https://ntrs.nasa.gov/search.jsp?

R=20020079889 2018-03-09T08:43:39+00:00Z

Advanced Nacelle Acoustic Lining

Concepts Development

G. Bielalc, J. Gallman, R. Kzrlzze, P. Mziway, J. Premo, and M . Kosalzchick

Boeing Co~?zwzercialAilplane Co~~zpany, Seattle, Washington

A . Herslz, J. Celano, and B. Walker

Herslz Acozistical Engineering, Westlalce Village, California

J. Yzi, H . W . Kzuan, and S. Chiozi

BF Goodrich, Chzrla Vista, California

J. Kelly, J. Betts, J. Follet, and X,Thol~zas

Virginia Consortiunz of Engineering G. Sciences Ulziversities, Hampton, Virginia

National Aeronautics and

Space Administration

Langley Research Center Prepared for Langley Research Center

Hampton, Virginia 23681-2199 under Contract NAS1-97040, Task 4

August 2002
Available from:

NASA Center for Aerospace Information (CASI) National Technical Information Service (NTIS)
7 121 Standard Drive 5285 Port Royal Road
Hanover, MD 2 1076-1320 Springfield, VA 22 161-2 17 1
(301) 62 1-0390 (703) 605-6000
ABSTRACT
The work reported in this document was conducted under NASA contract NAS 1-97040
Aircraft and Engine Noise - Task 4 (Advance Nacelle Acoustic Lining Concepts). The
task was managed by the Fluid Mechanics and Acoustics Division of NASA Langley
Research Center. Mr. Tony Parrott, and Mr. Mike Jones were the Technical Monitors for
this task. This work was a continuation of the AST lining development work started at
Boeing in 1994. Studies reported in this document include:

Analysis of Model Scale ADP Fan Duct Lining Data Boeing

Bias Flow Liner Investigation Boeing, VCES

Grazing Flow Impedance Testing Boeing, GE, NASA,

BFG

Development of Micro-Perforate Acoustic Liner Technology BFG, HAE, NG

Extended Reaction Liners Boeing, NG

Development of a Hybrid ActiveIPassive Lining Concept HAE

Page intentionally left blank
 TABLE OF CONTENTS
Page

List Of Figures v

List Of Tables vi

1.0 INTRODUCTION .................................................................................................. 1.

2.0 STATEMENT OF WORK ..................................................................................... 1.

2.1 Analysis of ADP model fan lining test data........................................................2.

. .
2.
2.2 Bias flow liner investigation ...............................................................................

2.
2.3 Grazing flow impedance testing .........................................................................

2.4 Development of micro-perforate acoustic liner technology .................................2 .

2.
2.5 Extended reaction lining study...........................................................................

2.6 Development of a hybrid activelpassive lining concept ......................................2.

3.0 SUMMARY OF ACCOMPLISHMENTS .............................................................3.

3.1 Analysis of Model Scale ADP Fan Duct Lining Data .........................................3.
. .
6.
3.2 Bias Flow Liner Investigation ................................................................................................

6.
3.2.1 Grazing flow test ..........................................................................................

3.2.2 Time-domain analysis..................................................................................

7.

3.2.3 Sub-contract with the Virginia Consortium of Engineering & Sciences

Universities ........................................................................................................
12.

3.3 Grazing flow impedance test data analysis.......................................................14.

3.4 Development of Micro-Perforate Acoustic Technology ....................................19.

20 .
3.5 Extended Reaction Liners .................................................................................

24 .
3.6 Development of a hybrid activelpassive lining concept ....................................

4.0 CONCLUSIONS & RECOMMENDATIONS ...................................................25.

5.0 REFERENCES ..................................................................................................27.
Appendicies

Appendix 1 - Analysis Of ADP Model Fan Lining Test Data

Appendix 2 - Grazing Flow Impedance Testing Of Bias Flow and High Temperature
Liners
Appendix 3 - Acoustic Testing Of Scaled Micro-Perforated Sheets

Appendix 4 - Extended Reaction Lining Study

Appendix 5 - Test and Analysis -Hybrid ActiveIPassive Lining Concept

LIST OF FIGURES

Fig. No. Fig. Title Page

Figure I ADP Cutaway Schematic 4

Figure 2 Resistance vs. Particle Velocity for the 8% Open Area Lining 8

Figure 3 Measured Resistance at Different Bias Flow Velocities 9

Figure 4 TD-Predicted Resistance at Different Bias Flow Velocities 10

Figure 5 Comparison of TD and FD Predictions of Acoustic Resistance 11

Figure 6 Effect Of Grazing Flow On Perforate Resistance - 8% OA 14

Perforate, d hole =.05 in. , bdy layer = I .O in. thickness.

Figure 7 Effect Of Boundary Layer Thickness On Grazing Flow

Resistance - 8% OA Perforate, Mg=.4

Figure 8 Effect Of Grazing Flow On Perforate Resistance - 8% OA

Perforate, d hole =.05 in., bdy layer =.26 in. thickness

Figure 9 Resistance vs. Orifice Flow and Grazing Flow For Baseline
Perforate For 8.7 % Open Area Perforate

Figure 10 Schematic of 2D Duct and Anechoic Chamber

Figure 11 NG Measured Insertion Loss

Figure 12. MELO Predicted Attenuations for the Test Samples

LIST OF TABLES

Table No. Table Title Page

Table 1 Test Configurations 4

Table 2 Bias Flow Liner Description for Grazing Flow Testing 6

Table 3 Heated Liner Description for Grazing Flow Testing 6

Table 4 Impedance Measurement with Grazing Flow Test Matrix 17

Table 5 Extended Reaction Test Samples 22

The Advanced Subsonic Technology Noise Reduction Program (AST) goal is to develop
technology to reduce aircraft noise by 10 EPNdB, relative to 1992 technology, by the year
2000. The technology development strategy is a coordinated effort among government,
industry and academia addressing engine source understanding and reduction, nacelle aero-
acoustics, enginelairframe integration and flight procedures. Interim goals were established
which includes a goal to validate concepts to improve nacelle treatment effectiveness by 25%
relative to 1992 technology by the second quarter of FY'97. A model fan rig test using the
NASAIPW 22 in. diameter ADP model fan was conducted in 1996 to provide data to
demonstrate achievement of the 25% goal. A final goal was established at 50% improvement
in nacelle attenuation by the year 2000. It is believed that this goal can be achieved with
improved nacelle design which goes beyond passive liner efficiency improvements. For
example a directional inlet with more efficient application of acoustic lining and active noise
control are being developed to help meet this goal. Under AST Noise Reduction Task 5,
Boeing has supplied design and analysis tools which substantiate the superior acoustic
performance of directional inlets (Ref. 1). Using internal funds, Boeing is aggressively
pursuing implementation of directional inlets and has designed and manufactured a full scale
test article for a 1998-99 AST Pratt & Whitney (PW) static engine test. The test objective
is to evaluate noise reduction concepts for demonstration of the 2000 AST goals.

Boeing has been working to improve nacelle acoustic lining technology under the AST
program since 1994. The basis for the early Boeing AST lining technology work, (NAS1-
200090 Task 1)1994-1996 (Ref. 2), was a Boeing study conducted in 1993-94 under a
preliminary NASAIFAA contract investigating broadband acoustic liner concepts
(summarized in Ref. 2). As a result of this work, it was recommended that linear double layer,
triple layer, parallel element and bulk absorber liners be further investigated to improve
nacelle treatment effectiveness. NASA Langley also suggested that "adaptive" liner
concepts, which would allow "in-situ" acoustic impedance control, be considered. As a result,
bias flow and high temperature liner concepts were added for investigation.

Using mostly existing tools and past Boeing experience with parallel element and double layer
liners, a number of linings were designed, manufactured and tested in Boeing's grazing flow
impedance measurement facility during the period 1994-1996. This facility propagates the
fundamental mode over the acoustic liner with a flow Mach number up to M=0.5 and
determines the effective liner impedance from measurement of the complex acoustic pressure
pattern over the length of the liner. The purpose of the testing was to verify the designed
impedance spectrum of the liners. In addition, an analytical, lining trade study was conducted
to evaluate the potential benefits of these advanced design liners to a mid-sized twin airplane.
Lastly, advanced double layer liners were designed for a model scale fan test using the
NASAIPW 22 in. diameter ADP model fan to provide data to demonstrate achievement of
the 25% goal.

This report presents the results of the Boeing AST nacelle aero-acoustics work conducted in
1997-2000 under NAS1-97040 Task 4. The statement of work is first presented, followed
by a summary of the accomplishments relative to the statement of work. More in depth
reports are contained as appendicies or referenced.

2.0 STATEMENT OF WORK

The objective for NASA Contract NAS1-97040 Aircraft and Engine Noise -Task 4,
"Advanced Turbofan Duct Liner Concepts," was to develop technical knowledge and design
concepts for advanced turbofan duct acoustic linings to help attain the AST goal of 50%
nacelle noise reduction capability relative to the year 1992 by the year 2000. The following
requirements are set forth in the Statement of Work:

2.1 Analvsis of ADP model fan linina test data.

Analyze the results of the model scale test of the advanced fan duct liner Boeing designed in
1995 for the P&W 22 in. ADP model fan tested at the NASA 9x15 acoustic wind tunnel in
1996.

Measure the effect of grazing flow on impedance in the frequency range 1000 to 6000 Hz for
bias flow and high temperature liners.

Develop a time-domain modeling process to analyze the effect of bias flow on liner
impedance.

Give a sub-contract to the Virginia Consortium of Engineering & Science Universities to

study the application of bias flow for controling perforate liner impedance.

2.3 Grazins flow impedance testing

Participate in the NASA/GE/Boeing grazing flow impedance measurement technique
evaluation study.

Study the possibility of using steady-flow measurements to determine frequency dependent

orifice impedance changes with grazing flow using time-domain analysis.

2.4 Development of micro-perforate acoustic liner technoloav.

Give sub-contracts to Hersh Acoustical Engineering and BF Goodrich to perform grazing flow
and normal incidence impedance measurements of small orifice perforates and develop
mathematical models for small orifice impedance based on their results.

Give a sub-contract to Northrop Grumman to investigate the application of ultra-violet lasers

for drilling micro-perforate holes for nacelle acoustic lining application.

2.5 Extended reaction linina study

Conduct studies with the Boeing analytical duct propagation computer program to design
porous core (extended reaction) liners to be built and tested in a rectangular duct with no
flow.

2.6 Development of a hvbrid activelpassive linina concept

Give a sub-contract to Hersh Acoustical Engineering to conduct analysis and testing to
improve their hybrid active/passive lining system in preparation for a full-scale engine
demonstration.
3.0 SUMMARY OF ACCOMPLISHMENTS.
The objective of Task 4 of the NASA AST contract, "Advanced Nacelle Acoustic Lining
Concepts," is to develop technical knowledge and design concepts for advanced turbofan duct
acoustic linings. This work included subcontracts to BF Goodrich, Hers11 Acoustical
Engineering and Virginia Coilsortium of Enginering & Science Universities. There are six sub-
tasks for Task 4.

o Analysis of Model Scale ADP Fan Duct Lining Data

0 Bias Flow Liner Investigation

0 Grazing Flow Impedance Testing

0 Development of Micro-Perforate Acoustic Liner Technology

Extended Reaction Liners

Development of a Hybrid ActiveIPassive Lining Concept

The following are summaries of the accomplishments for each task. More in-depth reports
are contained as appendicies and, for selected tasks, in references.

3.1 Analvsis of Model Scale ADP Fan Duct Lininq Data (Detailed report in
Appendix Ij
The NASA Advanced Subsonic Technology (AST) Noise Reduction Program set an interim
goal to validate concepts to improve nacelle treatment effectiveness by 25% relative to
1992 technology by the second quarter of FY'97. Analytical design studies were conducted
within the AST program to develop liner concepts to help achieve this goal. The objective
of the program described below was to use the results of the above design studies to design
liners for a specific fanlnacelle and demonstrate the resulting benefits.

The demonstration test took place in 1996. The fan used for this demonstration was a
NASA Lewis 22 in. diameter Pratt Whitney (PW) Advanced Ducted Propeller (ADP) model
fan rig with 18 blades and 45 stators. The model was equipped with fan duct acoustic linings
designed by Boeing using the most advanced design technology then available. Advanced
inlet and fan case linings were designed by PW as well. A schematic of the fan rig is shown in
Figure 1.
Figure 1: ADP Cutaway Schematic

The test matrix included a variety of lining configurations in the inlet, fan case and fan duct
sections of the rig, including hardwall (see Table 1). Several different lining types were used
in this test including single layer (single degree of freedom - SDOF), double layer (double
degree of freedom - DDOF), and bulk liners. The Boeing designed single layer and double
layer liners used woven wire for the resistive elements. Both of these liners were linear,
meaning that the impedance is nearly independent of sound pressure level (SPL).

The baseline liner for 1992 level comparisons, was a liner produced by Dynamic Engineering,
Inc. (DEI), which approximately scaled the liners tested by PW in the ADP engine
demonstrator test conducted in 1992. Table 1 lists the lining type and location for each
configuration to be discussed in this report.

Table 1: Test Configurations

Far-field measurements were made for each configuration tested. In addition, an aft barrier
was used to acquire inlet data for several key configurations, (Configurations #1, #3, #4, # 5 ) ,
to keep aft fan noise from contaminating the inlet radiated noise data. The aft barrier was
placed parallel and offset from the exterior nacelle so as not to affect the flow in the tunnel
around the engine model itself.
The aft fan liners consisted of six different liner sections, each with its own geometric
constraints. The scope of this report is limited to analysis of the performance of the Boeing
designed lining within the fan duct. Analysis of the PW designed inlet and fan case lining data
are not included in this report. A detailed description of the Boeing designed fan duct linings
and the design processes are contained in Ref. 2.

The model data was scaled to a hypothetical, full-sized engine chosen to fit on a Boeing 747
derivative. The conversion was from the 22" Fan Rig model to a 130" full-scale engine. PW
used the scale factor to shift the frequencies to full-scale as well as distance corrections to
radiate the levels to a 150' polar arc.

The evaluation of the nacelle liners relative to the 1997, 25% peak aft angle PNLT
attenuation improvement goal shows that the Boeing fan duct liner designs met the goal. At
the approach condition, both the Boeing single and double layer liner fan ducts showed a very
large percentage improvement (230%) relative to the DEI liner. At cutback the Boeing
double layer liner showed a 5 1% improvement relative to the DEI liner and a 35%
improvement relative to the Boeing single layer liner. At sideline the Boeing double layer
liner showed a 109% improvement relative to the DEI liner and 56% improvement relative
to the Boeing single layer liner.

There were some concerns associated with this evaluation of the Boeing liners however. The
main concern is associated with the constraints of model scale acoustic liner testing. Model
scale testing of nacelle liners is not an established procedure. It is not possible to scale the
full scale designs because of the small curvature radii in model scale systems, the strength
limitations for a scaled sheet thickness and the manufacturing limitations associated with the
small orifice holes used with nacelle liners. In addition, it is not possible to verify that the
liner built for the model test has the desired acoustic impedance spectrum. While care was
taken to minimize these concerns by using woven wire resistance elements without perforated
plate and careful manufacturing procedures, the lack of ability to confirm the design
impedance is a major difficulty. Therefore it is difficult to draw conclusions relative to full
scale lining capability from the model test results. In addition, a significant difficiency with
the DEI liner was that its tuning frequency was about two 113 octaves lower than optimum.
It is not known if this was a design choice or a manufacturing error. Again the lack of a
method to measure lining impedance at high model scale frequencies makes it impossible to
determine if a model scale liner is manufactured to meet the design impedance target.

Relative to the Boeing single layer liner fan duct, the double layer fan duct liner showed
consistent improvement between bands 25 and 30. However, the improvement was not as
large as predicted (except for the sideline condition). The high frequency (> band 30)
attenuation improvement predicted for the double layer liner, was not observed. In fact, in
the very high frequencies, bands 36-40, the single layer liner data show more attenuation
than the double layer liner data. The problem was that the very high frequency attenuation
of the double layer lining was much poorer than predicted when tested with the treated fan
case for the approach and cutback conditions.The single layer liner high frequency
attenuation was about the same as predicted at approach and cutback. (measured with a
treated fan case). When the double layer liner fan duct was evaluated with a hardwall fan case,
its high frequency attenuation was found to increase to the predicted levels. This can be
interpreted in terms of modal attenuation. If the fan case has already attenuated the more
easily attenuated modes, the fan duct lining efficiency drops because it must attenuate a larger
percentage of more difficult to attenuate modes.

The double layer liner fan case attenuation measured with hardwall fan duct showed more
attenuation at the higher frequency bands than with the treated fan duct. This is also probably
a modal effect. The fan case lining is not as important if another liner, the fan duct liner, is
available to attenuate the same modes that the case attenuates.

3.2 Bias Flow Liner Investisation.

3.2.1 Grazinq flow test (Detailed report in Appendix 2)

Boeing conducted grazing flow impedance testing in late 1996 to investigate effects of
acoustic lining impedance control through heating the acoustic lining as well as bias air flow
through the acoustic lining. The bias airflow lining panel consisted of a double layer
Helmholtz resonator configuration, with a 200 cgs Ray1 (4.8 pc) porous mesh skin replacing
the standard impervious backing sheet. The bias airflow was introduced via a plenum chamber
attached on the back side of the lining. The lining geometric parameters are shown in Table
2:

Table 2: Bias Flow Liner Description for Grazing Flow Testing

Face sheet 14.7% open area.

Septum 6.0% open area

Face sheet to septum 3.14"

Septum to back sheet 0.76"

The heated lining panel was a standard double layer configuration with the impervious
backing sheet replaced with a thick aluminum sheet. This aluminum backing sheet had slots
machined in one side to accept electric heating rods. Thermocouples were installed during
panel fabrication. These were located at four levels in the double layer, at six locations spread
over the lining, nominally in the centers of honeycomb cells. The levels were 1) just beneath
the face sheet; 2 & 3) just above and just below the buried septum; and 4) at the lining
interior surface of the heated aluminum backing sheet. The lining geometry parameters are
shown in Table 3 :

Table 3: Heated Liner Description for Grazing Flow Testing

The grazing flow test conditions included :

Up to four grazing flow Mach numbers (0.00., 025, 0.35, and 0.50)
 Two sound levels - maximum achievable -approximately 160 dB OASPL- and nominally
10 dB down from maximum . The spectrum consisted of a 1KHz tone and harmonics
superposed on the airflow broadband noise.

Bias air flow - nominal 120 cmlsec average flow through the liner (M=.004), both
pressure (air exiting the face sheet into the test duct) and vacuum (air entering the lining
from the test duct)

Temperature - maximum heating not to exceed the 350 OF limit of the liner bonding
adhesive.

At the high SPL's tested, 160 dB and 150 dB, the bias flow test liner impedance did not show
a strong dependence on changes in grazing flow up to M=0.5, or bias flow up to average
M=0.004. The small grazing flow dependence is predicted by the Boeing impedance model at
high SPL's. The DeanJHersh impedance model predicts stronger dependence on grazing flow
and bias flow than was measured. The bias flow test was of limited success primarily because
no data could be measured at lower duct SPL's similar to those within the engine at the
airplane landing condition (130 dB - 140 dB). This is where bias flow effects would be the
strongest and where its application as a nacelle design concept was envisioned. With the liner
design chosen for testing no effect of bias flow was observed for a mean bias flow Mach
number of .004 over a range of grazing flow Mach numbers from 0.0 to 0.5 and SPLs of
approximately 150 dB and 160 dB. It was recommended that more fundamental impedance
modeling data be obtained and incorporated into an improved computer model before
additional nacelle design studies using bias flow liners are considered. This was the objective
of the time-domain analysis and VCES studies discussed below.

The elevated temperature liner test qualitatively verified the pre-test calculation predicting
decreased liner resistance with increased liner temperature. The change in reactance however
was quite different from the shift with frequency expected and is not understood.

3.2.2 Time-domain analvsis (Detailed report in Ref. 3)

The time-domain impedance modeling of perforates uses a time marching scheme to solve
the nonlinear momentum equation coupled to an acoustic cavity. This is accomplished by
assuming the instantaneous unsteady resistance is the same as the steady-flow resistance at a
given in flow velocity. (i.e. a pseudo-steady assumption where the flow is always in
equilibrium across the orifice). This method may help one understand the physics associated
with current semi-empirical acoustic impedance models and has been extended to predict the
effect of bias flow. The objectives of the Boeing time-domain impedance modeling effort are
to: (1) Develop models to predict the impedance of perforate linings in the time-domain
using measured DC flow resistance data. (2) Use time-domain models to evaluate bias flow
and frequency scattering for improved liner performance. (3) Determine if the grazing flow
effect can easily be added to the time domain models.

A model has been implemented in Matlab code and testing has been conducted to measure
flow resistance and impedance data for comparison to the model. Two liners were
manufactured at the Boeing Wichita Nacelle Responsibility Center for testing. These liners
were made using the same manufacturing processes used for airplane engine nacelles.
Specifically, a honeycomb core was bonded to the facesheet using a core reticulation process
which is used for nearly all the nacelle lining applications at Boeing. Both 8% and 12% open
area (OA) samples were made. These open areas were chosen because they are representative
of engine nacelle applications. The flow bench was used to measure the resistance versus
particle velocity of a test sample. The resistance was measured in both directions through the
samples over the range of 25 to 250 cm/s. The acoustic impedance of the samples was also
measured with and without bias flow with an impedance tube. A test assembly was designed
and built to measure the normal impedance of the test assemblies with bias flow. The
objective of the test assembly was to pass bias flow through the liner while maintaining a
boundary condition on the back side of the liner that approaches that of a hard-wall. The
device is basically made up of a plenum with a highly resistive layer (nominally 190 cgs Rayls
at 105 cmls) that will pass air, a plenum to hold the air, an air supply, and a mass flow
measurement device. This allows a controlled mass flow and thus a bias flow velocity to be
put through the lining. A vacuum system was not available during testing so all the bias flow
testing was with flow exiting the lining. A typical plot of the impedance tube measured versus
TD-predicted acoustic resistance at 1000 Hz is shown in Figure 2 for no bias flow. For this
case the impedance measurements were made with a nonporous backwall.

Comparison of Measured and Predicted Resist

Test Data

TD Prediction
Using Steady
Flow Resistance

...'---
0.2 I-".. .-.----
.n

Velocity (cmls)

Figure 2: Resistance (pc) versus Particle Velocity (cmls) for the 8% OA Lining

This particular liner has 8% OA. This comparison is representative of other perforate
samples at 1000 Hz in that the resistance is under-predicted slightly, but the slopes of the
acoustic resistance versus particle velocity are well predicted. The good agreement in the
slope shows that there is not a strong frequency dependence on the discharge coefficient of
the orifice at 1000 Hz. The under-prediction of the resistance at 1000 Hz can be explained
best by examining the form of the semi-empirical frequency domain (FD) impedance
prediction models. The Boeing semi-empirical FD models are comprised of a DC viscous
term, an AC viscous term, and an AC inertial term. The AC viscous term accounts for the
frequency-dependent increase in resistance due to the reduction of the acoustic boundary
layer thickness in the orifice as frequency is increased. The TD formulation is based on the
zero-frequency, steady-flow resistance and thus has no measure of this effect. Assuming an
infinite duct with a boundary layer profile dependent on frequency, as described by Allard
(Ref. 4), the AC viscous increase in the acoustic resistance at 1000 Hz is predicted to be 0.07
pc. Adding this to the TD predicted data results in a prediction that better matches the
absolute levels of the measured data. Figure 3 shows the measured resistance for the different
bias flows for the 8% OA liner at 1000 Hz and Figure 4 shows the TD predictions of the same
lilling face sheet with the impedance of the porous backing sheet and back cavity accounted
for. The TD predictions clearly do well at representing the data.

