 62 and 103. Cataldi. 38.Europe’s vision for aviation”.J. Contractor Report NASA- CR-2000-210636. NASA..: “Acoustic absorption charac- teristics of an orifice with a mean bias flow”.K. 5(2):89–104. K. J. (Cited on page 4. 29. NASA.. 1994. M.: “High amplitude acoustic behaviour of a slit-orifice backed by a cavity”. (Cited on pages 34 and 54. (Cited on pages 26 and 270. (Cited on page 18.: “Measurement of the scattering-matrix of acoustical two- ports”. Bodén. Contractor Report NASA-CR-195010.. J. NASA.K. 2001. M. Gaeta. Contractor Report NASA- CR-2000-210635. (Cited on page 4.. (Cited on pages 17. 2000.K.S. M. Ap- plied Mathematics Series 55.A.. I. Report.) [10] Ahuja. 2000. 83(6):2429– 2438.) [6] Ahuja.. R.K. M. 2011.J. R.

49.) [18] Andreini. A. Copenhagen. (Cited on pages 45. 19. A.L. 2-6 July 2006. 58(1):42–48...J. Simonetti. 73. S..) [13] Allam. Texas. B. (Cited on pages 124 and 126. 15(2):175–196. 35. A. AIAA-2008-3074. I. 55. F..D. 55. 2005. C. and 75.: “Experi- mental and theoretical investigation of thermal effectiveness in multiper- forated plates for combustor liner effusion cooling”. Facchini. S. (Cited on page 73. Åbom. M.) . San Antonio. B. F. Turrini.: “Thermoa- coustic analysis of a full annular lean burn aero-engine combustor”. 22.Peschiulli: “Experimental investigation of effusion liner geometries for aero-engine combustors: Evaluation of global acoustic parameters”. 136(9):1–13. 55. 19.) [15] Anderson. Vancouver..: “Acoustic modelling and testing of a complex car muffler”. Journal of Sound and Vibration. GT2013- 94877. 49. and 71. 245. (Cited on pages 49 and 50.. F. L.. Physics Today. (Cited on pages 16. Bianchini. Canada. S. 2013. R. Åbom.O. F. 72. Journal of Sound and Vibration.. GT2012- 69853. 2014. Vancouver.. (Cited on pages 16.) [12] Allam. B. 2006. A. M.. 2012.: “Investigation of damping and radiation using full plane wave decomposition in ducts”. 3-7 June 2013... 1971. 73. GT2011-46303. Peschiulli. J. P.: “Performance of exhaust silencer com- ponents”. and 75.: “Assessment of numerical tools for the evaluation of the acoustic impedance of multi-perforated plates”. Vienna. 47. 2008. 262. Canada. A.. 49.: “Experimental characterization of acoustic liners with extended reaction”. (Cited on pages 45. Ferrari. Åbom. 2006. NASA. Lenzi. in 14th AIAA/CEAS Aeroacoustics Conference. 246. in ASME Turbo Expo 2013. 50. Contractor Report NASA-CR-111405. M. Journal of Turboma- chinery.) [17] Andreini. in ASME Turbo Expo 2011.) [20] Anonymous: “Acoustic liner feasibility program”. Facchini. A. 6-10 June 2011.) [19] Andreini. USA.. Simonetti. A. in ASME Turbo Expo 2012.. in International Congress on Sound and Vibration. B.. 292:519–534.. (Cited on page 53. (Cited on page 44.) [16] Andreini.: “Ludwig Prandtl’s boundary layer”. 1970. 2011. Picchi. A. G. Giusti.. L. 11-15 June 2012.298 bibliography [11] Alfredson. Davies.. Tarchi. Denmark. Facchini.) [14] Allam. ICSV13. 5-7 May 2008. (Cited on page 117. Turrini. Facchini. Austria. and 263.. 139.. Vitale.A.

H.: “Smoothing periodigrams from time-series with continu- ous spectra”. A..E. 3. in W. pp. L. Nature.J. Contractor Report NASA-CR-120848. 1958. (Cited on page 140. American Society for Testing and Materials. T... 53.): Marks’ Standard Handbook for Mechanical Engineers. (Cited on pages 37 and 38.: “Absorption of sound by air: High temperature predictions”. 88:2019–2021. 131. 34.B. H. 132.A.L. (Cited on page 130. Baumeister. Evans.) [30] Bass. 1981. Journal of Sound and Vibration. Sadegh. Y. A.) [31] Bass.E. NASA.) . 1948. 2012. Starobinski. Journal of the Acoustical Society of America.E. 12(3):72–82. H. and 202. 109(1):59–64. and 134. L.. 1990. H.. (Cited on page 32. Orlando.) [22] ASTM E 1050-12: “Standard test method for impedance and absorption of acoustical materials using tube.. Frequenz... (Cited on page 219. D. L. Journal of the Acoustical Society of America. Vol. M.. 2001. J. Journal of the Acoustical Society of America. R. Journal of the Acoustical Society of America. Leroux. 17. Evans. F. Piercy.. (Eds. Standard.: “Atmospheric absorption of sound: Update”.Principles and Methods. 1971. 2007. R. Sutherland.) [24] Auregan.S.) [23] Auregan.M.: “Acoustic grazing flow impedance using waveguide principles”.. (Cited on page 146. two microphones and a digital frequency analysis system”.) [25] Avallone. E. (Cited on pages 28. M.) [26] von Barthel. L. 45.: “Experimental evidence of an instability over an impedance wall in a duct with flow”. 69(1):124–138. FL.) [28] Bass. (Cited on pages 130 and 132. 44. 317:432–439. H.) [29] Bass. (Cited on page 32.E. McGraw Hill.: “Influence of grazing flow and dissipation effects on the acoustic boundary conditions at a lined wall”. Sutherland. Zuckerwar.N. Pagneux. bibliography 299 [21] Armstrong. Chap. 1984.).: “Atmospheric absorption of sound: Analytical expressions”.: “Untersuchungen über nichtlineare Helmholtzres- onatoren”.C.P. Mason.) [27] Bartlett. Thurston (Eds. 145–232. 2008.J. Bauer. 161:686–687.. Physical Acous- tics . 52(3):821–825. (Cited on pages 130. 11th ed. Academic Press. Y. 133. (Cited on page 130..C. New York. V. 1972.: “Absorption of sound by the atmosphere”.

53.) [35] Baumeister. Pfizenmaier.) [34] Bauer. 1980.: “On the use of Helmholtz resonators for damping acoustic pulsations in industial gas .: An introduction to fluid dynamics. D.: “Sound absorption caused by vorticity shedding.W.: “Atmosperic absorption of sound: Further developments”. E. AIAA-1979-0575. F. 16(9):873–874..E. P.J. Flohr.) [41] Bellucci. and 53.. Journal of the Acoustical Society of America. 211. 78..) [42] Bellucci. and 231. K. AIAA-1977-1278.) [38] Bechert. 70(3):389–405. in 4th AIAA Aeroacoustics Conference. 1977. in 5th AIAA Aeroacoustics Conference. 12-14 March 1979. (Cited on pages 39. 2000. 98. Sutherland. 1995. P.) [40] Bechert. (Cited on pages 45. D. 3-5 October 1977. Washington.B. (Cited on pages 28. Jour- nal of the Acoustical Society of America. and 284. 29. 74. AIAA Jour- nal.: “Experiments on the trans- mission of sound through jets”..) [37] Bechert.. (Cited on pages 45. A. L. Flohr.T.. Michel. Pfizenmaier. D. Atlanta.. Blackstock.: “Impedance theory and measurements on porous acoustic liners”. 73.: “Boundary layer attenuation of higher order modes in rectengular and circular tubes”. U. D.: “Sound absorption caused by vorticity shedding. 46. and 75. A. Journal of Sound and Vibration. AIAA Journal.C. R.W. Seattle.: “Numerical and experimental study of acoustic damping generated by perforated screens”. 97.300 bibliography [32] Bass. (Cited on pages 45 and 53.M. Cambridge University Press. (Cited on pages 46 and 53. Paschereit. Paschereit. V. 1950. 64. 35. (Cited on pages 6.J. 22(6):850–854. 76. Michel.O.) [36] Beatty. 77. 1977. Technical Memorandum NASA-TM-X-3288. D. C. 97(1):680–683. 2004. demon- strated with a jet flow”. Georgia.J. H. Magni. 1979.W. Cambridge. (Cited on page 121. V. 115.) [33] Batchelor.. UK. (Cited on page 50. NASA. 1978. E.. 14(8):720–728. U. Zuckerwar.: “Experiments on the trans- mission of sound through jets”. D. 68. 42(8):1543–1549. (Cited on pages 130 and 133.W.K...: “Visual study of the effect of grazing flow on the oscillatory flow in a resonator orifice”. C.) [39] Bechert.. G.. Journal of Aircraft. Rice.. E. 67. 114. demon- strated with a jet flow”. 79.E. 54.O. 1975. Hes- ter. and 203.

Fares.: Acoustics. 126:271–275.) [44] Bellucci. 4th ed. Kelly. (Cited on pages 27. Flohr. Woodbury. Follet. V. (Cited on page 227. sive de viribus et motibus fluidorum commen- tarii (Hydrodynamics.. Acoustical Society of America. Nowak.. H..F. Abhand- lungen der Berliner Akademie (1816-17).) [46] Bendali. B. Journal of Engineering for Gas Turbines and Power. (Cited on page 62. 1993. NY.: “Sulle funzioni bilineari”. 141. 1819. John Wiley and Sons.. bibliography 301 turbines”. SIAM Journal on Applied Mathematics. (Cited on page 110. 1873. C. Wood.S. 17- 19 June 2002. Piot. D.. 142.. Paschereit. A. and 71. 49–55. J.: “Methematical justifica- tion of the Rayleigh conductivity model for perforated plates in acous- tics”. M. Virginia Polytechnic Institute and State University. P. 48.) [43] Bellucci. 1738. Piersol.L. sic ut tem- poribus aequales altitudines emetiatur: & alterius cujusdam Problematis Propositio”. and 76. 73(1):438–459. 75.) [48] Beranek. in 8th AIAA/CEAS Aeroacoustics Conference. Paschereit. Giornale di Mathematiche ad Uso degli Studenti Delle Universita. J. (Cited on pages 28. Journal of Turbomachinery. D.I. 1690..) [49] Bernoulli. in 39th AIAA . V.O. S. L.G. A. AIAA-2002-2437. 2000... 2005.: Hydrodynamica.) [45] Beltrami... P. (Cited on page 39. Schuermans.) [50] Bernoulli. 2013. NJ.) [47] Bendat. Breckenridge. Tordeaux. 54. J.d. Ph. 68. C. (Cited on pages 6. 127(2):372–379. J. Colorado.) [52] Betts.: “Analysis problematis antehac porpositi. 45. 2004. (Cited on page 80. (Cited on page 101.) [51] Bessel. 65.: Random Data. J..: “An improved impedance model for perforates including the effect of bias flow”. 9:217–219. Hoboken.) [53] Betts. Acta Eruditorum. (Cited on pages 54 and 76. J..: “Analytische Auflösung der Kepler’schen Aufgabe”. 47. E. or notes on the forces and motion of fluids). and 143. 2010. E. 2002.: “Thermoacoustic modeling of a gas turbine combustor equipped with acoustic dampers”. 11:98–106. Flohr.W.. F.: Experiments and impedance modeling of liners including the effect of bias flow. de inventione linea descensus a corpore gravi percurrendae uniformiter. (Cited on page 145.F.: “Impedance of perforated screens with bias flow”. Strasbourg. thesis.O. Johann Rein- hold Dulsecker. pp.

26. 45.) [55] Bhayaraju.J. (Cited on page 147.. G. U.) [59] Bird. B.: “Effect of cool- ing liner on acoustic energy absorption and flame response”. 75. Murray... 65...: Transport Phenomena. J. Kwan.) [63] Blackman. and 269. (Cited on page 45. NASA. ARS Journal. NASA..302 bibliography Aerospace Sciences Meeting & Exhibit. J. D. Premo.) [62] Blackman. 51.: “Advanced nacelle acoustic lining concepts development”.A. J. J. J. 2002.S. 55.F. John Wiley & Sons. Kelly. 8-11 January 2001. 54. 12-14 June 2000. 1999. 2001. New York. in ASME Turbo Expo 2010. R..W. (Cited on pages 19. Follet. Contractor Report NASA-CR-1999-209002. Hersh. (Cited on pages 28. Kunze.B. and 79. 1957.. Yu.. and 54. P. in 6th AIAA/CEAS Aeroacoustics Conference. (Cited on page 44.W. Kelly. UK.. 76. AIAA- 2001-0990. S. 1960. Nevada.B.. Celano.: “Variable pressure wave absorption for combustion chambers”.W. 2nd ed. 14-18 June 2010. W.. Glasgow. GT2010-22616. 2002. (Cited on pages 32... J.N.. (Cited on page 97. Hawaii. 68. J.I. (Cited on page 141. 45.H. A. A.. J. 1960.B.W. 1996. Chiou.. and 202. 27. AIAA-2000-31118.. S. (Cited on pages 28. H. Follet. J.. E.) [57] Bielak. Hersh. 29:711– 714. Kashinath. Betts. Premo.. Philadelphia. Reno.. G. 1959. J. A. J. 66.) [56] Bielak.: “Effect of nonlinear losses on the design of absorbers for combustion instabilities”. Hochgreb.) [60] Björck. Wilson.) [54] Betts. Kosanchick.: “Acoustic impedance of a helmholtz resonator at very high amplitude”. 45.: “Evaluation of an impedance model for perforates including the effect of bias flow”.. Lightfoot. J.) [61] Blackman.. Society for Indus- trial and Applied Mathematics. Schmidt. Stewart. A. Lahaina. 30:1022–1028. Thomas. 2010. J.W. R. United Aircraft Corporation. (Cited on pages 45..: Numerical Methods for least squares problems.G.G. and 202. O.W. (Cited on page 54.. Walker. Contractor Report NASA-CR-2002-211672.E. Thomas. Journal of the Acoustical Society of America.. New York. Patent US2941356.: “Advanced turbofan duct liner concepts”. K.. Tukey. R. R.) . 2000.: The measurement of power spectra from the point of view of communications engineering. J. Dover. 35.) [58] Bies. A.

