DAFNOR Distributed Active Foils for NOise Reduction - aproject overview

K. Henrioulle, P. SasDepartment of Mechanical Engineering, division PMA, K.U.Leuven, Belgiume-mail : kris.henrioulle@mech.kuleuven.ac.be

AbstractThis paper investigates the feasibility of using piezoelectric polymer material (PVDF) as sensor and actuatormaterial in active noise control applications. Three different active panels based on PVDF material are designedand tested. The first panel is a loudspeaker panel that produces a relatively high sound power output at low fre-quencies. Secondly a smart foam was built with the PVDF material integrated in the passive polyurethanefoam. The third panel is a volume velocity sensor/actuator panel that controls the volume velocity through apanel. Finally some tests were performed with the EMFi flat acoustic actuator.Numerical and analytical models are developed to optimise the performance of the active panels in variousconfigurations.Experiments in a car with the active panels show a substantial reduction of the sound pressure level. A secondtest is the active control of sound transmission through a single panel partition according to the ISO 140 stan-dard, using the active panels. The tests show an increase in the sound transmission loss of 3dB to 6dB below300Hz, which is the frequency region where the passive sound transmission loss of a single panel partition isinsufficient.

1 Introduction

Researchers have spent much attention on the use ofpiezoelectric elements as sensors and actuators in sys-tems for active control of sound and vibration. Morespecifically on the so called d31 type of elements, thatare thin films of piezoelectric material that exhibit amain deformation in the plane of the film. This be-cause of the interesting properties of these elements.Piezoelectric elements are light weight, very thin andcan easily be integrated in existing structures whereclassical point actuators like shakers or loudspeakerscan not be used. As disadvantages of piezoelectricfilm actuators their high driving voltage and limitedtemperature range have to be mentioned. However itcan be expected that these shortcomings can be over-come in the future.

In the past piezoceramic PZT material has beenused mainly as actuator. Because of its high Young’smodulus, it can exert a relatively large force on astructure. Because it is a piezoceramic material, itis very brittle. PVDF (Polyvinylidene) on the otherhand is a piezoelectric polymer material, and has theadvantages of a polymer material: very flexible andreadily shaped in complex shapes to cover irregularstructures. It is produced in films of few micrometers

to millimeters thick. The main disadvantage is thatits piezoelectric constants are two orders of magni-tude lower than those of PZT. Because a thin film ofPVDF merely influences the behaviour of a structure,it is mostly used as sensor.

This paper will give an overview of the results ob-tained within the DAFNOR (Distributed Active Foilsfor NOise Reduction) project. The aim of the projectwas to explore the advantageous properties of PVDFas piezoelectric material in active noise and vibra-tion control applications, and to investigate the use ofPVDF as actuator. The objective was to develop lightweight actuators and sensors for active noise and vi-bration control. The project brought together three re-search institutes KUL (B), ISVR (UK) and FFA (SE).The material suppliers were Thomson Marconi Sonar(F) and VTT (FI). End users of the technology wereCRF (I) and G+H Montage (G). Three essentiallydifferent prototypes were built and tested based onPVDF piezoelectric material, and specific simulationtools were developed to optimise the design of theseprototypes. In addition an acoustic actuator based onEMFi material was included in the tests.

Section 2 to 5 discuss the design and results ofthree types of actuators based on PVDF material: anacoustic actuator, a smart foam actuator and a volume

velocity sensor/actuator panel, and the EMfi panel.Section 6 compares the performance of the four ac-tuator panels. Finally two practical case studies arepresented: the active control of noise in a car interiorin section 7 and the active control of sound transmis-sion through a panel partition in section 8 using thenewly developed active panels.

2 Acoustic actuator

2.1 Description

All prototypes have characteristic dimensions of300mm � 400mm, which is small compared to thewavelength of the sound that has to be controlled.

The first actuator is an acoustic actuator, designedto produce large sound pressure levels in the fre-quency region below 500Hz [1, 2]. A schematic viewof the actuator is given in figure 1.