Measured Resistance at Different Bias Flow Velocities

0 50 100 150 200 250 300

Particle Velocity (cmls)

Figure 3: Measured Resistance at Different Bias Flow Velocities

 Predicted Resistance at Different Bias Flow Velocities

2.5

Speed cmlsec

0 50 100 150 200 250 300

Particle Velocity (cmls)

Figure 4: TD Predicted Resistance at Different Bias Flow Velocites

Notice that both the measured data and the predictions show that the bias flow effect
diminishes with increasing RMS acoustic velocity. The data show that with bias flow the
acoustic resistance initially starts out flat (maybe even dropping slightly) and then starts to
increase as the RMS acoustic velocity increases. Clearly, the physics represented in the TD
model predict the effect of bias flow on the acoustic resistance.

The TD model, based on the non-linear differential equation (DE) from the momentum
equation, was used to make improvements to the semi-empirical orifice resistance FD models
generally used for nacelle lining design studies. The steady flow resistance model consists of
the sum of a linear (viscous) and non-linear (inertial) loss mechanisms. This gives a form
+
R(V) = &, I?,,,lVI that closely represents the measured steady-flow resistance data.

The semi-empirical FD impedance models predict &,, and I?,,, in terms of the hole diameter,
open area, plate thicltness, and frequency. The linear resistance for these models, is
composed of AC and DC components and the inertial term of the resistance, I?,,, , is based on
the DC non-linear component corrected for frequency. The results of a TD numerical study
show that the linear or viscous part of the TD-predicted acoustic resistance is nearly identical
to the current FD model.

The TD study also showed that the non-linear or inertial part of the resistance can be
modeled as approximately (1.15) times the RMS acoustic velocity times the steady-flow
slope, I?,,, . Therefore, a new model is proposed to predict the acoustic resistance using the
steady-flow parameters, &, and I?,,, , and the RMS acoustic velocity, y, :
The semi-empirical model of the acoustic resistance can be extended to include bias flow by
examining the behavior of the DE describing the perforate resistance. A scheme for
predicting the effects of both acoustic RMS and bias flow velocity based on the steady-flow
resistance is suggested:

Reference 4 plots the TD model and the resulting FD models with different bias flows. Note
the FD predictions are very close to the TD solutions except near yLz = 1, where they are
slightly high.

Figure 5: Comparison of TD and FD Predictions of Acoustic Resistance

A correction could be applied to the model near xc, = 1 to reduce the difference, but it was
determined that the models were close enough (within a couple of tenths of a pc).

3.2.3 Sub-contract with the Virainia Consortium of Enaineerina & Science

Universities (Detailed report in Ref. 5)
(The following is the Introduction from Ref. 5.) The initial purpose of this study is the
experimental determination of liner impedance and the validation of impedance models using
data obtained from the NASA Langley Normal Incidence Tube (NIT). An experimental
database was produced which included both fibermetal and perforate samples tested with and
without bias flow in a single-degree-of-freedom (SDOF) liner configuration. The database was
used to validate available impedance models and develop improvements in order to produce
the best available frequency domain impedance model that includes the effect of bias flow.
With this model, the ultimate goal is the design of double-degree-of freedom (DDOF) liners
representing optimum passive and bias flow liner configurations in the presence of grazing
flow. A multi-layer configuration will allow bias flow to affect the total resistance through
the resistance of the septum.

The surface impedance predictions were obtained with the NASA Langley Zwikker-Kosten
Transmission Line Code (ZKTL). This computer program is based on Zwiklcer and Kosten's
theory for sound propagation in channels. In general, the model is composed of continuous
arrays of multi-degree-of-freedom liner elements. The ZKTL program can compute
absorption coefficients, reflection factors and insertion losses for free-field excitation.
Matrix techniques are employed to compute the composite impedance due to the liner
elements. For a nonlinear liner, an iteration scheme determines the impedance based on a
given incident sound-pressure level (SPL). In this study, the NIT was treated as a distributed
element (channel) and the liner sheets with the backing cavity as lumped elements. The
different perforate impedance models were implemented in the lumped element part of the
modular structure of ZKTL, minimizing coding modifications. Approximately twenty
impedance models were used representing derivatives from five baseline models. For example,
one of the baseline models was the existing standard model in ZKTL. In the original ZKTL
code, a rigid terminating condition (infinite impedance) was imposed for each channel. For
this investigation, a finite impedance boundary condition was developed to simulate a porous
back plate that was used to introduce the bias flow.

The error criteria used was a least squares linear fit (with zero intercept) between the
experimental and numerical results. The slope of the fit indicated the difference between
experimental and numerical values, and the coefficient of determination (the correlation
squared) indicated whether they were following the same trend. A slope and coefficient of
determination of one indicated a perfect match between experiments and predictions.

The error criteria was the basis for a large search and optimization routine that was
programmed to evaluate the twenty impedance models against the more than one hundred
sets of experimental data. The routine was also used to evaluate improvements to the
impedance models as an optimum bias flow model was developed.

Initial results of the comparison between prediction and experiment indicated that the
resistance term needed more improvement than the reactance. The linear term of the
resistance in the standard ZKTL model became one focus of attention for correction. This
observation was due to the fact that the slope and coefficient of determination between
prediction and experiment improved with increasing incident SPL, which indicated that the
cause of the problem was the linear component. For example, sample 52 with thickness to
diameter ratio of 1.67 and percent open area of five produced results indicating a slope of
2.5 1 and 1.65 for sound pressure levels of 120 and 140 respectively at zero bias flow. The
coefficient of determination also improved from 0.07 to 0.92. These are significant
improvements for the slope and coefficient of determination for a 20 dB increase in SPL.
Based on these observations the resistance term in the standard model could be improved by
including a frequency dependent term similar to that found in the Crandall impedance model.
The result is a significant improvement in the prediction when compared to the experiment.
This illustrates the use of this data and error criteria in evaluating and improving the twenty
impedance models. Full results are shown in the report with rankings of the models and the
development of the impedance model that best predicts the behavior of the bias flow effect.

Experiments were performed to determine the impedance of lumped single degree of freedom
liners with bias flow. These liners were composed of nonlinear perforate facesheets followed
by a 1.7" cavity and a high resistance fibermetal backing. The high resistance fibermetal was
necessary in order to introduce the bias flow into the cavity. Data will show the
determination of the resistance of the fibermetal and the minimal effect on the impedance of
liner material as opposed to having a hardwall non-porous backing. The bias flow was fed into
a plenum chamber 3 inches in length with a cross sectional area 2x2 inches before flowing
through the high resistance fibermetal. Tlle perforate samples were structurally resonating in
the middle to upper range of the frequencies tested. A post was inserted through the
fibermetal to support the perforate sample in the center. Adding the post support shifted the
resonance above 3 kHz which is out of the frequency range of interest. A thin nut was used
to secure the fibermetal that also had resonant frequencies within the range tested. Other
than eliminating the structural resonance, the data showed that the post and the nut supports
had a minimal effect on the impedance measurements. This determination was achieved by
comparing the continuity in resistance and reactance of the experimental impedance results
(as a function of frequency),

Impedance was measured using three stationary microphones. The first microphone was used
to set the reference sound pressure level (SPL) at the sample, the other two microphones
measured the transfer function between two points on the standing wave produced by the
superposition of incident acoustic waves, produced by the acoustic drivers, and reflected
waves from the perforate sample. The transfer function was then used to create a least
squares fit of the standing wave to calculate the overall impedance of the sample and cavity.
Since only two points were used for the transfer function, the least squares fit reduced to a
deterministic closed form solution.

The perforate samples tested varied in open area from 1 - 15% with thickness to diameter
ratios from 0.71 to 1.8. Twenty three perforate samples were tested in all. The bias flow
velocities tested ranged from -25 to 600 cmls (negative indicating suction, positive, blowing)
in the cavity used with all samples. The frequency range tested was between 1000 to 3000
Hz, one tone at a time (i.e. did not study frequency interaction effects), in increments of 100
Hz. The reference SPL was set at 120, 130, and 140 dB for low flow rates and at 130 dB at
high flow rates where changing SPL had no effect on measured impedance based on a few
measurements over the SPL range.

Acoustic resistance and reactance were acquired for all samples. Overall, resistance increased
with bias flow for all samples. At zero and low bias flow, increasing the reference SPL
increased the resistance. Above a certain critical bias flow velocity, changing the reference
SPL had no effect on the resistance and the samples exhibited linear behavior with respect to
SPL. The reactance was minimally affected at low bias flow rates, but at high bias flow rates
the reactance was significantly reduced. The effect was most noticeable as the velocity in the
holes approached choked condition. The data is presented using the best non-dimensional
groupings of the parameters tested. In addition to the acoustic impedance, flow resistance was
measured for all samples. This was used to determine an effective discharge coefficient that,
together with microscopic measurements of the geometric parameters, provides a complete
specification of the samples.

Together this data set does provide the necessary careful parametric study that is used to
improve the ability to model the impedance of liner elements with bias flow. In turn, that will
create the ability to design optimum bias flow liners providing a means to in-situ control liner
impedance with no grazing flow.

3.3 Grazina Flow lmeedance Test Data Analysis (Detailed report in Ref. 6)

Boeing participated with NASA and GE in a grazing flow impedance measurement technique
evaluation study. This study was motivated by the large differences between AST
participants GE and Boeing in their models for the effects of grazing flow on acoustic
impedance of perforate acoustic liners. It is generally recognized that the acoustic resistance
of a perforate single layer liner is dominated by the grazing flow contribution. Double layer
liners are less affected by grazing flow but the effect cannot be ignored. A comparison of the
magnitude of the difference between the Boeing and GE estimates for the effect of grazing
flow on perforate resistance is shown in Figure 6.

Fig. 6 Effect Of Grazing Flow On Perforate Resistance 8% -

OA Perforate, d hole =.050 in,
Boundary Layer Thickness= 1.0 in. t (117 power law)

11
-+-- Boeing
+Heidelberg 1 ~

Figure 6: Effect of Grazing Flow on Perforate Resistance

At the time the study was planned GE was using the Heidelberg relationship shown in Figure
6. GE has since collected new data as part of the present study which results approximately
in the curve labeled "Syed (GE)". It is seen that the new Syed (GE) relationship deviates
from the Boeing relationship even more than the Heidleberg curve. A major parameter used
by both the Boeing and Heidelberg relationships is the boundary layer thickness. The Boeing
model, for low SPL, is of the form:
Where d is the orifice hole diameter, 6 is the boundary layer momentum thickness, M, is the
grazing flow Mach number, cd is the orifice discharge coefficient and o is the liner POA.

The Heidelberg model is:

Where 6* is the boundary layer displacement thickness.

The Syed model is:

The Syed (GE) model is independent of the boundary layer thickness and is only a function of
M, and o. The dependence on boundary layer (for a 117 power law boundary layer shape) for
these relationships is demonstrated in Figure 7 for M, =0.4.

Fig. 7 Effect Of Boundary Layer Thickness On

-
Grazing Flow Resistance 8%OA Perforate,
M, 1.4 (calculated assuming I17 power law)

+Boeing
-c Heidelberg
-& Syed (GE)

d BL I d hole

Figure 7: Effect of Boundary Layer Thickness on Grazing Flow Resistance

Note that the Boeing and Heidlelberg models approach each other at large "d BLId hole". It
is also noteworthy that (for M,=0.4 and below) the Boeing and Syed models approach each
other at small "d BLId hole". This is shown more clearly in Figure 8. Syed's data was
collected at relatively small values of momentum thickness (stated as 6=0.045 in.). The
major discrepencies between the Boeing calculation and Syed's are the dependence on
boundary layer and the high Mach number results.

Fig. 8 Effect Of Grazing Flow On Perforate Resistance -

8% OA Perforate, d hole =.050 in, Boundary Layer
Thickness= .26 in.

-t Heidelberg
- Syed (GE)

Figure 8: Effect of Grazing Flow on Perforate Resistance

The plan for the grazing flow study was for Boeing, GE and NASA to each measure the effect
of grazing flow at their test facilities for comparison. Table 4 is a listing of the test samples
chosen for testing.
 Table 4: Impedance Measurement with Grazing Flow Test Matrix

The effects of perforate sheet thickness, POA, hole diameter, core depth and face sheet
material are investigated with the geometries shown in the table. It is believed that the
boundary layer momentum thickness is relatively small at each of the facilities. A perforate
hole diameter variation at constant POA is included in the test matrix and could be used to
examine the effect of the ratio of the boundary layer thickness to hole diameter, but the
parametric change is not very large. To examine the boundary layer thickness dependance it
will be preferable that at least one of the facilities test with a thickened boundary layer as
well. Something of the order -9=0.1 in. is necessary to simulate the engine nacelle situation.
This was not part of the plan for the Boeing testing. Also, Boeing only tested a subset of the
configurations shown in Table 4. Specifically, Boeing tested the first 5 configurations and
the last configuration. All of these data were provided to NASA.

Boeing and NASA use very similar facilities for measuring grazing flow impedance; however
Boeing's testing process is very different from NASA's. The NASA test process is to run a
single tone (500 Hz to 3000 Hz in steps of 500 Hz) at 120 dB, 130 dB and 140 dB for
grazing flow Mach numbers of 0, . l , .3 and .5. The Boeing process used for this test ran
various multi-tone combinations at OASPL's of 140 to 160 dB for grazing flow Mach
numbers approximately 0, . l , .2, .3, .4, .5. The details of the tone combinations are
contained in Reference 6 which contains the test report documenting the Boeing
measurements.

Comparisons with NASA test data should be done versus acoustic particle velocity. This will
require calculation of the acoustic particle velocities using the deduced acoustic impedances
and the measured SPL's.

Boeing received data from GE for the baseline liner for steady flow resistance measured with
grazing flow. This data is shown in Figure 9. Also shown on the plot are estimates of the
acoustic resistance based on the steady flow resistance curves. One set is the GE (Syed)
estimate. The other was calculated by Boeing using the time-domain code discussed earlier
with the GE steady flow curves as inputs. Both estimates are seen to agree very well.
Preliminary measurements of acoustic resistance from the Boeing Wichita facility are also
shown. The orifice flow particle velocity was estimated for the Wichita data using the
OASPL at the leading edge of the 12 in. long test panel and the measured resistance. This
means that the estimated particle velocities for the Wichita data plots may be too high.
Given this concern it appears that the Wichita data may be somewhat consistent with the GE
acoustic curves for M, = 0.1 and 0.35 but the M, = 0.50 resistance values are clearly lower
than the GE values.

- - - - -

Wichita Momentum Thickness = .03 in.

Syed steady flow Syed Momentum Thickness= 04 in.
Premo TD -Syed SF data
Syed Acoustic
,=.\ ..-,,->e
a.... -

*.A
Mg=0 3
A
44
<*.C

(W-Mg 0.5)
Wlchlta Data (est. Vpart)
(W-Mg 0.35)
&'

Mean Orifice Flow - cmlsec

Figure 9: Resistance vs. Orifice Flow and Grazing Flow For 8.7% Open Area
Perforate
 r e ~ o r t sin Refs. 7 a n d 8 a n d Appendix 3)
The subcontract work done by BF Goodrich, Hers11 Acoustical Engineering (HAE) and
Northrop Grumman was aimed at developing micro-perforate face sheet and septa capability.
Micro-perforate resistance elements have acoustic characteristics which show small
dependence on local sound level and grazing flow conditions and give better high frequency
attenuation characteristics compared to conventional perforates. BF Goodrich is
investigating methods to obtain micro-perforate holes (order of 4 mil) in Aluminum as well
as developing a semi-empirical math model to predict the acoustic properties of micro-
perforates. An initial model has been completed by BF Goodrich based on testing with
Perfolin micro-perforate material. The BF Goodrich model is very similar to the Boeing
perforate impedance model. For micro-perforates, the primary resistance element is the
Poiseuille (P) flow term. For standard perforates Boeing increases this term by 75% relative
to basic P flow. For micro-perforates (d<.005in., t/d.>5) however the basic P form is
maintained. The main difficulty associated with calculating the acoustic impedance for
micro-perforates is specifying the perforate geometry. The P calculation is very sensitive to
orifice diameter (dA3)and an effective diameter is difficult to determine for irregular shaped
holes common with laser drilling. BF Goodrich used a process to estimate the perforate
diameter and percent open area using optical measurements and steady flow resistance data.
The BF Goodrich detailed report is contained in Reference 7.

HAE is also developing a math model for micro-perforates using more idealized test samples
with drilled holes in Plexiglas with thickness and hole diameters scaled up by a factor of
approximately 10. Examing HAE's test results it appears that t/d scaling was not sufficient to
reproduce the trends observed from the micro-perforate testing conducted in 1996 (Ref. 2).
Two reasons suggested for this are the (t/d2' dependency of resistance expected for Poiseuille
flow and the lack of scaling of the boundary layer thickness to orifice hole diameter. A first
cut model which includes the (t/d2' dependence (Appendix 3) has been completed by HAE and
reviewed by Boeing. Questions about the impedance model formulation have been raised and
passed on to HAE.

Northrop Gmmman was contracted to investigate the use of ultra-violet lasers for
manufacturing micro-perforates. Their report is contained in Reference 8. Ultra-violet lasers
have the potential for drilling much cleaner holes at much more rapid rates than is realized
with carbon dioxide lasers today. Also a free electron laser with very high energy density is
currently in development which offers even better potential. This project investigated the
feasibility of laser-machining high-aspect-ratio micro-holes in structural composite and
aluminum face-sheet skins. The holes were nominally 0.004 in. in diameter. The skins had
4% open area and were 0.040-in.-thick composite laminates, thermoset and thermoplastic,
and 0.040-in.-thick aluminum sheet. The diameter and thickness were selected to achieve
the desired 10: 1 skin-to-hole size ratio for improved attenuation performance.

The thermoset material was graphite-epoxy, and the thermoplastic material was glass-
reinforced polyetherimide. This thermoplastic material was selected because of its high
impact strength, its ability to be post-formed after laser drilling in the flat, and its ability to
achieve cauterized hole walls, thereby improving resistance to moisture and oxidation. The
original plan called for manufacturing test specimens for flow resistance and structural
testing. Only the initial manufacturing investigations were conducted however because of
funding limitations.
 3.5 Extended Reaction Liners (Detailed report in Appendix 4)
The extended reaction liner concept allows acoustic propagation within the liner. Current
liners are "locally reacting" in that this propagation is not present because it is blocked by
the non-porous honeycomb core. The purpose of this study is to investigate the potential for
attaining additional improvements in attenuation, relative to the locally reacting case, by
allowing acoustic propagation inside the liner. In a duct, a phase difference exists between
neighboring acoustic lining cells. The phase difference is a function of incidence angle
(effectively mode number for a given frequency, as each mode can be thought of as striking
the liner at a different incidence) and the spatial separation between cells. This phase
difference leads to a varying pressure distribution across the two neighboring cells, which, in
turn, will drive the 'in-liner' propagation for acoustically connected cells. Two approaches
have been adopted for this study. The first involves the validation and investigation of two
codes designed to predict the impedance of an experimental configuration, with two cells
linked by a resistive element, and the phase of the source to each being independently
controlled. These codes can then be used to model a finite length, non-locally reacting liner,
where, as previously stated, a phase difference exists between neighboring cells.

The second study is being performed using the Boeing liner optimization and duct attenuation
program (MELO). The code is being used to investigate the modal duct attenuation of various
configurations of extended reaction liners, with downstream propagation only, simulating a
fan duct.

The calculation of duct lining attenuation can be conceptualized by imagining a wave

reflecting from the lined walls of the duct as it propagates down the duct. At each reflection
some energy is absorbed by the lining reducing the downstream-propagated noise. The energy
fraction absorbed at each reflection (absorption coefficient) can be easily calculated if the
acoustic impedance at the lining face is known as well as the incidence angle of the duct
propagating wave. The acoustic impedance at the lining face depends on the manner in
which sound is propagated within the lining itself. The superposition of the sound incident
on the lining and the sound within the lining determines the lining acoustic impedance. For a
non -porous honeycomb core lining the sound propagation within the lining is limited to
being along the axis of the honeycomb down to the backsheet and reflected back to the face
sheet. The propagation within the lining is therefore the same no matter the angle of
incidence of the duct wave. This is called a "locally reacting" lining. The acoustic
impedance only depends on the local properties of the lining and the duct sound field.

For a porous core lining there is the opportunity for sound to propagate laterally within the
lining through the core as well as along the axis of the honeycomb. The lateral propagation
within the lining must be analytically "connected" to the duct wave since the duct wave is
continuing to influence the lateral wave as it propagates. The superposition of the duct wave
and the lining lateral wave at the lining face is therefore dependent on the angle of incidence
of the duct wave and the lateral wave propagation properties within the lining. This is called
an "extended reaction" lining. The Boeing analytical calculation for a duct with porous core
models the lateral propagation within the lining for each incident mode at a given frequency
(incidence angle) separately, calculating a unique lining impedance for each incident mode and
then calculating the attenuation of that mode as for a "locally reacting" lining. The
analytical "connection" of the duct wave with the lining lateral wave is obtained by forcing
the lateral wave numbers of the duct wave and lining wave to be the same (same lateral wave
length and propagation speed). The lumped resistive and reactive effects of the core pores
are used to determine properties of a homogeneous medium with equivalent dissipation and
propagation speed which is used to model the lateral wave propagation.
The report in Appendix 4 detailes the analytical extended reaction study. The objective is to
investigate the concept thoroughly through the validation of MELO with a 2D duct
propagation test scheduled at Northrop Grumman. Then the validated code can be used for
assessment of these liners for 'real' applications. The Northrop Grumman test was funded by
NG with NASA funded consultation by Boeing. Boeing defined the details of the two
dimensional propagation duct and anechoic section for measuring far-field propagation.

Figure 10 shows a schematic of the test set-up. The lined duct dimensions were chosen to
represent a typical fan duct. The duct height at 22" was chosen to conveniently mount the
four speaker horns necessary for the combined requirements of a broadband source and tones
with a preferred modal content. The lined duct length of 3' gives an L/H of 1.64. A higher
LID would compromise the overall length of the device, given the size of the anechoic
chamber. The duct width was chosen at 1.3" to be shallow enough to ensure the acoustic
pressure didn't vary in that plane over the frequency range of interest (500Hz to 4000Hz).

- (both tMti#]

Figure 10: Schematic of 2D Duct and Anechoic Chamber

The exponential horns on the speakers are designed to give a cut-off frequency well below
500Hz. Fiberglass wool is placed in the horns to reduce the interference between them. A
microphone is also inserted in each horn to ensure consistency in level between runs.

The test samples, both locally reacting and extended reaction, were constructed from wire
mesh facing sheets and, if applicable, wire mesh walls. Table 5 shows the specifications for
the test samples. Sample liner 2 is an optimized single layer liner for a representative fan
duct application. Sample liner 3 is an optimized porous core liner. The liner designs were
done with the Boeing duct propagation code MELO.
 Table 5: Extended Reaction Test Samples

Sample # Face Sheet DC Flow Core Depth (cm) Common Wall

Res~sta~lce
(cgs Rayls) Resistance (cgs Rayls)
1 50 32 20

2 90 32 Hard (I e locally react~ng)

3 50 32 40

4 SO 32 0 (1 e no cells)

5 50 32 Hard

Only insertion loss measurements were made, the mode measurement microphones shown in
Fig. 9 were not installed. The insertion loss for a given liner was obtained from the
difference of the logarithmic sum of the SPL's from the anechoic chamber microphones for
the hardwall duct and the lined duct. The measured insertion loss spectra are shown in Figure
11. This data was taken with the drivers driven by uncorrelated broadband noise sources.
The MELO predictions for the tested configurations are shown in Figure 12.