27-29 May 2013. 56(16):639–643. Labate.) [71] Boij. L.P. bibliography 303 [64] Blackstock. 60:565. Theory”. and 273.: “The acoustic reactance of small cir- cular orifices”. in 12th International Congress on Sound and Vibration.: “Low frequency sound absorption at sharp edges in flow ducts”.: “Reheat buzz: an acous- tically coupled combustion instability.) [73] Borth.H.: “Schwingungs.) [68] Bodén. A. 1986. H. 1988. (Cited on page 33. Journal of the Acoustical Society of America.. Portugal. in 19th AIAA/CEAS Aeroacoustics Conference. Ireland. (Cited on page 54. (Cited on pages 43 and 53. Lisbon. 1916. Journal of the Acoustical Society of America. 2012. Dowling.) [70] Bodén. S. 271. Germany.J. 2000. Åbom. Ingard. in 10th International Con- ference on Flow-Induced Vibration. S.) [66] Blokhintsev. NACA. 1992. R.) . (Cited on page 146. Journal of Fluid Me- chanics.I. and 202. 59. M. H. Universität Göttin- gen. (Cited on page 5. D.. 21(2):94–97.) [74] Brandes. 2005. Berlin. John Wiley & Sons..T.: Optimierung eines Meßvervahrens zur Bestimmung von akustis- chen Parametern im durchströmten Rohr. (Cited on pages 50. G. Diploma thesis. Technical Memorandum 1399.: Fundamental of physical acoustics. (Cited on page 155. H. 2-6 July 2012. M. AIAA-2013-2175. 193:445–473.: “Das Aehnlichkeitsgesetz bei Reibungsvorgängen”. (Cited on page 32. P. 79(2):541–549. W. H. 2013. originally published 1946 in rus- sian. (Cited on pages 139 and 147.. Zeitschrift des Vereins Deutscher Ingenieure. Zeitschrift des Vereines Deutscher Ingenieure.: “Acoustic properties of perforates under high level multi-tone excitation”. 1956. (Cited on pages 28.: “Acoustics of a nonhomogeneous moving medium”.) [65] Blasius..und Resonanzerscheinungen in den Rohrleitungen von Kolbengebläsen (Vibration and resonance phenomena in the ducts of piston blowers)”.) [67] Bloxsidge. 203.J. 1949.) [69] Bodén. ICSV12.. New York. (Cited on page 117. 45. 55.: “Acoustic properties of an in-duct orifice subjected to bias flow and high level acoustic excitation”. Dublin. 11-14 July 2005. 1912.) [72] Bolt. Zhou.: “Influence of errors on the two-microphone method for measuring acoustic properties in ducts”. D. 65. U. Langhorne. Part 2.

. Berlin..H. Doctoral thesis. 1987.. 1996.. NACA. 7th ed. Ferrante. Journal of Sound and Vibration. 37. B. and 15.: “Anti-screech inner body”. C. Patent US2934891. Zorumski.J.T. Musiol. 101. S. L.: “Boundary layer attenuation of higher order modes in waveguides”.G. M.) [76] Brater.) [80] Bräunling.N. and elliptical orifices at high pressure ratios”. Berlin.. K..) [77] Bronshtein. S. 1997. F. AIAA Journal.. (Cited on pages 27. Universität Göttingen. Germany. M. Röhle. Richter. United Air- craft Corporation. Springer.. Part 1: DLR tests and method- ology”. (Cited on page 219. 100. Bowden. W.: Handbook of hydraulics. and 38. P.M.) [85] Camporeale. 27-29 May 2013.: “Impedance eduction based on microphone measurements of liners under grazing flow conditions”. W. King. and 30.) [78] Brown... (Cited on page 44. J. and 142. M. Lahiri.W.: “Compara- tive study of impedance eduction methods. 9-13 June 2008... Fortunato. Mülig.A. 3rd ed. H.: “Prediction of ther- moacoustic instability in combustion chamber equipped with passive dampers”. 28. I. 2013.) [79] Bruneau. 1960. (Cited on pages 62. 2007. C. 5th ed. 119(1):15–27. Technical Memorandum NASA-TM-X-2361. Technical Note 1947. (Cited on page 38. L.G..F. (Cited on page 121. 2009.: “Investigation of flow coefficient of circu- lar. 115. 14. in ASME Turbo Expo 2008. E. 97.. Thiele. Bruneau. M.) [82] Busse-Gerstengarbe. E. Scofano.: Schallverstärkung in Strömungskanälen mit resonanzartiger Wandauskleidung.. W..E. (Cited on page 6. A.. F. NASA. Berlin. 1971.) .M.: Handbook of Mathematics. Berlin.. P.. square. Bake.) [84] Callaghan. 2012. H. Herzog... Kergomard. GT2008-51387.304 bibliography [75] Brandes. McGraw Hill. (Cited on page 32. M. Semendyayev. Springer. 50(4):867–879. Enghardt. AIAA-2013-2124. 1949. A. (Cited on pages 36. Boston. D.) [81] Budoff. Mastrovito. in 19th AIAA/CEAS Aeroacoustics Conference.: Flugzeugtriebwerke. Jones..) [83] Busse-Gerstengarbe. (Cited on pages 2. 2008.H. S.: “Flow resistance of perforated plates in tan- gential flow”. Germany.E. I. (Cited on pages 32 and 139... G. Enghardt.

(Cited on page 37. (Cited on pages 27.A. J. (Cited on page 146. H.: “On 36 forms of the acoustic wave equation in poten- tial flows and inhomogeneous media”. sciences. Washington. Bellevue. 1977. (Cited on page 101. lit- erature and general information. A.. bibliography 305 [86] Campos. 37(1):37–44.): Encyclopedia Britannica . Applied Mechanics Review.H. P.) [93] Christie. 44(1):98– 110. Maastricht.: “Enhanced sound absorption through negative bias flow”.) [88] Cerri.) [96] Chung. Journal of the Acoustical Society of Amer- ica.C. 10-12 May 1999.: “Measurement of the acoustic properties of a sound absorbing material at high temperatures”. Netherlands. 1976. 1999. Gaeta. 68(3):907–913.J. Ahuja. G.) [95] Chung.G.H. D.. 2001. I. H.: “Advances in ef- fusive cooling techniques of gas turbines”. 1911. 46(3):347–355. (Cited on pages 140 and 143. Kim. (Cited on page 147.: “Rejection of flow noise using a coherence function method”. (Cited on pages 26 and 266.A dictonary of arts.. Fedrizzi..) [91] Chisholm..W. Journal of Sound and Vibration.) [87] Cataldi. Encyclopedia Britannica Inc. W. New York.) [92] Cho. 11th ed.: “Experimental computational study of fluid dynamics and acoustics of perforate acoustic liner”.T.) . Journal of Sound and Vibration.B.Y. (Cited on page 13. R.: “Effects of geometry on the resonance frequency of Helmholtz resonators”. Blaser. in 5th AIAA/CEAS Aeroacoustics Conference. 1991... AIAA-2001-2264. D.A. J. 2001.C. JSME International Journal Series B Fluids and Thermal Engineering. Applied Thermal Engineering. B. L.. AIAA-1999-1879.K. 2007. Rhee. D.) [90] Chien. 1994.. 178(3):337–348. in 7th AIAA/CEAS Aeroacoustics Conference. 27:692–698. Noise Control Engineering Journal. K. 60:149– 171. Theory”. L. Battisti. E. Journal of the Acoustical Society of America.: “Transfer function method of measuring in- duct acoustic properties.) [94] Chu.: “Enhancement of film cooling perfor- mance using a shaped film cooling hole with compound angle injection”. 62(2):388–395. 28-30 May 2001. R. (Ed. 2007. and 54.: “Impedance tube measurements . (Cited on page 27.A comparative study of cur- rent practices”.M. 1980.R. J.) [89] Chanaud.. Giovannelli. (Cited on pages 17 and 234. (Cited on page 13. Yu. 34.. R.

38(1):149–155. 1834.G. 29.) .: “Mémoire sur la puissance motrice de la chaleur (Memoir on the motive power of heat)”. (Cited on pages 115. D. B. Journal of Sound and Vibration.O.) [106] Cummings. Journal of Engineering for Gas Turbines and Power. C. AIAA Journal. 2006.) [103] Cummings. 14:153–190. 48. Archiv der elektrischen Übertragung. F. A. 18. (Cited on page 98. and 202. 1926.B.E.. 1984. 22(6):786–792. A.: “Transient and multiple frequency sound transmission through perforated plates at high amplitude”.) [102] Culick. Journal de l’Ecole Royale Poly- techique. (Cited on pages 48. 73. (Cited on pages 32 and 67. pp. (Cited on page 26. 2005.) [99] Cosic. Paschereit. with superimposed mean flow”. NATO Research and Technology Organization. A. B. 79:942–951.: “Sound transmission at sudden area expansions in circu- lar ducts.: Theory of vibrating systems and sound. (Cited on page 53. Van Nostrand. (Cited on page 5. L. (english translation in: Scientific Memoirs.) [100] Crandall. 134:101503. 1986. 2012. T.: “Acoustic response of a Helmholtz resonator exposed to hot-gas penetration and high amplitude oscillations”.C.. 1837).) [101] Cremer. (Cited on pages 28. 1:347-376. and 73.) [98] Cohen. A. 117. J. 1975.) [104] Cummings.: “Unsteady motions in combustion chambers for propul- sion systems”. Banaszuk. 45. New York. in [305]. Acta Acustica united with Acustica..: “Über die akustische Grenzschicht vor starren Wänden”.: “The effects of grazing turbulent pip-flow on the impedance of an orifice”. A. and 62.: “Acoustic nonlinearities and power losses at orifices”. A. (Cited on pages 59. (Cited on page 203. 2:136–139. 1948. I. 1977. 61(4):233–242.: “High temperature effects on the radiation impedance of an unflanged duct exit”. Journal of Sound and Vibration. 50. 581–609.E. and 121. Reichel.306 bibliography [97] Clapeyron. 52(2):229–304. Journal of the Acoustical So- ciety of America. Chap.) [107] Cummings. Technical Report RTO-AG-AVT-039. 118. 1986.P.) [105] Cummings.M. Selected From the Transactions of Foreign Academies of Science and Learned Societies and From Foreign Journals. (Cited on page 6. 50. 61.: “Factors affecting the control of unstable com- bustors”.

(Cited on page 121.M. bibliography 307 [108] Cummings. A.L.D. 2008. Journal of Sound and Vibration. Journal of Sound and Vi- bration. Bhattacharya. AIAA-2008-3007. Histoire de l’Académie des sciences et belles-lettres. 72(4):539–542. (Cited on pages 121 and 122. NACA.) [117] Dean. Mendez. and 221. 45.L.L.) [114] Davies. (Cited on pages 100 and 109.. A. J. and 260. 1989. L. D. 1993.. 3:214–219. Vancouver.O. 2003. G. A.A.. 1954.: “Sound propagation in narrow tubes of arbitrary cross- section”. 1983. G.O. Björck... 53. (Cited on pages 53 and 68.D. 124(1):91–115. B.) [111] d’Alembert: “Recherches sur la courbe que forme une corde tenduë mise en vibration (Investigation of the curve formed by a vibrating string)”.. NASA.. Dover. (Cited on pages 122 and 123. 162(1):27–42. 5-7 May 2008. (Cited on pages 44. M.: “Slightly dispersive waves in pipes”. 1980.. W.L. (Cited on page 55. (Cited on pages 23. 1975. 132(1):169. Mineola. D.) [115] Davies.A. (Cited on page 123.) [118] Dean. F. P.: “Duct wall impedance control as an advanced concept for acoustic suppression”. Canada. 91(4):503–518. G.) . Tester. 259.: “Practical flow duct acoustics”. Contractor Report NASA-CR-134998. Journal of Sound and Vibration. Report 1192. Nicoud. S.) [112] Dasse.L.: Numerical Methods. NY. in 14th AIAA/CEAS Aeroacoustics Confer- ence.) [110] Dahlquist. (Cited on pages 145 and 146. NASA.: “Coupling of Helmholtz resonators to improve acoustic lin- ers for turbofan engines at low frequency”. P.W.J. 1976.: “Measurement of plane wave acoustic fields in flow ducts”. P.) [109] Cummings.) [113] Davies.: “High amplitude acoustic transmission through duct terminations: Theory”. Eversman.A.) [116] Davis. Contractor Report NASA-CR- 134912.. Stevens. 1988. P. Coelho. (Cited on page 74.: “Theoretical and ex- perimental investigation of mufflers with comments on engine-exhaust muffler design”. Moore.O. Stokes. J..B.: “Large-eddy simulation of the acoustic response of a perforated plate”. 1747. 67. Journal of Sound and Vibration. Journal of Sound and Vibration.

. S. 45.E.) [126] Dokumaci. Journal of the Acous- tical Society of America.P.) [123] Dickey.P. 210(3):375–389. E. Journal of the Acoustical Society of America. 1965..: “Discharge coefficients for combustor-liner air-entry holes. (Cited on page 48.: “Sound absorption by a screen with a regular array of slits”. J. M. Selamet. Stow.S.. NACA. Novak. 1977. Journal of Sound and Vibration. and 202. 1 . 1992. Proceedings of the Institution of Mechanical En- gineers.T. (Cited on page 32.) [129] Dowling.) [122] Dickey. 182(5):799–808.) [120] Delany.R. E. (Cited on page 113. step louvers.E. (Cited on page 36. (Cited on pages 130. 156(3):387–405.Circular holes with parallel flow”. B.) [127] Dokumaci. 110(5):2360–2370.: “Sound propagation in the atmosphere: A historical re- view”. 2003. Technical Note NACA-TN-3663. E. 2000. Acoustica. 38:201–223. Hughes.. Journal of Propulsion and Power. 1956. 2 . Technical Note NACA-TN-3924.T. and 132. Graves. 108(3):1068–1081. Journal of Sound and Vibration.) [121] Delany. Ciray. 1970. (Cited on page 38.) [128] Dokumaci. A. 2001. Bazley. (Cited on page 266.) [130] Dowling. A.E. 19(5):751–764.: “An investigation of steady compressible flow through thick orifices”. 1997. Journal of Sound and Vibration. Journal of Sound and Vibration.F.308 bibliography [119] Deckker.N. 180:312–323.C. M. NACA. E.) . Applied Acoustics. 1998.) [125] Dittrich.. (Cited on pages 113 and 123. Chang. 47. (Cited on pages 17.: “Discharge coefficients for combustor-liner air-entry holes. Y. 1995.. A.L. and scoops”. Selamet. (Cited on page 38.: “An experimental study of the impedance of perforated plates with grazing flow”.: “The effect of high-amplitude sound on the attenuation of perforated tube silencers”. I. N. A. N. 208(4):653–655. (Cited on page 123. 3(2):105–116.: “On transmission of sound in circular and rectangular nar- row pipes with superimposed mean flow”.: “Acoustic analysis of gas turbine combustors”.. 131.Flush rectangular holes.M. R.S. C.J. 1958.) [124] Dittrich.S..: “Sound transmission in narrow pipes with superimposed uniform mean flow and acoustics modelling of automobile catalytic con- verters”. and 54.: “A note on transmission of sound in a wide pipe with mean flow and viscothermal attenuation”. R. (Cited on pages 28. M.: “Acoustical properties of fibrous absorbent materials”..

