Вы находитесь на странице: 1из 9

Claus Lahiri

Establishment of a High Quality


Lars Enghardt Database for the Acoustic
Friedrich Bake Modeling of Perforated Liners
Institute of Propulsion Technology,
German Aerospace Center (DLR), Perforated liners, especially in combination with a bias flow, are very effective sound
10623 Berlin, Germany absorbers. When appplied to gas turbine combustors, they can suppress thermo-acoustic
instabilities and thus allow the application of new combustion concepts concerning
higher efficiency and lower emissions. While the successful application of such a damp-
Sermed Sadig ing concept has been shown, it is still not possible to accurately predict the damping
performance of a given configuration. This paper provides a comprehensive database of
Miklós Gerendás high quality experimental data. Variations of geometric, fluid mechanic, and acoustic
parameters have been studied, including realistic engine configurations. The results dem-
Combustor Aerothermal and Cooling,
onstrate each parameter influence on the damping performance. A low order thermo-
Rolls-Royce Deutschland Ltd. & Co. KG,
acoustic model is used to simulate the test configurations numerically. The model shows
15827 Dahlewitz, Germany
a good agreement with the measurements for a wide range of geometries and Strouhal
and bias flow Mach numbers. 关DOI: 10.1115/1.4002891兴

1 Introduction is illustrated in Fig. 1. It allows high precision acoustic measure-


ments of the damping performance of various liner configurations,
In order to achieve low emissions in a combustion system, the
including grazing and bias flow.
control of thermo-acoustic combustion instabilities is essential.
The test duct consists of two symmetric measurement sections
These instabilities arise from a feedback of acoustic pressure pul-
共section 1 and section 2 in Fig. 1兲 of 1200 mm length each. They
sations on heat release fluctuations. This phenomenon can be sup-
have a circular cross section with a diameter of 140 mm. In order
pressed by increasing the acoustic losses in the system. One pos-
to minimize the reflection of sound at the end of the duct back into
sibility is the use of Helmholtz resonators, which provide high
the measuring section, the test duct is equipped with anechoic
acoustic damping but only within a small bandwidth. Another
terminations at both ends 共not shown in Fig. 1兲. Their specifica-
approach is to use the broadband damping abilities of perforated
tions follow the ISO 5136 standard.
liners. While perforated liners are already installed in combustion
A total of 12 microphones are mounted flush with the wall of
chambers, they are optimized for cooling purposes and not neces-
the test duct. They are installed at different axial positions up-
sarily for acoustic damping. In recent years, the implementation of
stream and downstream of the damping module and are distrib-
modern combustion concepts, e.g., lean combustion, has increased
uted exponentially with a higher density toward the damping
the demand of further understanding and optimization of the
module. Two microphones are installed opposite each other at the
damping. Still, difficulties arise in the prediction of the damping
same axial position close to the signal source. As evanescent
performance of a given configuration.
modes become more prominent in the vicinity of the source, their
Theoretical analyses of the damping effect have been given in
influence is reduced significantly by using the average value of
Refs. 关1–3兴. In Ref. 关1兴, Howe developed an expression of the
these two microphones for the analysis. This technique helps re-
Rayleigh conductivity of a single orifice to give its acoustic be-
duce the errors for frequencies approaching the cut-on frequency
havior. Later, Eldredge and Dowling 关4兴 developed a method to
of the first higher order mode and thus extends the frequency
predict the damping characteristic of axial acoustic waves within a
range for accurate results.
cylindrical tube based on Howe’s definition of the Rayleigh con-
At the end of each section, a loudspeaker is mounted at the
ductivity. This approach was used by Macquisten et al. 关5兴 to
circumference of the duct 共A and B in Fig. 1兲. They deliver the test
define dampers and compare them with experimental results.
signal for the damping measurements. The signal used here is a
Based on these previous works, a one dimensional acoustic net-
multitone sine signal. All tonal components of the signal are in the
work tool was developed by Stow and Dowling 关6,7兴.
plane wave range. The signal has been calibrated in a way that the
Here, a comprehensive database of high quality experimental
amplitude of each tonal component inside the duct is about 102
data is established. The parameters include number, size, shape
dB.
and distribution of the orifices, wall thickness, cavity volume in a
The microphones used in these measurements are 1/4 in. 共6.35
double layer configuration, bias and grazing flow velocities, and
mm兲 GRAS type 40BP condenser microphones. Their signals are
signal frequency. The configurations are tested in an isothermal
recorded with a 16 track OROS OR36 data acquisition system
duct acoustic test rig providing a high accuracy of the results. The
with a sampling frequency of 8192 Hz. The source signals for the
database is used to test the prediction capability of the one dimen-
loudspeakers are recorded on the remaining tracks. The test signal
sional acoustic model proposed by Stow and Dowling.
is produced by an Agilent 33220A function generator. The signals
are fed through a Dynacord L300 amplifier before they power the
2 Experimental Setup Monacor KU-516 speakers.
The experimental study is performed in the duct acoustic test
rig at DLR Berlin at ambient conditions. The setup of the test rig
3 Experimental Analysis
For each configuration, two different sound fields are excited
Contributed by the International Gas Turbine Institute of ASME for publication in
the JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER. Manuscript received
consecutively in two separate measurements 共indices a and b兲.
August 5, 2010; Final manuscript received September 24, 2010; published online Speaker A is used in the first measurement and in the second
April 20, 2011. Editor: Dilip R. Ballal. measurement the same signal is fed into speaker B. Then, the data

Journal of Engineering for Gas Turbines and Power SEPTEMBER 2011, Vol. 133 / 091503-1
Copyright © 2011 by ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Damping
Section 1 Module Section 2
bias
A 1 2 3 4 5 flow 7 8 9 10 11 B

grazing flow ⇒

6 - +
12
x=0 x=0

Fig. 1 Schematic setup of the duct acoustic testrig with speakers A and B and microphones
1–12. The anechoic terminations at both ends are not shown.