Hard or No Cell Walls

1
a
z
i
U)
Z

0
1W
Frequency Hz

Figure 11: Measured 113 Octave Band Insertion Loss (with permission from NG)
 MELO Predicted Attenuations for the Test Samples

12 -
, - r Liner 1, 20 ray1 cell vralls

Frequency (Hz)

Figure 12: MELO Predicted I13 Octave Band Insertion Loss for the NG Test
Samples

It is clear that there are some major differences between the measured and predicted results.
The measured lining tuning frequencies were two 113 octaves higher than predicted and the
measured insertion losses were generally about half of that predicted. However, it appears
that the effects of using porous cell walls and eliminating the cell walls altogether were well
predicted. The reason for the prediction/measurement discrepencies are not known. An
independent prediction for the locally reacting lining using the Boeing C-Duct code tended to
verify the MELO code predictions. Therefore, there is probably something about the test
that is not being modeled. One possibility is that the modal assumptions used by the codes
are incorrect. Another is that the anechoic chamber microphones were not far enough from
the duct exit for the radiated energy calculation to be correct. However, it does not seem
likely that either of these could explain the tuning discrepency. In any case there does
appear to be a bandwidth benefit for a single layer porous core liner compared to a single
layer hardwall core liner with a small penality in peak attenuation for the aft fan duct
application. However, the improvement in attenuation bandwidth is not as great as that
obtainable with a conventional double layer liner.

No conclusion can yet be drawn for inlet application of extended reaction liners for inlets. It
is felt that there is the possiblity of using extended reaction to develop a liner which can be
optimized for a larger range of modes (incidence angles) but this was not investigated in the
current study.

3.6 Development of a hvbrid activel~assivelining concept (Detailed report in

Appendix 5)
Under sub-contract, Hersh Acoustical Engineering (HAE) conducted an analysis and testing
study to improve their hybrid activelpassive lining system in preparation for a full scale
engine demonstration. HAE undertook a four-part initial investigation of simplifications
that might be applied to active control of rotortstator tones in turbofan engine inlets. The
underlying concept upon which this investigation was based assumed that only a subset of the
total number of cut-on modes would have a significant impact on sideline tone nose. If this
subset could be controlled with a less extensive array of active control actuators than required
by the full sound field, reduced active control system complexity and spacelweight would be
achieved.
The four major parts of the project were as follows:
1. Design and construction of a static test facility for generating and measuring multiple
spinning order and multiple radial order propagation modes in a circular duct.
For this a rooftop static 48-inch test duct facility was designed, constructed and calibrated for
radiation of modes (2,O) - (2,2) either in isolation or in combination.
2. Establish microphone arrays and signal processing systems for simultaneous measurement
of in-duct and far-field radiated sound.
For this the radiation characteristics for the individual modes were computed in near and far
field
3. Design and fabricate a linear microphone array to be mounted alongside the duct inlet and
used as an error signal sensor for far-field radiation in specific directional sectors.
4. Install an activelpassive hybrid liner in the test duct and incorporate it with the linear
microphone array to assess feasibility of simplified ANC to suppress multi-radial order
tones.
Far-field sideline radiation suppression was demonstrated using a hybrid activelpassive liner
employing two rows of active Helmholtz resonators to control a four-radial in-duct sound
field.
4.0 CONCLUSIONS & RECOMMENDATIONS
To a large degree the work reported in this document was a continuation of the AST lining
development work started at Boeing in 1994. This is particularly true regarding the bias flow
and the micro-perforate development work. The ADP model scale, lining test was an
application of the design tools and ideas developed up to 1995. The GEIBoeinglNASA
grazing flow impedance measurement study was initiated when it was realized that GE and
Boeing used very different modeling relations for calculating the effect of grazing flow on
liner impedance. In addition Boeing had had difficulties with this measurement with liners
designed as part of 1995-96 AST investigations. The extended reaction liner study was new
to the Boeing AST program for this Task and was added after it became apparent that the
European RANNTAC program had claimed some success with this concept. The hybrid
activelpassive lining concept study was a continuation of work HAE had done under AST
contract for NASA Glenn.

The ADP model scale fan duct lining test demonstrated that linings could be manufactured to
meet a 25% improvement relative to the designated baseline liner (DEI). However, it may be
questionable whether the designated model scale baseline represents 1992 engine lining
technology. The DEI single layer liner was tuned to a lower frequency than the Boeing
optimized single layer liner. This was partly due to the difficulty in scaling an engine test
liner. The engine test liner that the DEI liner attempted to scale was manufactured with a
high POA perforate bonded to the underside of the face woven wire. The model scale liner
used a felt metal face sheet with properties very similar to the full-scale face sheet and
eliminated the perforate back sheet. This results in a higher mass reactance for the model
scale liner at a scaled frequency. The resulting tuning frequency of the model scale liner was
probably about 1 one-third octave lower than the full scale liner. Unfortunately this could
not be verified because of the difficulty in measuring lining impedance at model scale
frequencies.

While it is true that wire mesh, linear single layer liner technology was commonly used by
many nacelle manufactures in 1992, this was not the case at Boeing. In 1992 Boeing utilized
a double layer perforate liner with a laser-drilled septum and punched aluminum face sheet.
This liner is believed to give approximately 20% better noise attenuation for a fan duct
application than an optimized single layer wire mesh liner (Ref. 2).

The double layer wire mesh liner is believed to give approximately 5% improvement relative
to the Boeing double layer perforate liner (Ref. 2). Unfortunately, it was not possible to
manufacture a model scale double layer perforate liner for the ADP model fan test, so a
measurement of the benefit of the double layer, linear liner relative to the conventional
double layer, perforate liner was not conducted. The Boeing designed linear, double layer
liner showed about the expected improvement relative to the Boeing designed linear, single
layer liner at cutback, but showed slightly less than the expected improvement at approach.
One could therefore conclude that the test results demonstrated approximately 5%
improvement relative to 1992 technology if the Boeing double layer perforate liner is
assumed as the base. Some other significant technical conclusions from the ADP fan duct
test were:
o There is a need to be able to measure model scale liner impedance at frequencies to 20
kHz. The demonstrated capability today is about 12 kHz.
o The acoustic treatment in the fan case strongly affected the performance of the fan duct
liner. It appears that the fan case treatment attenuated noise in higher order modes and
the stators did not scatter the remaining energy back into higher order modes as it
propagated into the fan duct. Therefore, analytical lining evaluation assuming uniform
modal energy distribution results in optimistic expectations. This suggests that if an
efficient method of modal scattering near the stators could be devised, an increase in fan
 duct liner attenuation (closer to that predicted assuming equal energy distribution),
relative to that currently observed, would result.
Lining tone attenuation was significantly different than the broadband attenuation and
generally better. This suggests there is a different modal energy distribution for tones
compared to broadband noise.

Significant progress has been made on bias flow acoustic liners. Boeing and VCES have
developed similar analytical expressions for the effect of bias flow on the resistance of
perforates. Boeing conducted a limited test to verify its formulation. More extensive testing
was conducted by VCES. The VCES testing also showed the need to incorporate a reactance
dependence on bias flow impedance models. The effect of bias flow with grazing flow is still
undefined. VCES is planning a series of tests at the NASA Langley acoustics laboratory to
investigate this dependence.

Boeing used their bias flow formulation to examine the impact of bias flow through the water
drainage system found in an actual inlet acoustic liner. It was determined that the bias flow
has a minimal impact on the acoustic impedance for the double layer perforate liner
considered in the study. VCES will investigate the potential to utilize bias flow to change
nacelle liner acoustic impedance as the engine operating point or liner environment (grazing
flow or SPL) changes.

As part of their bias flow studies VCES made a series of impedance and flow resistance
measurements over a wide range of perforate geometries. This was done to establish a good
baseline for determining the effect of bias flow over a range of perforate geometries. This
data will be valuable for verifying existing non-bias flow impedance models.

No comparisons have yet been done for the grazing flow resistance and reactance
measurements made at the Boeing, NASA Langley and GE test facilities. One problem, which
will have to be addressed, is the fact that the boundary layer characteristics are probably
different for the three facilities. Data collected by Boeing a number of years ago showed a
significant dependence of the grazing flow resistance with boundary layer thickness. In
particular, Boeing uses the boundary layer momentum thickness to define this dependence.
Although not currently part of the defined study, this dependence will have to be further
investigated.

Ref. 2 demonstrated the linearity and independence to grazing flow benefits of micro-
perforates (d = 4 mil, th = 40 mil) compared to conventional perforates (d = 40 mil, th = 40
mil). A major problem observed, however, was hole shape inconsistency in the
manufacturing process. This makes it difficult to characterize the micro-perforate geometry
and construct an impedance model. BF Goodrich demonstrated that if the hole geometry is
well controlled, as is the case for the Electro-Deposit Nickel Micro-perforate plates they
tested, the acoustic resistance is well modeled by the Poiseuille flow equation. However they
found significant variable hole distortion with the CO laser drilled, as well as the electron
beam drilled, aluminum plate samples. Northrop Grumman has demonstrated potential for
using ultra-violet lasers for drilling quality micro-perforate holes in aluminum sheet. They
were able to develop a process for achieving maximum out of roundness of If: .OO 1 in and
maximum half-angle taper of 3" for holes with exit diameter of .004 rt- .0005 in. Eventually
the availability of the high power free electron laser (FEL) is expected to make drilling these
holes sufficiently inexpensive for production applications.

The analytical and experimental study of a porous core lining for fan duct application did not
show a significant noise attenuation improvement potential relative to conventional non-
porous core liners used today. Small improvement in attenuation bandwidth was both
predicted and measured. However, there were predicted vs. measured differences in the tuning
frequency and attenuation amplitudes that could not be explained. Engine inlet application
of porous core liners was not studied. There may be potential for optimization for a range of
modes which could be useful in inlets where many more modes can be propagated compared
to fan ducts.

5.0 REFERENCES
1. Dougherty, R. P., "Nacelle acoustic design by ray tracing in three dimensions", AIAA 96-1773, 1996.
2. Bielak G. W., Premo J. W., Hersh A. S., "Advanced Turbofan Duct Liner Concepts", NASA CR-
1999-209002.
3. Premo, J., "The Application of a Time-Domain Model to Investigate the Impedance of Perforate
Liners Including the Effects of Bias Flow", AIAA 99-1876, 1999.
4. Allard J. F., "Propagation Of Sound In Porous Media - Modelling Sound Absorbing Materials",
1993, Elsevier Science Publishers Ltd.
5 . Kelly J. J., Betts J. F., Follet J. I., Thomas R. H., "Bias Flow Liner Investigation", VPI - 434487,
Virginia Consortium of Engineering & Science University, Dec. 1999.
6. Gallman, J. M., Kunze, R., K., "Grazing Flow Acoustic Impedance Testing for the NASA AST
Program", 2002-2447,2002.
7. Yu, J., Kwan, H.W., Chiou, S., "Microperforate Plate Acoustic Property Evaluation",
99-1 880, 1999.
8. Laser Micromachining for Sound Attenuation Nacelle Face Sheets", NCTA '98 Technical Report,
NCTA-AS&T-BJC-983, Northrop Grumrnan Corporation Report, 1998.
 Appendix 1 - Analysis Of ADP Model Fan Lining Test Data

INTRODUCTION
The NASA Advanced Subsonic Technology (AST) Noise Reduction Program has set a
goal of a 50% improvement in nacelle acoustic liner effectiveness relative to 1992 levels.
Analytical design studies have been conducted within the AST program to develop liner
concepts to help achieve this goal. The objective of the program described below was to
use the results of the above design studies to design liners for a specific fan/nacelle and
demonstrate the resulting benefits.

The demonstration test took place in 1996. The fan used for this demonstration was a
NASA Lewis 22 in. diameter Pratt Whitney (PW) Advanced Ducted Propeller (ADP)
model fan rig with 18 blades and 45 stators. The model was equipped with fan duct
acoustic linings designed by Boeing using the most advanced design technology available.
Advanced inlet and fan case linings were designed by PW as well. A schematic of the fan
rig is shown in Figure 1.

The test setup included special features designed to help extract detailed acoustic lining
performance information, such as a barrier to shield the aft fan noise for inlet radiation
measurements. The test matrix included a variety of lining configurations in the inlet, fan
case and fan duct sections of the rig, including hardwall in each section, as illustrated in
the schematic in Figure 1. Several different lining types were used in this test including
single layer or single degree of freedom (SDOF), double layer or double degree of
freedom (DDOF), and bulk liners. The single layer and double layer liners both had
woven wire face sheets with the double layer also having a woven wire septum. Both of
these liners were linear, meaning that the impedance is nearly independent of sound
pressure level (SPL). The baseline for 1992 level comparison was a liner produced by
Dynamic Engineering, Inc. (DEI) which approximately scaled the liners tested by PW in
the ADP engine demonstrator test conducted in 1992. Three runs were made using the
DEI liners: a run conducted in 1995 with a slightly different fan design, a primary run
during this current test, and a re-run during the current test. PW specified the last re-run
of the DEI liners as the standard for 1992 technology. Table 1 below lists the lining type
and location for each configuration to be examined in this report.
 Table 1: Test Configurations
Configuration Treatment Type
Inlet Fan Case Fan Duct
1 HW HW HW
2 DEI DEI DEI
3 SDOF DDOF SDOF
4 DDOF DDOF DDOF
5 BULK DDOF DDOF
6 DDOF HW DDOF
7 HW DDOF HW

The interstage liner was not changed unless necessary since it required the most downtime
to replace. The last two configurations listed allow for insight into the effectiveness of
interstage liners. In addition to the normal far-field measurements for each configuration
tested, several runs using an aft barrier were done to keep aft fan noise from
contaminating the inlet radiated noise data (Configurations #1, #3-5). The aft barrier was
placed parallel and offset from the exterior nacelle so as not to affect the flow in the tunnel
around the engine model itself.

The aft fan liners consisted of six different liner sections, each with its own geometric
constraints. The scope of this report is limited to analysis of the performance of the
Boeing designed lining within the fan duct. Detailed descriptions of the Boeing designed
fan duct linings and the design processes are contained in Ref. I. The PW designed inlet
and fan case linings are not included in this analysis.

Since a scale model was tested, the data was scaled up to a hypothetical full-sized engine.
This "new" engine was chosen to fit on a Boeing 747 derivative described in Ref. 1. The
conversion was from the 22" Fan Rig model to a 130" full-scale engine. PW used the
scale factor to shift the frequencies to full-scale as well as distance corrections to radiate
the levels to a 150' polar arc. The effect of the scaling is to increase the amplitudes (SPL)
at a given measurement distance by 20 log of the scale factor and to shift one-third octave
frequency bands by - 10 log of the scale factor.

ENGINE NOISE MODELING

The ADP model test data received from PW was already scaled to the full-size engine.
Both one-third octave and narrow-band data were used by Boeing to create a noise model
estimating the contributions of the inlet radiated fan broadband and tone noise and the aft
duct radiated fan broadband and tone noise components. The data was sorted by
increasing power setting, and the angles (for one-third octave data) interpolated to 20" to
160" in increments of 10" since the model scale data was not measured at these 10"
increment angles. The angles used in the far-field were referenced to the midpoint of the
ADP fan rig which is downstream of the fan itself. This causes an offset at 90" of
approximately 10" when compared with using the fan nozzle as the origin and
extrapolating the data to 150 foot polar arc full scale; likewise, there is approximately 8"
offset at 120" and approximately 4" offset at 140". The scale factor for going from a 22"
fan to a 130" fan is 5.91. This scale factor was used to scale the model N l C (corrected
rpm) in the data file to reflect the full-scale NIC. The final pre-processing step was to
unbandshare the fan tones in the one-third octave data. This process removes any portion
of a fan tone which lies in adjacent bands and puts all the energy in one band. The data
was then ready to go through the Boeing ENgine MODeling (ENMOD) Process.

Two additional assumptions were used in the modeling process for this data. First, it was
assumed that there is no jet noise present in the test data. Pratt Whitney manipulated the
scaled one-third octave data such that the low frequency and very high frequency noise
was rolled off at 1.5 dB1octave band. The low frequency roll-off was performed to
eliminate jet noise, and the high frequency roll-off was done to eliminate what was
believed to be wind tunnel noise. Second, since the turbine and low pressure compressor
flow was ducted aft of the test region, the test data consists of only fan tones and fan
broadband noise. It should be noted that the model ADP fan rig runs only at subsonic
conditions, therefore buzzsaw tones, which normally occur at high power conditions are
not present in this test data. The wind tunnel flow was around 0.1 Mach number for the
conditions used in the models.

The first step in the modeling process takes the one-third octave data and separates it into
forward and aft components. This split is done on a directivity basis for each band where
a computer program identifies the minimum SPL angle and initially defines this angle as
the point where inlet and aft components are equal (see Fig. 2). Generic forward and aft
directivity shapes, which were derived from engine data with barriers, are then applied to
roll-off the inlet and aft spectra. The program also attempts to minimize the change in
split point angle from one third octave band to the next. After completing this step, the
forward and aft components are defined at each angle for each one-third octave band.

The next step involves separating out all of the fan tone harmonics. This is done using
both 113 octave and narrow-band data. The computer program identifies the tone
frequencies in the data through the use of the fan blade count and RPM as shown in Figure
5. The process first uses the one-third octave component defined previously to determine
tone level by subtracting broadband noise estimated using adjacent bands from the 113
octave SPL. If there is no identifiable tone in the one-third octave data, the relative
difference between the tone and broadband levels in the narrow-band is used (with
correction for bandwidth) for the third octave component. The tone is then separated
from the one-third octave band data resulting in identified tone and broadband
components for both inlet and aft fan noise. Tones present in the data but not identified as
fan harmonics are removed so that they are not included in the broadband component
model. With this separation completed, the forward and aft components (tones and
broadband) are summed to create the total one-third octave spectra again. The model is
then complete.
The next step is to verify the model. This is done by comparing the measured one-third
octave data with the modeled levels. If these two do not agree, some energy in the third
octave bands was added or lost in the modeling process. This is generally not a problem
as the computer program continually checks to confirm totals with original test data and
makes changes as necessary. Also, the aft barrier data which was taken during the test for
some configurations was directly compared with the inlet component from the forward
and aft split. An inaccurate choice of the inlet and aft radiated fan noise split point is the
most probable source of error.

This modeling process was repeated for each of the seven configurations listed in the
Table 1. An example of the resulting model is shown in Figs. 2 to 4 for the hardwall
configuration at the approach, cutback, and sideline conditions. Table 2 below shows the
kinds of plots contained in the figures. These models can then easily be used to compute
the attenuation due to the inlet, interstage, or aft liners for both tones and broadband for
each configuration.

Prefix

FB5
I IB 1 IModel Inlet Fan Broadband 1 AF5 I Model Aft Fan Total I
There are some general observations which resulted from going through the process to
create the noise models for each configuration. The first fan harmonic (2BPF) is fairly
dominant throughout the configurations and is especially large at aft angles. This is quite
evident in Figure 5 where the dashed lines represent the frequency for BPF and its
harmonics and the triangle symbols denote where a tone was identified. As expected, BPF
is cut-off and not readily identifiable in the narrowband data. The higher harmonics are
present generally up to the third harmonic (4BPF). There were some problems in the data
at certain angles. The double layer liner at the sideline power condition had bad data at
147" and 158" which is evident in model plots at 150" for successive power levels. In
addition, there was no sideline power condition run for the hardwall fan case configuration
(#6 in Table 1 ), and the baseline DEI liner run was not a complete power line although the
three conditions of approach, cutback, and sideline were included in the data set.

The directivity split point between dominant inlet and aft fan noise normally was around
70" to 80". The directivities for the ADP had two peaks with a distinct valley in between
as shown in Figure 6 where the inlet component is represented by the circular symbols and
the aft component by the square symbols. The solid line is the total and where the inlet
and aft are equal define the split point. This distinct valley does not occur in every
configuration (especially at higher power settings) which results in a 10" uncertainty in the
split point choice. The split point was also confirmed by comparisons with the aft barrier
data taken for certain configurations (hardwall, single layer, double layer, and bulk). For
the high frequency bands (above band 30) the agreement was quite good. The lower
frequencies however show a fairly large disagreement between the barrier data and the
inlet noise niodel results. Examples are shown in figure 7 -9 for the hardwall approach,
cutback and sideline conditions. The reason for this discrepancy is not known. Bands 21
and 22 were corrected for jet noise by PW using the 1.5 dBIoctave roll off. However
bands 23-25 also show a significant disagreement. If jet noise and its associated roll-off is
responsible for the apparent contamination of the inlet data in the low frequencies one
would expect the problem to worsen with increased power setting. This does not seem to
be the case. In any case, caution should be used in calculating inlet liner low frequency
attenuation.

LINING ATTENUATION OF AFT RADIATED FAN NOISE

Since Boeing was contracted to design the aft fan duct liners for the ADP niodel test, only
the aft fan attenuations were calculated and compared with the pre-test predictions in this
report. There were four different aft fan duct liners tested: hardwall, Boeing single layer,
Boeing double layer, and a DEI liner. The attenuations for the aft liners can be calculated
by comparing several different configurations to extract the effect of the aft liners. The
DEI liner was specified as the 1992 technology standard to be compared against the
current liners. The differences between the single and double layer liners also give an idea
of how much improvement is possible using current technology.

Fully Treated Nacelle Aft Attenuations

Attenuations for the basic configurations (#2-5 in Table I ) were calculated by simply
taking the desired treated configuration SPLs and subtracting them from the fully hardwall
case SPLs (#I in Table I). Attenuations were calculated with the aft fan broadband noise
models described above and are shown in Figures 10-21 for the four basic treated
configurations (#2-5 in Table I). These attenuations all contain the effects of the fan case
liner and thus cannot be directly compared to the pre-test predictions which did not
include this liner. The aft attenuations with the Boeing designed single layer liner fan duct
(#3) are similar to those with the DEI single layer liners (#2) except the Boeing single
layer liner fan duct configuration appears to show more high frequency attenuation than
the DEI liner. Also the DEI liner peak attenuation is slightly larger than for the Boeing
configuration at the cutback and sideline conditions. Note that the case liner was a single
layer liner for the DEI linings and a double layer for the Boeing configuration. The
Boeing double layer liner fan duct configuration (#4) appears to have slightly higher peak
attenuation than the Boeing single layer fan duct configuration (#3) but the bandwidth of
the double layer liner fan duct does not appear appreciably better then for the single layer
liner fan duct. The aft attenuations for configuration #5, labeled the bulk absorber
configuration in figs. 19-21, are expected to be essentially a repeat of configuration #4
since only the inlet liner is different. The peak attenuations appear to repeat well but there
are bandwidth differences which are not understood. Configuration #5 shows poorer high
frequency attenuation than configuration #4 for the approach and sideline conditions but
slightly better high frequency attenuation at the cutback condition.
 Aft PNLT Attenuations
The aft PNLT directivities and attenuations calculated by extrapolation of the noise
models to the FAR conditions for approach, cutback and sideline for the design airplane
are shown in figs. 22-24. Results are shown for the fully hardwall nacelle, the DEI liner
nacelle, the Boeing single layer fan duct nacelle and the Boeing double layer liner nacelle.
The Boeing designed liners showed significant improvement relative to the DEI liner.
Also the Boeing double layer liner showed significant peak angle PNLT attenuation
improvement relative to the Boeing single layer liner at cutback and takeoff.

Fan Duct Liner Attenuations

The last two configurations tested (#6-7 in Table 1) can be used to isolate the aft liner
attenuations which were predicted during the design process. For the aft single layer liner,
only one comparison can be made for attenuations while two different comparisons can be
made for the aft double layer liner. These comparisons are shown in Table 3. In addition,
the attenuation effects of the fan case liner on aft fan noise are isolated and examined as
indicated in Table 3 even though no predictions for this attenuation were made by Boeing
prior to the test.
Table 3: Comparisons for Attenuation Calculations
Attenuation Inlet Mid Aft Attenuation Inlet Mid Aft
Fan Duct HW DDOF HW Fan Case HW HW HW
SDOF DDOF SDOF HW DDOF HW
Fan Duct HW DDOF HW Fan Case DDOF HW DDOF
DDOF DDOF DDOF DDOF DDOF DDOF
Fan Duct HW HW HW
DDOF HW DDOF

The aft fan duct liner prediction comparisons are shown in Figures 25-30 using the aft fan
broadband noise models and calculated according to Table 3 above. The plots for the fan
duct attenuations are marked according to whether the fan case was hardwall or treated.
The data are compared to predictions made for bands 27-38.