: “Sur les vvibration des gaz dans des tuyaux cylin- driques. Boot. A.. Dowling.. Italy. Hofmans. 14:49– 110. Italy. (Cited on pages 6.P. bibliography 309 [131] Dowling. 17-19 June 2002. (Cited on page 101. Dowling. in 4th AIAA/CEAS Aeroacoustics Conference.P. 54. 2005. (Cited on page 54. 369–414.C. J. pp. Stow.P. M.D. 2-4 June 1998. Dowling. AIAA Journal.. Boij. Dowling. (Cited on page 54.D. Shoeybi.P. Dowling. A. A. 47.. 55.) [139] Efraimsson.) [138] Durrieu. Journal of Sound and Vibration. (Cited on pages 18. (Cited on page 55.: “Acoustic analyis of gas-turbine combustors”. 2004.: “Absorption of sound near abrupt area expansions”. 2007.: “Simulation of wave scat- tering at an orifice by using a navier-stokes solver”.J. Hirschberg.D.. coniques. France.) [140] Eldredge. (Cited on page 54.M. 1998.D. M. 510:303–331.... AIAA 2007-3683..) [134] Dupere. Toulouse. M. 245.P. Bodony. in 8th AIAA/CEAS Aeroacoustics Confer- ence. A. and 246. and 71.”.. 2002.: “The absorption of sound near abrupt area axpansions”. A. and 139.J.J.J.) . 239(4):709– 730.A.. Pieper. S.P. in 13th AIAA/CEAS Aeroacoustics Conference 21-23 May 2007. I. 2000. Journal of the Acoustical Society of America.M.) [141] Eldredge.) [137] Dupere. 2001.D. 21-23 May 2007. (Cited on pages 27 and 54. R. A...J. Auregan. 13. (Cited on pages 6 and 54. Rome. A.: “Quasisteady aero-acoustic response of orifices”. AIAA-1998-2303. G. (Cited on page 54. Journal de mathématiques pures et appliquées. I.R. J. Journal of Fluid Mechanics. AIAA-2002-2590. I. in 13th AIAA/CEAS Aeroacoustics Conference.D.) [135] Dupere.: “The absorption of sound by helmholtzs resonators with and without flow”. Chap. Beckenridge. Colorado..) [136] Dupere.J. D. G. Rome. 2001.: “The use of helmholtz resonators in a prac- tical combustor”. 2005. Peters. 110(4):1859–1872.. M.. S. in [305]. (Cited on pages 19.. 48.) [132] Duhamel.: “Numerical investigation of the acoustic behavior of a multi-perforated liner”. etc. 7. T. AIAA-2007-3402. Åbom.C.: “On the interaction of higher duct modes with a perfo- rated liner system with bias flow”. G. 1849. 2007. Journal of Engineering for Gas Turbines and Power. Ajello. I.) [133] Dupere. 38(2):193–202. P. I.: “The absorption of sound near abrupt ax- isymmetric area expansions”. 127:268– 275. Y.

UK.php. 156. Mémoires de l’académie des sciences de Berlin. 1766.D.P.: “Supplement aux recherches sur la propagation du son (Supple- ment to the research on the propagation of sound)”. J. Issue 6.eu/certification/type-certificates/engines. 19. 72.: “On semi-empirical liner impedance modeling with grazing flow”. 1757. Journal of Fluid Mechanics. 88. 227. L. Manchester. 47.) [143] Elnady. Uni- versität Göttingen.: “The absorption of axial acoustic waves by a perforated liner with bias flow”. pub- lished 11 December 2013. Bodén. 49. E.: “Impedance of SDOF perforated liners at high temperatures”.) [151] European Aviation Safety Agency: “E. 2003. 66. Dowling.) [144] Elnady. Mémoires de l’académie des sciences de Berlin. 25. 2004. 54. Engine Type Certificate Data Sheet. (Cited on page 32. 73. (Cited on pages 26 and 266. H. Mémoires de l’académie des sciences de Berlin. 10- 12 May 2004. 2010.. T. thesis. 15:185–209. 15:210–240. AIAA-2004-2842.J.012 Rolls-Royce plc. Bodén.: “Principes généraux du mouvement des fluides (general prin- ciples of the motion of fluids)”. 65. Sweden. 48. South Carolina.: “De la propagation du son (On the propagation of sound)”. Diploma thesis. 211. T. in 9th AIAA/CEAS Aeroacoustics Conference. and 75.) [145] Elnady. Stockholm. Kungliga Tekniska Högskolan. A. T.) [150] Euler. 11:274–315.) [146] Enghardt. (Cited on page 100. Doctoral thesis. (Cited on page 124. (Cited on pages 97 and 98. 12-14 May 2003. L. in 10th AIAA/CEAS Aeroacoustics Conference. (Cited on pages 16. and 246. Hilton Head.310 bibliography [142] Eldredge. 245.) [147] English. 30. 28. (Cited on pages 45.. T. 1766. L. 2003. (Cited on pages 100 and 109. 1992. (Cited on page 3.. 85. University of Southampton. 71. RB211 Trent 900 series engines”. 35. europa.) [148] Euler. 2004. (Cited on pages 139 and 154.) [149] Euler.: Bestimmung akustischer Streufaktoren im durchströmten Rohr - Optimierung von Mikrofonanordung und Messprozedur. Ph.: Modelling and characterization of perforates in lined ducts and muf- flers.) . AIAA-2003-3304. L. Kontio.: A measurement based study of the acoustics of pipe systems with flow.D. accessed online 27 February 2014: http://easa. 485:307–335.

: “Teoreticheskoe issledovanie provodimosti kruglogo otverstiya v peregorodke. (Cited on page 61. 2001. E. Journal of the Acoustical Society of America. Bass. 29. W. (Cited on pages 28. N.A. Pergamon Press.I. 49:1372–1380. W. Brown. Journal of Sound and Vibration. 48(2):425– 428.) [158] Feder. 45. 1970. 8-11 January 2001. 1941. A course of lectures on the theory of sound.J.C.) [159] Flightglobal Insight: “Commercial engines 2013 year-end update”.: “Improvements to acoustic liner broad- band absorption using bias flow”. 53.) [157] Farmer.F. accessed online 27 February 2014: http://www. Kelly. Rutherford. 44.D.) [156] Fahy. Elsevier Academic Press. Betts.) . 29. 1963.com/airspace/media/reports_pdf/ commercial-engines-2013-year-end-update-108241.B. Doklady Akademii Nauk SSSR. and 218.) [161] Follet. 2003. General Electric Company.: “The effect of mach number on the tuning of an acoustic lining in a flow duct”. J.: “Preferential multihole combustor liner”. 51(5):1565–1575. J.flightglobal. Nevada.E. 1972. and 211. (Cited on page 13. Special Report. J. Contractor Report NASA-CR-1373. 54.: “Effect of boundary layer on the transmission and atten- uation of sound in an acoustically treated circular duct”. D. Sutherland..: Foundations of engineering acoustics.: “The boundary condition at an impedance wall in a non- uniform duct with potential mean flow”. M.) [154] Eversman. F. 2001. 2001. H. 246(1):63–69. alternatively: D. 31(9):875–882. G. (Cited on page 32. Rschevkin.aspx. London. L. Dean.) [160] Fok. Patent US6655149.W.. (Cited on page 32.) [153] Eversman.: “Atmospheric absorption of sound: Theoretical predictions”. Reno.. in 39th Aerospace Sciences Meeting and Exhibit. NASA.. bibliography 311 [152] Evans. W. 1971. Lon- don. V. postavlennoi poperek truby (Theoretical study of the con- ductance of a circular hole in a partition across a tube)”. Journal of the Acoustical Society of America. (Cited on pages 130 and 131. 1969. AIAA-2001-0823.. published 11 December 2013. L.: “Analytical and experimental studies for predict- ing noise attenuation in acoustically treated ducts for turbofan engines”. Journal of the Acoustical Society of America. L. (Cited on page 64. (Cited on page 2. [155] Eversman. (Cited on pages 28..E..

in InterNoise 84.: “Experimental investigation of the dissipation of acoustic energy in perforated walls with bias flow”. Russell. Kelly. T.: “Suppression of combustion oscillations with mechanical damping devices”.. (Cited on pages 44 and 53.B... and 54.J.. and 141. Reno. 1969.. and 220.) [167] Fujimori.: “Experimental impedance of single liner elements with bias flow”.) [165] Fourier.J.W.C.) [169] Gaeta. (Cited on page 54. pp.K. Acta Acustica united with Acustica. 2000. 2003. Cranfield University.D. Pratt & Whitney Aircraft for NASA. C. in International Colloquium on Combustion and Noise Control. K.com/press/release.forecastinternational.A. Cranfield.) [168] Fukumoto. 12-14 June 2000.D.C. accessed online 26 February 2014: http: //www. Schnell. AIAA-2001-0662.: “The measurement of the acoustic radiation impedance of the open end of a thick walled tube with hot flow”.R. Betts. Paris. G. Ahuja. R.L.. J. J. Press Release. Takayama. H. 2001.: “Effect of orifice shape on acoustic impedance”. P. 1972.: “An automated measurement system of complex sound pressure reflection coefficients”. P. Baldwin. England. 45. (Cited on pages 17 and 235. (Cited on pages 110. J. (Cited on pages 18 and 54.: Théorie analytique de la chaleur (The analytic theory of heat). Russell. Thomas. M. (Cited on page 147. Physics of Fluids A. R. U. in 39th AIAA Aerospace Sciences Meeting and Exhibit. 1984.J.. A.H. 1009–1014. (Cited on pages 27.) [171] Garrison. 8-11 January 2001. AIAA-2000-1950. 1822. 1977. 1991...I. (Cited on page 26.D. In- terim Report PWA FR-3299..: “Vorticity production at the edge of a slit by sound waves in the presence of a low-mach-number bias flow”. G. Schnell. Pratt & Whitney Aircraft for NASA.. Michel. Nevada. Minura.) [163] Forecast International: “Forecast international expects power genera- tion to fuel increasing demand for industrial gas turbines”. A. 12-15 August 2003.cfm?article=272. F. published 12 November 2013. 3(12):3080–3082. N. in 6th AIAA/CEAS Aeroacoustics Conference. Hawaii.) [170] Garrison. 53.. 115. (Cited on pages 44. S. J.. Sato. Koenig.F. Final Re- port PWA FR-4993.. Lahaina.312 bibliography [162] Follet.) . C.. S.: “Suppres- sion of combustion oscillations with mechanical damping devices”. Y. 38(2):124–130.) [164] Forster. 67. (Cited on page 2. J.) [166] Fricker. Roberts. Didot.

J. Wilson.C. R.) [176] Goldstein. 31:155–160. Yale University Press.) [180] Gordon. 1974. Hamburg. F.: Vector analysis. 17(5):595–607.E..F. 120:549–556. H. (Cited on page 13. K.. (Cited on page 99. International Journal of Heat and Mass Transfer. oxygen. John Hopkins University Press.) [173] Gibbs. 1809. C. H. M.. (Cited on pages 26 and 270.) [181] Grace. (Cited on pages 32 and 67.) [183] Gritsch. Journal of the Acoustical Society of America. C. Eckert. 1996.B. Wittig..G. van Loan. 60(6):1397–1404. Besser.S. Journal of the Acoustical Society of America. 132:1–6. 1929. 12:335–351.) [174] Giese. New York...) . Journal of the Acoustical Society of America.F.: Aeroacoustics.: “Effects of hole geometry and density on three-dimensional film cooling”. Smith. Journal of Turboma- chinery. Howe..R. McGraw-Hill.. (Cited on page 132. (Cited on page 32.E. E.: “Effect of shear on duct wall impedance”.) [177] Goldstein.P. Journal of Engineering for Gas Turbine and Power. (Cited on pages 44. C. (Cited on page 147.H.: Theoria motus corporum coelestium in sectionibus conicis solem ambientum (Theory of the motion of the heavenly bodies moving about the sun in conic sections). (Cited on page 145... Journal of Fluids and Structures. Perthes and I. M.: “Acoustic losses of a resonator with steady gas flow”. R. 1998. Schulz.) [179] Golub. 1976. and 53.. A. Rice. New Haven. 1959.L.: “On the performance of porous sound absorbent material in high temperature applications”. 1973.) [182] Greenspan. E. S. bibliography 313 [172] Gauss.W. G.) [178] Goldstein. A. C. 1976.: “Adiabatic wall effectiveness mea- surements of film-cooling holes with expanded exits”. E.L. Jour- nal of Sound and Vibration. Eigenbrod. M. F. S.: Matrix Computations. 3rd ed.: “The influence of shape on the rayleigh conductivity of a wall aperture in the presence of grazing flow”. Panton.. 46.W. M. (Cited on page 97. (Cited on pages 17. Baltimore.J. 37(2):257–267. 2010. 1998. and 234... M. Ries. P. Horan.: “Rotational relaxation in nitrogen.) [175] Goldman. 30(1):79–84. and air”. (Cited on page 13. 67. 2nd ed. 1965.M.: “Measurement of the acoustic impedance of an orifice under a turbulent boundary layer”.

P. 1972. F.. E. C. McCormick. Copeland. 1975. 1886. D. D. 40(1):119–137. (Eds.: “Electromagnetic induction and its propagation”. 122:269–274. in 2nd Ini- tiative on Advanced Combustion (INCA) Workshop.. The Elec- trician. F.. 61-71. 54:1569. D. G. J.. 2000. F.. Enghardt.M. W.: “High dynamic range measurements of acoustic particle velocity and flow velocity for the application at liners”. AIAA-2013-2053.. Mendez. Journal of Turbomachinery.: “Multi-scale measurement of acoustic particle velocity and flow velocity for liner investigations”. (Cited on page 6. 1894.C. Schulz.. Czarske. J. 2009. Experiments in Fluids. Section 29.. D. Nicoud... and 67.314 bibliography [184] Gritsch.): Liquid propellant rocket combustion in- stability.. Enghardt.. O.: “Com- bustion system damping augmentation with helmholtz resonators”.) [187] Gullaud. 2001. (Cited on page 51. S.: “Method for correlating discharge co- efficients of film-cooling holes”.. F.: “Damping effect of perforated plates on the acoustics of annular combustors”.. in 19th AIAA/CEAS Aeroacoustics Conference. (Cited on page 59. (Cited on page 55. (Cited on page 51. M..: “Calculation of perforated plate liner parameters from spec- ified acoustic resistance and reactance”. E. Florida. reprinted in: Electrical Papers Vol.) [185] Gritsch. Mendez. NASA-SP-194. Pionsot. 2. A. 27-29 May 2013. T. Reardon.H.. A.: “Effect of multiperforated plates on the acoustic modes in combustors”. Germany.. (Cited on page 38.. 36(6):976–980.) [186] Guess.: “Effect of crossflows on the discharge coefficients of film cooling holes with varying angles of inclination and ori- entation”. Wittig. S. S.) [192] Haufe.T. D. A. S.. 2013. Fischer. Schulz. (Cited on page 5.. A. p. 2008. A.. Sensiau. C. L. NASA. Wolf. 2013.) [191] Haufe. 17:212. AIAA-2009-3260... Bake.. 123:781–787. L.. Washington. (Cited on page 55. Berlin. Schulz. D. Rouen. in 15th AIAA/CEAS Aeroacoustics Conference. Macmillan & Co. 11-13 May 2009. F.C.S. Miami. 1998. Nicoud. Sensiau.. 23-24 October 2008. A. Proscia. M. Bake.. Czarske.) [189] Gysling. Schulz. Fischer.) [193] Heaviside.W. AIAA Journal.) [188] Gullaud. (Cited on page 38. 65. Journal of Sound and Vibration. France.L..) [190] Harrje. Wittig. Jour- nal of Engineering for Gas Turbines and Power.) . (Cited on pages 64. A.