of section 1 and section 2 共indices 1 and 2兲 are analyzed sepa- and B. In order to calculate the reflection and transmission coef-
rately. This results in four equations for the complex sound pres- ficients r+, r−, t+, and t− from the sound pressure amplitudes, the
sure amplitudes for each section and measurement: following four relations can be derived:
+ −
p̂1a共x兲 = p̂+1ae−ik1 x + p̂−1aeik1 x 共1兲 p̂−1a = r+ p̂+1a + t− p̂−2a 共6兲
−ik+2 x ik−2 x
p̂2a共x兲 = p̂+2ae + p̂−2ae 共2兲 p̂−1b = r+ p̂+1b + t− p̂−2b 共7兲
+ −
p̂1b共x兲 = p̂+1be−ik1 x + p̂−1beik1 x 共3兲 p̂+2a = r− p̂−2a + t+ p̂+1a 共8兲
−ik+2 x ik−2 x
p̂2b共x兲 = p̂+2be + p̂−2be 共4兲 p̂+2b = r− p̂−2b + t+ p̂+1b 共9兲
+ −
p̂ and p̂ are the complex amplitudes of the downstream and The equations from both measurements are combined and solved
upstream traveling waves with their respective wave numbers k⫾. for the reflection
The recorded microphone signals are transformed into the fre-
quency domain using the method presented by Chung 关8兴. This p̂−1a p̂−2b − p̂−1b p̂−2a p̂+2b p̂+1a − p̂+2a p̂+1b
method rejects uncorrelated noise, e.g., turbulent flow noise, from r+ = r− = 共10兲
the coherent sound pressure signals. Therefore, the sound pressure p̂+1a p̂−2b − p̂+1b p̂−2a p̂+1a p̂−2b − p̂+1b p̂−2a
spectrum of one microphone is determined by calculating the and transmission coefficients
cross-spectral densities between three signals, where one signal
serves as a phase reference. In our case, the phase reference signal p̂+2a p̂−2b − p̂+2b p̂−2a p̂+1a p̂−1b − p̂+1b p̂−1a
is the source signal of the active loudspeaker. As a result, we t+ = t− = 共11兲
p̂+1a p̂−2b − p̂+1b p̂−2a p̂+1a p̂−2b − p̂+1b p̂−2a
obtain a phase-correlated complex sound pressure spectrum for
each microphone signal. In the downstream and upstream directions, respectively. The ad-
According to Eqs. 共1兲–共4兲, the measured acoustic signal is a vantage of combining the two measurements is that the resulting
superposition of two plane waves traveling in opposite directions. coefficients are independent from the reflection of sound at the
In order to determine the downstream and upstream propagating duct terminations. These end-reflections are contained in the
portions of the wave in each section, a mathematical model is sound pressure amplitudes but do not need to be calculated ex-
fitted to the acoustic microphone data. This model considers vis- plicitly.
cous and thermal conductivity losses at the duct wall. They are The dissipation of acoustic energy is expressed by the dissipa-
included in the wave number with the following attenuation factor tion coefficient. The dissipation coefficient can be calculated di-
␣ as proposed by Kirchhoff 关9兴: rectly from the reflection and transmission coefficients via an en-

冑 冉 冊
ergy balance:
1 ␩␻ ␥−1
␣= 1+ 共5兲 R⫾ + T⫾ + ⌬⫾ = 1
cr 2␳ 冑Pr 共12兲

with the duct radius r, the speed of sound c, the dynamic viscosity The energy of the incident wave is partly reflected, partly trans-
␩, the angular frequency ␻, the density of the fluid ␳, the heat mitted, and partly absorbed by the damping module. R and T are
capacity ratio ␥, and the Prandtl number Pr. As a result of this the power quantities of the reflection and transmission coeffi-
least-mean-square fit, the four complex sound pressure amplitudes cients, respectively, while r and t are the pressure quantities.
p̂+1 , p̂−1 , p̂+2 , and p̂−2 are identified at position x = 0 for both measure- Blokhintsev 关10兴 defined the acoustic energy flux I in a moving
ments. These sound pressure amplitudes are related to each other medium 共see as well in Ref. 关11兴兲
via the reflection and transmission coefficients of the test object. 1
This is illustrated in Fig. 2 for the two different measurements A I= 共1 + M兲2具p2典 共13兲
␳c
where 具p典 is the time-averaged acoustic pressure, ␳ is the density
A B
(a) (b) of the medium, c is the speed of sound, and M is the mean Mach
p̂+
1a t+ p̂+
2a p̂+
1b t+ p̂+
2b
number. Integrating over the duct cross section area A and using
re− r+ r− re+ re− r+ r− re+ the pressure amplitude yields a relation between the acoustic pres-
p̂−
1a t− p̂−
2a p̂−
1b
t− p̂−
2b
sure p and acoustic power P quantities:
Section 1 Section 2 Section 1 Section 2
A
P⫾ = 共1 ⫾ M兲2兩p̂⫾兩2 共14兲
Fig. 2 Illustration of the sound filed in the duct for measure- 2␳c
ments A and B by means of the sound pressure amplitudes p̂,
the reflection coefficient r, the transmission coefficient t, and Then, the energy coefficients can be given relative to the pressure
the end reflection re coefficients as

091503-2 / Vol. 133, SEPTEMBER 2011 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