The attenuations for the Boeing designed single layer fan duct liner are shown in Figures
25-27. At approach, the predicted levels are 2-3 dB higher than measured; at cutback
they are close to the measured attenuations; and at sideline, the predicted levels are well
below the measured data except at the very aft angles. The measured peak attenuation
frequency is generally one or two bands lower than the predictions; however, the general
shape and bandwidth of the predictions agree quite well with the measured values. The
shift in peak attenuation may be due to off-design cavity depths of the liner due to the
difficulties of production in model scale. The single layer predictions also show a distinct
secondary peak at band 37 which is most likely a cavity resonance. This resonance is not
consistently observed in the measured data.
Two independent comparisons can be made to calculate the measured attenuation for the
double layer liner. The comparisons are shown in Figures 28-30. As for the single layer
liner, the predicted double layer liner attenuations were higher than the measured data at
the approach condition, agreed reasonably well with the measured data at the cutback
condition and were significantly below the measured data at the takeoff sideline condition.
The two measured attenuation calculations closely match each other below band 34;
however, there are large differences seen in the higher bands where more attenuation is
measured with the hardwall fan case than the treated fan case. The slope of the
attenuation levels at the higher bands for the treated fan case seems to agree with the roll-
off of the predicted attenuations. This is clearly seen at the cutback condition (Figure 29)
where the attenuation levels at the higher bands with the hardwall fan case do not roll-off.
Unfortunately, there was no takeoff sideline condition run for the hardwall fan case
configuration. One interpretation of this data is that for the hardwall fan case configuration
there is more energy in the higher order modes as the noise exits the case region into the
fan duct than for the treated fan case configuration. This is consistent with the idea that
the case liner will more strongly attenuate higher order modes. The Boeing prediction
code assumes a modal energy distribution which is nearly constant until the modes
approach cutoff where the energy per mode falls off rapidly. It is not known if this
distribution more closely matches the hardwall or treated fan case situation at the entrance
to the fan duct.

The standard process at Boeing uses a correction factor of 1.O at approach, 0.8 at
cutback, and 0.6 at sideline applied to the attenuations predicted with the duct
propagation code. The 0.6 factor clearly made the predicted attenuations too low at the
sideline condition. It appears that a correction factor of approximately 0.8 would better fit
the data at all power conditions if the predictions are applied to the treated fan case
reference noise levels. For a hardwall fan case the Boeing predictions with a 0.8 factor
will under predict the higher frequency attenuations.

Sound power levels were calculated from the aft broadband noise models using SPL data
from 90" to 140°. Sound power level attenuations were then determined by taking the
delta between the same configurations listed in Table 3 for the fan duct double layer liner.
Figure 31 shows a contour plot of the sound power level attenuations for both the treated
and hardwall fan case comparisons over the entire power-line. The sound power
attenuations confirm the same observations made previously. Clearly, the levels are very
similar below band 33 for both comparisons. The peak power attenuation level of 7 dB is
around band 30 in both cases, and the same increase in high frequency attenuation for the
hardwall fan case is evident. Note that the sideline condition is not included in these plots
due to missing data and/or bad data points in the spectra.

Sinqle Laver Liner Attenuations vs. Double Laver Liner Attenuations

As shown in Ref. I, Boeing expected a significant improvement in attenuation for the
Boeing designed double layer liner fan duct compared to the Boeing designed single layer
liner fan duct. Comparisons of the measured double layer - single layer differences with
the pre-test predictions are shown in Figures 32-34. The pre-test predictions showed the
most improvement for the double layer liner over the single layer liner at the approach
condition. The double layer liner did show consistent improvement between bands 25 and
30 as was predicted although generally the improvement was not as large as predicted
(except for the sideline condition). The predicted high frequency attenuation improvement
shown by the second peak centered around band 35 for the double layer liner however
was not observed. In fact in the very high frequencies, bands 36-40, the single layer liner
data showed more attenuation than the double layer liner data. The reason for this is not
known. It should be pointed out, however, that there is no way of knowing what the
actual liner impedances were at these high frequencies. Frequency bands 36-40 full scale
actually covered the frequency range of approximately 21 kHz to 66 kHz in the model
test. No demonstrated ability exists to measure liner impedance at these frequencies.

Fan Case Liner Attenuations

The effect of the fan case lining on the aft radiated noise is shown in Figures 35-37. Two
sets of attenuation spectra are shown representing the last two sets of comparisons shown
in Table 3 . The benefit of the case liner was smaller with the treated fan duct than with
the hardwall fan duct In fact, the case lining appears to increase the high frequency noise
with the treated fan duct. One would expect the fan case liner to be less effective with a
lined fan duct than with a hardwall fan duct as is observed but the apparent noise increase
at 90" and 100" is not consistent with the strong attenuation measured with the hardwall
fan duct. Note that the larger high frequency case liner attenuation with the hardwall fan
duct is consistent with the observation above that the fan duct lining was more effective
with the fan case hardwall than with the fan case treated.

As with the fan duct, aft sound power attenuation levels were calculated for the fan case
liner configurations and are shown in Figure 38. The contour shows quite clearly the
larger high frequency attenuations associated with the hardwall fan duct.

Fan Tone Attenuations

The lining attenuations discussed in the above sections were all determined from the aft
fan broadband noise models. The estimated aft fan duct radiated tone attenuations are
shown in Figures 39-44 for the Boeing designed single layer and double layer aft fan
ducts. Non-zero data points are only shown for those tones which could be clearly
identified in both the hardwall and treated narrowband data, i.e. those tones not masked
by the broadband noise floor. Also shown for comparison are the measured and predicted
broadband attenuations discussed previously. It is clear that there are significant
differences between the measured tone and broadband attenuations. Although there are
some exceptions it appears that the overall tendency is for the tone attenuations to be
larger than the broadband attenuations. As noted previously, the first fan harmonic is
dominant in the aft arc for the hardwall spectra. These figures show that both liners have
effectively attenuated this tone. The double layer liner, however, tends to attenuate more
fan harmonics than the single layer liner. This is especially evident at the cutback
condition where several fan harmonics have been attenuated. The cutback condition also
shows much higher tone attenuation levels for the double layer liner than the single layer
liner. At approach and sideline however, the tone attenuations for the two liners are
similar in magnitude. Also note that the blade passage tone does not appear in the
measured tone attenuations because it has been cut-off as expected. The Boeing
predictions more closely match the measured broadband attenuations than the tone
attenuations.

CONCLUSIONS
The evaluation of the nacelle liners relative to the 1997 25% peak aft angle PNLT
attenuation improvement goal shows that the Boeing fan duct liner designs met the goal.
At the approach condition both the Boeing single and double layer liner fan ducts showed
a very large improvement (230%) relative to the DEI liner. At cutback the Boeing double
layer liner showed a 5 1% improvement relative to the DEI liner and a 35% improvement
relative to the Boeing single layer liner. At sideline the Boeing double layer liner showed a
109% improvement relative to the DEI liner and 56% improvement relative to the Boeing
single layer liner. Boeing considers the comparison of the Boeing designed single layer fan
duct with the Boeing designed double layer liner fan duct a reasonable comparison of
1992 technology with current advanced liners with some concerns. The main concern is
related to the constraints associated with model scale acoustic liner testing. Model scale
testing of nacelle liners is not an established procedure. It is not possible to scale the full
scale designs because of the small curvature radii in model scale systems, the strength
limitations for a scaled sheet thickness and the manufacturing limitations associated with
the small orifice holes used with nacelle liners. In addition it is not possible to verify that
the liner built for the model test has the desired acoustic impedance. While care was taken
to minimize these concerns by using woven wire resistance elements without perforated
plate and careful manufacturing procedures, the lack of ability to confirm the design
impedance is a rnajor difficulty.

Relative to the Boeing single layer liner fan duct, the Boeing double layer liner showed
consistent improvement between bands 25 and 30 as was predicted, although generally the
improvement was not as large as predicted (except for the sideline condition). The high
frequency attenuation improvement predicted for the double layer liner however was not
observed. In fact, in the very high frequencies, bands 36-40, the single layer liner data
showed more attenuation than the double layer liner data. It appears that the single layer
liner high frequency attenuation was better than predicted. When the double layer liner fan
duct was evaluated with a hardwall fan case its high frequency attenuation was found to
increase. This can be interpreted in terms of modal attenuation. If the fan case has
already attenuated the more easily attenuated modes the fan duct lining efficiency drops,
because it must attenuate a larger percentage of more difficult to attenuate modes.

The double layer liner fan case attenuation measured with a hardwall fan duct showed
more attenuation at the higher frequency bands than with the treated fan duct. This is also
probably a modal effect. The fan case lining is not as important if another liner, the fan
duct liner, is available to attenuate the same modes that the case attenuates. The apparent
increase in noise by the fan case with the treated fan duct is not understood in light of its
significant attenuation with a hardwall fan duct.
The standard process at Boeing uses a correction factor of I .O at approach, 0.8 at
cutback, and 0.6 at sideline applied to the attenuations predicted with the duct
propagation code. The 0.6 factor clearly made the predicted attenuations too low at the
sideline condition. It appears that a correction factor of approximately 0.8 would better fit
the data at all power conditions if the predictions are applied to the treated fan case
reference noise levels.

There are some general observations which resulted from going through the process to
create the noise models for each configuration. The first fan harmonic (2BPF) is fairly
dominant throughout the configurations and is especially large at very aft angles. As
expected, BPF is cut-off and not readily identifiable in the narrowband data. The higher
harmonics are present generally up to the third harmonic (4BPF).

REFERENCES
1. Bielak G. W., Premo J. W., Hersh A. S., "Advanced Turbofan Duct Liner Concepts",
NASA CR- 1999-209002
 PNLT, PNL, DBA DIRECTIVITIES
115

110

105

100
18 20 22 24 26 28 30 32 34 36 38 4C 95
+ PNL/T=101.4/101.4,DBA= 89.9,N34,T 0
0 PNL/T=101.4/101.4,DBA= 89.9,N34,T 0
20 40 6; 80 100 120 14i
EPNL SEL PKPNLT EMANG PKDBA E
+ 0.0 0.0 110.1 130.0 99.4 130.0
0 0.0 0.0 110.1 0.0 99.4 0.0
Metrics - First 2 RunIDs
sym runid altevo sl vtas £pa pae
FILE 1 t/mxk805O/nap3/mP/ob/ad~.hw.o . 0
+ NlC1135 16 -150 0 0.0 0.0
FILE 2 /nap3/ADP/model/hw/adp.hw.mode
o TOT1135 16 -150 0 0.0 0.0
o IB11135 16 -150 0 0.0 0.0
o IN11135 16 -150 0 0.0 0.0
A FB51135 16 -150 0 0.0 0.0
V AF51135 16 -150 0 0.0 0.0

One Third Octave SPL vs. Band No.

150 ft. Polar Arc

0 18 20 22 24 26 28 30 32 34 36 38 40
+ PNL/T=106.8/106.8,DBA= 95.6.N34,T 0 + PNL/T=108.4/108.4,DBA= 96.5,N35,T 0
0 PNL/T=106.8/106.8,DBA= 95.5,N34,T 0 _ 0 PNL/T=108.4/108.4,DaA=96.5,N35.T 0
Pad 1 User: mxk8050 Executed: 97/08/25 TIME: 09:56 Compiled 97.08.05 10:30 bh stplwX
Figure 4: Hardwall Configuration at Sideline Condition
 1OC
SPL dB
95

90

85

80

75

70

65

Figure 5: Hardwqll Narrow-band Data at 130 deg for Cutback

95

90
SPL dB
85

80

75

70

65

60

55

50r I

L 80 100
Angle (degrees)
Figure 6: Hardwall Directivity for Band 34 at Cutback
Figure 7: Hardwall F o m d i A f t Split Comparison with Aft Earrier Data at Approach Conditio~l
43
Figure 8: Hardxrill FomardiAft Split Point. Comparison vfitJi Aft Barrier Data at Catback Conditiol:
44
IC, Q LC> a LC- CLJ
Figure Y: Hardv,rdll Forv~djAft.Split Cornparison vsiUi Aft. Earrier Data at. Sideline Condit.ior~
45
15 20 25 30 35 40
BAND

Figure 10: DEI Liner Aft Fan Broadband Attenuation at Approach

 15 20 25 30 35 40
BAND

" ,
15 20 25 30 i5 40
BAND
20

16
m
E!
12
3
C
.g 8
s
C
a,
4 4

0
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 11: DEl Liner Aft Fan Broadband Attenuation at Cutback

 15 20 25 30 35 40
BAND

1.5 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 12: DEl Liner Aft FanI Broadband Attenuation at Sideline

 15 20 25 30 35 40
BAND

15 20 25 30 35 40
BAND
20

16
,--.
-U7 12
m

4
s
2 8
m3
c
a,
44
" 7- 1 0
15 20 2'5 3b i5 40 15 20 25
BAND
30 35 40
BAND

Figure 13:Single Layer Liner Aft Fan Broadband Attenuation at Approach

 20 20

-m 16 16

-$
'u
12
5-
S
$12
-
aI
, -
aI
,
s c
.g 8 .g 8
m m
C C
a, a,
24 24

0 0
15 20 2 15 20 2
20 20

-mv 16
- 16

-$ 12
s7 12
-
aI
, a,
_I
c C
.O 8
4-
.O 8
4-
m m
C
3 s
3
a, a,
p4 p4

0 0
15 20 2 15 20 2
20 20

-mv 16
-m 16

-F 12
9
7 12
a,
_I a,
_I

C s
.g 8 .g 8
'cs m
3
8
a, a,
g4 4 4

0 0
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 14: Single Layer Liner Aft Fan BroadbandAttenuation at Cutback

15 20 25 30 35 40
BAND

Figure 15: Single Layer Liner Aft Fan Broadband Attenuation at Sideline
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 16: Double Layer Liner Aft Fan Broadband Attenuation at Approach
 15 20 25 30 35 40
BAND
20

16

-7
m
-0
12
3
c
g 8
s
a,
4 4

0
15 20 25 30 35 40
BAND

Figure 17: Double Layer Liner Aft Fan Broadband Attenuation at Cutback
Figure 18: Double Layer Liner Aft Fan Broadband Attenuation at Sideline
 15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND
20 - I 20
8=130'

16-
-m 16

-7
5-
v
12
U
7 12
2
a, a,
_I

.sm 8
8
.sm 8
LI

3 3
s s
a, a)
4

0
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 19: Bulk Absorber Aft Fan Broadband Attenuation at Approach

 15 20 25 30 35 40
BAND

Figure 20: Bulk Absorber Aft Fan Broadband Attenuation at Cutback

Figure 21: Bulk Absorber Aft Fan Broadband Attenuation for Sideline
90 100 110 120 130 140 150 160
Angle (degrees)

Figure 22: PNLT Calculations at Approach Condition

 Angle (degrees)

Figure 23: PNLT Calculations at Cutback Condition

Figure 24: PNLT Calculations at Sideline Condition
 20

-m 16

-7
U
12
a,
J
t
.g 8
98
a,
24

0
15 20 25 30 35 40
BAND
20 I
0=120'

16
65-
9
7 12
a,
-I

-
c
.O 8
m
t
3
\':

15 20 25 30 35 40
BAND

Figure 25: Single Layer Liner in Aft Fan Duct at Approach

 20

16
3
9
$12
w
_I

C
g 8
9s
a,
4 4

0
15 20 25 30 35 40
BAND
20

16

-
@
$12
a,
_I

s
.g 8
2
!
s
w
2 4

0
15 20 25 30 35 40
BAND
20

16
3
9
12
3
c
2 8
m
TIT
s
w
4 4

0
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 26: Single Layer' Llner in Aft F:an Duct at Cutback

 1 5 20 25 30 35 40
BAND
20

16
65-
U
3 12
-
aI
,
s
8
m3
s
a,
4 4

0
1'5 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 27: Single Layer Liner in Aft 1=an Duct at Sideline

15 2b 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 28: Double Layer Liner in Aft Fan Duct at Approach

Figure 29: Double Layer Liner in Aft Fan Duct at Cutback
 " ,
15 20 2'5 3b 3'5 40
BAND

15 20 25 30 35 40
BAND
20

16
-al
5 12
-2
c
.-0

-
a,
z
8

0
15 20 25 30 35 40
BAND

Figure 30: Double Layer Liner in Aft Fan Duct at Sideline

a) Treated Fan Case
20
I " " " '

j roo0 DPWR (dB)

Band

b) Hardwall Fan Case

20

20 30
Band

Figure 31 :Aft Broadband Power Attenuations

Double Layer Fan Duct Liner
 I
Predi l i o n
0=90' Differance

1
I
,-.,,
I
,
,P\/ -
/-\
1.;
'

Data ,fl
Differe lces

BAND BAND

-10
15 20 25 30 35 40 15 20 25 30 35 40
BAND
10

- 5
m
4!
-
>
a,
4 o
-
c
.-0
m
-5
3

F
-5

-10
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 32: Single Layer and Double Layer Liner Comparison

at Approach Condition
1'5 20 25 30 35 50
BAND

15 20 25 30 35 40

15 20 25 30 35 40
BAND

Figure 33: Single Layer and Double Layer Liner Comparison

at Cutback Condition
 15 20 25 30 35 40
BAND

Figure 34: Single Layer and Double Layer Liner Comparison

at Sideline Condition
 BAND BAND

15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 35: Fan Case Attenuations at Approach Condition

 15 20 25 30 35 40
BAND
10

-10
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND
10

- 5
5
-i
a,

S 0
C
.-+
0
m

-f
2

F
-5

-10
15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 36: Fan Case Attenuations at Cutback Condition

1'5 io 2'5 3b i5 io
BAND BAND
10

15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

Figure 37:Fan Case Attenuations at Cutback Condition

a) Hardwall InleUFanduct

DPWR (dB)

1 :a4

Band

b) Treated Inlet(Fanduct

20

Band
Figure 38: Aft Broadband Power Attenuations
Fan Case Liner
 20

-1 6

-
.--a

g 12
111
-I
c
'= 8
0
nj
3
c
111
24
0
15 20 25 30
BAND BAND
20

-1 6
'i!
-
w

;12
-I
c
.-+
0
8
4
c
m
24
0
15 20 25 30 35 40 15 20 25 30 35 40
BAND
I
8= 130

.
tk.,
.
w1'1 'A

15 20 25 30 35 40
BAND

F i g u r ~39: Aft Fa11 Tone and Brctidbar~dAttenuation for2

Single Layer Linet- at Approach Condition
 I BPF 2 X 3 X 4 X I I BPF 2X 3 X 4 X
I
. . . .

15 20 25 30 35 40
EAND

Figure 40: Aft F m Tone m d E3t-uadba-d Attenuations for2

Single Layer Liner at Cutback Condition
 I BPF 2X 3 X 4 X
I I BPF 2X 3 X 4 X
I

Fig1~t-e41: Aft F a i Tor~eatid Bwadband Atterruatiotis for?

Single Layer Li net- at Sicideline Condition
 'rrri 'nri
BPF 2X 3X 4 X BPF 2X 3X 4X

. . . .

BAND

15 20 25 30 35 40
-- BAND BAND

BAND BAND

Figure 42: Aft F a i Tone Emadband Attenuat ions for?

Duuble Lajfet-Liner at Akppt-ctach Condition
 I BPF 2X 3 X 4 X
I I BPF 2 X 3 X 4 X I

15 20 25 30 35 40 15 20 25 30 35 40
BAND BAND

BAND BAND

3
15 20 25 30 35 40
BAND BAND
Figure 4'3:Aft FXI Tone a - ~ Bwa4krand
d Attet~uationsfelt-3
T ) n ~ l hIlaiiet-
~ I i n ~at
r ~71lthack
(7nnditint-I
 I BPF 2X 3X4X
I I BPF 2X 3X4X
I

-- BAND

BAND

BAND

Figur-e 444: Aft Fai Tone x l c f B t - ~ ~ i d h aAttenuations

-~d for-2
Double Layer Liners at it i cdelitie Condi t ions
 -
Appendix 2 Grazing Flow Impedance Testing .Of Bias Flow
and High Temperature Liners

BACKGROUND
Grazing flow impedance testing was accomplished in late 1996 to investigate effects of
acoustic lining impedance control through heating the acoustic lining or bias air flow
through the acoustic lining. This report presents results of the data analysis accomplished
to extract lining impedance from the test data and compares the results to pre-test
predictions.

TEST PANEL CONFIGURATIONS

Bias Flow Double Laver Confiauration

The bias airflow lining panel consisted of a double layer Helmholtz resonator
configuration, with a 190 cgs Ray1 @ 105 crn/s porous mesh skin replacing the standard
impervious backing sheet. The bias airflow was introduced via a plenum chamber attached
on the back side of the lining.

A schematic of the lining geometry is shown in Figure 1.

The lining geometric parameters are :

Face sheet: 14.7 % Percent open area.

Septum : 6.0 % Percent open area

Core depths- Face sheet to septum: 3.14"

Septum to back sheet 0.76"

Heated Double Laver Confiquration

The heated lining panel was a standard double layer configuration with the impervious
backing sheet replaced with a thick aluminum sheet. This aluminum backing sheet had
slots machined in one side to accept electric heating rods. Thermocouples were installed
during panel fabrication. These were located at four levels in the double layer, at six
locations laterally, nominally in the centers of honeycomb cells. The levels were 1) just
beneath the face sheet; 2 & 3) just above and just below the buried septum; and 4) at the
lining interior surface of the heated aluminum backing sheet.

A schematic of the lining geometry is shown in Figure I.

The lining geometry parameters are:

Face sheet: 6.9 Percent open area

Septum : 4.2 % Percent open area

Core depths- Face sheet to septum: 1.31"

Septum to back sheet 2.63"

TEST CONDITIONS
The test conditions included :

Up to four grazing flow Mach numbers - 0., .25, .35, and .5

Two sound levels - maximum achievable (approximately 160 dB) and nominally 10 dB
down from maximum (approximately 150 dB)

Bias air flow - nominal 120 crn/sec average flow through the liner (M=.004), both
pressure (air exiting the face sheet into the test duct) and vacuum (air entering the lining
from the test duct)

Temperature - maximum heating not to exceed the 350 degree F limit of the liner bonding
adhesive. Thennocouples at the heated backing sheet monitored for this condition.

TEST GEOMETRY
The high frequency grazing flow duct was used for all testing. The duct flow cross section
is 2" x 2". The installed liners comprise the top wall of the test section. The microphone
axial traverse data are taken over the 16" length of the test liners.

IMPEDANCE EXTRACTION ANALYSIS

The basic test measurement is the transfer function between a fixed (reference)
microphone and a traversing microphone. The microphones are located in the duct wall
opposite the installed liner. The reference microphone is located at the approximate start
of the liner (the upstream end). The traverse microphone begins at the same axial position,
and a transfer function is acquired over the length of the liner, in 0.2" increments.

Within the extraction process, any interim impedance (some 'guess' at the start of the
process) is used together with duct conditions to obtain wave equation modal eigenvalues
and propagation constants. Modal amplitude coefficients are obtained via imposing
acoustic pressure and velocity matching over duct cross sections at the beginning and end
of the liner. Additional conditions involved relate to source and duct exit conditions. The
source is assumed to be a plane wave for frequencies below second mode cut-on. Above
cut-on, the second mode is included, and the complex modal amplitude leads to two more
independent variables in the overall process. At the duct exit, a rho-c termination is
currently assumed, leading to the absence of any upstream direction propagation
downstream of the lining. In general, seven upstream propagating modes and seven
downstream propagating modes are utilized in the three sections of the analysis -
upstream hardwall, lined section, and downstream hardwall. This allows for the higher
modes created at the hardllined interfaces due to scattering, and these modes are necessary
to satisfy matching near these interfaces. The special cases of the input source (upstream
hardwall with one or two downstream modes) and the downstream hardwall section (rho-
c termination leads to no upstream direction modes here) were discussed above.

With the modal amplitude coefficient solutions, a set of analytical transfer functions can
be generated at the same locations and frequencies as the test data. Minimization of the
residual (least squares) is then the control for the iteration process. The resultant final
impedance, and the second incident mode amplitude coefficients for frequencies above
cut-on, are the solution set.