Sweden... T. Pardowitz. (Cited on page 55. Aigner.. in 16th AIAA/CEAS Aeroacoustics Conference. and 203.. Röhle. Y. Hirschberg. Bake. (Cited on pages 18 and 54.) [202] Heuwinkel. AIAA- 2009-3110. Boot. H. Fischer.. 14-18 June 2010.) [199] Heuwinkel. C. C. and 211. in ASME Turbo Expo 2010. Rice. Piot. S. (Cited on page 131. Durrieu.. B. 2001. 55.. C. Bake... L. M. Röhle..J. Italy.. C.: Experimentelle Untersuchung der akustischen Eigenschaften perforierter Liner..) [200] Heuwinkel. 1980. Germany. 21-23 May 2007.: “Aeroacoustics response of a slit-shaped diaphragm in a pipe at low Helmholtz number.: “Characterization of a perforated liner by acoustic and opti- cal measurements”. Bake. 9-13 June 2008... Journal of Sound and Vibration.) [195] von Helmholtz. 54. 1: Quasi-steady results”. S. Röhle. 2008. Litovitz. B. in ASME Turbo Expo 2008.C. FL. E. B.A. (Cited on pages 34 and 52. F. Busse. Enghardt. R. Academic Press. L. I. C.. 30. 1959. (Cited on pages 51. Sadig. C. AIAA- 2007-3525. F. Mühlbauer. 56(17):257–260. I. in 15th AIAA/CEAS Aeroacoustics Conference. A. A. 2010. AIAA-2010-3765. New York.. 1863. Röhle. I. 2010.) [201] Heuwinkel..J. NASA.: “Experimental evaluation of a spinning-mode acoustic-treatment design concept for aircraft inlets”. 11-13 May 2009. GT2010- 22329. 2006. Auregan. 244(1):35–56.) [196] Herzfeld. Noll. L. GT2008-50585. E. Enghardt. Enghardt. Enghardt. L. 2009. L. Miami. 139.M. Technical Paper NASA-TP-1613. 2007.. 55.: “Es- tablishment of a high quality database for the modelling of perforated liners”.J.: “Comparison of experimental and numerical re- sults concerning the damping of perforated liners with bias flow”. L.F.P. Micheli. G. (Cited on pages 16. Technische Universität Berlin. bibliography 315 [194] Heidelberg. and 281. (Cited on pages 62 and 114.... I.) [197] Heuwinkel.J. P. Stockholm. Gerendás.) .: “Über den Einfluss der Reibung in der Luft auf die Schallbewegung”..: Absorption and dispersion of ultrasonic waves. 7-9 June.J.) [198] Heuwinkel. UK. Diploma thesis. Glasgow. (Cited on pages 28. in 13th AIAA/CEAS Aeroacoustics Conference. (Cited on page 67.... Enghardt.: “Experimental investigation of the acoustic damping of perforated liners with bias flow”.: “On the excitation of a zero mass flow liner for acoustic damping”.) [203] Hofmans. Rome. M. Verhandlungen des Naturhistorisch-Medizinischen Vereins zu Heidelberg. F. (Cited on pages 54 and 161. Berlin... F..J. 202. K.

) [209] Howe.: Ermittlung der aerodynamische Lärmquellen und Berechnung des abgestrahlten Schall mittels der im Nahfeld gemessenen Druckschwankungen am Beispiel eines Triebwerksmodells.: “The influence of grazing flow on the acoustic impedance of a cylindrical wall cavity”. M. Bodén. 127. M. 1997. Auregan. 1979. 1979. Journal of Sound and Vibration.) [210] Howe.S. A. Proceedings of the Royal Society of London. Jour- nal of the Acoustical Society of America. G. M. M.S. Düsseldorf. (Cited on pages 47 and 53. 2: Unsteady results”. M.) [212] Howe. Cambridge. 1975. 98(3):1723–1730.) [214] Howe.. Journal of Sound and Vibration. 85.) [213] Howe. Journal of Sound and Vibration. 74. 73.S.: “On the absorption of sound by turbulence and other hydro- dynamic flows”. A. M. Cambridge University Press.: “Influence of cross-sectional shape on the conductivity of a wall aperture in mean flow”. Journal of Fluid Mechanics. Ranucci. 2011. (Cited on pages 47 and 100.S. 2001. and 271. 48. VDI. (Cited on pages 35. Doctoral thesis. and 129. 330:6336– 6354. 69.S.S.: “The damping of sound by wall turbulent shear layers”. (Cited on pages 32 and 67.: “On the theory of unsteady high reynolds number flow through a circular aperture”. 71:625–673. (Cited on pages 35. 1995. 91(2):209–229.. M. 244(1):57–77. and 129. 100. 32:187–209. 127. G. (Cited on page 101.) [211] Howe.. 47.. published in: Fortschritt-Berichte VDI. 68. Hirschberg.) [207] Howe.: “Contributions to the theory of aerodynamic sound. Series 7. 70. 118. UK. Åbom. 1995.: “Accurate experimental two-port analysis of flow generated sound”.: “Attenuation of sound in a low mach number nozzle flow”..: Acoustics of fluid-structure interactions. H. F. 58. Y. 1998. 47. Journal of Sound and Vibration.S. 207:601–616.) [208] Howe. 1995.316 bibliography [204] Hofmans. 1984. (Cited on page 126. with application to excess jet noise and the theory of the flute”.C. M. 126. M.) . (Cited on page 124. No. Journal of Applied Mathematics. 1979.S. 366:205–223. 68.: “Aeroacoustics response of a slit-shaped diaphragm in a pipe at low Helmholtz number.. Techische Universität Berlin.) [206] Holste. (Cited on pages 126.J. Ajello. Journal of Fluid Mechanics. 53. 272.) [205] Holmberg. (Cited on page 18. 69. 67(4):533–544. M. (Cited on pages 17 and 18.

252. H.: “On the theory and design of acoustic resonators”. C.: “Schallabsorption in durchströmten Rohrmündungen”. (Cited on page 53. Massachusetts Institute of Technology.J.S. 493–496.: Handbook of hydraulic resistance.) [222] Igarashi. 88. 71. 35. Proceedings of the Royal Society of London. (Cited on page 294. I. 202. 452(1953):2303–2317. (Cited on page 146. Dowling. 28. (Cited on page 32. 54.: “Attenuation and regeneration of sound in ducts and jet dif- fusers”.) [225] Ingard.D.. 77. University of Tokyo. pp.) [219] Hughes.) [217] Hu. 64. 53. I. 63. Part 1: The- ory and experiments of acoustic low-pass filters”. Diploma thesis. München. (Cited on pages 17. bibliography 317 [215] Howe.) [224] Imelmann. Berlin. 218:299–335. (Cited on page 219. Mumbai. 211.: Einfluß der Strömung auf den Schalldurchgang durch Rohrmün- dungen (Influence of flow on the sound transmission through orifices). 1980. Series A. 56(4):258–269. Sipcic..) . (Cited on pages 122 and 284. 1996. Interna- tional Civil Aviation Organization.: “Fundamentals of acoustic silencers. and 217. Report 339.) [220] ICAO: “Manual of the ICAO standard atmosphere”. in Fortschritte der Akustik. U.) [221] Idelchick.: “The absorption of sound by perforated lin- ings”. C. Journal of the Acoustical Society of America. (Cited on pages 25. 31(9):1202–1212. S. 1958.: “Untersuchungen zum akustischen Meßleitungsverfahren mit festen Meßorten”. Jaico Publishing House. 51.P. Journal of Fluid Mechanics.: “The influence of tangential mean flow on the rayleigh conductivity of an aperture”. M. 1984. 1959. and 233. India. Scott. 47. M.: Développement de panneaux hybrides passifs/actifs pour l’acoustique (De- velopment of hybrid passive/active acoustic panels).: Flow generated noise of acoustical duct liners. (Cited on pages 44. 1990. 2010.. Toyama. 1953. (Cited on page 80. 74.. Doc 7488/3. 1993..) [218] Hudde.I. (Cited on page 38. Journal of the Acoustical Society of America.E. 3rd ed. (Cited on page 53. 25(6):1037–1062. DAGA ’80. 65. 59. 45. Technische Universiät Berlin. Ecole Cen- trale de Lyon. Master’s thesis.) [226] Ingard. VDE-Verlag. J. M. 2011.R.. Letens. U. J. 1978. U.) [216] Hrubes.) [223] Imelmann. A. 227. Acustica. 1977. Doctoral thesis. Y. and 271.

Journal of the Acoustical Society of America. Geneva. Standard. Journal of the Acoustical Society of America. International Organization for Standardization.: “Acoustic circulation effects and the nonlinear impedance of orifices”. 2003.Attenuation of sound during propagation outdoors .) [229] Ingard.. 1974. Standard. (Cited on pages 28. S. 42:6–17. (Cited on page 126. In- ternational Organization for Standardization. Geneva. V. and 202. Switzerland. Press Release 67. (Cited on pages 19 and 36. 1967.) [235] ISO 80000-8:2007: “Quantities and units . In- ternational Organization for Standardization. (Cited on page 113.org/ pressroom/pr/pages/2013-12-10-01. Labate. International Organization for Standardiza- tion. Singhal.Part 2: Transfer- function method”. 45. Re- port.) [236] ISO 9613-1:1993: “Acoustics . Geneva. U.) [228] Ingard. 67.: “Sound attenuation in turbulent pipe flow”. 65. International Organization for Standardiza- tion.aspx. Standard.) [233] ISO 5136:2003: “Acoustics . accessed online27 February 2014: http://www.) . Switzerland.. (Cited on page 1. 1993. 22(2):211– 218. 65. Switzerland.Determination of sound absorption coefficient and impedance in impedance tubes . 44.K. H. and 202.) [232] ISO 10534-2:1998: “Acoustics .pdf. Switzerland.In-duct method”. 2007.. and 134. (Cited on pages 130. Journal of the Acoustical Society of America.: “Acoustic nonlinearity of an orifice”. Ising. (Cited on page 165. (Cited on page 2. 45. U. 133.Determination of sound power radiated into a duct by fans and other air-moving devices . published 2013.Part 1”. 55(3):535–538.Part 8: Acoustics”. org/publications/freepublications/publication/KeyWorld2013. Geneva.318 bibliography [227] Ingard. accessed online 25 February 2014: http://www. (Cited on pages 28. Switzerland.iata. (Cited on page 146. 1950.) [234] ISO 5167-2:2003: “Measurement of fluid flow by means of pressure differ- ential devices inserted in circular cross-section conduits running full. 53.) [230] International Air Transport Association: “Airlines expect 31 % rise in passenger demand by 2017”. 2003. Report.iea. Part 2: Orifice plates”. published 10 De- cember 2013. Standard.) [231] International Energy Agency: “Key world energy statistics 2013”. U. Geneva. 1998.

Physics of Fluids. Richards. 93. (Cited on pages 25. (Cited on page 147. Germany. 34. 1999.: “Decoupling approach to modeling perforated tube muffler components”. 1989.) . S. bibliography 319 [237] Jang. (Cited on pages 55 and 249. 29. 11-15 June 2012. Ih. in Proceedings of ASME Turbo Expo 2012.M. D. W. X.G. 227.) [239] Jayaraman.: “On the multiple microphone method for measuring in-duct acoustic properties in the presence of mean flow”. Q. 69(2):390–396. P. 88.: “Comparative study of impedance eduction methods.) [242] Jing. (Cited on pages 32 and 67. 38(9):1573–1578. Denmark. C.: “Effect of grazing flow on the acoustic impedance of an orifice”.. (Cited on pages 27 and 28. 271. W.. Rubini.. Journal of the Acoustical Society of America. M.: “Absorption of normal-incedence acoustic waves by double perforated lin- ers of industrial gas turbine combustors”. 68.... and 202. 71. 39(8):1478–1484. USA. 54. X.) [244] Jing.: “Evaluation of a multi-point method for determining acous- tic impedance”. 1998. X. Orlando.) [240] Jayatunga.. (Cited on page 146. Busse-Gerstengarbe. X. Krebs. X.. 77.. 103(3):1520–1526. 1981.) [241] Jing. 48. in ASME Turbo Expo 1997. G. Journal of the Acoustical Society of America. Journal of the Acoustic Society of America. 212. Wu. 27-29 May 2013. K. 45. Meng. 63:1455–1460. Sun. and 280. GT1997-266.M. 2001.: “Experimental investigation of perforated liners with bias flow”. V. 1997.E. B. AIAA-2013-2125. Sun. S. M. Yip. Yam.: “Measurement of acoustic reflection from an obstruction in a pipe with flow”. You. (Cited on page 82. (Cited on page 5.) [247] Jones. Qin. Florida. 47.: “Effects of ambient conditions and fuel composition on combustion stability”. Journal of the Acoustical Society of America.) [243] Jing. AIAA Journal. Part 2: NASA tests and methodol- ogy”.. Copenhagen.H.J.. 211. 2002. Sun. 14(1):268–276. 72. 2-5 June 1997.G. (Cited on page 147. 1978.. (Cited on pages 48. 2000.) [238] Janus..: “Sound-excited flow and acoustic nonlinearity at an ori- fice”.) [245] Johnston.) [246] Jones. in 19th AIAA/CEAS Aeroacoustics Conference. X. X. 2013. GT2012-68842. M. 72. M.. 2012. 251. Berlin. J. Schmidt. Mechanical Systems and Signal Processing. 54. K. AIAA Journal.. 3(1):15–35. Howerton. Sun.. 106(5):2436–2441. Sanderson.: “Effect of plate thickness on impedance of perforated plates with bias flow”.P. J.C.. J. K. X. and 231. (Cited on pages 28..

Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen (Mathematisch- Physikalische Klasse). Tech- nische Universität Berlin. (Cited on page 127. Comptes Rendus de l’Académie des Sciences.) [251] Jörg. in ASME Turbo Expo 2012. M. GT2013-94843.: “Experimental investigation of the acoustic reflection coefficient of a modeled gas turbine impingement cooling section”. 2011. Sattelmayer. (Cited on page 55. (Cited on page 108. Portland.G.. pp. J. 5-8 June 2011.H. Journal of the Acoustical Society of America. USA. Paris..Texas.S. (Cited on page 26.: Discharge coefficient and jet deflection studies for combustor liner air entry holes.) [252] Jörg. T. C.: “Numerical investigation of the plane-wave reflection coefficient of an exhaust pipe at elevated tempera- tures using linearized Navier-Stokes equations”. J. Cranfield University. 2012. (Cited on page 145. 122:153–160.320 bibliography [248] Jones. M. W. 1964. C. 75(1):58– 62.) [254] Kaddah. 3-7 June 2013. 58–76. 1930.: “Sur la réduction des formes bilinéaires”. Sattelmayer. C. D.) [249] Jordan.) [257] Keefe. Oregon. 1874.R. AIAA-2011-2865. T. GT2013..) . (Cited on page 13.) [255] Kameier.: “Effects of orientation angles on film cooling over a flat plate: Boundary layer temperature distributions and adiabatic film cooling effectiveness”. 19:35–54.. K.: “Mémoire sur les formes bilinéaires”. in 17th AIAA/CEAS Aeroacoustics Conference. 1984. San Antonio. I. F.: “Acoustical wave propagation in cylindrical ducts: Trans- mission line parameter approximations for isothermal and nonisothermal boundary conditions”. T. Journal of Turbomachinery. 2000. Diploma thesis. Doctoral thesis.. (Cited on page 38. 78:614–617. Copenhage.) [250] Jordan.) [256] von Kármán. Journal de Mathématiques Pures et Appliquées. C. Deuxiéme Série. 2013. Gikadi. in ASME Turbo Expo 2013. 11-15 June 2012. (Cited on page 145.S.. (Cited on pages 113 and 121. Lee.: “Mechanische ähnlichkeit und Turbulenz”.S. Watson. (Cited on page 30.: “On the use of experimental methods to im- prove confidence in educed impedance”.: Experimentelle Untersuchung zur Entstehung und Minderung des Blattspitzen-Wirbellärms axialer Strömungsmachinen. Denmark. Wagner.) [253] Jung. 1874. GT2012-68916. 1993.