PVDF layer

PVDF layer Carrier structure

V~

V~

400mm

300mm

Figure 1: Schematic view of acoustic actuator

The actuator consists of a carrier structure of lowbending stiffness, i.e. honeycomb material of 6.4mmthickness. It is covered on its entire upper and lowersurface with a PVDF foil of 0.5mm thickness. ThePVDF foils are driven in opposite phase, thus excit-ing a bending motion in the actuator assembly. Thisinduces the out-of-plane displacement that causes thesound radiation.

2.2 Analytical model and simulation re-sults

The displacement of the actuator can be describedwith an analytical model based on the classical plateequation in which the piezoelectric constitutive equa-tions are inserted [1]. The design was optimisedto yield a maximum sound power output in the fre-quency region from 30 to 500Hz. Figure 2 shows thecalculated and measured sound pressure level at 1 me-ter distance from the actuator for an excitation volt-age of 1000V. A first resonance frequency lies around180Hz, it is the (1,1) bending mode of the actuator,and the most efficient radiating mode. The secondresonance is the (1,3) bending mode. The effect of

simulation

measurement

50 100 150 200 250 300 350 400 450 50050

55

60

65

70

75

80

85

90

95

100

frequency Hz

soun

d pr

essu

re (

dB r

ef. 2

e−5

Pa)

sound pressure at 1 m from the acoustic actuator

Figure 2: Simulated and measured sound pressurelevel from the acoustic actuator

errormicrophones

incidentpressure waves

~V

passiveplate

acousticactuator

cavity

Figure 3: Set-up used for the simulations

the PVDF foils is to exert a line moment at the edgesof the plate, and only the odd-odd – efficiently – radi-ating modes are exited.

This analytical model was extended with a modelof a double panel partition of which one of the platesis the acoustic actuator. A linear quadratic controllerwas added to the model.

A series of simulations were performed to inves-tigate how this actuator can increase the sound trans-mission loss through a single or double panel parti-tion [2]. The set-up in which the actuator is placed atthe side where the disturbing sound field is incidentis depicted on figure 3. Besides this configuration,the acoustic actuator was tested as single transmissionwall, in a configuration where it is placed at the sidethat radiates the sound in the far field and in a doublepanel partition placed in the cavity between two steelplates.

Different configurations for the error sensors weresimulated. The error sensors were microphonesplaced in the far field, microphones in the cavity be-tween the two plates of the transmission wall or ac-

without control

with control

0 50 100 150 200 250 300 350 400 450 5000

10

20

30

40

50

60

70

80

frequency (Hz)

soun

d po

wer

(dB

ref

. 1e−

12 W

att)

Radiated sound power

Figure 4: Sound power radiated by the double panelpartition

celerometers on the plate that radiates the sound inthe far field.

A typical result is shown in figure 4. The config-uration corresponds to the one depicted in figure 3.The acoustic actuator is placed at a distance of 70mmfrom a steel plate with the same dimensions as theactuator, and of 1mm thickness. The control systemminimises the response in a large number of error mi-crophones placed in the far field. In the low frequencyregion, that is below 250Hz, the vibration pattern ofthe acoustic actuator is dominated by the (1,1) bend-ing mode. This mode can be controlled effectively bythe PVDF layers. Likewise the first structural modeof the actuator couples well to the first acoustic modein the cavity between the steel plate and the actua-tor. Therefore this acoustic mode can be controlledby the actuator which explains the nice control per-formance below 250Hz. Above this frequency moremodes contribute to the sound power radiated, amongwhich even modes that cannot be controlled by thisactuator. However the control was aimed at a fre-quency region below 300Hz.

It was found that the most practical and effec-tive configuration was the one where one error mi-crophone is placed in the centre of the cavity. Thismicrophone will measure only those acoustic modesthat can be controlled by the acoustic actuator. Thisconfiguration was selected to be used in later experi-ments, as described in section 8.