6 These results prove the high accuracy and the reliability of the
M=0 experimental data.
5 M = 0.05
Dissipation Error [%] M = 0.1
5 Thermoacoustic Network Model
4
A low order one dimensional thermo-acoustic network model is
3 used to simulate the experimental setup of the DLR duct acoustic
test rig numerically. This model is based on the linear theory of
2 sound propagation and is able to predict combustion instabilities
with linear and nonlinear flame transfer functions in the frequency
1
and the time domain. A detailed description of the low order
thermo-acoustic network model was published in Refs. 关6,7兴. A
0
one dimensional thin annular geometry is assumed, i.e., axial and
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 circumferential modes will be regarded. Radial variations are not
kr represented. This simplification is valid either in an axially long
geometry, which is typically for industrial combustors, or in a
Fig. 3 Error of the dissipation coefficient in the experimental narrow annular cross section, which is typically for aero-engines.
results at three grazing flow Mach numbers The details of the geometry can be represented in acoustic mod-
ules. The acoustic network tool offers a wide range of modules,
e.g., straight ducts, area changes, combustion zones, Helmholtz
P−1 共1 − M 1兲2 + 2 resonators, and damping holes. The main modules to represent the
R+ = = · 兩r 兩 共15兲 test rig configuration are the straight duct and the damping holes.
P+1 共1 + M 1兲2
They will be described in more detail. A perfect gas is assumed
P+2 共1 + M 2兲2 − 2 and the flow variables will be split into a mean part denoted by a
R− = = · 兩r 兩 共16兲 bar and a perturbed part denoted by a prime: ␾ = ␾ ¯ 共x兲
P−2 共1 − M 2兲2
+ ␾⬘共x , ␪ , t兲.
P+2 A2 ␳1c1 共1 + M 2兲2 + 2 The straight duct module is described in detail in Ref. 关6兴. It is
T+ = = · 兩t 兩 共17兲 the main module in the acoustic network and characterizes the
P+1 A1 ␳2c2 共1 + M 1兲2 geometrical dimension. Flow features with a significant axial
length can be simulated with a straight duct. All other features are
P−1 A1 ␳2c2 共1 − M 1兲2 − 2 treated to be acoustically compact. Wave propagation is used to
T− = = · 兩t 兩 共18兲
P−2 A2 ␳1c1 共1 − M 2兲2 relate the perturbations at the start of the duct to the perturbations
where the indices 1 and 2 refer to section 1 and section 2 of the at the end of the duct.
duct as illustrated in Fig. 2, respectively. With A1 = A2, ␳1 = ␳2, The axial and circumferential perturbations can be written as a
sum of four waves: pressure and entropy waves as well as axial
c1 = c2, M 1 = M 2 = M, and Eq. 共12兲 follows the definition of the
and circumferential velocity perturbations. A thin annular geom-
energy dissipation coefficient:

冉 冊
etry is assumed with a constant cross section. A constant subsonic
共1 ⫿ M兲2 ⫾ 2 mean flow is used.
⌬⫾ = 1 − · 兩r 兩 + 兩t⫾兩2 共19兲 The damping effect of perforated liners on acoustic waves has
共1 ⫾ M兲2
been analyzed by several authors, e.g., Refs. 关4,12–14兴. In Ref.
This is an integral value of the acoustic energy that is absorbed 关4兴, Eldredge and Dowling presented a model, which is able to
while a sound wave is passing the damping module. The dissipa- predict the absorption coefficient of a screen of orifices for axial
tion coefficient is used to evaluate the damping performance of acoustic waves. The representation of the damping holes in the
the test object. thermo-acoustic network model is based on the description of a
single hole and will be discussed in more detail.
4 Accuracy of the Experimental Results Howe 关1兴 developed an expression of the Rayleigh conductivity
The test facility, the measurement techniques, and the data of a single aperture to give its acoustic behavior. This is defined as
analysis have been optimized to improve the accuracy of the re- the ratio of the fluctuating volume flux through an aperture to the
sults 关12兴. In order to provide a quantitative value for the accu- fluctuating difference in pressure across the aperture
racy, a reference measurement without a damping module has

been performed. The two sections of the test duct have been K = − ik 共21兲
joined together without the damping module in-between. For this p̂u − p̂d
configuration, the dissipation coefficient is expected to be zero, The model describes the viscous effects, which cause the shedding
i.e., no acoustic energy is absorbed. The deviation from this ideal of vortex structures from the rim of an aperture in an infinitely
value is the total error resulting from the experiments and the thin wall. These vortices dissipate the acoustic energy into heat.
analysis. The dissipation error is given as the deviation from the Other viscous effects are only considered by the adaptation of the
ideal value in percent. The downstream and upstream directions discharge coefficient. The Rayleigh conductivity in Howe’s ex-
have been combined in an arithmetic averaged error of the dissi- pression is described by the local pressure ratio, which does not
pation coefficient: take into account the propagation direction of the waves. Hence,
⌬err =
1
2
冑共⌬ref
+ 2
兲 + 共⌬ref
− 2
兲 共20兲 Howe’s model is independent of the shape of the mode. But due to
the assumptions, the network model is restricted to axial and cir-
The results at three grazing flow Mach numbers are shown in Fig. cumferential wave propagations.
3. The dissipation error is plotted over the Helmholtz number Hughes and Dowling 关13兴 extended Howe’s approach to give
He= kr formed with the duct radius r. The dissipation error is an expression of a homogeneous screen of apertures. Their de-
mostly below 1%, with a slight increase at the lower frequencies. scription is assuming a large distance s between neighboring ap-
As kr approaches the cut-on frequency of the first higher order ertures compared with their diameter 2R and an acoustic wave-
mode 共kr = 1.84兲, the error increases rapidly. This is due to the length ␭, which is much larger than the interaperture distance.
influence of evanescent modes that are becoming more prominent In the case of a single aperture and based on Ref. 关1兴, the
close to the cut-on frequency. Rayleigh conductivity can be expressed by

Journal of Engineering for Gas Turbines and Power SEPTEMBER 2011, Vol. 133 / 091503-3