D. Auregan..: Acoustical response of shear layers.) [268] Kooijman. M. in 11th AIAA/CEAS Aeroacous- tics Conference. Ronneberger. in 7th AIAA Aeroacoustics Conference. AIAA-80-0990. Doctoral thesis. 1995..) [267] Kooijman.B. 61. (Cited on pages 19. 1869. A. A. Yoon. Journal of Sound and Vibration. AIAA- 1981-1998. S.) [265] Kompenhans. AIAA-2005-2857. (Cited on page 32. John Wiley and Sons. Journal für Math- ematik. 183(1):115–127.: “Zur Theorie freier Flüssigkeitsstrahlen”. bibliography 321 [258] Keller.: “Über den Einfluss der Wärmeleitung in einem Gase auf die Schallbewegung”. D. in 6th AIAA Aeroacoustics Conference. 114. 1972. Connecticut. Frey.... and 239. 22(2):193–210. Journal of Sound and Vibration. E. (Cited on pages 59. Annalen der Physik und Chemie.. 54. Hartford. (Cited on page 37. (Cited on page 23.) [263] Knutsson.J. (Cited on page 126.H. 1981. Y.) [262] Kirchhoff. J. 63. 5-7 October 1981. G. Y. 4-6 June 1980. G. 238. 2007.L. Coppens. 329:4719–4739..) . 110. 1995. (Cited on pages 62. 23-25 May 2005. Hirschberg.) [260] Kinsler. 118. Åbom.: “An experimental study on the acoustic character- istics of perforated pipe in terms of wavenumber and porosity”. 121. 79. 2000. Sarin. New York.: “On the use of Helmholtz resonators as sound attenuators”. J. G. California.. 2005. A. Zeitschrift für angewandte Mathematik und Physik. 46:297–327. Palo Alto. J.: “An experimental study of the impedance of helmholtz resonator arrays under a turbulent boundary layer”. Zauner. 70:289–298. and 212.. 120.) [261] Kirchhoff. California. (Cited on page 18. Sanders. 210(6):177–193. 1868.H. 4th ed. 115. J.) [259] Kim.: Fundamentals of acoustics.. Technis- che Universiteit Eindhoven. 1980. G.: “Sound attenuation in acoustically lined circular ducts in the presence of uniform flow and shear flow”. Monterey. S..: “The effect of turbulence damping on acoustic wave propagation in tubes”. L. M. and 131. (Cited on pages 32 and 67.: “Orifice impedance under grazing flow: Modal expansion approach”. 2010. (Cited on pages 32 and 217. [266] Kooi.) [264] Ko.W. Journal of Sound and Vibration.: “The acoustic impedance of orifices in the wall of a flow duct with a laminar or turbulent flow boundary layer”. 78.

J. L. F. Journal of Sound and Vibration. W.: “Acoustic measure- ments of perforated liners in hot and pressurized flow”.. 4-6 June 2012.322 bibliography [269] Kooijman. W. Contractor Report NASA-CR-1999-209120 Vol. D. A.) [278] Lahiri. USA. J. L. Flohr. in 18th AIAA/CEAS Aeroacoustics Conference. 20(10):657–674. 333(15):3440–3458. (Cited on page 46. K. Armitage. GT2013-94674. Enghardt. P. Enghardt. and 55. (Cited on page 115. C. and 75. B. Johnson.... 2014.W. Journal of the Acoustical Society of America.) [279] Lahiri.H. Annalen der Physik. Knobloch. 2012.) [274] Kurze. C..... Sat- tinger. Golliard. 1999. in AIAA Meeting Paper Archive 172. Prade.: “Attenuation of sound in wide ducts with flow at elevated pressure and temperature”. 2: Advanced treatment impedance models for high frequency ranges”. K... Yu. Yu.. U. S.: “Influence of cavity design on the performance of hot-stream bias flow liners”. J.: “Acoustical response of ori- fices under grazing flow: Effect of boundary layer profile and edge geom- etry”. (Cited on pages 32 and 217. Habilitation thesis. 1953. Texas. 2..S. Hirschberg. Bethke. A. 1868.W. in ASME Tubo Expo 2013. Beer. NASA.. C.: “Acoustic treatment design scaling meth- ods. F. C. AIAA-1999-1939. 1902. H. 135(11):337–372. 66. (Cited on pages 34.. S.) [270] Kovasznay.) [272] Krebs. den diese dabei erfahren.. Allen.) [276] Kwan. 315:849–874. Colorado. S. Sadig. 2005. Bake.: “Untersuchungen über die Schallgeschwindigkeit der Luft in Röhren”. (Cited on pages 61.) [277] Lahiri.) ..: “Influence of flow and high sound level on the attenuation in a lined duct”. 89–112. L. San Antonio.. (Cited on pages 190 and 191. 5.. 2008. Technische Hochschule München.G. Journal of the Aeronau- tical Sciences. 2013. 3-7 June 2013. Vol. R. 62. and 54. (Cited on page 19. H. B. 52.: Über die Strömung einer Flüssigkeit um in sie versenkte zylin- drische Flächen und den Druck. 65. 1999. (Cited on page 15. pp. Bake. Colorado Springs. 682. Kwan. Journal of Sound and Vibration.: “Turbulence in supersonic flow”. Bake.) [275] Kutta.J. (Cited on pages 16. J. in [305]. (Cited on pages 114 and 190..) [273] Kundt.. 45. AIAA- 2012-2200. (Cited on pages 52 and 190.: “Thermoacoustic design tools and passive control: Siemens Power Generation approaches”. 1971.. Lepers.E.) [271] Kraft.: “Bias flow adaptive acoustic liner”... 49(5):1643–1654. Chap.M. C. Knobloch. K. G. Knobloch. F.

S. Preclik. F. and 263. and 72. 71. Selamet. (Cited on pages 22. 743:448–480. Pierro. I. 114.: “Lower and upper bounds for the Rayleigh conductivity of a perforated plate”. (Cited on pages 22. and 238.) [288] Lee.: Acoustic characteristics of perforated dissipative and hybrid silencers. R.. Tordaux..) [286] Laurens. B. 14.T. S.M. S.) . (Cited on pages 39. Pongratz.B.. B.R. Enghardt.. Röhle.) [283] Landau. Tordeux. and 115.. AIAA Journal. (Cited on page 202.) [281] Lahiri. C. Chap. M....d. S. Piot. 25. 1966.H. F. P. Portland. in [497]. I. Kwon. (Cited on pages 6.D.) [285] Laurens. Gerendás.) [287] Lee...: Fluid Mechanics. AIAA-2011-2725. Bendali. Bendali. and 262. Pergamon Press. 2014.. Bake. A. L. A. 2011..: “Excitation of a zero mass flow liner for acoustic damping”.P. 47.: “Impact of perforation impedance on the transmission loss of reactive and dissipative silencers”. M. (Cited on pages 47 and 246. Journal of Engineering for Gas Turbines and Power. 52. R... E. (Cited on pages 19 and 59. 05 . 2011. D. bibliography 323 [280] Lahiri. 34.. 19. D. C. (Cited on pages 32.. in 17th AIAA/CEAS Aeroacoustics Conference. I. Journal of Fluid Mechanics. Oxford.) [289] Lee.: “The application of an aeroacoustic actuator in a zero mass flow liner for acoustic damping”. 65. L. 48.. Pardowitz.: “Estimation of the absorption performance of mul- tiple layer perforated panel systems by transfer matrix method”. Y. thesis. A. Bake. 47(6):1691–1712. Ph. (Cited on pages 17. 2006..) [282] Lahiri. 262. N... Kotiuga. 2013. 23. Huff. M. 27.: “Experimental proce- dures aiding the design of acoustic cavities”. 19.. Pardowitz. E.. Fares. Enghardt.. ESAIM: Mathematical Modelling and Numerical Analysis.: “Establish- ment of a high quality database for the acoustic modeling of perforated liners”. S. 3rd ed. Bake. 133:1–9. Oregon. 109.: “Effective conditions for the reflection of an acoustic wave by low-porosity per- forated plates”. 2005. E. (Cited on pages 34 and 52.) [284] Laudien. 55. 52. 49(3):513– 519.. 377– 399. 2011. 278:847–860. L.. 262. Fares. and 263. L. Lifshitz. Enghardt. Journal of Sound and Vibration.. Ohio State University. Sadig. C. 2004. 35. 120(6):3706–3713. Journal of the Acoustical Society of America.. F. pp. 1995.08 June 2011.

) [294] Lei. E. Ih. D. Journal of the Acoustical Society of America.: “The role of acoustic absorbers in pre- venting combustion instability”.. and oxygen from 60 to 2000 K at pressures to 2000 MPa”..M. 2007.S. 212. Utah. F.W.. (Cited on pages 3.. (Cited on page 147. J....) [293] Legendre. 1973. 26. 23:623–630. Ballal.G.H. 2004. 2007. L. (Cited on pages 284 and 285. R.M.G. 2000. Peat. Jacobsen.T. K. Paris. Levine. (Cited on pages 44 and 220.G. 3rd ed. (Cited on pages 284 and 285. (Cited on pages 16.W. R. Li.: “In-duct orifice and its effect on sound absorption”.C.. Journal of Fluid Mechanics. R. (Cited on pages 32. and 75. H.) [295] Lemmon.324 bibliography [290] Lee.. 2010. 299:990–1004.H..) [291] Lee.M.D.) [296] Lemmon.C. Xiaofeng.G. So. Salt Lake City. Wang.) [297] Leppington. S. 24:199–215. 29(3):331–385.: Nouvelles methodes pour la determination des orbites des cometes (New Methods for the Determination of Comet Orbits). and 280. Z. G.G. Garrison. Math- ematika. Ih. (Cited on page 55.: “Emperical model of the acoustic impedance of a circu- lar orifice in grazing mean flow”. E. F. S.D.K. Friend. Didot. and air”.G. J.H. Boca Raton. in AIAA/SAE 7th Propulsion Joint Special- ist Conference. CRC Press.: Gas Turbine Combustion .: “The effective compliance of perforated screens”. 67.H. 1805..: “A passive method to control combustion instabilities with perforated liner”. 114(1):98–113. S. International Journal of Thermo- physics. oxygen. Journal of Physical and Chem- ical Reference Data.) [299] Leung. 1977. 4. Penoncello. and 55. 25(1):21–69. Journal of Sound and Vibration. (Cited on pages 55. argon.) .Alternative Fuels and Emissions..R. Chengyu. 303:741–752.) [292] Lefebvre. FL.. 12. Chinese Journal of Aeronautics. (Cited on pages 47 and 227.: “Reflexion and transmission at a plane screen with periodically arranged circular or elliptical apertures”. M. 93. 61(1):109–127. Zhihui. S.: “Viscosity and thermal conductivity equa- tions for nitrogen.. D. G. F. (Cited on page 227. and 15.) [298] Leppington.T. 2010. X. R. 2003..14-18 June 1971.. AIAA-1971-0699.) [300] Lewis. A. A.. Jacobsen.: “Ther- modynamic properties of air and mixtures of nitrogen. 13.: “A model of acoustic impedance of perfo- rated plates with bias flow considering the interaction effect”. argon. G. Journal of Sound and Vibration. 68. 1971.

106:13–17. A. 1934. (Cited on pages 43 and 53. 2005.H. 26. (Cited on page 36. 27. GT2006-90874.: “A time-averaging transient testing method for acoustic properties of piping systems and mufflers with flow”. 7. A.) [311] Maier.C. 2012.): Gas turbine emissions. AIAA. (Cited on page 37. A. (Eds. Cambridge Univer- sity Press. (Cited on page 3.G. 27. UK. New York. 8-11 May 2006.. Journal of Mechanical Engineering Science. J. 2006. (Eds.) .: “Resonanzerscheinungen in den Rohrleitungen von Verbrennungskraftmaschinen (Resonance phenomena in the ducts of com- bustion engines)”. A. and 343. bibliography 325 [301] Lewis Laboratory Staff: “A summary of preliminary investigations into the characteristics of combustion screech in ducted burners”. 54. T. T.: “Pas- sive damper lp tests for controlling combustion instability”.) [302] Lichtarowicz. T. 1954.: “On the Rayleigh conductivity of a bias-flow aperture”. E. R. T. 73:867–876. Journal of Fluids and Structures.C. and 269. Barcelona. Berichte aus dem Laboratorium für Verbrennungskraftmaschinen der Technischen Hochschule Stuttgart. in ASME Turbo Expo 2006. M.) [309] Lutz. J. V... NY. (Cited on page 154... Berichte aus dem Laboratorium für Verbrennungskraft- maschinen der Technischen Hochschule Stuttgart.C.Y.: “Resonanzschwingungen in den Rohrleitungen von Kolben- maschinen (Resonant oscillations in the ducts of reciprocating engines)”. O. 55.): Combustion instabilities in gas turbine engines. Cambridge. and 203. (Cited on pages 5. 211. 1934. Lienhard.A. 74. Yang. Holt.. Lutz..) [306] Lieuwen.. M.. Journal of the Acoustical Society of America.H... 1984. Moran. Journal of Fluids Engineering.) [303] Lienhard. 2013. 1983.J. 73. Howe. Cambridge University Press.) [305] Lieuwen.) [307] Lung. Reston. J. (Cited on pages 43 and 53. (Cited on page 15. W. (Cited on page 15. O. R. (Cited on pages 16. 7(2):210–219. McGowan.: “Discharge coefficients for incompressible non-cavitating flow through long orifices”. 25. 309. 328.. (Cited on pages 48.) [310] Macquisten. Yang. National Advisory Committee for Aeronautics.. T. Rupp. M.S. 306. 72.: Unsteady Combustor Physics. V.K. Markland.) [308] Luong. Doige.S. Spain. 3:17–110. 1965. Technical Report NACA-TR-1384. Duggins. 2005. 322.) [304] Lieuwen. Whiteman. 3:1–16.: “Velocity coefficients for free jets from sharp-edged orifices”. 15. VA. 21:769–778.