It was concluded that the sound transmission lossthrough a single and double panel partition can be in-creased considerably in the low frequency range withthis type of actuator. For a single panel configuration,

the actuator can be placed at the side where the dis-turbing sound is incident, or at the side that radiatesthe sound in the far field. For both configurations, theactive performance is comparable, but when the ac-tuator is placed at the radiating side, the control volt-age required is lower. However it is more practical toshield the actuator from the environment by placingit at the incident side. At low frequencies, the vibra-tion pattern of the radiating plate is dominated by thefirst bending mode of the radiating plate. Thereforea good control performance is observed when using asingle error accelerometer placed in the centre of theradiating plate, or with one error microphone placedin the centre of the cavity.

Although the increase of the sound transmissionloss in the low frequency range is considerable, it isrestricted by the presence of some even modes thatcannot be controlled by the actuator. With a minorchange in the design, that is by dividing the elec-trode that covers the actuator in four independent sec-tions, the actuator can control some of the even modesas well. Consequently, this configuration shows thebest performance in increasing the sound transmis-sion through a panel partition.

2.3 Experimental results

The acoustic actuator was built and used in a numberof tests. Figure 5 shows the actuator placed in a ductto test the active control of sound transmission lossthrough the actuator. The actuator is placed in front ofa passive plate at the side where the disturbing soundfield is incident. The passive plate has been removedon the picture to show the 9 accelerometers placedon the actuator surface, and the 3 microphones placedin the cavity. The microphone in the centre of thecavity is used as error sensor for the filtered-X LMSfeedforward control algorithm.

Figure 6 shows the sound pressure level measuredin the receiving side of the duct for different controlstrategies. For a single error microphone placed atthe receiving side, a reduction of about 15dB is ob-served at low frequencies. A single microphone inthe centre of the cavity yields comparable results. Thefirst transverse acoustical mode of the duct is excitedabove 340Hz, which excites an even (uncontrollable)mode of the actuator. This explains the poor reductionabove this frequency. The acoustic actuator with fourindependent elements can control the sound transmis-sion in a broader frequency range.

Figure 5: Picture of the acoustic actuator

Figure 6: Measured sound pressure level at the receiv-ing side of the duct

3 Smart foam actuator

The second type of actuator is called the smart foamactuator. It is inspired by the work of Fuller [3] (andlater work) who tried to enhance the low frequencysound absorption of a foam material by inserting ac-tive material in it.

3.1 Description and numerical model-ing

This prototype is composed of one PVDF sheet ofthickness 0.5mm embedded in a soft foam. ThePVDF sheet has a specific profile: it has the shape ofsix long half-cylinders next to each other, oriented up

PVDF FOIL

484

400

384 302

foam

Figure 7: Schematic view of the smart foam actuator

and downwards. The diameter of the half cylindersis 66.6mm, length is 300mm and width is 400mm.The overall thickness of the actuator is 95mm. Thesmart foam actuator is shown on figure 7. Whether itcomes to active transmission or active absorption con-trol it is fundamental that the secondary sources canreach output levels of the same magnitude as the pri-mary sound. The relatively small displacements thatcan be generated by PVDF foils (e.g. compared withan electro-dynamic loudspeaker) directed the foil de-sign towards a maximised volume velocity output.This situation implied a need for strong geometricalcoupling between the in-plane and out-of-plane dis-placements of the foils and a surface covering mostof the foam surface. The most effective configurationamong analysed suggestions came out to be a cor-rugated PVDF-foil in the middle of a porous foamblock. In order to obtain the desired dynamic be-haviour of the foil the direction in which the largeststrain take place, the d31-direction, should be ori-ented along the half-cylinder axis. When every otherhalf-circle profile act in extension and the oppositeones in compression, the corrugated foil deforms ina shape corresponding to the (1,1)-mode of a free-free plate. This displacement shape dominates the re-sponse throughout the studied frequency range 30 -410Hz and gives a smooth frequency response. Withthis shape of the active foil the displacements alsoshowed to be quite insensitive to the foam stiffness.