ANALYSIS RESULTS - IMPEDANCE EXTRACTION

Bias Flow Panel

The impedance extraction results for the bias flow panel for the condition of bias flow
into the back of the panel with an average speed of 120 cmlsec (M=.004) and exiting the
face sheet into the duct (pressure case) are shown in Figure 2 for the maximum SPL
condition with grazing flow Mach numbers O., 0.25, 0.35, and 0.5. Figure 3 is for the
same bias flow speed and direction for the minimum SPL condition with grazing flow
Mach numbers O., 0.2, and 0.5. Maximum SPL conditions were approximately 160 dB
overall, with a spectrum comprised of a 1kHz. tone and resultant harmonics,
superimposed on the flow noise. Minimum SPL conditions were a reduction in the tone
level of approximately 10 dB relative to maximum conditions (OASPL of approximately
150 dB).

For the condition of bias flow from the duct into the face sheet with an average speed of
120 cmlsec (M=.004) and evacuated through the back sheet (vacuum case), the results for
the maximum SPL are shown in Figure 4, for duct grazing flow Mach numbers O., 0.24,
0.35, and 0.5. Results for the minimum SPL are given in Figure 5, for duct grazing flow
Mach numbers O., 0.2, and 0.5.

The bias flow panel was also tested under zero bias flow conditions. The maximum SPL
data are shown in Figure 6 for duct Mach numbers O., 0.24, and 0.5. Minimum SPL
conditions are given in Figure 7 for duct Mach numbers 0 and 0.5.
 Heated Panel
The heated panel results are shown in Figure 8. These results are for the panel heated to
approximately 350 degree F, maximum SPL conditions, for grazing flow Mach numbers
O., 0.25,0.35, and 0.5. The reduced SPL condition was only tested at Mach 0, and is
shown in Figure 9. The maximum SPL condition from Figure 8 is included here to show
the significant impact of sound level not only on the resistance magnitude, as expected,
but also on the frequency characteristics of both resistance and reactance. ,

ANALYSIS RESULTS - PREDICTIONS AND DISCUSSION

Bias Flow Liner

The bias flow liner impedance measurements shown in figures 2-7 do not appear to show
significant differences. The resistance values near 2000 Hz show the most changes but
this is an anti-resonance region where the impedance measurements are most susceptible
to error. Also there seems to be a trend for the reactance curves for the M=.5 grazing
flow conditions to show lower values in the mid frequency range. Otherwise, in spite of
changes in SPL by approximately 10 dB, grazing flow Mach numbers from 0-.5 and bias
flow mean Mach numbers of 0 and .004 the impedance curves are remarkably similar.
Fig. 10 shows predictions of the bias flow liner impedance spectra for the test conditions
with no bias flow using the DeadHersh prediction procedure of reference 1. While these
predictions are similar to those measured they appear to show larger SPL and grazing
flow dependence than the measurements. Fig. 11 shows the same predictions using the
Boeing orifice impedance code. The Boeing code predictions show better agreement with
the no-bias flow data with smaller SPL and grazing flow dependence.

Figs. 12-13 show predictions of the bias flow effects on the liner impedance spectra for
the test conditions using the DeanIHersh prediction procedure. The predictions show
stronger bias flow effects than measured for the 150 dB SPL condition. Smaller effects are
predicted at 160 dB as was measured. It is unfortunate that data was not measured for
smaller SPL's. In fact, the concept for application of bias flow for nacelle linings being
considered, is to use bias flow at the landing condition where the SPL's are of the order of
140 dB. Fig. 14 shows a prediction of the effect of bias flow on the test liner at 140 dB
and the test bias flow Mach number of .004 using the DeadHersh procedure. A fairly
large bias flow effect is predicted for no grazing flow with a lesser effect predicted with a
grazing flow Mach number of .5.

Boeing now believes that using the DeanIHersh impedance model to design a nacelle
candidate bias flow liner for testing may have been premature. More fundamental testing
aimed at accurate impedance modeling of the bias flow impact and its dependence on SPL
and grazing flow for an individual resistance element should be done before a more
complex double layer liner is designed. The DeadHersh impedance model may not be
sufficiently accurate for designing a bias flow double layer liner. Also, as reported in
reference 2, the basis for the Boeing bias flow test liner design used a version of the
DeadHersh impedance model with a code error. The above comparisons corrected this
error and show similarities to the measured data but indicate that the model is still in need
of improvement.

Elevated Tem~eratureLiner
Unfortunately we were not able to determine the impedance at ambient temperature with
the flow duct test data. A measurement with the normal incidence impedance tube at
ambient temperature however was made with the SPL approximately 150dB. Also there
was data measured at 150 dB in the flow duct at high temperature and zero grazing flow
(fig.9) . If this data is compared with the ambient temperature normal incidence
impedance tube data in fig. 15 one may get an estimate of the effect of increasing the lining
temperature. The high temperature resistance appears to have decreased relative to the
ambient temperature data as was predicted prior to the test. This is believed to be due to
the reduced air density at the lining with increased temperature. The reactance change
seen in figs. 9 and 15 associated with the lining temperature change is not understood.
The anti-resonance seen in the ambient temperature impedance tube data would be
expected to move to higher frequency by the ratio of the square root of the temperature as
appears to be the case but the sign change of the reactance is not predicted. Fig. 15 shows
predictions of the effect of grazing flow and elevated temperature for the test liner at 150
dB. The predictions show the reduced resistance and reactance shift and also show a
reduced effect of grazing flow at the higher temperature. Although the data is not totally
clear, fig.8 does not appear to show a resistance change with grazing flow as predicted but
this is data measured at 160 dB where we have seen a small dependence on grazing flow
with the bias flow liner.

CONCLUSIONS
At high SPLs, 160 dB and 150 dB, the tested bias flow liner impedance did not show a
strong dependence on changes in SPL, grazing flow up to M=.5, or bias flow up to
average M=.004. The small SPL and grazing flow dependence is predicted by the Boeing
impedance model. The DeadHersh impedance model predicts stronger dependence on
SPL, grazing flow and bias flow than was measured. The value of the test data is
significantly compromised due to the lack of lower SPL data. Background noise restricted
the testing to high SPLs. Testing of more basic liners, single layers, is needed to develop
better impedance modeling before a complex nacelle lining design is again attempted.

The high temperature liner test was not conclusive because of the lack of reference data at
ambient temperature. Limited data was measured at ambient temperature with a normal
incidence impedance tube which indicated that the liner resistance was reduced with
increased temperature as predicted. The reactance change observed with this data,
however, was not understood.
 Bias Flow Liner Elevated Core Temperture Double Layer

6.9%

3.14"
Core 250 deg F

x-Thermocouples

76 '

180 Rayls
@ 20 Heating I
cm/sec elements

Figure 1: Adaptive Liner Test Designs

 R1, X I M=O.OO Test
R2, X2 M=0.25 Test
R3, X3 M=0.35 Test
R4, X4 M=0.50 Test

2
Resistance 1
0
Reactance -1
-2

2 3 4 5 6
FREQUENCY - kHz

Figure 2: Bias Flow Panel, Pressure Condition, Max SPL

 R1, X I M=O.OO Test
R2, X2 M=0.25 Test
R3, X3 M=0.50 Test

FREQUENCY - kHz

Figure 3: Bias Flow Panel, Pressure Condition, Min SPL

 R1, X I M=O.OO Test
R2, X2 M=0.25 Test
R3, X3 M=0.35 Test
R4, X4 M=0.50 Test

Resistance

Reactance

FREQUENCY - kHz

Figure 4: Bias Flow Panel, Vacuum Condition, Max SPL

 R1, X1 M=O.OO Test
R2, X2 M=0.25 Test
R3, X3 M=0.50 Test

5
4
3
2
Resistance 1
0
Reactance -1

:
-4
-5 1.
lzlsBI 2 3
FREQUENCY - kHz
4 5 6

Figure 5: Bias Flow Panel, Vacuum Condition, Min SPL

 R1, X I M=O.OO Test
R2, X2 M=0.24 Test
R3, X3 M=0.50 Test
5

Resistance 1
0
Reactance -1

2 3 4 5 6
FREQUENCY - kHz

Figure 6: Bias Flow Panel, No Bias Flow, Max SPL

 R1, X1 M=O.OO Test
R2, X2 M=0.50 Test
5
4
3
2
Resistance 1

0
Reactance

-2
I
-3 . I
-4
-5 -
1 2 3 4 5 6
FREQUENCY - kHz

Figure 7: Bias Flow Panel, No Flow, Min SPL

 R1, X1 M=O.OO Test
R2, X2 M=0.25 Test
R3, X3 M=0.35 Test
R4, X4 M=0.50 Test
5
4
3
2
Resistance 1

0
Reactance

2 3 4 5 6
FREQUENCY - kHz

Figure 8: Heated Panel, With Heat, Max SPL

 R1, X I M=O.OO Max Test
R2, X2 M=O.OO Min Test

5
4
3
2
Resistance 1

0
Reactance -1

2 3
-
FREQUENCY kHz

Figure 9: Heated Panel, Mach 0, Max vs. Min SPL

 4.5

4.0

3.5

a,

-.-g
m
(I)
3.0

3 2.5
n
7J
- 2.0
a,
N
.-
2
zb 1.5

1.o

0.5

0.0
0
5

a,

.2m
0
2

n
B 1
7J
a,
- 0
.N
m
E
2 -1

- t l o
-4 8 0 9 01000
Frequency Hz

Fig. 10 Bias Flow Acoustic Liner Impedance Predictions With

DeanIHersh Procedure - No Bias Flow
 4.5

4.0

3.5

-.-gm
V)
3.0

1 2.5
n
u
-R 2.0
.-
2
b
z
1.5

1.o

0.5

0.0
0
5

-g
m
0
2

a 1
u
a,
-
.E 0
(0
E
2 -1

-2

-3

-4

Frequench Hz

Fig. 11 Impedance Predictions For Bias Flow Test Liner With Boeing
Orifice Impedance Models - No Bias Flow
 Frequency Hz

Fig. 12 Bias Flow Liner Predicted Impedance DeanIHersh

Impedance Model Effect of Bias Flow At SPL = 160 dB
 Frequency Hz
5
I I I

Frequency Hz

Fig. 13 Bias Flow Liner Predicted Impedance DeanIHersh

Impedance Model Effect Of Bias Flow At SPL = 150 dB
 Freauence Hz

Frequence Hz

Fig. 14 Bias Flow Liner Predicted Impedance DeanIHersh

Impedance Model Effect Of Bias Flow At SPL = 140 dB
 R1, X1 M=O.OO Max Temp Model
R2, X2 M=O.OO Min Temp Model
R3, X3 ZTUBE DATA

Resistance 1

0
Reactance -1

-5
1 2 3 4 5 6
FREQUENCY - kHz

Figure 15: Heated Panel Impedance Tube Data vs. Model

 3.5

-
m
.-
(0
3.0

n
8 2.5
U
d 2.0
.-
-
m
2 1.5

1.o
Temperature = 350 def
0.5

0.0
0
Frequence Hz

Fig. 16 High Temperature Liner Impedance Prediction Boeing

Prediction Model Effect Of Temperature At SPL = 150 dB
 -
Appendix 3 Acoustic Testing of Scaled Micro-Perforated Sheets

1. INTRODUCTION
The aircraft industry's current interest in using Helmholtz resonators constructed with cavity-
backed micro-diameter orifices to achieve linear or near linear sound absorbing liners
motivated the development of a preliminary design, one-dimensional impedance model. The
intent of the model is to provide the aircraft engine nacelle liner designer with a reasonable
accurate, yet simple means of specifying resonator geometry to achieve desired wall
impedance in aircraft engine applications. An experimental investigation was conducted to
measure the effects of faceplate thickness-to-orifice diameter ratio, SPL and grazing flow
speed on the impedance of Helmholtz resonators. The effects of grazing flow boundary-layer
thickness were not investigated. The measurements were used to calibrate unknown empirical
parameters derived in the one-dimensional impedance model.
2. MODEL DERIVATION
The derivation of the impedance prediction model is based on applying conservation of
unsteady mass and vertical momentum flux across the control volume of the resonator orifice
configuration sketched in Figure 1. During the inflow half-cycle, the acoustic volume flow
entering the resonator orifice through the upper control surface is denoted by uoNSON . The
quantity uvcNSvcNrepresents the sound particle volume flow exiting the control volume lower
surface, where SvcNrepresents the so-called "vena contracta" area. The volume flow VSVN
represents the grazing volume flow deflected into the resonator cavity by the driving sound
pressure field PD. The quantity HN represents the orifice lumped element inertial length. The
faceplate thickness is denoted by r, the cavity depth by Lcavand the cavity cross-sectional area
by Scav
The in-flow model shown in Figure 1 is valid only during the half-cycle when the incident
acoustic velocity is pumped into the resonator cavity - it is not valid during the other half-cycle
when the acoustic velocity is ejected from the resonator cavity. The restriction of the model to
the in-flow half-cycle is not unduly limiting because the particle volume flow pumped into and
out-of the resonator cavity must be constant over a dynamically steady-state sound period.
Thus an empirical solution over the in-flow half-cycle should result in an empirical solution over
the entire cycle.
Conservation of Mass. Assuming HN << A, the conservation of mass flux within N perforate
control volumes (all assumed equal) may be written',

Equation (1) shows that to first-order, the pumping of volume flow into and out-of a resonator
orifice is governed by unsteady, incompressible motion. This makes sense because acoustic
changes can occur only over scale lengths on the order of an acoustic wavelength.
Conservation of Vertical Momentum. The conservation of momentum in the vertical direction
may be written,

The various terms in Eq. (2) are described below:

The first term on the left-hand-side (LHS) represents the rate of increase of
momentum stored in the control volume. The lumped element inertial length
Hersh Acoustical Engineering, Inc.
Appedix 3

parameter HNis unknown and must be determined experimentally. The quantity

NSoNrepresents the open area of the N orifices.
0 The second term on the LHS represents a Bernouilli type of nonlinear increase in
momentum flux across the control surfaces SONand SvcN.
The first term on the right-hand-side (RHS) represents the incident acoustic driving
force acting on the control surface area NSON.
The second term on the RHS represents the cavity restoring force acting upon the
fluid entering the resonator cavity.
The third term on the RHS represents the momentum loss from frictional wall shear
stresses zwNdistributed over the face-plate thickness wetted area SWN= d ~ z .
Model Simplifications. The non-linearity of Eq. (2) prevents an analytic solution. It must be
solved numerically to achieve a dynamically steady-state solution, followed by a Fourier
transform to calculate the fundamental harmonic velocity component. Although this procedure
is numerically straightforward, it greatly complicates the design of sound absorbing liners.
Since the goal of this effort is to derive a reasonable accurate, but simple impedance model,
the following sixsimplifications are introduced to derive an analytic solution.
Discharge Coefficient. The first simpification introduces an acoustic discharge coefficient,
CDNanda grazing flow deflection coefficient, CvN defined as
cDN -- SvcN .
SON ' SON
The parameter CDN governs the average sound particle volume flow rate entering the
resonator cavity. It is the acoustic equivalent to the discharge coefficient concept used in
steady-state pipe The parameter CvNgoverns the steady-state grazing flow volume
flow rate deflected into the resonator orifice. Substituting Eqs. (1) and (3) into Eq. (2) results in
the following simplification,

In deriving Eq. (4), steady-state terms associated with the deflection of the grazing flow into
the cavity were ignored and only acoustic terms retained. It is intuitively clear that at low
values of SPL, resonator non-linear resistive losses become negligibly small. Under these
conditions, the non-linear term (I-CDN) should also become negligibly small. Test data,
presented later, shows that indeed CDN-+ 1 when SPL becomes small.
Cavity Pressure. The second simplification assumes that the cavity pressure can be
accurately modeled by solving the one-dimensional wave equation resulting in the following
expression,
Pea,= -ipocooCDN
cot
Scav
Non-Linearity. The third and fourth simplifications address the non-linearity of Eq. (2). These
simplifications are based upon the experimental findings of lngard and lsing4. Near resonance
and at high SPL, lngard and lsing used hot-wires to measure the amplitudes of higher
harmonic velocity components. Test data showed that the higher harmonic velocity
Hersh Acoustical Engineering, Inc.
Appedix 3

components were small relative to the fundamental. This permits the replacement of the
nonlinear term ( ~ , , ~ ~ ewith
" ' ) the
~ approximate expression,

It also permits the following harmonic relationship between uoand Po,

( t )= DvcNexp ( j o t + 8,)
uVcN (7)

where Bp is an unknown phase shift between uvcN and Po. If sound frequencies are restricted
to be at or very near resonance, then 6, = 0. To simplify the notation, the symbol (") is deleted
in the remainder of this report and it is understood that only acoustic amplitudes are
considered.
2
Equation (6) forces the nonlinear velocity term uVcNto oscillate harmonically. Physically, this
can be interpreted as the loss of higher harmonic acoustic energy - most of the harmonic
acoustic energy is retained. Since we are interested in using micro-diameter orifice liners to
achieve linear or near linear sound absorbing liners, this simplification is reasonable.
Viscous Scrubbing Losses. The fifth simplification addresses the wall shear stresses z,~ and
assumes that they are generated by steady-state and unsteady viscous scrubbing losses. A
simple model, based upon dimensional analysis, is proposed. Steady-state shear stresses are
assumed to be proportional to ,ffoCDN~vcd8avN where 8aVNis an orifice faceplate averaged
boundary-layer thickness and CDNuvcN is an orifice area averaged acoustic particle velocity.
Because aaVN is unknown, it is assumed to be proportional to doNso that steady-state shear
stresses are proportional to , u ~ C ~ ~ U ~ Acoustic
~ & ~ . shear stresses are derived assuming
"Stokes-like" axially uniform diffusion of vorticity over the orifice thickness so that it is
proportional to p o C , u V c N ~ . With these assumptions, the wall shear stress rwNiswritten,

where K s sand
~ K a care
~ unknown parameters that must be determined experimentally.

Resistance/Reactance De-Coupling. The sixth and final simplification is introduced to

achieve a simple, practical solution to Eq. (4). This simplification is motivated by the
impedance measurements shown in Figure 2 for a resonator constructed with a faceplate
thickness-to-orifice diameter ratio, ddoN>1. The measurements show resistance increases
dramatically and reactance decreases moderately with grazing flow speed. These
measurements suggest that a resistance model can be derived that is de-coupled from the
effects of reactance. Because of the relatively modest effect of grazing flow speed on
reactance, well-known lumped element reactance equations are used, modified to account for
the effects of grazing flow speed and SPL, to model reactance. In Section 3, empirical
corrections will be used to generalize the resistance and reactance models over a wide range
of frequencies below and above resonance.
Substituting Eqs. (5 - 8) into Eq. (4) and dividing by NfiSoN yields
Hersh Acoustical Engineering, lnc.
Appedix 3

Tuned Resistance. Near resonance, the real part of Eq. (9) simplifies to,

The solution to Eq. (10 ) is

where a',introduced to simplify notation, is defined as

With some modest algebra, the following tuned resistance (R,,, e P A a v ) of resonators
constructed with N circular orifices backed by a common cavity is derived,

where a is defined as,

In deriving Eq. (14), the coefficients 2 and 4 in Eq. (12) were absorbed into the parameters
CVN,K s sand
~ Kac~.
Linear Resistance. Combining Eqs. (13) and (14) and assuming V = 0 and SPL low,
resonator viscous resistive losses may be approximated as,

Since at low SPL, resistance is independent of SPL and de-coupled from reactance at all
frequencies, Eq. (15) is not restricted to frequencies near resonance.
Equation (15 ) can be further simplified by assuming that the resonator face-sheet open area is
constant, independent of number of orifices. This leads to the following relationship between
single and multiple orifice face-sheet configurations,
2
ndoN
NSoN= N--So, - ~ 4+,dON=-do,
=-
4 4 JN
Hersh Acoustical Engineering, lnc.
Appedix 3

Substituting Eq. (16) into Eq. (15) results in the following simplification,

N K +
POCO 0 cod07
=

Equation (17) shows that steady-state shear stresses are proportional to N and the acoustic
shear stresses are proportional to v/N. This suggests that resonators constructed with
sufficently large numbers of orifices such that N >> ( K ~ ~ N / K , , N ) ~ ( ~ ~will
N ~have
/ V ~ ) resistive
losses that are nearly constant, insensitive to frequency and sound amplitude!
It is of interest to compare the steady-state resistive losses predicted by Eq. (17) for the case
N = I to the steady-state losses predicted from fully-developed laminar pipe

" - 32' -
POCO 0
(0

cod01
I$')
Comparing these expressions yields KssN= 32. Thus, it is reasonable to assume that KssN +
32 for resonators constructed such that z/dol > I . Combining Eqs. (13-17), the final expression
for the resonator-tuned resistance is written,

--
Rres [

Poco
D

CON
N PD
P O C Z ~ * 20
R~
2 ~ 0 ~ 0 20
,) +[c,~+R)
2

2Poco
(19)

Observe that when M = V/co is large, RreJpoCo -+ CVNMICX

Reactance. The derivation starts from the following well-known lumped element (low SPL)
expression for the reactance of a single orifice Helmholtz resonator,

-Ode1 cot(kLcav)
POCO oco
For resonators constructed with shallow depth cavities, lngard5 derived the following
approximate expression for the orifice inertial length parameter dell

Motivated by the data shown in Figure 2 and the suggestion by lngard and !sing4that nonlinear
orifice jetting reduces orifice end correction, a simple reactance model is proposed by
replacing del in Eq. (20) by an unknown inertial length parameter HN,

- OH' cot (kL,)

POCO ace
Here HNis a function of orifice number, resonator geometry, SPL, frequency and grazing flow
speed. Empirical solutions for HNare presented below in Section 3.
Hersh Acousfical Engineering, Inc.
Appedix 3

3. EXPERIMENTAL PROGRAM
An experimental program was undertaken to provide a data base in order to generate
empirical curve-fits of the five unknown parameters KssN,KacN,CDN,CVNand HN. These
parameters are assumed to be independent of time and hence frequency. Table I below
defines the parameters of six multiple orifice resonator configurations used to derive curve-fits
to the above parameters where the subscript N denotes the number of orifices enclosed by a
cavity. The bold numbers of the N = 36 resonator denotes a deeper cavity depth relative to
the other cavities. Detailed impedance measurements of the six resonator configurations were
obtained using Dean's two-microphone method6.