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


l la that the length of the perforation l p is given by the longest axial
lp
distance between two orifices. For a regular perforation, the po-
rosity is defined as the ratio of the area of one orifice to a norm
θ θ area around it. The size of the norm area is given by the distance
in axial sx and circumferential s␪ directions between two adjacent

orifices. The dotted lines in Fig. 4共a兲 indicate the norm areas. One
(a) sx x 2R (b) x norm area is highlighted in gray and the corresponding orifice area
is shown in black. The result is a normalized porosity independent
Fig. 4 Definition of the perforation parameters of the liner length: ␴ = ␲R2 / sxs␪. This definition is called open
area ratio. Here, normalized porosity used as open area ratio might
be confused with the ratio of the open areas of two liners in a
Ka = 2R共␥ + i␦兲 共22兲 double skin configuration.
where ␥ is the inertia of the aperture and ␦ denotes the resistance Two configurations 共10 and 13兲 have an irregular perforation
of the flow, which determines the acoustic losses. Both coeffi- pattern 共see exemplary illustration in Fig. 4共a兲兲. As the norm areas
cients depend on the Strouhal number associated with the convec- have different sizes, the definition above cannot be applied here.
tion velocity of the vortices shed from the aperture. Howe ap- Instead, an active length la is defined. The ratio of the total open
proximated this velocity with the half of the mean velocity area to the active area defines an averaged normalized porosity for
through the aperture the irregular perforation: ␴irr = n␲R2 / la2␲r, where n is the number
of orifices, R is the radius of the orifices, and r is the duct radius.
␻·R For a regular perforation pattern, the average normalized porosity
St = 共23兲
v̄/2 coincides with the normalized porosity. Table 1 lists the averaged
normalized porosity.
In this case, it is assumed that the discharge coefficient of the hole
It should be noted that it is possible for a large sx to have the
is CD = 0.5.
active length la become longer than the actual length l. This can
In this expression the effect of the thickness of the liner is not
happen for both regular and irregular perforation patterns.
represented. This can be done by extending Eq. 共22兲 with the liner
The geometric parameters of the configurations are given in
thickness t as proposed by Jing and Sun 关15兴 and applied by Stow
Table 1. Configurations 3 and 4 are reference configurations. Their
and Dowling 关6兴
porosity is kept constant while the orifice diameter and the num-
Ka ber of orifices are modified. Basically, configuration 4 has fewer
K= 共24兲 but larger orifices compared with configuration 3. Both configu-
t
1 + Ka rations have a regular perforation pattern with cylindrical orifices
␲R2 in a single skin arrangement.
The same approach but extended for a perforated liner was used in Configuration 5 keeps the orifice diameter from configuration 4
Ref. 关4兴 to derive an integrated module of damping screens. A and the number of orifices from configuration 3. These relations
comparison of both models has proven to give the same results. are inverted in configuration 6, keeping the orifice diameter of
The presented approach is only capable of describing the linear configuration 3 and the number of orifices of configuration 4.
sound absorption. Nonlinear effects as the saturation of energy Naturally, the porosity changes between these configurations.
losses at high amplitudes as described by Rupp 关16兴 are not cap- Configuration 7 changes the wall thickness of configuration 3
tured. The thermoacoustic network model implements Eq. 共24兲 from 1 mm to 3 mm, leaving all other parameters unchanged.
without any further adjustments. Configurations 8 and 9 apply a special geometry to the orifices
The mass flux, temperature, and pressure are specified at the themselves. In configuration 8, the standard circular orifice is re-
inlet and outlet. The mean quantities are derived from these val- placed by a cross-shaped orifice with a nearly identical open area.
ues. The boundary conditions for the acoustic perturbations at the A detail of configuration 8 is given in Fig. 5共a兲 with units in
inlet and outlet are set to be either an open end 共p̃ = AE = AV = 0兲, a millimeter. The open area yields 5 mm2 compared with the area
closed end 共ũ = AE = AV = 0兲, a semi-infinite pipe 共with no reflected of 4.91 mm2 of a circular orifice of configuration 4. Configuration
acoustic waves and AE = AV = 0兲, or a chocked inlet 共m̃ = 0兲. These 9 has a circular shape, but both edges are chamfered with a radius
boundary conditions apply to all circumferential and axial waves. of 0.5 mm. A section of one orifice is shown in Fig. 5共b兲.
Single and multiflow paths are possible. Flows may split from In configurations 10 and 13, the perforation is irregular, yield-
the main path or may originate from secondary inlets. They can ing a nonuniform porosity. Configuration 10 varies the porosity in
terminate at dead ends or join into other paths. Holes can be the circumferential direction. The porosity changes symmetrically
included between paths. from high porosity at ␪ = 0 to low porosity at ␪ = 180. The porosity
Equation 共24兲 does not contain a term to couple the heat release distribution in the axial direction is kept constant. The averaged
in the combustor with the Rayleigh conductivity of the hole and normalized porosity of this configuration matches with configura-
hence with the dissipation characteristic. To assess the impact of tion 4. In configuration 13, the porosity varies along the axial
the heat release on the dissipation in a real combustor is a com- length. This is illustrated exemplarily in Fig. 4共b兲. The variation is
plex topic of its own and is not addressed in this paper. symmetric to l / 2 with a high porosity in the middle of the liner
and a lower porosity toward the edges. To enable a more effective
axial variation, this configuration is nearly 6 times as long as the
6 Configurations other configurations. This explains the low porosity and large vol-
A total of 231 combinations of geometric and flow parameters ume stated in Table 1.
are studied to establish a diversified database. Eleven liner con- Configurations 11-3 and 12-3 are double skin arrangements
figurations are studied at seven bias flow conditions and three with configuration 3 as the damping liner. Here, the cavity volume
grazing flow conditions. The damping performance of each com- of configuration 3 is divided by a second liner, the metering liner.
bination is measured at 24 frequencies. This creates two separate volumes, one between the damping liner
Figure 4 illustrates the perforation parameters for a regular 共Fig. and the metering liner and the other one between the metering
4共a兲兲 and an irregular 共Fig. 4共b兲兲 pattern with the axial coordinate liner and the back-wall of the cavity. The diameter of the metering
x and circumferential coordinate ␪. The length l is the length of liner is modified between configurations 11-3 and 12-3 so that the
the liner defined by the length of the liner’s cavity. In most cases, volume of the damping liner is small for configuration 11-3 and
the perforation is not spanning over the total length of the liner, so large for configuration 12-3. As the diameter of the back-wall is

091503-4 / Vol. 133, SEPTEMBER 2011 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 1 Overview of each configurations geometric parameters