Figure 8: Schematic view of the set-up for experi-ments with active control of sound absorption

The alternative is to orient the most sensitivePVDF axis perpendicular to the half-cylinder axis.In that case, the excitation voltage would cause theradius of the upper half cylinders to increase, whilethe radius of the lower half cylinders would decrease.However, this yields a lower volume velocity outputcompared to the configuration discussed above – atleast for this configuration.

3.2 Experimental validation

The smart foam actuator was used to increase thesound absorption coefficient of an acoustic lining.The experiments are performed on a small scale smartfoam in a one-dimensional waveguide of diameter16cm as shown on figure 8.

Two control algorithms where implemented on aTMS320C40 DSP.

The first is the controlled time delay method,which is based upon the work presented in [4]. Thecontrol strategy has been developed for use in a one-dimensional waveguide only. The waveguide is as-sumed to be terminated on one side by a primarysound source and on the other side by the smart foamactuator (figure 8). The plane sound waves prop-agate from the primary source to the smart foam,which acts as a boundary where the incident soundwaves are partially absorbed and partially reflected.If the smart foam would be a perfectly absorbingboundary, then the sound field inside the waveguidewould be constituted of incident (plane) sound wavesonly. For two microphones (1) and (2), mounted inthe waveguide (microphone (1) is closest to the pri-mary source), 100% absorption implies that the mi-crophone (2) output must be exactly equal to the mi-

crophone (1) output, measured � seconds before. �

is the plane wave traveling time between microphone(1) and (2). Defining e(t) as :

e(t) = p2(t)� p1(t� �) (1)

where � = d

cwith d the distance between both micro-

phones, and c the speed of sound. Perfect absorptionis realised when e(t) = 0. The smart foam is drivensuch that the target signal e(t) is minimised.

The second method is the wave decompositionmethod. It can be shown that in the case of aone-dimensional waveguide, the measurement of thesound pressure level at two closely spaced locations,can be used to decompose the sound field into an inci-dent component and a reflected component. With thisinformation, it is possible to determine the absorp-tion coefficient of the reflecting medium (the bound-ary), as shown in [5]. From a control point of view,this implies that a two-microphone sensor system canbe used to control the absorption coefficient of theboundary directly.

For both techniques a filtered-X LMS feedforwardcontrol algorithm was used to drive the error signalto zero. However the calculation of the error sig-nal requires quite some of the calculation capacity ofthe DSP. Moreover due to their implementation bothmethods require a high sampling frequency (about7kHz). Therefore the wave decomposition methodwas implemented in an analog electronic circuit thatprocessed the signal from the two microphones in anincident and reflected component. The signal of thereflected component is the error signal fed to the con-troller. In this case the error signal is not calculated bythe DSP and the length of the control filters could beincreased resulting in a better performance. A typicalresult is shown on figure 9 in which the absorptioncoefficient of a smart foam is controlled by the wavedecomposition method at discrete frequencies. Thefigure shows that the sound absorption coefficient canbe increased to almost 100%. Due to a resonance inthe duct at 200Hz, the absorption coefficient at thatfrequency is limited. Similar results are obtained withbroadband excitation.

4 Volume velocity sen-sor/actuator panel

A detailed analysis of this volume velocity sen-sor/actuator panel is presented by Gardonio [6]. Hereonly the main principles are outlined.

Figure 9: Absorption coefficient of the smart foamwith and without control

Figure 10: Schematic view of the volume velocitysensor/actuator panel

The sound radiated by a panel can be decom-posed into independent contributions from the ra-diation modes [7]. Radiation modes consist of aweighted contribution of the structural modes of thepanel. In the low frequency range, the sound powerradiated is dominated by the first term in the de-composition, which is the volume velocity mode orpiston-like mode of the panel. The volume velocity ofthe panel is obtained by integrating the velocity nor-mal to the panel surface over the panel.

The volume velocity sensor/actuator panel con-sists of an Aluminium plate, equipped on each sidewith a PVDF foil. The electrodes of the sensor foilhave a quadratic shape such that its current outputis proportional to the volume velocity of the plate[8]. The electrodes of the actuator foil have the sameshape, such that they exert a uniform force on theplate and thus excite the volume velocity of the plate.This way a collocated sensor/actuator pair is formedthat has some advantageous properties for control.