(Cited on page 32. Maastricht.) [318] Mazdeh.. 11-15 June 2012. GT2012-69727. 10(2):208–226. Report AMRL-TDR-62-140(1). Denmark. 2001. 1”.: “Research on sound propagation in sound-absorbent ducts with superimposed air streams. Beyer. 1960. NASA. AIAA-2001-2193.: “Schalldämpfung und schallverstärkung in luftströmungen durch absorbierend ausgekleidete Kanäle”.. 28-30 May 2001.326 bibliography [312] Malmary.: “Absorption of sound in fluids”. P. J. P. and 212. thesis. 55. Parson. J. Kashani. R. 10(3):133–148.P. 1969.: “The impact of perforation geometry on acous- tic damping attributes of a perforated liner with bias flow”.) [315] Mason.L.C. GT2011-46649. C.: “Distributed parameter acoustic modeling of a perforation with bias flow”. Vol. in ASME Turbo Expo 2011.: “Some experiments on the propagation of sound along a cylindrical duct containing flowing air”. 1966. A. 1928. S.) [316] Mason. Physical Review. F. 6-10 June 2001. Canada. 212:918...A.. Schnell.) [314] Markham. and 71.. A.J. (Cited on pages 32 and 217.B. Auregan. G. Wright-Patterson Air Force Base. W.C. (Cited on pages 47. C.) [317] Mazdeh. 1962. Acustica.Sc.: “Hysteresis in the flow through an orifice”. 31:283–295. (Cited on pages 47. Contractor Report NASA-CR-90094. (Cited on page 131. 55. (Cited on page 190. Netherlands.) [322] Mechel. and 71.T. 51. Lindsay. 53.: The influence of high speed flow on the bahavior of acoustical elements.: “A study of the suppression of combustion oscillations with mechani- cal damping devices”. Y. in ASME Turbo Expo 2012. (Cited on page 36. Journal of Sound and Vibration. M. F. 1967. (Cited on pages 220 and 221.. Russell.) . Bohn. R. A.. (Cited on page 103. R. (Cited on pages 43.E. 1951. Copenhagen. N. Massachusetts Institute of Technology. V.) [313] Marino. Nature. Reviews of Modern Physics. 2012. Pagneux. Kashani. Aerospace Medical Research Laboratories.) [320] McVeigh. 23(4):353–411.L.) [319] McAuliffe. V.: “Acoustic impedance measurement with grazing flow”. 2011. in 7th AIAA/CEAS Aeroacous- tics Conference.. Carbonne.) [321] Mechel. 51... 1950. Vancouver. R.: “The propagation characteristics of sound tubes and acous- tic filters”. (Cited on pages 217 and 219.

Schilz. published 26 May 2011. T. Journal of the Acoustical Society of America. P. P. 29(1):1–65. 81. Mertens.. A.. (Cited on pages 44 and 53. Nantes. and 55. 2nd ed. Tonon.co. in Acoustics 2012.) [333] Mongia. Eldredge. W. (Cited on pages 80. 50.. R. (Cited on page 103. 15:199–206. F.. Berlin.. 2009.) [325] Mechel. (Cited on pages 55 and 246.: “Interaction between air flow and airborne sound in a duct”. Journal of Sound and Vibration. S. F. Report AMRL-TR-65-53. 2008.) [329] Meyer. 62. 75. Hirschberg. Wright-Patterson Air Force Base. 77. Acustica. J. 23-27 April 2012..C. Springer. 1965. Held. 19.) [328] Mendez. and 85. Journal of Computational Physics.M. Pandalai.. Journal of Sound and Vibration. (Cited on pages 18. Kurtze..600 °c turbine inlet tem- perature in test operation of world’s highest thermal efficiency j-series gas turbine”. (Cited on pages 28.) [324] Mechel.jp/en/news/story/1105261435.. and 67. Schilz. F.. T. Aerospace Medical Research Laboratories.: “Incorporation of combustion instability issues into design process: GE aeroderivative and . 45. (Cited on page 217. Hsiao.: “Experiments on the influence of flow on sound attenuation in absorbing ducts”.. 1973. Mechel.D. 228(13):4757–4772. and 227. 64. E. 34. 82. W. Dietz. D. 2012. H.) [326] Mechel. 63.) [332] Moers.: “Akustische Impedanz einer luftdurch- strömten Öffnung”. G. Mertens. 61. 1963. 134(2):203–234.) [327] Melling.) [330] Michalke. (Cited on pages 2 and 14.: “The acoustic impedance of perforates at medium and high sound pressure levels”. html.) [331] Mitsubushi Heavy Industries: “MHI achieves 1. accessed online 27 February 2014: http://www. A. 1965. 53.H. J. F. 1435. Press Release No..: Formulas of acoustics. G. 1989.: “Sound absorption by perfo- rated walls with bias/grazing flow: Experimental study of the influence of perforation angle”.. 13(3):154–165.: “On the propagation of sound generated in a pipe of circu- lar cross-section with uniform mean flow”. Acustica. bibliography 327 [323] Mechel. F.J.: “Acoustic modeling of perforated plates with bias flow for large-eddy simulations”..mhi. 30:165.: “Schallausbreitung in absorbierend ausgeklei- deten Strömungskanälen bei hohen Windgeschwindigkeiten”. 202. 65. (Cited on pages 44. France. (Cited on pages 32 and 217.T. 1958.C. E.P.

(Cited on page 81.D. (Cited on page 103.: “On the reciprocal of the general algebraic matrix”...) .) [336] Morfey.) [344] Munjal.L. 1969. 24-26 March 1975.: “Theory of a two source-location method for direct experimental evaluation of the four-pole parameters of an aeroacoustic element”. 1990. 98. P. 82. Sahasrabudhe. 1987. 234.) [340] Morse. Ingard. 100. 46. Virginia. 3. 102. Journal of Sound and Vibration. (Cited on page 82.L.: “Velocity ratio-cum-transfer matrix method for the evalua- tion of a muffler with mean flow”. 114(2):173– 188. 1975.L. New York. 1975.) [345] Munjal. (Cited on pages 80. (Cited on pages 17.: “Aeroacoustic analysis of perforated muffler components”. Journal of Sound and Vibration. McGraw-Hill. Hampton. 26(9):394–395. Chap. C. Bul- letin of the American Mathematical Society.: Vibration and sound. Jour- nal of Sound and Vibration.) [335] Morfey. 103. 111.: “Acoustic properties of openings at low frequencies”. Academic Press. New York. (Cited on page 101. 2nd ed. M. C.: Dictonary of acoustics..: Acoustics of ducts and mufflers. K. P. 1948. 9(3):357–366. C. McGraw-Hill. 1987.G.: Theoretical acoustics. 100. 110. E.: “Sound transmission and generation in ducts with flow”. 14(1):37–55.U. in [305]. 1:60–87.L. 1920. 109. Journal of Sound and Vibration. and 118. 100. A. New York..) [337] Morfey. Journal of Sound and Vibration. P. 1968. M. 58.M. Doige. 101. in 2nd AIAA Aeroacoustics Conference. 141(2):323–333.L. 61.) [341] Mungur. John Wiley and Sons. AIAA-1975-0494.L. (Cited on pages 66.L. Whitesides. (Cited on page 32. J.: “Influence of grazing flow on duct wall nor- mal impedance”. Journal of Sound and Vibration. 2001.. (Cited on page 59. M. C.M.L.) [342] Munjal. 2005.) [339] Morse.H. 1971. Rao. 1964. 81. M. 39(1):105– 119.: “Rotating pressure patterns in ducts: Their generation and transmission”. pp.) [334] Moore. and 155.328 bibliography aero engine experience”. (Cited on pages 6 and 15. and 111. (Cited on page 154.N. (Cited on page 145. 43–63.L.) [338] Morfey. (Cited on pages 37. A. K.) [343] Munjal.. and 235.

. 1994. UK.. AGARD-CP-390. W. R. Report DLR-IB-22314-94/B5. Journal of Sound and Vibration.L. R.: “The propagation of sound in cylindrical ducts with mean flow and bulk reacting lining: III. bibliography 329 [346] Munt.) [350] Nayfeh. 34(4):505–517.R. J. E.: Philosophiae Naturalis Principia Mathematica.M.. (Cited on page 36. I. (Cited on page 53.: “Acoustic transmission properties of a jet pipe with subsonic jet flow”. (Cited on page 13. 1977. 10-12 May 2004. AIAA-2004-3033. Step discontiuities”. (Cited on page 77. Journal of Fluid Mechanics. (Cited on pages 26 and 53..) [349] Nayfeh.) [348] Navier.. Journal of the Institute of Mathematics and its Applications.) [347] Munt. A. (Cited on page 26. Journal of Sound and Vibration. 1989. P. 2004. in Heat transfer and cooling in gas turbines.) [353] Newton. C. (Cited on page 217. O. A. Kaiser.A. Advisory Group for Aerospace Research and Development.: “Acoustic radiation from a circular pipe with an infinite flange”. Michel.E. 1822. 1990. U. Sheng. Manchester.H.L. S.: “Acoustics of aircraft engine- duct systems”. Cantabrigiæ.: “Development of new liner technology for application in hot stream areas of aero-engines”. 13(2):130–153.M. 25. J. 27:105–131. 1974.) [356] Norris. in 10th AIAA/CEAS Aeroacoustics Conference. Brander. A. D. 1985. C. Mongia. (Cited on page 53.. 1713. AIAA Journal. 142(3):413–436.. Journal of Sound and Vibration. H. B. 1981. 83:609–640. (Cited on page 114.: “Effect of transverse velocity and temperature gra- dients on sound attenuation in two-dimensional ducts”. (Cited on page 268. 135:85–93.) [354] Nilsson. Sun. Reider. UK.: “The interaction of sound with a subsonic jet issuing from a semi-infinite cylindrical pipe”. Mémoires de l’Académie des sciences de l’Institut de France. 6:389–440. Chap...) [355] Nordin.: “Memoire sur les lois du mouvement des fluides”. 1975.B. Cambridge. Telionis.N.: “Alternate cooling configura- tions for gas turbine combustion systems”. DLR. D. (Cited on page 101.. Rademaker.P. S.M.) [352] Neise.H. Sarin.) [351] Nealy.) .C.H.: “Aerodynamic noise of turbomachines”.

A. J. and 82. Polifke.: “A numerical decoupling analysis of perforated pipe silencer elements”. T.. Contractor Report NASA-CR-115087.. D. 123(2):199–212.) . and 126. Ih.: Die galvanische Kette. 1984. (Cited on pages 19. (Cited on page 57.L. 54.) [358] Ohm..S. A. Journal of the Acoustical Society of America..: “Effects of high-wave amplitude and mean flow on a helmholtz resonator”.: Messung des Schalldurchgangs an Diskontinuitäten in durch- strömten Kanälen.) [363] Peat. 1971.O. 1976. 114(6):3076–8086. 2003.: “Evaluation of acoustic cavities for combustion stabilization”.L. 93(4):473–480. 1993. S. R. NASA. W.: “Measure- ments of transfer matrices and source terms of premixed flames”. 2002. 1827. Mattson. 73.M.330 bibliography [357] Oberg. 60:1390–1396..H.) [362] Paschereit. 51(3):406–413. Journal of the Acoustical Society of America.M. T. 256:499– 534. 122. C... and 202. R. Berlin.) [360] Pallek. (Cited on pages 28.: “Damping and reflection coefficient measurements for an open pipe at low mach and low helmholtz numbers”. Reijnen. Technical Memorandum NASA-TM-X-1582. N. B.H. Riemann. 1984.: “The acoustic impedance of a circular orifice in grazing mean flow: Comparison with theory”.: “A generalized inverse for matrices”..) [359] Page.. 1955. NASA.S. M. G. (Cited on page 44.J.) [364] Peat. A.) [361] Panton. 45. B. (Cited on page 145.) [365] Penrose. 124:239–247. Jour- nal of Engineering for Gas Turbine and Power.W.. 1968.: “Wall effects on sound propagation in tubes”. Doctoral thesis.J. Hirschberg. O.: “Correlation of nonlinear orifice impedance”. 1988. D. Jour- nal of Sound and Vibration.C. Journal of Fluid Mechanics.J. Goldman.. (Cited on pages 54.P. C.. and 139. A.L. Journal of Sound and Vibration. Mathematical Proceed- ings of the Cambridge Philosophical Society.G.L. Mee. Wong. (Cited on page 113. Wijnands. K. 124.) [367] Phillips. K. Schuermans. (Cited on pages 77. (Cited on page 80.A. Ford. (Cited on pages 32 and 67. Universität Göttingen.) [366] Peters. W. Lee.S. mathematisch bearbeitet (The galvanic circuit investigated mathematically). (Cited on pages 26 and 53.

109.A. Demmenie.R.: “Über Flüssigkeitsbewegung bei sehr kleiner Reibung”.: “The application of a time-domain model to investigate the impedance of perforate liners including the effects of bias flow”. Transactions of the American Society of Mechanical Engineers. Journal of Fluid Mechanics. statically and in flight”. 1984.P. 1971. E.) [371] Powell. 132. Prague. Internationalen Mathematiker-Kongresses. B. L. Russell. (Cited on page 44. pp. Journal of Sound and Vibra- tion. 22-25 August 2004. New York.) [370] Pinker.D. 134.: Acoustics . and 135.C.: Combustion-driven oscillations in industry.: “Theory of vortex sound”. 97. 54. in 33rd International Congress and Exposition on Noise Control Engineering.) .S. W. 99. Elsevier. 1976.: “A study of burner oscillations of the organ- pipe type”. 2004. A. A. 100. and 68. Czech Republic.A. E. J. M.: “On the design of acoustic liners for rocket engines: Helmholtz resonators evaluated with a rocket combustor”. 110. (Cited on pages 155 and 217.A. in Verhandlungen des III. N. Acoustical Society of America. Journal of the Acoustical Society of America. New York.) [369] Pierce. 116.: “Sound propagation in a fluid flowing through an attenuating duct”. NLR-TP-2004-146.) [378] Rademaker. (Cited on page 219. 131. (Cited on pages 5 and 6.: “In-situ acoustic impedance measure- ments on air-injected liners under grazing flow”.: “On the production and absorption of sound by lossless liners in the presence of mean flow”. 80. Hannum. (Cited on page 26. 36(1):177–195. D. 484–491. R. (Cited on page 54.An introduction to its physical principles and applica- tions.) [374] Pridmore-Brown. Howe.. 1999. Technical Note NASA-TN-D-5171. 1964. AIAA-1999-1876.) [373] Premo. bibliography 331 [368] Phillips. (Cited on pages 60. 4:393–406. 1969. 73:15–28. Dennis. 115.A. in AIAA Meeting Paper Archive 711. 1904. AIAA-1976-581. (Cited on page 117..: “The radiation of plane-wave duct noise from a jet exhaust. 1958. 130. in 3rd AIAA Aeroacoustics Conference..M..D..) [372] Prandtl. 98. M. (Cited on pages 45. 121. (Cited on page 47. 1989. Bryce.) [376] Putnam. NASA. W. 97(1):1–9. A.C. L. 1953. A..) [375] Putnam. (Cited on page 5. 118.) [377] Quinn.