Single skin Double skin


Orifice Number of
diameter orifices Porosity Porosity
number of orifice Wall Cross- Chamfered ␪- x- Small Large
Ref. 1 Ref. 2 orifices diameter thickness shape 共R = 0.5兲 variation variation volume volume
Configuration 3 4 5 6 7 8 9 10 13 11-3 12-3

Damping Liner

Orifice diameter 2R mm 1 2.5 2.5 1 1 Ⳏ2.5 2.5 2.5 1 1 1


Number of orifices n — 364 60 364 60 364 60 60 60 364 364 364
Porosity ␴ % 1.1 1.0 6.8 0.2 1.1 1.0 1.0 1.0 0.2 1.1 1.1
Wall thickness t mm 1 1 1 1 3 1 1 1 1 1 1
Cavity volume V cm3 1757 1757 1757 1757 1703 1757 1757 1757 8201 280 780

Metering Liner

Orifice diameter 2R mm 1 1
Number of Orifices n — 108 108
Porosity ␴ % 0.4 0.3
Wall thickness t mm 1 1
Cavity volume V cm3 1447 941

kept constant, the second volume also changes with the diameter offset between them. The energy absorption is higher for waves
of the metering liner. The open area of the two metering liners is propagating against the flow direction 共see Eq. 共19兲兲. For clarity,
kept constant, but the porosity changes due to the modified liner only the results for ⌬− are presented in this paper.
diameter. As the numerical model is limited to describe linear acoustic
The bias flow is defined by the pressure ratio across the liner in behavior, care has to be taken in choosing the pressure amplitudes.
%. The pressure ratio is varied over a wide range from high values As mentioned by Rupp 关16兴, there is a transition point into non-
of 3% to 0%. The latter means that there is no additional air linear absorption, which corresponds to a velocity approaching
supplied through the liner orifices. The air supply is controlled via zero for parts of the unsteady cycle in the plane of the orifice.
the mass flow rate, which means the pressure ratios can vary a bit Relating to the pressure level used in this investigation, nonlinear-
between configurations. The approximate pressure ratios are 0%, ity can be expected for pressure ratios smaller than 0.02%.
0.01%, 0.05%, 0.1%, 0.2%, 0.3%, and 3%. However, the static Figure 6 presents the influence of the bias flow on configuration
pressure on both sides of the liner was recorded during the mea- 3 at a grazing flow M = 0.05. Dissipation curves are plotted for the
surements so that the exact values are available. A realistic pres- pressure ratios 0%, 0.1%, 0.3%, and 3%. As the model is only
sure ratio across the combustor walls of an engine is around 3%. valid for the linear range, i.e., with bias flow, no numerical results
It should be noted that in a double skin arrangement, the engine are given for dP = 0%. While the good damping performance
parameter is the total pressure ratio across both liners dPT, without bias flow is limited to a small frequency range, a broad-
whereas the damping parameter is the pressure ratio across the band damping is achieved by the bias flow configurations. The
damping liner dP. In a single skin arrangement, there is only one damping performance is better for small pressure ratios, where the
pressure ratio dP = dPT. Strouhal number of the flow decreases and the resistance of the
A typical value for the grazing flow Mach number within a flow increases. The results of the numerical model are in good
combustor is around M = 0.1. Here, three grazing flow velocities agreement with the experiments. While the damping is predicted
have been applied: M = 0, 0.05, and 0.1. quite accurately for the lower range of Helmholtz numbers, it is
slightly underestimated in the higher range, especially at the low
7 Results pressure ratio.
The limited space in this paper only allows a presentation of
some selected results. The damping performance of the selected
configurations is described by a dissipation curve. Here, the dis- 0.6
sipation coefficient is plotted over the Helmholtz number, formed dP 0%
with the length of the perforation 共see Fig. 4兲 as the characteristic 0.5 dP 0.1%
length He= kl p. The dissipation coefficient is defined in Eq. 共19兲. dP 0.3%
As all liner geometries are symmetrical in the axial direction, ⌬+ dP 3%
Dissipation ∆

0.4
and ⌬− are identical if no grazing flow is present in the duct.
However, even under the influence of the grazing flow, the dissi- 0.3
pation curves for ⌬+ and ⌬− show a similar behavior but with an
0.2

1
0.1
(a) radius = 0.5 (b)
0
1 t 0.2 0.4 0.6 0.8 1 1.2
kl
p
2R
Fig. 6 Influence of the bias flow on configuration 3, grazing
Fig. 5 Special orifice geometries flow M = 0.05. Measurements: symbols, model: lines.

Journal of Engineering for Gas Turbines and Power SEPTEMBER 2011, Vol. 133 / 091503-5

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


0.6
0.6 dP 0%
dP 0.1% 0.5 conf. 3: σ = 1.1 %
0.5 dP 0.3%
conf. 4: σ = 1.0 %

Dissipation ∆
dP 3% 0.4
conf. 5: σ = 6.8 %
Dissipation ∆

0.4
0.3 conf. 6: σ = 0.2 %

0.3
0.2
0.2
0.1

0.1 0
0.2 0.4 0.6 0.8 1 1.2
0 kl
p
0.2 0.4 0.6 0.8 1 1.2
kl
p Fig. 8 Influence of the porosity, bias flow dP = 0.2%, and graz-
ing flow M = 0
Fig. 7 Influence of the bias flow on configuration 5, grazing
flow M = 0.05. Measurements: symbols, model: lines.