Table I. Summary of Small Diameter Resonator Geometries

7

N d~ z &ON Lm SN&N CT
(inches) (inches) (inches)
1 0.375 0.500 1.333 0.850 NA 0.035

3.1 Determination of Parameters K s s ~ CDN,CVNand HN

KacN,
,
Empirical curve-fits to the above parmeters were obtained by dividing them into low SPL, high
SPL and grazing flow parts. The following notation is introduced to distinguish low from high
SPL values of HN. The subscript L denotes linear or low values of SPL, the subscipt NL
denotes non-linear or high values of SPL. Thus HLNdenotes the low SPL values of the inertial
length parameter and HNLNdenotes its corresponding high SPL values. Finally, the notation
HNVis introduced to account for the effect of grazing flow. The frequencies at which
resonance occurs as a function of SPL and grazing flow are denoted as FLN,FNLN and FNV.
Linear Parameters Kss~and KacN. When SPL is low andlor orifice diameter is sufficiently
small that d d o >>
~ 1, viscous scrubbing losses become very large relative to non-linear and
grazing flow resistive losses. Under these conditions, resonator resistance is approximated by
Eq. (17). Hersh, Walker and celano7derived the following expression for the parameter KssN,
-1.44

Ksr = 13 + 1 0 . 2 3 [ 2 ) (23)

Equation (23) shows that resonators constructed with very small values of z/do have viscous
scrubbing resistive losses that are much larger than fully-developed pipe flow resistive losses.
These results are consistent with orifice entry and exit boundary-layer thicknesses that are less
than orifice radius. Conversely, resonators constructed with large ratios of z/dohave resistive
losses substantially less than the fully developed pipe flow valued of Kss~= 32. This was not
expected and is not understood. The derivation of the fully developed pipe flow resistance
model assumes that entry flow effects are negligible. For steady state laminar flow, pipe entry
lengths as long as 150 - 300 pipe diameters have been These very long entry
lengths suggest that three-dimensional effects ignored in the derivation of fully-developed
Hersh Acoustical Engineering, Inc.
Appedix 3

laminar resistance model may play an important role in explaining the data. For example,
instantaneous laminar separation bubbles formed near the orifice entry during inflow could
explain why the asymptotic behavior of KssN< 32. Here, the orifice wetted area shear stresses
in the laminar separation bubble would be opposite to the wall stresses on the remaining
orifice wetted area thereby reducing the net stress and hence Kse

The data shown in Figure 3 was used to derive the following curve-fit to the parameter KacN,

Linear Parameter FLN.The effect of orifice number on resonator tuned frequency is displayed
in Figure 4 for the resonator configurations shown in Table I. By normalizing the data as
shown, the following curve-fit was derived,

Here FLNrepresents the tuned, low SPL, frequency for a resonator constructed with N orifices
normalized by the tuned frequency for a resonator constructed with a single orifice ( F L ~and
)
having the same open area. Equation (16) connects the dimensions of the N orifices to an
equivalent open area single orifice.
Linear Parameter HLN.The effect of orifice number on resonator inertial length parameter can
be calculated from Eqs. (22) and (25). At resonance, HLNbecomes

where a,,= 2xFLN. The RHS of Figure 5 shows that the model predicts reasonable accurately
the low SPL, zero grazing flow impedance data for the resonator configurations described in
Table I.
Nonlinear Frequency FNLN. Although it is difficult to see, the reactance data displayed in
Figure 2 shows the resonator tuned frequency to decease with increasing SPL in accord with
the earlier measurements by lngard and lsing4. The nonlinear tuned frequency is plotted in
Figure 6 as a function of faceplate thickness and sound pressure amplitude for the resonator
configurations of Table I. FNLNwas correlated in terms of the ratio of the following acoustic
velocity parameters, qN HLNand [P~~/P~]'/*,where Ppk = d2pDrepresents the incident sound
pressure peak value,

The parameter Vnonis important because a H L N characterizes the effects of resonator

geometry and [ ~ ~ k ~characterizes
~ ] ~ ' ~ the effect of sound pressure amplitude on the acoustic
behavior of Helmholtz resonators.
Hersh Acoustical Engineering, Inc.
Appedix 3

The data shown in Figure 6 suggest that the effects of orifice number and hence the ratio of
faceplate thickness-to-orifice diameter are small. The following curve-fit of FNLNwas derived,
normalized by FLN,,

where F Lis~defined by Eq. (25).

Non-Linear Inertial Length Parameter HNLN. The inertial length parameter HNLNwas
determined from experimental data by setting the reactance X = 0 (resonance) in Eq. (22)
yielding

where F Nis ~defined

~ by Eq. (28).
A computationally simpler way to calculate HNLNis shown in Figure 7. Here H N L ~ H L isNplotted
in terms of the ratio FNLN/FL~ for the resonator configurations of Table I. The data shows that at
HNLN/HLN -+ 1 when F N L N / F+~ ~1 which occurs when SPL becomes low as shown in Figure 6.
Conversely at high SPL, HNLN/HLNdecreases with SPL and is insensitive to faceplate thickness-
to-orifice diameter ratio. The following correlation of HNLN/HLN in terms of FNLN/FL~was derived,

Non-Linear Discharge Coefficient Parameter CDN. Setting V = 0 in Eq. (19), the resonator
tuned resistance becomes

With Rresand RL measured, Eq. (31) can be solved for the discharge coefficient parameter
CDNres
at resonance resulting in the expression,

'DNres
1
= -----
. yres -- P O ~ O ( P O CPOCO)
O
1+ Yres p~k

The effects of SPL and resonator geometry on CDNres was determined at resonance from
experimental data using Eq. (32). The effect of faceplate thickness and SPL on CDNres is
+ 1 as SPf. + 0. The following curve-fit to CDNres
shown in Figure 8. Observe that that CDNres
was derived,
Hersh Acoustical Engineering, Inc.
Appedix 3

The rather complicated curve-fit of the effect of frequency on the discharge coefficient data
was also derived,

'DNres + acdfrh ., fnon ---IfNL

=
+ bcdfkn + accffim f
where

Z
a,, = 18.81--57.11
do

and

Grazing Flow Parameter CVN. The grazing flow parameter CVNwas calculated from
experimental data as a function of grazing flow speed, SPL, cavity geometry and number of
orifices. The analysis of the data was substantially simplified by assuming that CDNwas
independent of the grazing flow. Physically, this ignores the complexity of non-linear
Hersh Acoustical Engineering, lnc.
Appedix 3

interaction between the sound pressure field and the grazing flow. This interaction is highly
three-dimensional and beyond the scope of the one-dimensional model.
Restricting the frequencies to resonance, the following expression for CVNres
was derived from
Eq. (19),

where R,denotes resistance measured at a function of grazng flow speed, M denotes grazing
flow Mach number, R, denotes resistance in the absence of grazing flow and RLdenotes linear
resistance. The results of applying Eq. (37) to the data are shown in Figure 9. The data was
correlated in terms of the parameter [p,~2/~pk]"2.Figure 9 shows that when [ p o ~ 2 / ~ ~ p2k10,
]'/2
Cv~~~~asymptotically appraches a maximum that is dependent upon the ratio ddoN. Further, the
asymptotic value of CvNres decreases with increasing %IdoN.
The following curve-fit of the data shown in Figure 9 was derived,

where

Grazing Flow Tuned Frequency and Inertial Length Parameters FNVand HNv. The effect of
grazing flow on the the resonator tuned frequency FNV and the inertial length parameter HNV
are plotted in Figures 10 and 11 respectively. A close examination of Figure 10 shows that the
tuned frequency of the single orifice configuration was the most sensitive to grazing flow,
increasing by almost 20% relative to its non-grazing flow, low SPL value. The tuned
frequencies of the remaining resonators, constructed with N 2 4, were insensitive to grazing
flow. The following, rather complicated, curve-fit of the resonators tuned frequencies was
derived,

The inertial length parameter HNVis determined from experimental data by setting the
reactance X = 0 (resonance) in Eq. (22) yielding
Hersh Acoustical Engineering, lnc.
Appedix 3

where FNv is defined by Eq. (40).

A computationally simpler way to calculate HNv is shown in Figure 11. Here HNdHLN
is plotted
in terms of the ratio FNV/FNL.
The data shows that at HNV/HLN -+ 1 when FNV/FNL
+ 1 which
occurs when SPL and grazing flow become small as shown in Figure 10. Conversely at high
SPL and high grazing flow speeds, HNV/H~~ decreases as shown. The following curve-fit of
HNV/HL~ was derived,

3.2 Model Predicted vs. Measured Impedance

HN, CDNJand CVNwere used to predict
The empirical curve-fits of five parameters KssN, KacNJ
the impedance of the resonators of Table I. Since the curve-fits derived for these parameters
were based upon the impedance measurements of these resonators, it is reasonable to expect
good agreement between model predicted and measured impedance. This will be shown to
be true. The model needs to be validated by comparing its prediction to other impedance
measurements published in the open literature.
The comparisons between predicted and measured impedance are displayed in Figures 12 -
14. The zero grazing flow data, summarized in Figures 12(a-f), show that non-linear Bernouilli
resistive losses occur over the entire orifice number range even for the N = 64 ( ~ / =d 10.67)
~~
configuration, although the effect of SPL is substantially less than at lower values of N. The
data shows that reactance is insensitive to the effects of SPL. The model predicts both
resistance and reactance quite accurately over the entire frequency, SPL and orifice number
range tested.
The effects of grazing flow on resistance are displayed in Figures 13(a-f). The grazing flow
resistance data is not understood. The data shows inconsistent trends with frequency - it
increases and/or decreases with frequency. Although this behavior suggests that the
measurements may not be accurate and shoud be repeated, we believe the measurements
may be accurate near resonance. The measurements do suggest, nontheless, that the effects
of grazing flow are small for resonators constructed with z/doN > 9.
The effects of grazing flow on reactance are displayed in Figures 14(a-f). The data shows that
reactance is fairly insensitive to orifice number and SPL and only moderately sensitive to
grazing flow. The model predicted reactance is in good agreement with data.
4. CONCLUDING REMARKS
The model satisfactorily predicted the effect of orifice number, SPL and uniform grazing flow
on the impedance of multiple orifices backed by a common cavity. The effects of boundary-
layer thickness was not studied. These findings, however, are valid only for the single orifice-
to-orifice spacing of SN/dN= 2. The effect and hence the importance of relative spacing has
not been studied. In a effort to at least to identify the importance of relative spacing, two
resonators with N = 25 were constructed with spacings of S25/d25= 1 and 3 as summarized in
Table II below. The S25/dz5= 1 spacing was as close as practical (without orifices merging)
and the S25/d25= 3 spacing as far as practical (to avoid cavity wall interaction effects). The
effect of array spacing on resistance is displayed in Figure 15 and the corresponding effect on
reactance is displayed in Figure 16.
Hersh Acoustical Engineering, lnc.
Appedix 3

Figure 15 contains four graphs. The LHS graphs labeled "a" and " b compare the effects of
array spacing for SPL = 1201140 dB at V = 0. These graphs show a negligible effect of
spacing on resistance. In contrast, the RHS graphs labeled "c" and "d" show a large effect of
array spacing on resistance at V = 80 mls. Figure 16 shows that the corresponding effect of
array spacing on reactance is much less dramatic. The data shows clearly, however, that
increasing array spacing decreases orifice inertial reactance. The decrease appears to be
insensitive to grazing flow speed and SPL. These tests show clearly that the effects of orifice
spacing with grazing flow have an important on resonator resistance. Further tests are
required to establish a larger data base to better understand the effects of orifice array
spacing.

Table II. Resonator Used in Array Spacing Tests

N do^ z Lcav SN&N CT
(inches) (inches) (inches)
25 0.075000 0.50 0.85 1.O 0.035
25 0.075000 0.50 0.85 3.0 0.035

The above test program was conducted with the orifice arrays orientated perpendicular to the
incident grazing flow. A test was conducted to measure the effects of rotating the array 45'
with respect to the grazing flow. Test results, displayed in Figures 17 and 18, show that the
effects of resonator orientation have only a small affect upon resonator resistance and
reactance. This suggests that orifice interaction in the form of upstream wake impingment can
be ignored for grazing flow speeds below 80 meterslsecond.
The effects of orifice spacing and orientation with respect to grazing flow are important issues
that were not included as part of this study. The limited data show, however, that orifice
spacing is important and the effects of array orientation to grazing flow may be important at
grazing flow speeds in excess of 80 meters/second. Their effects should be understood and
incorporated into a liner design prediction code. Finally, as mentioned earlier, the model
should be validated by comparing it to impedance data in the open literature.
The authors acknowledge the insightful and important contributions made by Mr. Gerald Bielak
of the Boeing Commercial Aircraft Company.
REFERENCES

1. Temkin, S. Elements of Acoustics, John Wiley & Sons, pg. 179, 1981.
2. AS ME. Fluid Meters - Their Theory and Applications. AS ME, 1959.
3. Schlichting, H., Boundary-Layer Theory, McGraw-Hill 'Book Company, pgs 176-178, 78-
80,560, 1968.
4. Ingard, U. and Ising, H., "Acoustic Non-Linearity of an Orifice", JASA, Vol. 42, No. 1. 6-17.
1967.
5. Ingard, U., "On the Theory and Design of Acoustical Resonators", JASA, V. 25, pp. 1037-
1062,1953.
6. Dean, P. D., "An In-Situ Method of Wall Acoustic Impedance Measurement in Flow Ducts",
JSV, Vol. 34, No. 1, p. 97, 1974.
Hersh Acoustical Engineering, lnc.
Appedix 3

7. Hersh, A. S., Walker, B. E., and Celano, J. W., " Helmholtz Resonator Impedance Model.
Part 1. Non-Linear Behavior", submitted to AlAA Journal, October, 2001.
N
-
.
5
73
'-t
m"
m
0
Q)
.-
0
E
6
0
C,
(i,
V)
Q)
s
Y
.-0
z
I-
Q)
-
CI
([I
n
Q)
0
([I
LL
rC
0
Figure 9. Effect of Orifice Number and Grazing Flow on Resonator Grazing Flow Parameter C,,
Figure 11. Correlation of Inertial Length Parameter in Terms of Tuned Frequency Parameter
 0
C""""""' 4m
X
c
U
ffl
u
2
U.
X
U
ffl
u
9
LL
 SPL = 120 dB SPL = 130 dB SPL = 140 dB
"575r-----~

Frequency - Hz Frequency - Hz

Figure 14b. Measured and Predicted Effect of Grazing Flow on Reactance of N = 4 Resonator
Non-Dimensional Resistance - R/poCo Non-Dimensional Resistance - fl/poco

Non-Dimensional Resistance - Rip,c,

Non-Dimensional Resistance - Rip,c,
4 a. SPL = 120 dB b. SPL = 130 dB c. SPL = 140 dB
4
47T- 1

I
CI

-7-d"
200 300 400 500 600 700 800 900 1000

Frequency - Hz Frequency - Hz Frequency - Hz

Figure 18. Effect on Rotating Array 45' on Reactance of N = 4 Resonator Configuration: V = 80 mls
Appendix 4 - Extended Reaction Lining Study

INTRODUCTION.
Current nacelle acoustic liners are designed around the assumption of local reaction. That
is, the impedance at a given point on the liner only responds to the acoustic and flow
conditions in its immediate vicinity (with the region small compared to the acoustic
wavelength). This assumption has been shown to hold for presently installed liners which
are typically made up of a face sheet and one or more layers of honeycomb cells.

A correctly designed single layer locally reacting liner will provide the best possible
attenuation for a single mode at the design frequency. The attenuation bandwidth can then
be increased by adding additional layers, but the locally reacting liner can only be
optimized for one mode (or incidence angle).

The purpose of this study is to investigate the potential for attaining additional
improvements in attenuation, relative to the locally reacting case, by allowing acoustic
propagation inside the liner. In a duct, a phase difference exists between neighboring cells.
The phase difference is a function of incidence angle (effectively mode number at a given
frequency, as each mode can be thought of as striking the liner at a different incidence)
and the spatial separation between cells. This phase difference leads to a varying pressure
distribution down the two neighboring cells, which, in turn, will drive the 'in-liner'
propagation for acoustically connected cells.

The ideal outcome would be for any additional attenuation gains attained by extended
reaction to compensate for the replacement of mesh facing sheet single layer liners with
more durable perforated facing sheet liners. Locally reacting mesh liners have better
(broader) attenuation profiles than perforate liners.

The questions investigated by this work are two-fold, with the common goal of designing
a test to investigate the most promising extended reaction designs. These are,

1. Does the extended reaction concept exhibit any potential for liner optimization for
more than one mode, given that different modes will have different phase relationships
between neighboring cells (because of their differing incidence angles)?

2. Is there a potential for increased attenuation bandwidth?

Two approaches have been adopted. The first involves the validation and investigation of
two codes designed to predict the impedance of an experin~entalconfiguration, with two
cells linked by a resistive element, and the phase of the source to each being independently
controlled. These codes can then be used to model a finite length non-locally reacting
liner, where, as previously stated, a phase difference exists between neighboring cells.

The second study is being performed using the Boeing liner optimization and duct
attenuation program (MELO). The code is being used to investigate the modal duct
attenuation of various configurations of extended reaction liners, with downstream
propagation only.

NORTHROP-GRUMMAN DOUBLE IMPEDANCE TUBE STUDY.

Test Setup
Northrop-Grumman has designed a dedicated test to investigate the acoustic behavior of
non-locally reacting cells, using a concept suggested by Ron Olsen of Boeing. They have
constructed a double impedance tube arrangement which can determine the acoustic
impedance of two neighboring cells with the common wall linked via a resistive mesh. The
cell samples are 1.5" square and 1.42" deep. Fig 1 shows a sketch of the set-up.

The SPL and phase of the sound incident on each cell can be independently controlled. A
phase difference in the incident sound at each cell leads to a pressure difference across the
common wall, varying with frequency. This pressure gradient will drive sound through the
common wall resistive layer, leading to a change in the surface impedance of each cell.

Test Results.
Northrop-Grumman has supplied impedance data (as a function of frequency and phase
difference) for three configurations. All three have 40 cgs Rayl face sheets. The
configurations are,

1. Hard common wall.

2. 40 cgs Rayl common wall.

3. 80 cgs Rayl common wall.

The raw data are plotted in figs 2 to 9 for the cases with the resistive common walls. A
positive phase difference indicates that cell 1 leads cell 2. For ease of interpretation of the
sensitivity to phase angle, the raw data has been plotted at 20" intervals.

Inspection of the plots provides a high degree of confidence in the data quality. This is
observed via the consistent trends of the data between the 40 cgs Rayl and 80 cgs Rayl
common wall samples as a function of frequency and phase difference.

The 40 cgs Rayl common wall liner has zero reactance (a resonance) at - 2100Hz, at zero
phase (where the liner behaves as if it is locally reacting). At IOOOHz, the cell I resistance
shows almost no variation with phase angle, whereas the cell 2 resistance increases by
-0.6pc at a phase angle of 60". A small shift in frequency to 800Hz results in a significant
change in the impedance behavior with phase, probably associated with an anti-resonance.

At 800Hz, the zero phase or "locally reacting" resistance measured for both cells shows a
sizeable jump from that expected, i.e. - 1 . 7 8 ~instead
~ of .96pc (i.e. 40 cgs Rayls). The
resistance measurement is much more susceptible to error near anti-resonance so the
validity of the 800 Hz data is somewhat questionable.

However, even though the 'baseline' (i.e zero phase or locally reacting) value of - 1 . 7 8 ~ ~
is high, there is still a significant systematic variation in impedance with phase angle at 800
Hz. Cell 1 and cell 2 exhibit opposing trends in resistance - cell 1 resistance decreases
from the baseline level to - 1 . 3 5 ~at~60°, whereas cell 2 increases rapidly to -13pc at the
same phase angle. Hence, if these effects are real, and not resulting from experimental
inaccuracies, the predictions may be able to pick it up. Prediction/measurement agreement
under these conditions (with the rapid fluctuations) would verify both the data and the
predictions. This question is addressed in Impedance Predictions section.

At frequencies above the resonance, e.g. at 4000Hz, the resistance of cell 1 increases by
-0.4pc and the resistance of cell 2 varies to a much smaller degree (actually reducing by
- 0 . 1 5 ~ ~This
) . is somewhat opposite behavior to that at around 800Hz.

The 800Hz data still easily exhibit the most phase-sensitive trends. Unfortunately, it
becomes progressively more difficult to test at lower frequencies to see if these trends are
continued. This may be worth further inspection by attempting perhaps 600Hz. Hopefully,
any inaccuracies in the data will not mask the trends of impedance with phase angle.

The reactance data exhibit a different trend to that for the resistance, in that there is little
or no effect with phase at the higher frequencies (i.e. typically above 3000Hz), with an
increasing sensitivity as frequency reduces. Below the resonance frequency, there is an
appreciable change in reactance with phase for both cells. At IOOOHz, the reactance of cell
1 increases by -0.7pc at 60°, whereas the cell 2 reactance decreases by -1.2pc at the
same phase angle. Again, a small shift to the lower frequency of 800Hz has a dramatic
effect. Cell 1 exhibits similar but slightly larger magnitude phase behavior as that for
1000Hz. However, cell 2 varies extremely rapidly with phase, to the point that reactance
goes through a large negative value at 40° , before quickly reversing in sign to go positive
above -60" and then continuing to go further positive at higher phase angles (not shown).
This is associated with a rapid change in the resistance at this frequency.

The 80 cgs Rayl common wall data (figs 6 to 9) show the same trends (with frequency and
phase angle) as the 40 cgs Rayl data. The variations in impedance, however, are reduced
for a given phase angle. This is consistent with intuitive expectations where, in the limit of
increasing comnlon wall resistance, a hard wall is reached and the liner reverts to the
locally reacting case.

In general, in order to attain a greater bandwidth at low frequency, a less negative

reactance is preferred. Therefore, on fxst inspection, the results from this impedance test
are not very promising. Certainly, from lkHz to 2 kHz the cell 2 behavior is in the wrong
sense in that the reactance gets more negative as the phase difference increases. The
picture is more complex with these liners, though, in that resistance also varies with phase
angle and frequency. In general resistance varies little with frequency for single layer
locally reacting linear liners, thereby making reactance the most flexible design parameter
in obtaining an impedance match. Further investigation is required, following validation of
the prediction routines, to see if,

1. the impedance change of one cell favorably outweighs that of the other cell.

2. the behavior is different for a larger number of linked cells.

3. the variation of impedance with phase can be exploited to achieve a better match to
the optimum impedance for more than one mode (incidence angle) at a given
frequency.

Impedance Predictions.
Two impedance computer models have been constructed for prediction of the double
impedance tube data. The first of these, Zmatrix, was developed by Fred Hutto of Boeing,
while working at Northrop-Grurnrnan. The second is an extension to the existing Boeing
impedance prediction code, Ymod, and was developed by Bielak and Premo. Both codes
were used to predict impedance as a function of phase angle for the discrete frequencies
tested by Northrop-Grumman (800Hz and 1,2,3,4KHz).

The two models use the 'lumped element' technique to model an acoustic lining. This
treats each element of a liner separately (e.g. face sheet, cavity) and, assuming plane wave
propagation, solves a matrix equation for nodal pressures and particle velocities. Both
codes allow propagation between two neighboring cells with a variable phase difference
(and SPL) between the sources.

As previously stated, the variation in phase at the face sheet leads to varying spatial
pressure distributions down into the two cells. This variation is modeled by splitting the
cavity into a number of sub-cavities. Five sub-cavities have been used in these models but,
in the limit, there will be a continuous pressure distribution down each cell.

Figs 10 to 19 show the impedance predictions against the raw data for the 40 cgs Ray1
common wall. The unconnected points represent the data and are plotted at the actual
phase angles tested between the drivers, not interpreted values as in the previous section.
The predicted degree of variation in impedance with phase is in broad agreement with the
data, for both models, i.e., limited variation around the resonance (e.g. 2,3kHz) and larger
variations away from the resonance. However, although qualitatively good, the
quantitative agreement is not.

Note that the agreement with the models would be better if the impedance tube measured
resistance was at the nominal value of 1pc. This is the DC flow resistance value but was
-
measured higher at some frequencies (- 1.8pc at 800Hz and 1.2pc at 1KHz) by the
impedance tube with a cavity backing. This phenomena has been discussed earlier. At
higher frequencies, the impedance tube measured value was nearly equal to the DC flow
resistance value of lpc.
At 800Hz, Ymod predicts little variation in impedance with phase angle. However,
impedance varies significantly with phase for both cells (in particular, cell 2) and especially
at the higher phase angles. At this frequency, Zmatrix agrees with the data much better
than Ymod, predicting a much larger excursion in resistance and reactance than the latter.
Zmatrix predicts a negative shift in cell 2 reactance with increasing phase angle followed
by a sign reversal at -90'. Though underpredicting the severity of the measured data
behavior, the ability of Zmatrix to predict this general trend lends weight to the confidence
in the low frequency data. The cell 1 impedance, while varying to a much lesser extent
than cell 2, is also better predicted by Zmatrix.

At IKHz, Zmatrix also shows the better agreement with the data. At 2KHz and 3KHz,
both methods predict the data to an acceptable degree of accuracy. At 4KHz, Ymod
provides the better prediction.

To summarize, both methods exhibit potential. However, the first question which needs to
be addressed is why the models differ in their estimations. This is particularly true at low
frequencies (I1000Hz). Some investigations have been carried out but will require further
work. Once this discrepancy is clarified, the best model (or either if they agree), could be
used to try to exploit the behavior with phase angle for acoustic advantage. Also, the
observed reactance behavior at low frequencies and for a larger set of connected cells
remains to be fully addressed. In order to perform the former task, validation of one of the
models is important, given the lower limit of the tube is -800Hz.