A schematic drawing of the volume velocity sen-sor/actuator panel is shown on figure 10.

A main problem was encountered when measuring

the transferfunction between the sensor and actuatorfoil, as shown on figure 11. The solid line shows thetransferfunction when only the bending motion of thepanel is considered, and exhibits a alternating pole-sero pattern. However the actuator and sensor foil arenot placed symmetrically with respect to the neutralaxis of the panel, and due to the non perfect clamp-ing conditions, the plate can extend as well. The ex-tension produces a constant charge in the sensor foil,and thus a current proportional to the excitation fre-quency. This extension causes a feedthrough termin the transferfunction. In case the extension is in-cluded, the simulated transferfunction (dotted line)agrees well with the measured one (solid gray line).

The study of the panel is being continued, and lat-est results are reported in [9].

5 EMFi electro-mechanical filmactuator

This loudspeaker panel was not developed in theDAFNOR project. It is based on the electrostaticworking principle [10]. A thin charged film is vi-brating in an electric field. The panel can be used asacoustic actuator and sensor. The EMFi panel has thesame dimension as the other active panels. The thick-ness is 7mm, and the latest generation of the panel hasa thickness of a few millimeters.

When used as an actuator, the frequency responseis quite flat with a roll-off towards low frequencies.Still the low frequency output is enough to allow ac-tive noise control experiments.

6 Comparison of the three actua-tor prototypes

In principle, the three actuator prototypes were de-signed for various applications, and therefore a com-parison is not possible. However in order to havesome idea of the efficiency of the prototypes in con-verting the electrical energy in sound power output,figure 12 compares the sound pressure measured at1 meter distance from the prototype above the actu-ator centre. The actuators are mounted in the semi-anechoic room at the KUL acoustical laboratory, inthe opening above the sending room. To completethe figure, the same measurement was repeated forthe EMFi panel. The actuators are excited with themaximum allowable voltage. This is 200V for thePVDF actuator which is the limit of the power am-

102

103

−240

−230

−220

−210

−200

−190

−180

−170

−160

−150

−140

Gai

n (d

B)

Freq (Hz)

(a) magnitude

102

103

−200

−150

−100

−50

0

50

100

150

200

Phas

e (°

)

Freq (Hz)

(b) phase

Figure 11: Transferfunction from voltage excitation to current output from the volume velocity sensor/actuatorpanel

EMFi

smart foam

acoustic actuator

volume velocity panel

50 100 150 200 250 300 350 400 450 500 55030

40

50

60

70

80

90

100

frequency (Hz)

soun

d pr

essu

re (

dB r

ef. 2

e−5

Pa)

Maximum sound pressure output

Figure 12: Sound pressure level at 1 meter distancefor maximum excitation voltage

plifier used. One should notice that this is well be-low the maximum voltage that PVDF can withstandbefore breakdown. For practical and safety reasons, alimit of 200V amplitude is reasonable. To prevent dis-tortion of the acoustic output, the voltage to the EMFipanel is limited to 30Vrms. In recent developed EMFipanels, the excitation voltage can be increased and thesound power output of the panel has been increasedconsiderably.

The sound pressure output from the acoustic actu-ator shows a first resonance at a frequency of about

180Hz, and a second resonance around 430Hz. Thefirst actuator mode coincides with a vertical mode inthe sending room at 130Hz. This actuator yields asound pressure in the frequency region below 200Hzwhich is about 20dB higher than any other actuatorthat was tested, which is very important frequency re-gion when dealing with noise control problems. Itwill be especially interesting in applications wherethe sound field is dominated by a few harmonics, suchthat the design of the acoustic actuator can be tuned tothe frequency of the disturbing sound field. This tun-ing can be realised by choosing an appropriate thick-ness for the honeycomb structure.

The smart foam actuator is a weaker source, asalready expected from the simulations, especially atvery low frequencies.