A previous conclusion from Fig. 7 was the existence of an


optimum pressure ratio/bias flow velocity. Together with the find-
Figure 7 illustrates the results for configuration 5 at the same ings above, it follows that the combination of pressure ratio and
flow conditions as above. The damping behavior of configuration porosity is crucial to achieve maximum damping performance.
5 共high porosity兲 is very different to configuration 3. There is only The limiting parameters in an engine are the available mass flow
a poor absorption without bias flow. The bias flow increases the rate and the given pressure ratio.
damping performance significantly so that up to 60% of the inci- Figure 9 plots the influence of the bias flow on configuration 7
dent sound energy is absorbed. It becomes obvious that an opti- at a grazing flow M = 0.05. The liner in configuration 7 is 3 times
mum bias flow setting exists where the damping performance is at as thick as configuration 3. Comparing its results to configuration
a maximum for a wide range of Helmoltz numbers. Configuration 3 in Fig. 6 reveals the influence of the wall thickness on the
5 is able to reach maximum absorption for a range of pressure damping. The general behavior is very similar to configuration 3.
ratios 共0.1–0.3%兲. The damping performance drops for higher While their characteristic is identical at dP = 3%, the discrepancy
pressure ratios. Nevertheless, the absorption at 3% is still 3–4 grows for smaller dP. The damping optimum is at a slightly lower
times higher than for configuration 3. However, it needs to be Helmholtz number and extends over a smaller range. The ability
considered that the mass flow rate is also increased drastically. of broadband absorption is reduced by the thicker wall. This is
The maximum absorption can be found at a different Helmholtz reflected by Eq. 共24兲, where the Rayleigh conductivity decreases
number compared with configuration 3. with a thicker wall. At higher frequencies, the resistance of the
Two anomalies can be observed in Fig. 7. First, the scattering of flow decreases and the correction term becomes more pro-
the dissipation coefficient at 3% and second, the dissipation coef- nounced. This case is handled well by the model. The numerical
ficient becomes negative for high Helmholtz numbers at 0%. Both results are in good agreement for all pressure ratios and for the
have the same origin. Sound is produced by the damper. In the range of Helmholtz numbers.
first case, it is broadband noise generated by the high mass flow The influence of the orifice geometry is presented for a bias
rate through the orifices that interferes with the measurements. In flow of dP = 0% and a grazing flow of M = 0.1 in Fig. 10. The
the second case, a tone is produced at the liner by the grazing reference configuration 4 with circular-shaped, sharp-edged ori-
flow. This tone can be found in the spectrum at kl p ⬇ 1.2. This is in fices is compared with configuration 8 with cross-shaped, sharp-
the frequency range, where the dissipation coefficient becomes edged orifices and configuration 9 with circular-shaped, cham-
negative. The negative dissipation coefficient confirms that in- fered orifices. While there is no remarkable difference between the
stead of absorbing the incident sound, additional sound is gener- cross-shaped orifices and the reference configuration, the cham-
ated. fered orifices yield a slightly higher dissipation. Measurements at
The numerical results reproduce the damping characteristics other bias flow and grazing flow settings confirm this behavior. It
found by the experiments. They slightly overestimate the damping can be concluded that the shape of the orifices can be disregarded,
at low Helmholtz numbers, whereas they tend to underestimate for
higher Helmholtz number. As before, the discrepancies are larger
for lower pressure ratios.
0.6
A direct comparison of configurations 3–6 is given in Fig. 8.
dP 0%
This comparison reveals the influence of the orifice diameter, the
0.5 dP 0.1%
number of orifices, and the porosity. Configurations 3 and 4 keep
dP 0.3%
a similar porosity, ␴ = 1.1% and ␴ = 1.0%, respectively, while the dP 3%
Dissipation ∆

0.4
orifice diameter and the number of orifices are varied. The varia-
tion has little impact on the absorption, so that they yield a similar
0.3
dissipation curve. Other measurements from this study 共not
shown兲 contradict these findings and show a clear distinction be- 0.2
tween configurations 3 and 4 at pressure ratios below dP
= 0.15%. However, the influence of the diameter and number of 0.1
orifices for measurements at higher pressure ratios dP ⬎ 0.15% is
negligible and the porosity seems to be the relevant parameter. 0
Configuration 5 has a higher porosity of ␴ = 6.8%, while con- 0.2 0.4 0.6 0.8 1 1.2
figuration 6 has a lower porosity of ␴ = 0.2%. A higher porosity kl
p
yields a better damping performance. It must be considered that a
higher porosity demands a higher mass flow rate in order to Fig. 9 Influence of the bias flow on configuration 7, grazing
achieve the same pressure ratio. flow M = 0.05. Measurements: symbols, model: lines.

091503-6 / Vol. 133, SEPTEMBER 2011 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


0.6 0.6
dP 0%
0.5 0.5 dP 0.01%
dP 0.1%
dPT 3%
Dissipation ∆
0.4

Dissipation ∆
0.4

0.3 0.3

0.2 0.2
conf. 4: circular, sharp−edged
0.1 conf. 8: cross, sharp−edged 0.1
conf. 9: circular, chamfered
0 0
0.2 0.4 0.6 0.8 1 1.2 0.2 0.4 0.6 0.8 1 1.2
kLp kl
p

Fig. 10 Influence of the orifice geometry, bias flow dP = 0%, Fig. 12 Influence of the bias flow on configuration 11-3, graz-
and grazing flow M = 0.1 ing flow M = 0. Measurements: symbols, model: lines.