MELO STUDY

Introduction.
MELO (Multi-Element Liner Optinizer) is a liner optimization and duct attenuation
calculation program developed by Boeing. The program will optimize a given liner
construction to attain the maximum attenuation of the modes propagating downstream in
a given duct geometry.

The purpose of this part of the extended reaction study is to investigate the attenuation of
various non-locally reacting geometries and compare them to what is achievable with
locally reacting liners. MELO has initially been used to look at two rectangular duct
geometries. The first was 5cm x 20cm (height x length), the second 50cm x 200cm.
-
Primarily, only the lowest order mode propagates (cut-off frequency is 3400Hz) for the
smaller duct. Running the code, for this geometry, enabled familiarity to be gained and
confidence to be established in the predictions for extended reaction configurations.

The second geometry is more representative of a full scale bypass duct application. Given
the large LIH of most bypass ducts, higher order modes are generally well attenuated and
lining optimization tends to focus on maximizing the attenuation of the lowest order
mode. Hence, when considering bypass ducts, the emphasis for extended reaction liners
will be on achieving greater bandwidth rather than multi-modal attenuation. Inlets should
be studied with multi-modal optin~ization.
The prime target for this study is to evaluate the potential attenuation gains achievable
with extended reaction liners in a fan duct. This will be achieved by validating MELO.
With this in mind, a duct propagation test was devised. For convenience of manufacture
and analysis, this was a 2-D test and was performed by Northrop Grumman. Once
validated, MELO can then be used for full scale applications.

Initial MELO studies

Predicted Attenuations.
A single layer, locally reacting, linear liner was chosen as a datum. For convenience, this
liner was optimized for attenuation of the lowest order mode at 2000Hz. The exercise was
then repeated for the larger duct geometry.

In addition to the locally reacting liner, the following cross-section of extended reaction
configurations were also assessed,

1. 4% open area perforated core (i.e. 4% POA).

2. 8% open area perforated core (8% POA).

3. Icm flute @ O0 to the duct axis. A flute here is used to describe a channel within the
liner.

4. lcm flute @ 90° to the duct axis (essentially locally reacting, given a constant phase
and amplitude across the duct).

5. Icm flute 0 45' to the duct axis.

6. 40 cgs Rayllcm bulk absorber.

7. 80 cgs Rayllcm bulk absorber.

8. 40 cgs Rayllcm bulk absorber, locally reacting.

The results presented here are for non-locally reacting liners having the same depth and
face sheet resistance as the locally reacting liner. Further studies have optimized the
individual extended reaction designs for test in the 2-D flow duct.

5 cm Duct.
The locally reacting datum liner for this duct is a single layer liner with a 9.9 cgs Ray1
linear face sheet and a core depth of 3.8cm. The attenuation comparison for the lowest
order mode is plotted in fig 20. The y-axis has been limited to 30dB, in order to identify
the bandwidth behavior more easily.

The results for this duct provide confidence in the MELO predictions, in that,
1. The locally reacting liner provides the best attenuation at the design frequency, 2KHz
(the liner was designed with the Cremes optimum impedance at this frequency).

2. As expected, the attenuation characteristics of the flute at 90' to the duct axis
approximate those of the datum liner.

3. The bulk absorber shows a broad attenuation bandwidth, with the locally reacting bulk
absorber performing slightly better in this respect. This is consistent with expectations.
It is worth noting that the bulk liner attenuation peaks at -4000Hz and significantly
exceeds the locally reacting attenuation, which dips at the anti-resonance at this
frequency.

4. The 4% POA perforated core produces attenuation levels closer to the locally reacting
case than the 8% POA core. In the limit (i.e. 0% POA), they would be identical.

5. As the flute angle becomes more aligned with the duct axis, the observed attenuation
decreases.

In addition,

6. Although the peak attenuation of the 4% and 8% perforated core liners was lower
than for the locally reacting liner, both comfortably exceed 30dB and the attenuation
away from the design frequency improved.

50cm duct.
The locally reacting datum liner for this duct is a single layer liner with a 118 cgs Ray1
linear face sheet and a core depth of I. lcm. This duct has multiple modes propagating.
Figs 21 to 23 show the modal attenuation for modes 1 to 3. As expected, the higher order
modes are more efficientIy attenuated (note that mode 3 is cut-off at 500 and 630Hz).
When comparing the lowest order mode attenuation spectra for both ducts, some points
are notable,

1. The attenuation characteristics are different because the behavior depends on h/H. H
has increased by a factor of 10 for the larger duct. When the wavelength becomes
small relative to the duct height, a significant degree of beaming occurs and the wall
amplitudes are reduced.

2. The bulk absorbers do not provide the same relative gains in attenuation bandwidth
especially at high frequencies (2 3 150Hz). However, their attenuation levels are much
closer to the locally reacting levels around the tuned frequency and are almost
symmetrical about this frequency.

In addition, in the larger duct where multiple modes propagate,

3. The attenuation spectra retain a fairly constant profile from mode to mode. The peak is
shifted up by 113'~octave as mode number increases.
4. As the perforated core POA increases, the attenuation peak reduces and moves to
higher frequencies.

5. As the flute angle, of the fluted core, moves towards the duct axis (i.e. towards 0°),
the attenuation peak also reduces and shifts to higher frequencies.

These predictions do not provide convincing evidence for any improvements in attenuation
from the extended reaction liners modeled using MELO, except from the increased
bandwidth observed from the bulk liners, but primarily for the 5cm duct. However, these
liners were all assessed with the face sheet resistance and cell depth of the optimized single
layer locally reacting liner. A full optimization, for example, on perforated core POA, flute
angle and bulk absorber density should lead to more promising results.

This study satisfied its objective of assessing the general behavior of the MELO
predictions. Finally, bandwidth comparisons should be made against a double layer locally
reacting liner, the liner type most commonly installed for enhanced bandwidth. These
studies are discussed later where MELO predictions are computed for 'real' fan duct
applications.

Duct Propaqation Test

Introduction
MELO has been used initially to evaluate the attenuation behavior of extended reaction
configurations. This code has not been validated against test data for these liners. Hence, a
duct propagation test is necessary in order to assess the accuracy of the attenuation
predictions.

It is known that MELO performs better for a 2-D rectangular duct. Hence, a 2-D duct
propagation test will most easily investigate the accuracy of the predictions. In addition, a
2-D duct is easily manufactured and the modal attenuation analysis is greatly simplified
(because, in one plane, only the lowest order mode exists).

Liner and Test Definition

Fig 24 shows a schematic of the test set-up. The lined duct dimensions were chosen to
represent a typical fan duct. The duct height at 22" was chosen to conveniently mount the
four speaker horns necessary for the combined requirements of a broadband source and
tones with a preferred modal content. The lined duct length of 3' gives an L/H of 1.64. A
higher LID would compromise the overall length of the device, given the size of the
anechoic chamber. The duct width was chosen at 1.3" to be shallow enough to ensure the
acoustic pressure didn't vary in that plane over the frequency range of interest (500Hz to
4000Hz).
The exponential horns on the speakers are designed to give a cut-off frequency well below
500Hz. Fibreglass wool is placed in the horns to reduce the interference between them. A
mic is also inserted in each horn to ensure consistency in level between runs.

The samples, both locally reacting and extended reaction, were constructed from wire
mesh facing sheets and, if applicable, wire mesh walls. Given the goal of replacing wire
mesh single layer liners with perforate liners, the latter were not used because the
relatively low duct SPLs would drive down the facing sheet POA, leading to unacceptably
high facing sheet mass reactance.

The purpose of the test is to fully validate the MELO code, so it was desirable to design a
range of samples which would cover the full spectrun~,of the concept. MELO was used to
optimize locally reacting and extended reaction mesh single layer liners for maximum
OASPL attenuation and maximum PNLT attenuation of a typical mid-twin wide chord
cutback spectra at 1 10'. The optimized locally reacting and extended reaction liners
designed have been simplified for ease of manufacture and are listed in table 1. These
liners present a satisfactory cross-section of liners to explore the concept of extended
reaction fully and are adequately different to hopefully be able to easily distinguish the
attenuation spectra.

Two types of measurement are proposed for the samples. In the fxst phase of testing,
insertion loss measurements will be performed by taking a log average of the SPLs of the
far-field mics in the anechoic chamber, and subtracting this from a similar measurement for
either a datum sample or a hard wall test section. This can then be used to validate MELO
predictions of the overall attenuation for all propagating modes.

The second phase of testing will depend on the outcome of the insertion loss
measurements. If these show promise, modal measurements will be made. This will entail
using 4 rows of 12 mics inserted 5" from the start and end of the lined section ( i.e. 2 pairs
of 2 rows). These are needed to identify the modal content of forward and aft radiated
sound. The speakers can be driven in specific combinations of phase relationship to
preferentially excite a given mode. The measurements will provide further information on
the mode-by-mode (and hence incidence-dependent) attenuation, which MELO is also
able to predict.
Table I Optimized Samples

Sample # DC Flow Resistance Depth (cm) Common Wall

(cgs Rayls) Resistance (cgs Rayls)

1 SO 3.2 20

-
7 90 3.2 Hard (i.e locally reacting)

3 50 3.2 40

4 50 3.2 0 (i.e. no cells)

5 50 3.2 Hard

Notes:

Samples 1 and 2 approximate the MELO optimized output

Sample I - Max PNLT attenuation, extended reaction

Sample 2 - Max PNLT attenuation, local reaction

Samples 3,4, 5 - Additional samples to complete a parametric set

Attenuation Predictions
Fig 25 shows the MELO attenuation predictions for the samples listed in table I . As
expected from their parametric nature, the attenuation profile varies substantially from
sample to sample. The predictions indicate that the difference in profiles should make it
possible to measure the influence of extended reaction. The 'optimized' (approximate, as
samples are a parametric set) extended reaction liner has a 0.51dB increase in PNLT
attenuation over the optimized locally reacting liner.

An attempt was also made to optimize extended reaction liners, both mesh and perforate
for a 'real' application, i.e. with flow. The attenuation predictions are shown in fig 26. The
PNLT attenuation gains exhibited for mesh extended reaction liners over mesh locally
reacting liners with zero flow, are reduced substantially with a grazing flow Mach number
of 0.51 (now 0.22dB instead of 0.5 1dB). However, the gain from extended reaction is of
the same order as the gain of the mesh over the perforate (single layer) facing sheet.
Unfortunately, the gain from extended reaction shown for mesh liners does not carry
through for perforates. At M =0.5 1, perforate extended reaction liners do not show any
attenuation gain over local reaction (and hence this liner plot is superfluous).

The extended reaction single layer mesh lining PNLT attenuation still lies well below the
double layer capabilities, which are 1.12dB (perforate) and 1.31dB (mesh) higher.

These predictions, which will be evaluated fully following the duct propagation test,
indicate that extended reaction will not exhibit any significant potential for 'real'
applications in the fan duct (nil gain for perforates). However, this is not necessarily true
for inlets, where multiple modes are propagating. The wavefronts are also more normal to
the liner. Hence, the gain in attenuation can be expected to be greater than for fan ducts
(given extrapolation from M =-0.5 1 through M=O.O to positive Mach numbers). This will
be investigated to see if any gains in the inlet justify the incorporation of extended reaction
liners.

CONCLUSIONS AND RECOMMENDATIONS.

This report has detailed the status of the extended reaction study. The objective is to
investigate the concept thoroughly through the validation of MELO with a 2-d duct
propagation test scheduled at Northrop Grumman. Then the validatd code can be used for
assessment of these liners for 'real' applications.

Early work concentrated on the validation of three prediction schemes, two of which are
aimed at understanding the fundamental process of propagation between cells while the
other optimizes extended reaction configurations for maximum attenuation in a given duct
geometry.

The two double cell prediction schemes show good qualitative agreement with the
Northrop-Grumman experimental data but need further investigation. The predictions for
this unique 2 cell arrangement show the largest phase-dependent effects at low frequency.

A 2-D duct propagation test has been defined for the evaluation of the MELO extended
reaction prediction capability. This were performed by Northrop Grumman in October
1998. A group of liners were designed to cover the full spectrum of extended reaction wall
resistances. The test data in general agreed well with the predicted spectral variations for
the different samples except that the attenuation magnitudes were about half the predicted
values. The reason for this discrepency is not known.
Prelininary predictions for fan ducts do not show large attenuation gains from extended
reaction designs over locally reacting liners, with the only benefits being observed for wise
mesh face sheets and connecting cavity walls.

Studies should continue to investigate the variations in impedance with mode order
(effectively phase angle) to identify if potential exists to optimize for more than a single
mode, as currently occurs with present locally reacting liners. This will be more easily
qualified for an inlet, as opposed to a fan duct, where higher order modes are less well
attenuated.
 Drivers Fig. 1 Schematic of double impedance
tube.

Impedance tubes

40 Rayls 40 Rayls

Cell 1 Cell 2

/\

I
Common Wall:-
i) Hard Wall
ii) 40 Rayls
iii) 80 Rayls
Fig.2 Double Impedance Tube Resistance, 40 Ray1 Face Sheet and Wall, Cell 1

+Phase = 0 deg.
+20deg

+6Odeg.

04 I
0 500 1000 1500 2000 2500 3000 3500 4000
Frequency (HZ)

Fig. 3 Double Impedance Tube Resistance, 40 Ray1 Face Sheet and Wall, Ce11.2

+Phase = 0 deg.
+ZOdeg

+-6Odeg.

I
0 500 1000 1500 2000 2500 3000 3500 4000
Frequency (HZ)
Fig. 4
Double Impedance Tube Reactance, 40 Rayl Face Sheet and Wall, Cell 1

-
2 Odeg

500 1000

Frequency (Hz)

Fig.5 Double Impedance Tube Reactance, 4 0 Rayl Face Sheet and Wall, Cell 2

Frequency (Hz)
 Fig. 6
Double Impedance Tube Resistance, 40 Rayl Face Sheet, 80 Rayl Wall, Cell 1

14-80 deg

04 I
0 500 loo0 1500 2000 2500 3000 3500 4000
Frequency(Hz)

Fig. 7
Double lmpedance Tube Resistance, 40 Rayl Face Sheet, 80 Rayl Wall, Cell 2

+Phase= 0 deg
-%- 20 deg
+40 deg
-60 deg

Frequency (Hz)
Fig. 8
Double lmpedance Tube Reactance, 40 Rayl Face Sheet, 80 Rayl Wall, Cell 1

+Phase = 0 deg

-*- 40 deg
+-60 deg

Frequency (Hz)

Fig. 9
Double lmpedance Tube Reactance, 40 Rayl Face Sheet, 80 Rayl Wall, Cell 2

+Phase = 0 deg
- 20deg

Frequency (Hz)
Fig. 10
Predictionsfor Double ImpedanceTube, 40 Rayl Face Sheet and Wall, 800Hz

7
-- -$
7
Grumman, cell 1
e z m a t r i x , cell 1

4matrix, cell 2
- Ymod, cell 1
- Ymod, cell 2

Phase Angle (deg)

Fig. 11 Predictionsfor Double ImpedanceTube, 40 Rayl Face Sheet and Wall, 800Hz

m &ZmaUx, cell 1

Phase Angle (deg)

 Fig. 12
Predictionsfor Double Impedance Tube, 40 Rayl Face Sheet and Wall, 1000Hz

Grumman, cell 2
-7mabix. cell 1
--e-- zmabix. cell 2
- -
e Ymod, cell 1
- -7- Ymod. cell 2

I
0O
. 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Phase Angle (deg)

Fig. 13

Predictions for Double lmpedance Tube, 40 Rayl Face Sheet and Wall, IOOOHz

-
A

05

.f 0

$ -05 - 4- yrnod, cell 1

-1

-1.5

-2

-2 5

Phase Angle (deg)

 Fig. 14
Predictions for Double Impedance Tube, 40 Rayl Face Sheet and Wall, 2000Hz

"0° T

Grumman. cell 2

e z m a t r i x , cell 2
-.o -Ymod, cell 1
- - mod, cell 2

Phase Angle (deg)

Fig. 15

Predictions for Double Impedance Tdbe, 40 Rayl Face Sheet and Wall, 2000Hz

--:: Grumman, cell 2

b
z 0.00
70.0 80.0 so 0
i4
a CT----a -0- Ymod,celll
I --w----u----- - a - Ymod, cell 2

-0.50

Phase Angle (deg)

 Fig. 16
Predictions for Double Impedance Tube, 40 Rayl Face Sheet and Wall, 3000Hz

Grumman, cell 2
-2mabix. cell 1
-- Zmahx, cell 2
- *- Ymod, call 1
- - - Ymod, cell 2

0.004 I
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Phase Angle (deg)

Fig. 17 Predictionsfor Double ImpedanceTube, 40 Rayl Face Sheet and Wall, 3000Hz

Grumman, cell 2
-Zmatrix, cell 1
-ZmaW!x, cell 2
- 9-Ymcd, cell 1

PhaseAngle (deg)
 Fig. 18
Predictionsfor Double lmpedance Tube, 40 Rayl Face Sheet and Wall, 4000Hz

GNmman, cell 2

3--Zmatrix,cell 2
-4- Ymod,eelll
- "-

0.00 4 I
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 60.0 90.0 100.0
Phase Angle (deg)

Fig. 19 Predictionsfor Double lmpedance Tube, 40 Rayl Face Sheet and Wall, 4000Hz

- Gwmman. cell 2
e 0.00 -- -+-Zmatrix. cell 1
10.0 20.0 30.0 40.0 50.0 80.0 70.0 80.0
-0.50 -- -+- Yrnod,celll

-1.00 --

-1.50 --

-2.00 --

-2.50 --

-3.00 --

-3.50 --

-4.00 -
PhaseAngle (deg)
 Fig' 20 Mode 1 Attenuation for 5cm x 20cm Duct

20
m456 Perf Core
E .Bob Perf Core
3 lcm FluteOdeg (parallel to duct axis)
c
%a 15 H lcm Flute9Odeg
l c m Flute 45deg
i2
4
m40Rayllcm Bulk
~BORayllcmBulk
10

0
0 0 0 0 0 0 0 0 0 0 0 0
i 2 z % g In " g 0
i z $ g g
Frequency Band (Hz)

Fig. 21 Mode 1 Attenuation for 50cm x 200cm Duct

.4% Perf Core

H8% Perf Core
lcm FluteOdeg(parallel to duct axis]
nlcm Flute9Odeg
Q lcm Flute 45deg
m40RayVcrn Bulk
HBORaylIcm Bulk

500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300
Frequency Band (Hz)
 Fig. 22
Mode 2 Attenuation for 50cm x 200cm Duct

.
.4% Perf Core
8% Perf Core
13lcm Flute Odeg (parallel to duct axis)
1 lcm Flute 90deg
El lcm Flute 45deg
140RaylIcm Bulk
.80RayVcm Bulk

500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300
Frequency Band (Hz)

Fig. 23 Attenuation for Rectangular 50cm x200cm Duct

1 4 % Perf Core
1 8 % Perf Core
lcm Flute Odeg (parallel to duct axis)
lcm Flute 9odeg
El lcm Flute 45deg
W40Rayllcm Bulk

500 630 800 1000 1250 1600 2000 2500 3150 4000 5000 6300
Frequency Band (Hz)
 Insertion Loss Measurement ' - A
Microphones A

Mode Measurement 1
Microphones ,

.-I
Fig. 24 . (both
Test Set-Ug

Samples

I
.
+Sample 1, 5ORayl f s , 20 Rayl Wall,
6.19dB
+sample 2, 90 Rayl f s , l o c a l l y
reacting, 5.68dB
+Sample 3, 50 Rayl f s , 40 Rayl wall,

l+~ample 4, so Rayl f s , no c e l l s , 2.3546

+sample 5, 50 Rayl f s , l o c a l l y
reacting, 4.95dB
Fig. 26
PNLT Optimized Liners for M0.51 Mid T w i n Wide Chord Fan Duct

+Double Layer Mesh (5.81dB)

+Double Layer Perforate (5.62dB)

+Single

+Single
Layer Mesh (4.28dB)

Layer Perforate (4.04dB)

I
+Single Layer Mesh Extended
Reaction (4.50dB)
 - -
APPENDIX 5 Tests and Analysis Hybrid ActiveIPassive Lining Concept
Hersh Acoustical Engineering, Inc.

Introduction

Under Boeing Purchase Order #JR5935 (Task 4 under NASA Contract NASI-97040) ,
Hersh Acoustical Engineering, Inc. (HAE) has undertaken a four-part initial investigation
of simplifications that might be applied to active control of rotorlstator tones in turbofan
engine inlets. The underlying concept upon which this investigation was based assumed
that only a subset of the total number of cut-on modes would have a significant impact
on sideline tone nose. If this subset could be controlled with a less extensive array of
active control actuators than required by the full sound field, reduced active control
system complexity and spacelweight would be achieved.
The four major parts of the project were as follows:
1. Design and constuct a static test facility for generating and measuring multiple
spinning order and multiple radial order propagation modes in a circular duct.
2. Establish microphone arrays and signal processing systems for simultaneous
measurement of in-duct and far-field radiated sound.
3. Design and fabricate a linear microphone array to be mounted alongside the duct
inlet and used as an error signal sensor for far-field radiation in specific directional
sectors.
4. Install an activelpassive hybrid liner in the test duct and incorporate it with the linear
microphone array to assess feasibility of simplified ANC to suppress multi-radial
order tones.

Test Facility Design and Construction:

48-Inch Test Duct:

The test duct, modeled after the ANCF inlet at NASA Glenn Research Center, has a
48-inch inside diameter and a total length of 96 inches, divided into two 48-inch sections.
Each duct section is constructed of M-inch thickness high density resin-impregnated
cardboard cylinder, intended for use as concrete forms. Mating flanges were added to
allow ease of fitting up and reinforcement of the circular cross section. To minimize
shape distortion due to gravity and to allow free-field measurements, the duct is oriented
vertically, with the inlet opening though the lab roof as shown in Figure 1.
Although it is a no-flow facility, an inlet bell-mouth was incorporated to provided a
realistic inlet radiation termination. To maintain similarity with the ANCF inlet, the mold
used to make the ANCF bell-mouth was shipped to HAE and used as a form for an
epoxy-covered foam unit of the same shape as the fiberglass unit at ANCF. The shape
of the bell-mouth is shown in Figure 2.
Mode Synthesizer:

The useful operating range of the actuators employed in the hybrid liner is 400-1,250 Hz.
In order to simplify existing active control systems, it was decided that a sound field of
four radial orders should be addressed with an active liner element consisting of two
rows of controlled sources. This combination of frequency range and mode requirement
dictated that circumferential modes m = Ior m = 2 should be used. Characteristics of
these modes (plus the first cut-off mode) are shown in Table 1. Here %, are the
hardwall duct eigenvalues, F, is the modal cutoff frequency and @ is the effective
propagation angle relative to the duct axis at a preselected frequency of 1228 Hz
(corresponding to 2300 CRPM at ANCF).

-
Table 1 Mode Characteristics Based on 1228 Hz Nominal Test Frequency

The variation in propagation angle for mode m = 2 was selected for initial investigation,
but capabilities for synthesis of both modes were incorporated into the system design.
In order to generate four radial order modes, a disk was designed containing five circular
arrays of high quality 3-inch electrodynamic transducers (loudspeakers). The highest
circumferential order mode that will propagate in the 48-inch duct at 1228 Hz is m = IfIl1.
Aliasing considerations therefore dictate that for excitation of m = 2, a minimum of 14
actuators must be provided around each circular array. In practice, 16 actuators were
provided in the three outer circles of the source array, as shown in Figure 3. However,
for the two inner circles, the number was reduced to 8 actuators per circle, resulting in
aliased radiation into mode m = -6, for which two radials are cut on.
The radial sound pressure distributions of target mode (2,O) and potential aliased mode
(6,l) are presented in Figure 4. Near the center, the mode strength of (6,l) is weak
compared to that of (2,0), although for the second loudspeaker circle (rlR = .3958) the
difference is only about 1.7 dB. Thus, a mode combination that relies heavily on this row
would be expected to have significant contamination by mode m = -6 and should be
flagged for avoidance.
Signal distribution to the five loudspeaker circles is accomplished with a computer-
controlled analog network as shown in Figure 5. A master oscillator signal is split into
quadrature components by a 90" offset network. Analog four-quadrant multipliers
(Analog Devices AD532) are used to establish the real and imaginary components of the
array drive signals under DC control from a digital to analog converter. The drive signals
for the individual arrays are again split into quadrature components and then further split
into 0°, 90°, 180" and 270" busses by means of inverting amplifiers.
Each loudspeaker is driven by an individual 15-watt amplifier with a summing junction.
Input summing resistors of conductance proportional to Icos(m0)l and Isin(m0)l (where 0
is the angular position of the speaker and m is the circumferential mode order) are
connected to the appropriate signal buss to achieve the eimespinning excitation for the
array. An example for achieving a 315" loudspeaker output signal is shown in Figure 6.