The volume velocity sensor/actuator panel wastested using both the PVDF layers as actuator drivingthem in opposite phase, in the same way the acousticactuator is operated. The results show that this actua-tor has a higher modal density below 500Hz. The areaof the PVDF elements is smaller than for the acous-tic actuator due to the shaped electrodes. Due to theAluminium plate, the actuator has a higher bendingstiffness, which will result in a lower vibration ampli-tude and a lower sound pressure. If one PVDF sheetis used as actuator, the sound pressure output will belower than the results shown in figure 12.

The EMFi panel shows a rather flat frequencyspectrum above 500Hz. Only above 150Hz, the sound

Figure 13: Picture of the two acoustic actuators in-stalled in the back of the car

pressure output is sufficient for active noise controlapplications.

It has to be concluded that none of the actua-tors can produce the high sound power output andthe flat frequency spectrum of a classical voice coilloudspeaker. However, these actuators can be appliedin configurations where classical loudspeakers cannotbe implemented due to their size or weight.

7 Active control of noise in a carinterior

In a first practical set-up, the acoustic actuator wasused as secondary loudspeaker in a control system toreduce the noise level in the interior of a Fiat Punto.

Two acoustic actuators are placed in the back ofthe car, as shown on figure 13. In addition, two EMFipanels are placed in the front doors. The resonancefrequency of the acoustic actuator is well tuned to thefrequency of a low frequency booming noise in thecar interior. The EMFi panels have a more flat spec-trum, and will reduce the sound at higher frequencies.In each head rest of the front seat, a microphone isinserted. Two more microphones are located in themiddle of car compartment near the internal rear mir-ror and in the rear of the car compartment.

Figure 14 shows the result of the active controlsystem when controlling the second engine harmonicin all four microphones, while exciting all four ac-tuator panels. A reduction of 5dB is observed atseveral frequencies, especially at a frequency around4500rpm in the error microphones. This reductioncontributes considerably to the sound comfort insidethe car.

Figure 14: Sound pressure in two microphones beforeand after control

8 Sound transmission loss mea-surement according to ISO 140

A series of tests were performed controlling the soundtransmission loss through a single and double panelpartition [11]. The tests were performed accordingto ISO standard 140 in the measurement facilities ofG+H Montage in Ludwigshafen, Germany. From lab-oratory tests, the most promising configuration wasselected, which is a set-up similar as shown on figure3 with one error microphone in the centre of the cav-ity. However the active transmission wall consists offour active panels, placed two by two aside in frontof a steel plate of dimensions 800mm � 1000mm.The error microphones are placed in the cavity be-tween the active panels and the steel plate, each mi-crophone in the centre of an active panel. On fig-ure 15 two active panels are removed to show the er-ror microphones. The control system is a filtered-XLMS feedforward control algorithm, implemented ina Digisonix control system. The controller is a di-agonal type, each active panel controls the sound inthe microphone placed right in front of it, and not inthe other microphones. This means that 4 indepen-dent Single Input/Single Output control systems arerealised. This can be implemented in an analog feed-back controller for future applications, one controllerper active panel.

The result of the experiments is shown on figure16. The active panels are EMFI panels, and the set-up is excited with a band-limited sound field in thefrequency region between 100Hz and 500Hz of 85dBSPL. A comparison is made with previous measure-ment on a standard passive transmission wall as it isproduced by G+H Montage (shown in solid line). The

Figure 15: Picture of the set-up; two actuators areremoved to show the error microphones in the cavity

system without control (dashed line-circles) showsvery good agreement with this measurement. In thevery low frequency range, below 120Hz, the resultdepends on the acoustical properties of the rooms.At frequencies above 300Hz, the mineral wool in thecavity of the standard transmission wall influencesthe result. The active system (dashed line-squares)yields an increase of 6dB in the sound transmissionloss at low frequencies, which is a considerable im-provement of the characteristics of the passive stan-dard transmission wall.

Figure 16: Measured sound transmission loss of theactive wall made up of four active panels

9 Conclusions and future re-search

At the end of the project, the feasibility of usingPVDF material in active noise control applicationshas been demonstrated.