while the design of the orifice edge affects the dissipation. How- = 280 cm3 and the maximum absorption is assumed to be beyond
ever, the orifice edge has only a minor impact compared with the the measured range of Helmholtz numbers kl P ⬎ 1.3. Configura-
porosity or the bias and the grazing flow. tion 12-3 has a damping volume of VD = 780 cm3 and the maxi-
Configurations 10 and 13 have a nonuniform porosity. In con- mum damping appears at kl P ⬇ 0.8. These configurations can be
figuration 10, the porosity varies around the circumference of the compared with configuration 3 with a damping volume of VD
liner. Figure 11 compares this configuration to the reference con- = 1757 cm3 and an absorption maximum at kl P ⬇ 0.5. The volume
figuration 4 with uniform porosity. Both configurations produce seems directly related to the Helmholtz number at which maxi-
similar results, except that there is a peak in the dissipation curve mum damping occurs. A larger volume shifts the maximum to
at kl p = 0.55 for the nonuniform case. This peak corresponds to the lower Helmholtz numbers. However, the different bias flow rates
cut-on frequency of the first azimuthal mode inside the damping do not affect the Helmholtz number of the maximum.
volume. The azimuthal mode is excited by the azimuthal structure The results of the double skin arrangements without bias flow
of the perforation and leads to a resonance effect at its cut-on 共dP = 0%兲 reveal a specific feature: A second damping maximum
frequency. appears. This second maximum is related to the second volume
In configuration 13, the porosity varies along the axial length of behind the metering liner. The low porosity and therefore high
the liner with a high porosity in the middle and lower porosities pressure ratio of the metering liner suppresses acoustic communi-
toward the edges 共see Fig. 4共b兲兲. Due to its extended length, a cation through its orifices already at a small pressure ratio of the
comparison to the other configurations is difficult. Only one dis- damping liner. The maxima corresponding to the metering volume
tinctive feature should be discussed without respective plot. The are at kl P ⬇ 0.25 and kl P ⬇ 0.3 for configuration 11-3 with V M
dissipation curve shows two peaks similar to the resonance effect = 1447 cm3 and configuration 12-3 with V M = 941 cm3, respec-
in Fig. 11. Here, an axial resonance inside the damping volume is tively. This demonstrates the same trend that a larger volume gen-
responsible. The frequency of the first peak corresponds to a erates damping at lower Helmholtz numbers. However, it does not
␭ / 2-resonance and the second peak to a ␭-resonance. This effect follow the sequence found for the damping volume. The damping
cannot be observed in the other configurations. They are much volume of configuration 3 is larger than the metering volumes in
shorter, so that axial resonances would occur at frequencies be- configurations 11-3 and 12-3 and yet its influence can be observed
yond the investigated range. at a higher Helmholtz number. This means that the effects of the
Figures 12 and 13 present results of the two double skin con- damping volume and any other adjacent volume have to be stud-
figurations, configuration 11-3 and configuration 12-3, respec- ied separately.
tively. Although both configurations share the same inner liner and The model can successfully predict the effects of different
flow conditions, their damping characteristic is different. It is damping volumes and finds the maximum damping at the appro-
mainly driven by the damping volume between both liners. Con- priate Helmholtz numbers.
figuration 11-3 has the smallest damping volume of VD Figure 14 presents configuration 12-3 under the influence of a
grazing flow of M = 0.1. Comparing these results to Fig. 13 reveal
0.6
conf. 4: uniform 0.6 dP 0%
0.5 conf. 10: θ−variation
dP 0.01%
0.5
dP 0.1%
Dissipation ∆

0.4
dPT 3%
Dissipation ∆

0.4
0.3
0.3
0.2
0.2

0.1 0.1

0 0
0.2 0.4 0.6 0.8 1 1.2 0.2 0.4 0.6 0.8 1 1.2
kLp klp

Fig. 11 Influence of the circumferential porosity distribution, Fig. 13 Influence of the bias flow on configuration 12-3, graz-
bias flow dP = 0.01%, and grazing flow M = 0.05 ing flow M = 0. Measurements: symbols, model: lines.

Journal of Engineering for Gas Turbines and Power SEPTEMBER 2011, Vol. 133 / 091503-7

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


0.6 the porosity at a given mass flow rate will lower the pressure ratio
dP 0% and thus not necessarily improve the damping performance.
0.5 dP 0.01% Increasing the wall thickness shows mainly negative effects on
dP 0.1% the damping. It is demonstrated by the experiments and repro-
dP 3%
Dissipation ∆
0.4 T duced by the model that the frequency bandwidth of the damping
is reduced for the thicker wall.
0.3 The Helmholtz number at which the damping has a maximum
is mainly determined by the damping volume. The volume and the
0.2 Helmholtz number of maximum absorption have a reciprocal re-
lationship. Enlarging or reducing the volume increases or de-
0.1 creases the Helmholtz number accordingly.
Double skin configurations can be successfully used to improve
0 the damping at a large given total pressure ratio, which usually
0.2 0.4 0.6 0.8 1 1.2
kl exists in gas turbine applications. The metering liner is used to
p
create a large pressure loss, so that the pressure ratio of the damp-
ing liner can be optimized to yield a high absorption. If a bias flow
Fig. 14 Influence of the bias flow on configuration 12-3, graz-
ing flow M = 0.1. Measurements: symbols, model: lines. is present 共dP ⫽ 0%兲, the Helmholtz number of the maximum
absorption is dependent on the inner damping volume alone, while
the outer metering volume adds an additional maximum for dP
the effect of the grazing flow on the absorption. While the maxi- = 0%.
mum damping occurs at the same Helmholtz number kl P ⬇ 0.8 if a The extensive database of experimental results was compared
bias flow is present, the maximum is shifted to kl P ⬇ 1 without with results from a thermo-acoustic network model. In general,
bias flow. The maximum damping value is reduced for the no the thermo-acoustic model has proven to be a powerful tool in the
共dP = 0%兲 and low 共dP = 0.01%兲 bias flow cases, where it is in- prediction of the absorption characteristic of a single and double
wall damper for a wide range of operating conditions and geo-
creased at higher bias flow rates. The model predicts a rise of
metrical variations. It has successfully demonstrated to provide
absorption for all cases. Thus, it predicts a larger dissipation at
reliable results within the range of its limitations, i.e., 1D geo-
dP = 0.01%.
metrical changes, no circumferential variations, and linear absorp-
Figure 15 illustrates the improvements of the double skin con-
tion.
figurations at realistic engine conditions. A typical pressure ratio
The experiments reveal the influence of various geometric,
across a combustor wall is dP = 3% with a grazing flow of about aerodynamic, and acoustic parameters on the liner performance in
M = 0.1. With an additional liner of low porosity, the pressure ratio detail and with high accuracy. The impact of the combustion on
across the damping liner is reduced, keeping a total pressure ratio the results cannot be reproduced by these isothermal and atmo-
of 3%. The damping performance of the double skin arrangements spheric tests. However, the behavior of the liner under high tem-
is doubled compared with configuration 3. The results could be perature and high pressure conditions is of great interest. In order
further improved by adjusting the open area ratio between the two to extend our database to include these additional parameters, a
liners, targeting the optimum pressure ratio for the damping liner. new test facility has been established. The hot-acoustic-testrig
共HAT兲 will be able to provide high accuracy results at almost
8 Conclusions realistic engine conditions.
The comprehensive experimental study has revealed the influ-
ence of many parameters on the damping performance of perfo- Acknowledgment
rated liners. This work is part of the GerMaTec project 共Contract No.
While the number of orifices and their diameter only have a 20T0602兲 within the LuFo 4 framework. It is funded in part by the
minor effect on the absorption, the porosity and the pressure ratio Federal Ministry of Economics and Technology 共BMWi兲 and ad-
共or bias flow velocity兲 are the dominant parameters. First, there ministered by the German Aerospace Center 共DLR PT-LF兲. Their
exists an optimum pressure ratio where maximum damping per- support is gratefully acknowledged.
formance is reached, and second, higher porosity yields a better
damping performance. For the design of a liner, it is crucial to
regard the combination of these two conclusions. Just increasing Nomenclature
M ⫽ mean grazing flow Mach number
K ⫽ Rayleigh conductivity
0.6 R ⫽ energy reflection coefficient, orifice radius
conf. 3: single skin T ⫽ energy transmission coefficient
0.5 conf. 11−3: double skin, small volume V ⫽ cavity volume
conf. 12−3: double skin, large volume c ⫽ speed of sound