J. Poland. 56(2):107– 114. and 235. 59. Dover.W. 1986. Journal of Noise Control Engineering. 57.G. 121. S. (Cited on page 32. 1876.) [382] Rämmal.) [381] Rademaker. R. in 16th International Congress on Sound and Vibration. (Cited on page 65.: “On the passage of waves through apertures in plane screens. 58. J.M.S. 2008. J.) [380] Rademaker. Lavrentjev.) . R.: “Notes on hydrodynamics”. E. Munjal. 1945.M.) [390] Rayleigh. M. van der Wal.W. 918. van der Wal.: “Hot-stream in-situ acoustic impedance measurements on various air-filled cavity and porous liners”.: Perforated panel absorber with viscous energy dissipation en- hanced by orificed design.W. 233. (Cited on pages 267 and 268. Philosophical Magazine Series 5. Nijboer.G. H. and allied problems”.) [384] Rao.: The theory of sound. (Cited on page 26.J. Philosophical Magazine Series 5. 2:441–447. 2000.. Report NLR- TP-2003-202.) [388] Rayleigh. 59. (Cited on page 37. H. J. E. Geurts. Geurts. 1897.: “On waves”.) [386] Rayleigh.: “Sound refelction at an open end of a circular duct exhausting hot gas”. S.) [387] Rayleigh. 1871. (Cited on page 101. 108(2):283–295. Kraków. 1876.L.W. 18(455):319–321.. 5-9 July 2009. 2011. NLR.. Vol.S. H.: “A new facility for hot stream acoustic liner testing”. National Aerospace Laboratory (NLR).T.R.: “On the theory of resonance”. 120.S. (Cited on pages 17. 43(263):259– 272.M.. Na- ture.W.M. 58.: “Hot-stream in-situ acoustic impedance measurements on various air-filled cavity and porous liners”.N.N. K. (Cited on pages 5.) [389] Rayleigh. Idzenga. (Cited on page 26.R.M. 2. Report NLR-TP-2009-142.) [385] Rayleigh. J. (Cited on page 17.. Doctoral thesis. H.: “The explanation of certain acoustic phenomena”. 2003. E.T. Huisman.332 bibliography [379] Rademaker. J. E. ICSV16.R.S..S. E. 1878. 2009.. and 235. (Cited on page 26....W. 233. J. Sarin.L.S. Philosophical Magazine Series 5. Philosophical Transactions of the Royal Society of London. New York. Journal of Sound and Vibration. (Cited on page 5.: “Experimental evaluation of impedance of per- forates with grazing flow”. Norwegian University of Science and Technology.) [383] Randeberg.M. 1(4):257–279. 57. 161:77–118.

in Fortschritte der Akustik.: “Schalltransmission durch Blenden und Lochbleche im Durchströmten Rohr”. D.J.) [394] Rice. 14-17 May 2007. GT2007-27319. Ronneberger. Berlin. Canada. C. VDE-Verlag. in Fortschritte der Akustik. Journal of Engineering for Industry.: Messung der akustischen Streumatrix durchströmter Lochblenden. Technical Memorandum NASA-TM- X-67950.) [398] Richter.) [396] Richter. NASA.: “Eine Methode zur Messung akustischer Grössen in Kanälen”. München. Technische Universität Berlin.J.) [397] Richter. DAGA ’80. (Cited on pages 53 and 139. Straub. pp. K. E. (Cited on pages 47 and 71. bibliography 333 [391] Rehman.: “A model for the acoustic impedance of a perforated plate liner with muliple frequency excitation”.) [393] Rice. (Cited on page 74. (Cited on page 32.. 1971.: “A small strouhal number analysis for acoustic wave-jet flow-pipe interaction”.) [400] Rienstra. 2009.: “Spinning mode sound propagation in ducts with acoustic treatment”. 1978. S. 1976... Journal of Propulsion and Power..H. E. (Cited on page 27. (Cited on pages 53 and 139.. (Cited on pages 67 and 221. NASA.D.) [401] Rienstra. S. 1975. DAGA ’78. Bochum. (Cited on page 77. S.A. 86(4):539–556. D.: “On the acoustic implications of vortex shedding from an exhaust pipe”.) [392] Rice. K. VDE-Verlag. Montreal. 819–822.) . Journal of Sound and Vibration.W. 103(4):378–384. Diploma thesis. 1981.) [395] Richards. 1980. in ASME Turbo Expo 2007. Universität Göttingen. Ronneberger.W. 1978. K. 2007. D. Technical Memorandum NASA-TM- X-71903.: Liner impedance modeling in the time domain with flow. Berlin. E. 2003. NASA. D..: “Passive control of combus- tion dynamics in stationary gas turbines”. 19(5):795–810.J. Pallek. Robey. (Cited on page 6. 375–378. J. Doctoral thesis. G. (Cited on page 53. Technical Note NASA-TN-D-7913.: “A theoretical study of the acoustic impedance of orifices in the presence of a steady grazing flow”. E. [399] Richter. 1983. Eldredge.: “Numerical investigation of a bias-flow per- forated liner for damping of thermoacoustic instabilities”. pp.L.F.

E. Paris.C. 1935.. 155. T. Technical Note NASA-TN-D-5467. 1987. Virginia.R. D. 1972.. 139. Report. (Cited on page 17. 139. 1975. Hampton. Herman. and 221. J. and 147. Richards.) [403] Ries.T. USA. 24-26 March 1975. H. 123. G. Kroll. Hersh. (Cited on page 26. Journal of Engineering Mathematics. UK. 231(5):1221–1242. 19:222–235.) [407] Rohde.: “Sound attenuation in tubes due to visco-thermal effects”.: “Experimentelle Untersuchungen zum akustischen Reflexionsfaktor von unstetigen Querschnittsänderungen in einem luft- durchströmten Rohr”. (Cited on page 53. 53. in 2nd AIAA Aeroacoustics Conference. Singh. 1996.) [408] Rolls-Royce: The jet engine. (Cited on pages 122.334 bibliography [402] Rienstra. Journal of Sound and Vibration. (Cited on pages 32. Texas. NASA.: “Discharge coefficients for thick plate orifices with approach flow perpendicular and inclined to the orifice axis”. 1969.. AIAA-1975-0493.W. 2013. M. 59:451–475. Rolls-Royce plc.. Metger. (Cited on page 165. Universität Göttingen. 24(1):133–150.: “The acoustical impedance of holes in the wall of flow ducts”.) [412] Ronneberger. (Cited on page 131. E. 2000.: “Theoretische und experimentelle Untersuchung der Schallausbreitung durch Querschnittssprünge und Lochplatten in Strö- mungskanälen”. and 15. (Cited on pages 19. Eigenbrod. Miller..) [409] Ronneberger.: “On the performance of porous sound absorbent ceramic lining in a combus- tion chamber test rig”. (Cited on pages 12. San Anto- nio. Journal of Sound and Vibration. C. 5th ed. D. 54. N..) [411] Ronneberger.) [404] Rocard. 1967/68. Hrnjak.V. (Cited on page 113.) [410] Ronneberger. Derby.) [405] Rodarte. 1975. 50.: Genaue Messung der Schalldämpfung und der Phasen- geschwindigkeit in durchströmten Rohren im Hinblick auf die Wechselwirkung zwischen Schall und Turbulenz. H. Y..) [406] Rogers.. Universität Göttingen. 124. G.S.: Propagation et absorption du son. A. P. and 240. S. Rezwan. Carlesson.. Acustica. (Cited on page 38. 3-7 June 2013. 2007. D.W.) ..: “Acoustic scattering at a hard-soft lining transition in a flow duct”. K.: “The effect of grazing flow on the steady state resistance of square-edged orifices”.. 147. in ASME Turbo Expo 2013. 14. GT2013-95492. S. D. Habilitation thesis.

14-18 June 2010. 2009. (Cited on pages 15. and 284. 5-7 October 1981. 6-10 June 2011.: “Interaction between the acoustic pres- sure fluctuations and the unsteady flow filed through circular holes”. J.: “Wall shear stress caused by small am- plitude perturbations of turbulent boundary-layer flow: An experimental investigation”.: “The use of perforated damping liners in aero gas turbine combustion systems”. New York. Spencer. GT2010-22178. J. J. 1981. Carrotte. K. Glasgow. Journal of Fluid Mechanics. 2012. (Cited on pages 49.: “The use of perforated damping liners in aero gas turbine combustion systems”. 2009. 283. Journal of Engi- neering for Gas Turbines and Power. (Ed.. 2010. in ASME Turbo Expo 2009. in ASME Turbo Expo 2010. in ASME Turbo Expo 2014. Macquisten. Vancouver.. A. USA.. 25.. 91(4):479–502... J.D. J. and 203. 203. 135. Journal of Engineering for Gas Turbine and Power.: “Acoustic power dissipation on radiation through duct terminations: Experiments”. Peacock. (Cited on pages 123 and 126.. G. 2011. Ger- many. M. GT2014-26907. T.) [417] Rupp. bibliography 335 [413] Ronneberger. 2014. J. (Cited on pages 55 and 203. 47. and 281. AIAA-1981-1978. J. Springer. 49. 2010. J.) [418] Rupp.K.: “Assessment and pre- diction of helmholtz resonator performance within gas turbine combus- tion systems”. 1977. Ahrens. J. M. C. Carrotte.) . 49. J. M.. Düsseldorf. 35. K.K. 29. 55. (Cited on page 53.. Canada.) [415] Rupp. FL. G. 2007. J.. Carrotte. (Cited on pages 133.: “Methodology to identify the un- steady flow field associated with the loss of acoustic energy in the vicinity of circular holes”. in 7th AIAA Aeroacoustics Con- ference.experiments”. Palo Alto. J. Journal of Sound and Vibration.: “Acoustic power dissipation on radiation through duct terminations . (Cited on pages 28.: “Interaction between the acoustic pressure fluctua- tions and the unsteady flow field through circular holes”. GT2009-59263. M. 16-20 June 2014.) [420] Rupp. in ASME Turbo Expo 2011. June 8-12.. GT2011-45488.) [416] Rupp. 83(3):433–464. 71. (Cited on page 53. 132:1–9. Macquisten. A. Carrotte. Ahuja.) [421] Salikuddin.) [419] Rupp. Regunath. 134:1–10. 1983..): Springer Handbook of Acoustics.D. D. (Cited on page 55. California. Orlando. Ahuja. Carrotte. Carrotte... (Cited on page 55. 55.) [422] Salikuddin.. 55.) [414] Rossing. UK. and 203. Spencer.

331:276– 290. M.. 1990. M.: “A LES based sound absorption analysis with high-amplitude waves through an orifice with bias flow”.D. E.: “Two-port in- direct acoustic impedance eduction in presence of grazing flow”.. (Cited on pages 28. 2012... A. (Cited on pages 25 and 45.: “Acoustic characteristics of per- forated sheets with throughflow in high itensity noise environment”.. (Cited on page 178.) [424] Salikuddin.. P. (Cited on pages 50 and 55.. S. Ducuix. A.. 2013.) . M. 1994. 71. Mungur.: “A novel design method for ro- bust acoustic dampers with perforated plates backed by a cavity operat- ing at low and high Strouhal numbers”. Portland.) [432] Schlichting. 29. Walter de Gruyter. Oregon. 2006. in ASME Turbo Expo 2011..) [431] Schade. H. Schuller. in 17th AIAA/CEAS Aeroacoustics Conference. GT2011-45639. A. Denmark.. Schuller. 50. 2011. 5-8 June 2011. Kunz. 72. and 212. Springer. Vancouver. S. Brown. 6-10 June 2011.: “Optimal and off-design oper- ations of acoustic dampers using perforated plates backed by a cavity”. (Cited on pages 50. W. 55..: “Acoustic characteristics of per- forated sheets with throughflow in high itensity noise environment”. T. T. Ducruix. N. S. P. AIAA-1990-3931. (Cited on pages 25 and 50. 2012. 55. T. W.) [426] Santana.. T.. Tran. Syed.A. in 13th AIAA Aeroacoustics Conference. P. A.: Strömungslehre. AIAA- 2011-2868. (Cited on page 32. GT2012-69249. Berlin. W.H.) [430] Scarpato.. in ASME Turbo Expo 2012. Berlin. 11-15 June 2012. and 72. Desmet.. Syed. 332:4856–4875. and 54.) [428] Scarpato. Florida. (Cited on page 54. (Cited on pages 47.) [429] Scarpato. Journal of Sound and Vibration. Jour- nal of Sound and Vibration.. de Roeck. 3rd ed...) [427] Scarpato.336 bibliography [423] Salikuddin. 2007. L. Tallahassee. 10th ed. Schuller. Canada. K..) [425] Salikuddin. 2011. Jour- nal of Sound and Vibration. Copenhagen.: “Non-linear effects in finite amplitude propagation through orifice plate and perforated plate terminations”.: “Modeling the damping properties of perforated screens traversed by bias flow and backed by a cavity at low Strouhal number”. A. H. Ducruix. Journal of Sound and Vibration. Mungur. 1990.A. 169(2):145–177. Schuller. A. Ferrante. Ducruix. 22-24 October 1990.: Grenzschichttheorie. Gersten.. 139(3):383–405. (Cited on pages 117 and 127.

and 55. (Cited on pages 25. D. E. (Cited on page 146.siemens. 304:91–109.: “Experimental determination of acoustic proper- ties using a two-microphone random-excitation technique”. Teil: Entwicklung willkürlicher Funktionen nach Systemen vorgeschriebener”. published 19 May 2011.) [439] Siemens: “Trail-blazing power plant technology”. Czarske.F. 2010. T. 2007.. D. S. Bhatia. Report. K. Kashinath..pdf.) [442] Smith. Journal of Sound and Vibration. A. R. 27-29 May 2013..) [441] Sivian. bibliography 337 [433] Schmidt. Haufe. W. B.) [434] Schmidt. Sheplak. 1977.) . L. Germany.) [435] Schultz. AIAA-2013-2177. (Cited on page 2. 55. Enghardt. Journal of the Acoustical Society of America. 65.. (Cited on page 38.: “Acoustic impedance of small orifices”. 1923. Fischer. U. Walker. (Cited on page 145.) [440] Sirignano.: “Zur Theorie der linearen und nichtlinearen Integralgle- ichungen. J. 63:433–476. K..: “Acoustic damping analysis of bias flow liners based on spectral flow char- acteristics”..J. Molavi. Hochgreb. 1935. (Cited on pages 51. L. (Cited on page 146. Bhayaraju. and 202. Ross. and 281.) [437] Seybert. M. (Cited on page 5.) [436] Schulz. 104(1):23–36.. 7:94–101.. A. 1981. AIAA-2010-3929. Cattafesta.: “Liquid fueled ramjet combus- tion instability: Acoustic and vortical interactions with burning sprays”. I.. Stockholm.. 51.: “Error analysis of spectral estimates with ap- plication to the measurement of acoustic parameters using random sound fields in ducts”.A.: “Orifice flow”.J.: “Investigation of the effect of combustor cooling geometry on acoustic energy absorp- tion”. Soenarko. accessed online 27 February 2014: http://www. Berlin.. 1990..) [438] Seybert.com/press/pool/de/pressemitteilungen/2011/ fossil_power_generation/EFP201105064e. 67. 2013. 1907. Journal of the Acoustical Society of America.. J. F. 7-9 June 2010. Proceedings of the Institution of Mechanical Engineers. (Cited on pages 28. Press Re- lease. A. 45.F. D.: “Uncertainty analysis of the two-microphone method”.. Bake.N. A. 69(4):1190–1199. Mathematische Annalen.F. Journal of the Acous- tical Society of America.. Sweden. in 19th AIAA/CEAS Aeroacoustics Conference. L. W. University of California. (Cited on page 146. in 16th AIAA/CEAS Aeroacoustics Conference. 63. 61(5):1362–1370.