The acoustic actuator can produce a relatively highsound power output compared to other flat actuators.By changing its design, its first resonance frequencycan be tuned to the frequency of the disturbing noise.Both on simulation and on experiments it was shownthat it can be applied successfully to increase thesound transmission loss through a panel partition.

The smart foam can increase the sound absorptioncoefficient of a foam material in the low frequencyrange to more than 90%. Simulations demonstratethat it can also be implemented in a sound transmis-sion wall. Its sound power output remains limited.

The volume velocity panel is capable of control-ling the sound transmission through a panel by using asimple single input/single output control system. Dueto mechanical feedthrough, the panel showed a differ-ent behaviour than expected. This problem has beenstudied and clarified.

The most promising control strategy and actua-tor configuration was demonstrated on a actively con-trolled transmission wall. An increase of 6dB was ob-served below 300Hz, which is an important improve-ment in the characteristic of the transmission wall.

The potential of using PVDF for active noise con-trol was demonstrated in three prototypes. The nextstep is further development of these panels to an in-dustrial product.

Acknowledgements

The work reported herein was related to the ECBrite/Euram Research Project ’DAFNOR’ (undercontract BRPR-CT96-0154). The project is supportedby the Directorate-General for Science; Research andDevelopment of the CEC. Partners in this project are:KULeuven (B), FFA (SE), ISVR (UK), Thomson (F),VTT (FI), CRF (I) and G+H MONTAGE (G).

The authors would like to thank all project part-ners for their contribution to the project and to thispaper, especially P. Gardonio, S. Debost, G. Roux, H.Nykanen, H.-J. Milz, G. Toniato and U. Tengzelius.

References

[1] Henrioulle K., Dehandschutter W., and Sas P.Design of a distributed actuator for active noisecontrol. 4th European conference on smartstructures and materials, Harrogate, UK, pages139–146, 1998.

[2] Henrioulle K., Dehandschutter W., and SasP. Active control of sound transmission lossthrough a single panel partition using distributedactuators. Part I : Simulations. ISMA23 Interna-tional conference on noise and vibration engi-neering, pages 797–804, September 1998.

[3] Fuller C.R., Bronzel M.J., Gentry C.A., andWhittington D.E. Control of sound radia-tion/reflection with adaptive foams. Noise-Con94 Ft. Lauderdale, Florida, pages 429–436,May 1994.

[4] Orduna-Bustamante F. and Nelson P.A. Anadaptive controller for the active absorption ofsound. J. Acoust. Soc. Am., 91(5):2740–2747,May 1992.

[5] Elliot S.J. A simple two microphone methodof measuring absorption coefficient. Acousticletters, 5(2):39–44, 1981.

[6] Gardonio P., Lee Y.S., Elliot S.J., and DebostS. Active control of sound transmission througha panel with a matched PVDF sensor and ac-tuator pair. ACTIVE99, Fort Lauderdale, USA,December 1999.

[7] Johnson M.E. and Elliot S.J. Active control ofsound radiation using volume velocity cancel-lation. J. Acoust. Soc. Am., 98(4):2174–2186,October 1995.

[8] Elliot S.J. and Johnson M.E. Radiation modesand the active control of sound power. J. Acoust.Soc. Am., 94(4):2194–2204, October 1993.

[9] Gardonio P., Lee Y.S., Elliot S.J., and DebostS. Design of a panel with matched PVDF vol-ume velocity sensor and uniform force actua-tor pair. ISMA25 International conference onnoise and vibration engineering, Leuven, Bel-gium, September 2000.

[10] Nykanen H., Antila M., Kataja J., Lekkala J.,and Uosukainen S. Active control of soundbased on utilizing EMFi-technology. Active99,

Fort Lauderdale, Florida, USA, pages 1159–1170, 1999.

[11] Henrioulle K. and Sas P. Experiments on theactive increase of sound transmission loss us-ing distributed actuators. 137th meeting of theacoustical society of America and the 2nd Con-vention of the European acoustics association :forum acusticum Berlin, March 1999.