Dissipation ∆

0.4 dP damping liner pressure ratio


dPT ⫽ total pressure ratio for a double skin
0.3 configuration
k ⫽ wave number
0.2 l ⫽ length of the liner
la ⫽ active length of the perforation
0.1
lp ⫽ length of the perforation
p̂ ⫽ complex sound pressure
0
0.2 0.4 0.6 0.8 1 1.2 r ⫽ reflection coefficient, duct radius
kl
p
s ⫽ orifice spacing
t ⫽ transmission coefficient, liner thickness
Fig. 15 Influence of the double skin arrangement at realistic x ⫽ axial coordinate
engine condition, bias flow dPT = 3%, grazing flow M = 0.1. Mea- ⌬ ⫽ energy dissipation coefficient
surements: symbols, model: lines. ␭ ⫽ wavelength

091503-8 / Vol. 133, SEPTEMBER 2011 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


␳ ⫽ density Coupling due to Helmholtz Resonators,” ASME Paper No. GT2003-38168.
关7兴 Stow, S. R., and Dowling, A. P., 2004, “Low-Order Modelling of Thermoa-
␴ ⫽ normalized porosity coustic Limit Cycles,” ASME Paper No. GT2004-54245.
␪ ⫽ circumferential coordinate 关8兴 Chung, J., 1977, “Rejection of Flow Noise Using a Coherence Function
␻ ⫽ angular frequency Method,” J. Acoust. Soc. Am., 62共2兲, pp. 388–395.
a,b ⫽ subscripts referring to measurements A and B 关9兴 Kirchhoff, G., 1868, “Über den Einfluss der Wärmeleitung in Einem Gase auf
1,2 ⫽ subscripts referring to sections 1 and 2 die Schallbewegung,” Ann. Phys. Chem., 210共6兲, pp. 177–193.
关10兴 Blokhintsev, D. I., 1956, “Acoustics of a Nonhomogeneous Moving Medium,”
+,− ⫽ superscripts referring to the downstream and NACA Technical Memorandum 1399, originally published in 1946 in Russian
upstream directions language.
关11兴 Morfey, C. L., 1971, “Acoustic Energy in Non-Uniform Flows,” J. Sound Vib.,
References 14共2兲, pp. 159–170.
关12兴 Heuwinkel, C., Enghardt, L., and Röhle, I., 2007, “Experimental Investigation
关1兴 Howe, M. S., 1979, “On the Theory of Unsteady High Reynolds Number Flow of the Acoustic Damping of Perforated Liners With Bias Flow,” The 13th
Through a Circular Aperture,” Proc. R. Soc. London, Ser. A, 366, pp. 205– AIAA/CEAS Aeroacoustic Conference, Paper No. AIAA-2007-3525, Rome,
223.
Italy, May 21–23.
关2兴 Bechert, D. W., 1980, “Sound Absorption Caused by Vorticity Shedding, Dem-
关13兴 Hughes, I. J., and Dowling, A. P., 1990, “The Absorption of Sound by Perfo-
onstrated With a Jet Flow,” J. Sound Vib., 70共3兲, pp. 389–405.
关3兴 Rienstra, S. W., 1981, “On the Acoustic Implications of Vortex Shedding From rated Linings,” J. Fluid Mech., 218, pp. 299–335.
an Exhaust Pipe,” ASME J. Eng. Ind., 103共4兲, pp. 378–384. 关14兴 Bellucci, V., Paschereit, C. O., and Flohr, P., 2002. “Impedance of Perforated
关4兴 Eldredge, J. D., and Dowling, A. P., 2003, “The Absorption of Axial Acoustic Screens With Bias Flow,” The Eighth AIAA/CEAS Aeroacoustic Conference,
Waves by a Perforated Liner With Bias Flow,” J. Fluid Mech., 485, pp. 307– Paper No. AIAA-2002-2437, Breckenridge, CO, June 17–19.
335. 关15兴 Jing, X., and Sun, X., 1999, “Experimental Investigation of Perforated Liners
关5兴 Macquisten, M. A., Holt, A., Whiteman, M., Moran, A. J., and Rupp, J., 2006, With Bias Flow,” J. Acoust. Soc. Am., 106共5兲, pp. 2436–2441.
“Passive Damper LP tests for Controlling Combustion Instability,” ASME Pa- 关16兴 Rupp, J., Carrotte, J., and Spencer, A., 2009. “Interaction Between the Acous-
per No. GT2006-90874. tic Pressure Fluctuations and the Unsteady Flow Field Through Circular
关6兴 Stow, S. R., and Dowling, A. P., 2003, “Modelling of Circumferential Modal Holes,” ASME Paper No. GT2009–59263.

Journal of Engineering for Gas Turbines and Power SEPTEMBER 2011, Vol. 133 / 091503-9

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 06/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

Вам также может понравиться