In-Duct Mode Measurements:

A microphone rake was designed and fabricated for measuring the radial, azimuthal and
axial distribution of sound pressure in the test duct. Ten miniature condenser
microphones are located on a radial arm at equally spaced radii 3 to 23 inches. This
arm pivots about the duct center and can be positioned axially via a track system.
Signals from the microphones are digitized and cross-correlated with quadrature
reference signals to obtain absolute amplitude and phase as shown in the block diagram
of Figure 7. A single measurement consists of recording radial pressures at 24 angles
on 15" increments. Circumferential modes at each radius are computed by a spatial
Fourier transform. Radial decomposition of each circumferential mode is computed as a
least squares fit to the basis functions

where h,are the hardwall mode eigenvalues.

In principle, the ten microphones would allow resolution of up to ten radial orders.
However, only six radials (one more than the maximum cut on for any circumferential
order) are included in the least-squares computation.

Far Field Meas

As shown on Figure 1, the far field measurement is accomplished on the lab roof with a
single microphone moved along an arc at a distance 6 radii (12 it) from a reference point
near the inlet bell-mouth to duct transition at the center of the duct. The measurements
are taken using a Rion NA-29 precision sound level meter with M-inch condenser
microphone. The signal output of the NA-29 is digitized and cross-correlated with
quadrature reference signals to obtain amplitude and phase as for the in-duct
microphones. Each measurement is an average of three to five individual readings to
allow for minor variations due to atmospheric disturbances. Measurements are taken on
an arc from 0° to 105O in 5" increments.

Mode Synthesis System Calibration:

For reference, the four radial modes at m = 2 targeted for radiation by the mode
synthesis system are shown in Figure 8. The mode synthesis system was calibrated for
generation of specific radial order modes by the following procedure:
1. The rotating microphone rake was positioned at an axial reference 84 inches from
the plane of the synthesizer source.
2. Each of the five loudspeaker rings on the source was activated in sequence with
control amplitude (5.0 + 0.01) and the amplitude and phase of the radiated modes
was measured with the rake.
3. A system of 5 equations in N (c5) unknowns was solved to provide a minimum norm
ring excitation vector for each of the Ncut-on radial modes.
4. The modal drive vectors were stored in the synthesizer control program.

Observe in Figure 8 that the ring at r/R = .39 is located close to a nodal circle for mode
(2,3)and the ring at r/R = .75 is close to a nodal circle for mode (2,l). Poor modal
coupling is expected for these combinations.
Measured modal amplitudes (complex) are shown in the following tables. A summary,
showing SPL only, is presented in Figure 9.

Table 2 - Radial Mode Amplitudes for m = 2 1228 Hz, Ring at r/R = ,9167
Radial Order Real Imaginary
0 0.46314689 2.1 170874
1 -1.8544016 0.46812057
2 -1.2366257 1.7544158
3 0.63846207 -3.0299787
4 -0.069316579 0.036991828
5 -0.075379865 -0.002928982

-
Table 3 Radial Mode Amplitudes for m = 2, 1228 Hz, Ring at r/R = ,7500
Radial Order 1 Real lmaainarv

Table 4 - Radial Mode Amplitudes for m = 2,1228 Hz, Ring at r/R = 5833
Radial Order Real Imaginary
0 0.38172216 0.96099738
1 2.0703876 -0.66168278
2 0.6721825 -1.50431 12
3 -0.61601518 -4.1768641
4 -0.032007551 0.03986563
5 -0.071255363 -0.044649256
 Table 5 - Radial Mode Amplitudes for m = 2, 1228 Hz, Ring at r/R = .3958
Radial Order Real Imaginary
0 0.1748744 0.2071 8574
1 1.0486539 -0.50273155
2 -0.68697006 1.47451 5
3 -0.007250865 -0.34829738
4 -7.77628E-05 -0.008053407
5 -0.000728595 0.004134448

-
Table 6 Radial Mode Amplitudes for m = 2,1228 Hz, Ring at r/R = .2708
Radial Order Real Imaginary
0 0.1 4893129 0.25747607
1 0.79314236 0.35062669
2 -1.8952537 0.62643183
3 -2.999751 4 2.9749977
4 0.14678428 0.02820488
5 -0.055210458 0.10412994

In Figure 9 are seen the modal amplitudes (in dB re .00002 Pa) produced by driving
individual source rings. One may observe the "dropouts" of (2,l) for r/R = .75 and (2,3)
for r/R = -39 as predicted in Figure 8. Also observe the general weakening of (2,O) as
r/R decreases and the generally (except at r/R = .39) high value of (2,3), owing to its low
cutoff ratio. Cutoff modes (2,4) and (2,5) are typically suppressed 30 dB. The residual
is due to minor alignment fluctuations in the rake axis.
The result of the modal excitation vector computation is presented in Table 7. Linear
combinations of these individual modal drives may be applied to provide any
combination of mode amplitudes and phases desired at any point in the duct.

-
Table 7 Speaker Ring Drive Vectors for 1 Pa Mode Amplitude at O0 Phase, Axially
84 inches from Source Plane

Figure 10 shows a general indication of the modal isolation achieved by the above
procedure. These are radial mode separations from a single rake angle (i.e., no m-order
separation) and therefore include some effects of modal spillover. It is still clear that the
modal isolation procedure is effective and will provide a means for evaluating liner and
control function using mode mixture as a test parameter. Figure 11 shows the full mode
measurement for excitation of Mode (2,O). Note that on a sound power basis, the target
mode is isolated by nearly 30 dB from the other m = 2 radials. Spurious m-orders are
suppressed approximately 20 dB and are partly the result of rake-axis fluctuation.
Modal Propagation Investigation:

Setting the mode mixture at arbitrary axial position requires verification of modal
propagation characteristics within the duct. To accomplish this, the outer source ring
(r/R = .92) was driven and the full mode measurement procedure was undertaken at
several axial positions of the rake. The amplitude and phase of each radial order mode
was plotted (Figure 12 and Figure 13) and a regression analysis was used to determine
the wave number for each mode. In Figure 12, the mode amplitudes are seen to be
reasonably independent of axial position and the two cut-off modes are suppressed 25
dB. Figure 13 shows that the phase vs axial distance is virtually a perfect straight line
with all collapsing to nearly 0° at the origin (note that the phases of the odd n modes
were offset 180" to allow easier comparison). Wave-numbers computed from the
measured phase rates and predicted from the duct geometry are shown in Table 8.
Agreement is within 1% except for mode (2,3), which is barely cut-on and therefore most
sensitive to errors in temperature measurement or small non-zero duct boundary
admittance.
These results verify both that propagation in the test duct exhibits circular duct spinning
mode characteristics and that the bell-mouth termination as installed at ANCF and on
the test ducts essentially eliminates reflections.

-
Table 8 Measured and Predicted Axial Wavenumbers at 1228 Hz

Radiation Directivity:

Since the program objective is to develop an error sensing array that allows far-field
directivity control, a prediction and measurement was made to determine far-field
radiation properties of the test facility. In addition, as an aid in the design process, the
sound field was computed in the proposed error array location to determine the
relationships between the quasi-near-field and far-field.
The predictions were computed by W. Eversman. Complex sound pressure amplitudes
were tabulated for radii 4 to 6 ft in 3-inch increments and then at radius 12 ft.to represent
the far field. Approximately 800 angular positions were computed for each radius, from
0° (duct inlet axis) to 120'. A sampling of the prediction results (r=4, 6 and 12 ft) is
presented in Figure 14
From Figure 14 it is clear that the directivity pattern evolves from the near to far field, so
that error measurements taken within the 4 to 6 ft radius zone will require processing to
allow use for controlling far-field radiation patterns.
As an initial check on the predictions and the roof-top measurement system, the mode
synthesizer was set to generate the modes (2,O) - (2,3) in succession and
measurements of SPL at 12 ft radius were taken on 5" increments from 15" to 105".
(Angles 0° - 10" were omitted due to difficulties with the positioning system.)
 The predicted and measured SPL were plotted as shown in Figure 15. Although there
are some deviations, the general trends agree very well. Some of the variations in detail
are attributed to the imperfect radial mode isolation and contamination by spurious m-
orders, which are suppressed by only 20 dB.

Boom Microphone Design:

A linear microphone array was designed using an airfoil shaped tube as a mount and six
miniature condenser microphones spaced over a 6-foot span. Various spacing patterns
are being investigated. Preliminary measurements were taken with uniform spacing, but
a better mode separation capability appears to result from a logarithmic spacing as
shown in Table 9 and Figure 16.

Table 9 - Example of Microhone Weightings For Radial Mode Separation from

Linear Boom Array
Mic Elev\Mode (2,o) PSI) (22) (2,3)
2 1.OOOO -0.6790+0.5544i 0.2637+0.54421 -1.2664-0.20431
2.64 -0.0694-0.41 09i 0.3322+0.2686i 0.3227-0.2754i -0.3709+0.9409i
3.48 -0.0940-0.9019i 0.4332+0.5527i 0.2252+0.1089i 0.7404+0.46271
4.59 0.3950-0.1 980i -0.3506+0.35721 -0.3397+0.19261 -0.8708+0.2756i
6.06 1. I 929-1.I 9331 -0.2064+1.0080i 1.0121+0.1048i -0.21 81+I .0388i
8 -1.3353+0.0405i 0.4996-1.3 168i -0.4780-0.01 64i 0.2316-0.3856i

Hybrid Liner Installation:

The hybrid liner used for the tests has been extensively described in (Parente eta0
"Hybrid Active/Passive Jet Engine Noise Suppression System," NASA CR-1999-208875,
February, 1999. It is a 24-inch long, 48-inch diameter duct section divided into two
segments. Segment 1 is two rows of 16 each active Helmholtz resonators, with 2-inch
diameter orifices spaced axially 5.5 inches and circumferentially 22.5'. Segment 2 is a
14-inch long linear passive liner with a wire mesh face sheet and a 2.2-inch deep
honeycomb core. The liner is installed in the test duct with the active segment facing the
source.
The original design of the hybrid liner was directed at controlling modes (4,O) and (4,1) at
1000 Hz. The "upstream" active segment redirects energy for more effective absorption
by the passive segment. For the two-mode environment, attenuation of both modes by
over 30 dB has been achieved.

Active Control System Implementation:

Active control for the study was implemented using a variation of the Adaptive
Quadrature algorithm. At the system excitation frequency, transfer functions from the
actuators in the active resonators to the error microphone array(s) are measured directly
(Figure 17) and then inverted to determine an excitation matrix to minimize the error
signal.
In order to simplify control processing, signals to the resonator actuators are pre-
processed to modal phase relationships (45" retard per actuator for 16 resonators at
mode m = 2). This allows the controller to "view" each resonator row as a single source
in the axisymmetric environment. If multiple circumferential order modes were to be
controlled, a separate distribution network would be applied for each mode and the
signals summed at the actuator amplifier.

Far Field Control Demonstration:

As an initial demonstration, a far field microphone was positioned at radius 12 feet, polar
angle 45" and azimuth approximately 270" relative to the duct inlet reference. The linear
"boom" microphone array was set up parallel to the duct axis 4 feet to the side at an
azimuth angle of 190" and with the uppermost microphone on the array 6 feet above the
duct highlight. The far field and boom microphones were placed at different azimuths to
minimize the probability of a false success indication from interference between spill-
over m-order modes.
The control system was set up using the upper three microphones in the linear array as
independent error signals. The mode m=2 drives to each of the two active resonator
arrays in the hybrid liner were used as control signals.
The mode synthesizer system was set to excite predominantly mode (2,2), which is well
cut on and has a strong radiation lobe in the direction 40 45". Measurements of the SPL
at 1228 Hz were made at the far field microphone and at each of the upper three boom
microphones for control on and control off conditions.
Preliminary indication was that with 2x2 control (upper two boom microphones only), the
control system reduced the SPL at the error sensors by over 30 dB and reduced the far-
field radiated tone SPL by 7-8 dB. Final tests, conducted with gusty winds affecting the
microphone positions and propagation, showed 18-22 dB tone reduction at the two
upper boom microphones and 2.0-3.3 dB tone reduction at the far field microphone.
For reference, the results of the final roof-top demonstration test are shown in Table 10

-
Table 10 Summary of Far-Field Attenuation Test Results
Condition Far Field SPL Boom Mic Boom Mic SPL
Control Off 70.5 B1 77.1
Control 3x2 68.5 B1 57.1
Control Off 70.1 B2 75.3
Control 3x2 68.0 B2 58.8
Control Off 70.2 B3 72.8
Control 3x2 67.5 B3 73.4
Control 2x2 67.2 B1 55.0

Summary and Conclusions:

The roof-top static 48-inch test duct facility was designed, constructed and calibrated for
radiation of modes (2,O) - (2,2) either in isolation or in combination.
Radiation characteristics for the individual modes were computed in near and far field.
A linear microphone array and signal processing system were designed to sample the
near sound field and provide control system error signals relating to far-field radiation.
Far field sideline radiation suppression was demonstrated using a Hybrid active/passive
liner employing two rows of active Helmholtz resonators to control a four-radial in-duct
sound field.

Concept Extension and Refinement:

Under NASA Langley NASl 99059, the mode generation and sound field sensing
concepts developed and studied for feasibility under this program are being extended to
include full use of the near-radiation field mode separation and to control multiple
circumferential modes and BPF harmonic tones simultaneously.
 C

Sound Absorbing Treatment

\ Mode Synthesis
Array

-
Figure 1 Schematic of Roof-Top Measurement System with Vertical Axis Duct

-
Figure 2 Section Through ANCF Bell-Mouth
 -
Figure 3 Mode Synthesizer Transducer Layout Showing Radius Ratios of
Mounting Circles

2 0.5
I
m
V)

2
a
'0
C
1
0
0
V)
0

-
.L
*a
0
-0.5

I
i
-1
0 0.2 0.4 0.6 0.8
Radius Ratio
- Mode (2,O)
- Mode (6,l)

-
Figure 4 Comparison of Modes (2,O) and (6,l) re Coupling Near Origin
 -
Figure 5 Modal Synthesis Signal Distribution Network

-
Figure 6 Summing Input Power Amplifier Example for 315O
 Reference System
Oscillator Excitation

1 Hilbert
Transform

Analog
to
Microphone Digital
Signal (Typ.) Converter

-
Figure 7 Block Diagram of Microphone Signal Processing Channel

g 0.5
2
m
2
a
'z
O
m
d)
>
-
*
.3
d
d)
-0.5

-1
0 0.2 0.4 0.6 0.8
Radius Ratio
- Mode (2,O)
- Mode (2,l)
- Mode (2,2)
Mode (2,3)

-
Figure 8 Mode Shapes for m = 2 Cut On Radials
 0.9167 0.7500 0.5833 0.3958
Ring Radius Ratio r/R

-
Figure 9 Relative SPL of Radial Modes for m = 2 at 1228 Hz as a Function of
Excitation Radius
 Driven Mode = (2, 0) Driven Mode = (2, 1)
120 120
I
115- - - - - I -
I
---
I
- ' - - - -1-
I I
---A - - - - - 1151 - - - -1-
I
-- - -1-
I
-- - .I-
I
-- - A- - -- -
I I

80
0 1 2 3 4 0 t 2 3 4
Measured Radial Order Measured Radial Order

Driven Mode = (2, 2) DnvenMode = (2,3)

120 120,
I I
I I I I I
115- - - - -1- --- -'- - - - -'- -- - 1 -- - -
I I

---- ---

I
I - - - - -

80
1 2 3 4
Measured Radial Order Measured Radial Order

-
Figure 10 Modes Measured at Single Rake Angle vs Modal Excitation Vector
 1-------

r------ --- I-------

-10 5 0 10
C~rcurnfnren(ill
M o b Order - rn

120
i I I II I
130 , l i i l I I I I
I I I II I I I I I I I 150- - - - I - --4--- C - - -1- ---
- I
110 - - - - -r -
I I II
--
T - -1-
I
- - m
420 -----
I I
T--r
I I
-l--T--
I I
I---
I l l 1

m I I I I I I I I I I I I
0 I I I I I I - I I I I I I
~ l m - - ~ _ _ ~ - - L - - l - - L - - l - - - %ilo--I--I--L-~--~--~---
_1 I I I I I I
I l l 1 1 I I I I I
I I I I I I O I I
-T --r-l-- --r-- -T--r -7--T--
1 1 1 1 1 I I I I I
! I 1 1 1 I I l l I
- I - _ L - J - - L - - C - -
1 I I I l
I I I I I I I I I I
I I I I I
-T--r-l--T--l---
I
I
I I
I
I
l
I
l
1 1 1
-150 - - - -1-
1
--A
1
- - - L - - -1-
1 1
-- -
I I I / I I I I
0 1 2 3 4 5 70 0 1 2 3 4 5 0 1 2 3 4 5
m = 2 RadlalMode Order- n m = 2 RadialMatie 0-1- n rn = 2 RadlalModeOrder -n

-
Figure 11 Isolation of Mode (2,O) and Measured m ~ r d e Spillover
r
 Mode m=2 1228 Hz Radial Order SPL vs Axial Distance from Source Plane

30 40 50 60 70 80 90
Distance from Source in Inches

Figure 12 - Amplitudes of m = 2 Radial Modes vs Distance from Source Plane

Mode m=2 1228 Hz Radial Order Phase vs Axial Distance from Source Plane
0

-500

-1 000
ln

&
',
0"
C

1
-1500

I
C
a
-2000
y = - 2 8 9 7 x + l 85

-2500

-3000
30 40 50 60 70 80 90
Distance from Source in Inches

Figure 13 - Measured Propagation Rates for m = 2 Radial Order Modes at 1228 Hz

-

(2.0). r/R = 2, Max SPL = 90.8 dB (2.1), dR=2,MaxSPL= 90.1 dB (2.2), dR = 2, Max SPL = 84.7 dB (2.3). r/R = 2.Max SPL = 80.6 dB

21

270 270 270 27 0

(2.0). r/R = 3 , Max SPL = 89.9 dB (2.1). r/R = 3. Max SPL= 84.8 dB (2.2), dR = 3 . Max SPL = 79.8 dB (2.3). r/R = 3. Max SPL = 75.3 dB
90 50

270 270 270 270

(2.0). dR = 6. MaxSPL = 85.3 dB (2.1). rlR = 6, Max SPL= 79.6 dB (2.2). rlR = 6. Max SPL= 74.6 dB (2.3). r/R=6.MaxSPL=67.1 dB
90 50

,
21 21
'\
1
24 00
270 270 270 270

-
Figure 14 W. Eversman-Computed Directivity of Modes vs Distance at 1228 Hz
 - -
Figure 15 Measured and Predicted Directivity at r/R = 6 for Modes (2,O) (2,3)
 Real Part
1 1 1 I I I I I 1

i
1
I I I I I I I I I I
025 L - - - - 1- - - - -1 - - - - - I -l
- - - - - L - - - - iiiiii -- -- - - - - - 1- - - - 1- - - - - I ----
I I I I

1 I I

: I

;
, ---------- ----- / I (

I I I 1
I I
-1 0 1 2 3 4 5 6 7 8 9 10
Boom Elevation above Duct Reference - ft

Imaginary Part
I I I I [ I I [ I
I I I I I I I I I I

0.25 - - - - - i - - _ - - I ----- ----- i - - - - l------ - - - - I - - _ - -L - - - - J -----

I I I I I I - 1 I I

I I I

I I I I I I i

-0.2 - - - - - ---- ,- - _ - -' - - r - - - -

- -
I I
-----
I - 8

I I I I I

-1 0 1 2 3 4 5 6 7 8 9 10
Boom Elevation above Duct Reference - ft

-
Figure 17 Predicted Transfer Functions from Radial Modes to Boom Mic at 4 ft to
Side of Duct Axis
 REPORT DOCUMENTATION PAGE I
I
Form Approved
OM5 NO.0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing inst~ctions,searching existing data sources,
- -
aatherina and maintainina the data needed. and com~letinaand reviewino the collection of information. Send comments reaardino this burden estimate or anv other aspect of this
collection of information, yncluding suggestions for reducingthis burden, to Washington Headquarters Services, Directora ,;or lnf&mation Operations and ~eports,1215
Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reductio Project (0704-0188), Washington. DC 20503.
1. AGENCY USE ONLY (Leave blank) 2. REPORT DATE 3. REPORTTYPE AN1 DATES COVERED
August 2002 Contractor Repc
4. TITLE AND SUBTITLE 5. FUNDING NUMBERS
Advanced Nacelle Acoustic Lining Concepts Development
NAS 1-97040
Task 4
6. AUTHOR(S)
G. Bielak, J. Gallman, R. Kunze, P. Murray, J. Premo, M. Kosanchick,
A. Hersh, J. Celano, B. Walker, J. Yu, H. W. Kwan, S. Chiou, J. Kelly,
J. Betts, J. Follet, R. Thomas
7. PERFORMING ORGANIZATION NAME@) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
Boeing Commercial Airplane Company REPORT NUMBER
Seattle, Washington 98124-2207

I
I

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10.SPONSORlNGlMONlTORlNG

AGENCY REPORT NUMBER
National Aeronautics and Space Administration
Langley Research Center
Hampton, VA 2368 1-2 199
I
11. SUPPLEMENTARY NOTES
NASA Langley technical monitor: Tony L. Parrott

12a. DlSTRlBUTlONlAVAlLABlLlTYSTATEMENT

Unclassified-Unlimited
12b. DISTRIBUTION CODE
I
Subject Category 71 Distribution: Nonstandard
Availability: NASA CASI (301) 621-0390
I
13. ABSTRACT (Maximum 200 words)
The work reported in this document consisted of six distinct liner technology development subtasks: 1) Analysis of
Model S C ~ ~ ~ A P Lining Data (Boeing): An evaluation of an AS;^‘ ~ i l e s t o n eexperiment to demonstrate
FanDDuct
1995 liner technology superiority relative to that of 1992 was performed on 1:5.9 scale model fan rig (Advanced
Ducted Propeller) test data acquired in the NASA Glenn 9 x 15 foot wind tunnel. The goal of 50% improvement
was deemed satisfied. 2) Bias Flow Liner Investigation (Boeing, VCES): The ability to control liner impedance by
low velocity bias flow through liner was demonstrated. An impedance prediction model to include bias flow was
developed. 3) Grazing Flow Impedance Testing (Boeing): Grazing flow impedance tests were condttcted for com-
parison with results achieved at four different laboratories. 4) Micro-Perforate Acoustic Liner Technology (BFG,
HAE, NG): Proof of concept testing of a "linear liner." 5) Extended Reaction Liners (Boeing, NG): Bandwidth
improvements for non-locally reacting liner were investigated with porous honeycomb core test liners. 6) Develop-
ment of a Hybrid ActiveIPassive Lining Concept (HAE): Synergism between active and passive attenuation of
noise radiated by a model inlet was demonstrated.

14. SUBJECTTERMS 15. NUMBER OF PAGES

Acoustic liner concepts, Aircraft Nacelles 205
16. PRICE CODE

OFTHIS PAGE OF ABSTRACT OF ABSTRACT

NSN 7540-01-280-5500 Standard Form 298 (Rev. 2-89)

Prescribed by ANSI Std. 239-18
298-102