Atlanta. G. 114.: “On the effect of the internal friction of fluids on the motion of pendulums”. B. (Cited on page 121.) [447] Stewart.P.) [452] Stow.R. 8:287–319. E. 1886. and 131.G. S. M. (Cited on pages 48 and 80.R.G. 1987. Acta Acoustica United with Acoustica. S. GT2003-38168. Transactions of the Cambridge Philosophical Society. GT2001-0038. in ASME Turbo Expo 2003. A.: “Über Schallgeschwindigkeit in Luft bei gewöhnlicher und bei hoher Temperatur und in verschiedenen Dämpfen”. (Cited on page 38.W. A. Louisiana. (Cited on pages 97. in ASME Turbo Expo 2001. 1902..R. (Cited on page 190.: “Modelling of circumferential modal coupling due to helmholtz resonators”. 1851. 2003.: “The propagation of plane waves in narrow and wide circular tubes. H. Chapman & Hall. New York..G.) [445] Stevens.: “On the theories of the internal friction of fluids in motion.) [448] Stinson. 114. and through open conduits and pipes. (Cited on pages 101 and 103. over weirs. Lindsay. and 131. John Wiley and Sons. New Orleans.H. 2001. 312(2):285–320. 89(2):550–558. G. . R. (Cited on page 131.) [449] Stokes. 1931.W.) [453] Stow. June 16-19. G. 9(2):8– 106.) [446] Stewart. G.: “Experimentelle Untersuchung zur Schallerzeugung durch die Turbulenz in einer Rohrströmung hinter einer unstetigen Querschnittser- weiterung”.: Acoustics: A text on theory and applications. Journal of the Acoustical Society of America. USA. 1991. 1:305– 317. and of the equilibrium and motion of elastic solids”.) [451] Stokes. G. (Cited on page 58. Philosophical Magazine Series 4. (Cited on pages 62. 4-7 June 2001.. 1992. London.: “An examination of the possible effect of the radiation of heat on the propagation of sound”. Technical Report TR-92-31 and TR-2855. 1851.) [450] Stokes.: Hydraulics .) [444] Stahl. and generalization to uniform tubes of arbitrary cross- sectional shape”.P. GA. (Cited on page 145. Dowling. University of Maryland. 1845. Transactions of the Cambridge Philosophical Society.The flow of water through orifices. Dowling. USA.B. 63(1):42–59.338 bibliography [443] Smith.: “Thermoacoustic oscillations in an annu- lar combustor”. Annalen der Physik. 2003.: “On the early history of the singular value decomposi- tion”.

2004. J. [455] Stow.: “A time-domain network model for nonlin- ear thermoacoustic oscillations”. Feng. 2009. J..: Fluid Mechanics.. 1889. A. (Cited on pages 48 and 80. J. bibliography 339 [454] Stow. Chen. (Cited on page 82.: “Effect of grazing-bias flow inter- action on acoustic impedance of perforated plates”. 3rd ed. J. 2002. Messenger of Mathematics.) [457] Strouhal. X. 66(3):772–778.) [465] Sylvester. Theory”. 1982. and 222. J. Journal of the Acoustical Society of America. (Cited on pages 54. Crocker. Applications”.W. Journal of Engineering for Gas Turbines and Power.) [460] Sullivan.) [461] Sullivan. V. 1962. J. . H. K.. Y.W. 67. Jing. J. abstract only.R. 1878. Journal of the Acous- tical Society of America. 267.. Journal of the Acoustical Society of America.: “A new proof that a general quadratic may be reduced to its canonical form (that is. McGraw-Hill.: “Ueber eine besondere Art der Tonerregung”. a linear function of squares) by means of a real orthogonal substitution”.J. H. Dowling.J.) [456] Streeter.) [458] Sullivan..: “Analysis of concentric-tube resonators having unpartitioned cavities”. S. 71:221– 235. Journal of Sound and Vibration. (Cited on page 271.P. Shi.W. 2004. 19:42–46.J.L.) [463] Sun. A. II. (Cited on page 82.: “Low-order modelling of thermoacoustic limit cycles”.. 254(3):557–573. New York. 131:1–10.R. 1978. Vienna.P.. and 82. V..) [459] Sullivan. 71:S78.. Journal of the Acoustical Society of America.: “Sound absorbing characteristics of fibrous metal materials at high temperatures”. X. Applied Acoustics. S. 5(10):216–251. I.: “A method for modeling perforated tube muffler compo- nents. 221. 2010. Annalen der Physik und Chemie. 66(3):779–788.. (Cited on pages 26.. (Cited on pages 44. 1889. (Cited on page 37. Austria.) [462] Sun. GT2004-54245. Dowling. M. Doige.: “On the reduction of a bilinear quadric of the n-th order to the form of a sum of n products by a double orthogonal substitution”. in ASME Turbo Expo 2004. Zhang.: “A method for modeling perforated tube muffler com- ponents.W.G. (Cited on page 145.) [464] Sylvester. June 14-17. 19:1–5. 1979. 1979. Messenger of Mathematics. and 268. F. Wu. 64(1):207–215.: “Measurement of acoustic impedance of per- forated plates in the presence of mean through flow”. (Cited on page 53. A.

121... in 15th AIAA/CEAS Aeroacoustics Conference. 1979.. Raitor. (Cited on pages 28. 11-13 May 2009. Kurbatskii.. Florida. Kurbatskii. K. Lambert.: “Nonlinear prop- erties of circular orifices”. 1965. (Cited on page 18.: “A numerical and experimental investigation of the dissipation mechanisms of resonant acoustic liners”. (Cited on page 145. H. (Cited on page 154.S. AIAA-2009-3288.W. 12-16 April 2010.. R. 245(3):545–557.W. B. 25(1):26–31. A. Martin. Paris. G. Massachusetts. (Cited on pages 32 and 217. and 202. Journal of Sound and Vibration. Enghardt. 8- 10 May 2006. Journal of the Acoustical Society of America. C.. 108:651–653. and 202.) [471] Tapken. Gaeta.F.: “Influence of shear flow on sound attenuation in a lined duct”. D. 120.) [468] Tam.) [470] Tapken. Journal of the Acoustical Society of America. Comptes Rendus de l’Académie des Sciences. Miami. U.H.: “On the propagation of sound waves in cylindrical tubes”.) [476] To.) [469] Tam.: “Sur la réduction biorthogonale d’une forme linéo-linéaire à sa forme cannonique”. 2006. 45. U. 29(9):992–1001.D. L. in 12th AIAA/CEAS Aeroacoustics Conference. and 122.B.) [467] Tack. J. 45. Cook.: “A transient testing technique for the determina- tion of matrix parameters of acoustic systems.K.. R. (Cited on pages 148 and 166. L. 38(8):1331–1339.B. 39(1):1–33.K.) [475] Tijdeman.) [473] Thurston. K.: “Periodic fluid flow through cir- cular orifices”. in 10éme Congrés Francais d’Acoustique. (Cited on pages 28. Enghardt. Lyon. (Cited on page 65. J. P. J. 1889..) [474] Thurston. Hargrove.: “Microfluid dynamics and acoustics of resonant liners”. AIAA Journal.A. 2009. 1975.A.J. Ahuja. 1953. Logan E.: “Hole interaction effects under high and medium sound intensities for micro-perforated panels design”. (Cited on pages 113..340 bibliography [466] Sylvester.. 1957. Journal of Sound and Vibration.) [472] Tayong. C.) . 38(4):655–666.: “Optimization of sensor arrays for radial mode analysis in flow ducts”. (Cited on page 148.G.: “Tonal noise radiation from an UHBR fan . J. 2000.. C. Journal of Sound and Vibration.W.K. France. Journal of the Acoustical Society of America. 2010. Doige. Leclaire.. T. Cambridge. K. R. AIAA-2006-2638. G. (Cited on page 18. 2001. J. Charles E. 62(2):207–222.Optimized in-duct radial mode analysis”.. I: Theory and principles”.

3(7):1039–1045.H..W.. 38. Journal of Spacecraft and Rockets.H. W. 104.A. Journal of Sound and Vibration.W.. 2002. Pruss. 62(2):223–233. (Cited on pages 44 and 53.: “The IAPWS formulation 1995 for the thermody- namic properties of ordinary water substance for general and scientific use”. AIAA-1965-0585.G. AIAA-1970-0128.. 32. W. II: Experimental procedures and results”. J. 1970.) [486] Wagner.T. in AIAA Propulsion Joint Specialist Conference. SAE Transaction.) [482] Tyler.: “Computational aeroacoustic characterization of different orifice geometries under grazing flow conditions”. and 269. T. (Cited on pages 53 and 65. Roeck. D. 31(2):387–535.. 2009. W. 791. Hirschberg. 19-21 January 1970.W. Doige. A... and 55. Schuller. D. T. H.J. 1993. 50...: “Evaluation of absorption lin- ers for suppression of combustion instability in rocket engines”. 27. Denayer.”. and 53.. A. Ducruix. Sirignano. H. Moers..: “Passive control of the inlet acoustic boundary of a swirled burner at high amplitude combustion instabilities”.. NY. 19.) [484] Utvik. W. T. Blackman. 44. Journal of Physical and Chemical Reference Data. (Cited on pages 25. in 8th AIAA Aerospace Sciences Meeting. 70:309–332. (Cited on page 134. 1979. 1965.: “Axial flow compressor noise studies”. 131:1–7.) [479] Tonon.S. D. (Cited on page 154. Journal of Physical and Chemical Reference Data. Journal of Engineering for Gas Turbines and Power. Revised according to the international temperature scale of 1990. (Cited on pages 18.J. Ford. H. in Proceedings of ISMA2012/USD2012.: “A transient testing technique for the determina- tion of matrix parameters of acoustic systems. New York..M. Sofrin.: “Evaluation of absorption lin- ers for suppression of combustion instability in rocket engines”.) [478] Tonon. A.M. Journal of Sound and Vibration. A.S. 26. 55.D. 50.) [485] Wagner. W... 203. 1962. 332:1654–1673. 2012. E.: “International equations for the saturation prop- erties of ordinary water substance. N. (Cited on page 134.: “Quasi-steady acoustic re- sponse of wall perforations subject to a grazing-bias flow combination”.) [480] Toulorge. T. Desmet. A.) . C.) [483] Utvik. 1966. S. bibliography 341 [477] To. Ford. 28.) [481] Tran. A. Colorado Springs. (Cited on pages 19 and 32. 34. Pruss. 202. 22(3):783–787.: “The nonlinearity of acoustic liners with flow effects”. Col- orado. Blackman.. (Cited on pages 26. 2013.G. 14-18 June 1965. (Cited on pages 101. and 106.

Ahuja. C. (Cited on page 140. 53. (Cited on page 145. Journal of the Acoustical Society of America. (Eds. Proceedings of the Physical Society.: “Acoustical impedance and the theory of horns and of the phonograph”.E. T. and 121. (Cited on pages 5 and 323. 1953. Massachusetts Institute of Technology.: “The theory of the propagation of plane sound waves in tubes”. F.) [490] Weng. W. 2nd ed. P.J. (Cited on pages 113. Anderson. 66:695–709. Mathematische Annalen. Y. 1995. K.: “The attenuation of sound by turbulence in internal flows”. Journal of the Acoustical Society of America..) [489] Wendoloski. 133:3764–3776.) [492] Weston. S.) [495] White. Proceedings of the Royal Society A. modified periodi- grams”.M. 2007.d. 5(7):275–282. 86(1):99–105. and 161. and 251.: Fluid Mechanics.. McGraw-Hill. M. McGraw-Hill. T. IEEE Transactions on Audio and Electroacoustics.) [497] Yang.) . V.D.) [494] Whiffen. 1998. 1991. 1967. Hanifi.M.K.C..) [498] Yazaki. 115. Journal of Sound and Vibration.. (Cited on page 126. (Cited on page 33. Boij.: The interaction of a finite amplitude acoustic wave with small obstacles and orifices. 1919. 1912. A.: Viscous Fluid Flow. 15(2):70–73.: “An improved schlieren system and some new results on acoustically excited jets”. AIAA. 463:2855–2862. A.) [491] Westervelt. 2013.. F. 104(1):122–132.) [496] White.: “The use of fast fourier transform for the estimation of power spectra: a method based on time averaging over short.C. Ph.342 bibliography [487] Webster.. H. thesis. 1998. 1983. (Cited on page 54. New York. (Cited on pages 114.. Tashiro.): Liquid rocket engine combustion instability.: “Das asymptotische Verteilungsgesetz der eigenwert linearer partieller Differentialgleichungen (mit Anwendung auf der Theorie der Hohlraumstrahlung)”. 4th ed. Section B. and 71. (Cited on page 60. P.. Reston. J.) [488] Welch. Biwa.: “Sound absorption by an orifice plate in a flow duct”. 54. 1951. VA. (Cited on pages 44.G. (Cited on pages 47. (Cited on page 113. 71:441–479.) [493] Weyl. 120. D. Proceedings of the National Academy of Sciences of the United States of America.: “Measurements of sound propagation in narrow tubes”.

New York. H. 2009. (Cited on pages 28.. (Cited on pages 113.: “Combustion instabilities: Basic concepts”. Journal of Sound and Vibration. Chap.) .: “Time-domain characterization of the acoustic damping of a perforated liner with bias flow”. Kosten. 74.. 2011.) [507] Zorumski.: “Tuned passive control of perforated liners”. 13(3):347–356. and 256. (Cited on pages 50 and 55. L. 11-13 May 2009. Miami. 1999. Morgans.. in 15th AIAA/CEAS Aeroacoustics Conference. No. 1. Berlin. (Cited on page 65.W. 47/2012. (Cited on page 55. Florida. 68. 1949. published 15 November 2012. (Cited on page 2. B. D. accessed online 27 February 2014: http://www. 2012. W. D.. A. Journal of the Acoustical Society of America. AIAA-2013-2128.. and 123..: “Tuned passive control of acoustic damping of perforated liners”. 45. Zhao.: Sound absorbing materials.S. A. (Cited on pages 16.de/2012/47/ Energiewende-Deutsche-Begriffe-Englisch. 1970.) [503] Zhong. 65. 1971.: “The effect of combined high level acoustic exci- tation and bias flow on the acoustic properties of an in-duct orifice”. 2005. 67.) [501] Zhao. Bodén. AIAA-2009-3406. Lieuwen.C. in 19th AIAA/CEAS Aeroacoustics Conference. B. Technical Note NASA-TN-D-6196. 27-29 May 2013.T. 37(7):825–831.: “Nonlinear acoustic theory for rigid porous materials”. H. Elsevier. D.zeit.P. Germany. C. bibliography 343 [499] Zeit Online: “The Energiewende”.L.) [500] Zhao.) [505] Zinn. T. NASA. 3–26.) [508] Zwikker.S. Parrott. Dowling. AIAA Journal.) [506] Zinn. in [305]. Morgans. pp..T. (Cited on page 55. X.E.) [504] Zhou. (Cited on page 5. 132(1):271–281.. C. 122.) [502] Zhao. T.: “A theoretical study of non-linear damping by helmholtz res- onators”. and 202. 2013. AIAA Journal..: “Active control of wall acoustic impedance”. Z. 55. A. 49(4):725–734. (Cited on page 54. Sun.