Patrizia

AERODYNAMIC AND JET NOISE

Measurement and Control of Flow Generated Noise

P. O. A. L. Daviesa, K. R. Holland

a and D. C. van der Walt

b

aInstitute of Sound and Vibration Research, University of Southampton, Southampton SO17 1BJ, UKbBosal Afrika, Central Development Plant 10, PO Box 1652, Pretoria, South Africa

This contribution reviews progress in the technology for assessing the silencing performance of piston engine intake andexhaust systems on a running engine, with particular reference to flow noise. The new developments include theapplication of selective averaging, order tracking and optimised sampling rate methods to clearly identify and quantify thesmall fraction of the total fluctuating wave energy that is being propagated across each discontinuity along the flow paththrough a highly reactive flow duct system. Such measurements then quantify the interaction between the aeroacousticsources of excitation and the associated local acoustic characteristics that govern the generation, transfer, propagation andemission of wave energy to the environment.

INTRODUCTION

Intake and exhaust systems are designed toreduce engine breathing noise emissions tolevels that comply with legislation, provide asound quality and internal vehicle climate thatmeets customer expectations, while maintainingoptimum fuel efficiency and vehicleperformance. This diversity of function,coupled with operational, system layout andspace allocation constraints, together with thebenefits of rapid prototyping, underline thepractical advantages of a design methodologythat is firmly based on realistic predictivenumerical modelling [1] of acoustic andoperational performance. Appropriate existingsoftware [2] can adequately describe the passiveacoustic and resonant behaviour ofgeometrically complex flow duct systems, butdoes not yet include any influence of flow noisesources on predicted acoustic performance.This deficiency is of practical importance, sinceflow generated noise is commonly the majorcontributor [1-3] at higher engine speeds to ICengine breathing noise spectra above 200 Hz.

FLOW NOISE AND ITSMEASUREMENT

The sustained excitation of a tuned resonatorby shed vorticity in a separating shear layer [4]has been exploited empirically for makingmusical and other sounds from timeimmemorial. In the normally highly

acoustically reactive intake and exhaust systems suchmechanisms [1-4] are seen to generate both coherentand broadband sound, or noise, by a flow drivengenerator that is highly nonlinear, exciting a resonantsystem with effectively linear acoustic behaviour.Similar mechanisms [1, 2] can also cause selectiveamplification of incident sound. The dominantaeroacoustic mechanism of concern here [4] is theaction of a fluctuating Coriolis force working on thefluctuating flow. However, with some specificexceptions [1, 4], the relevant controlling features ofthe flow cannot be specified significantly explicitly,due to current ignorance of many essential details ofthe associated vortical or turbulent fluid motion. Thusone must normally rely on measurement to establishquantitative descriptions of aeroacoustic sources inrelation to the associated boundary geometry, acousticclimate and fluid motion. Intake and exhaust system geometry consistsessentially of silencing and other component elementsconnected in sequence by lengths of uniform pipe.Wave reflections at the junctions, combined with anylocal sound generation, produces an acoustic climatecomprised of both standing and progressive waves.Superimposed on the time averaged flow, the unsteadyfluid motion includes both the acoustically relatedpotential fluctuations combined with vorticaldisturbances cast off with the separating shear layers[1, 4] or generated at the boundaries, that travel withthe mean flow. New robust procedures [2-4] usingcross-power spectral analysis of the signals from pairsof flush wall mounted pressure transducers were usedto quantify the incident and reflected wave amplitudespectra at appropriate positions along the flow path.The application of selective averaging, order tracking

and optimised sampling rate methods [2-4]provided sufficient precision to identify clearlythe small fraction of the total fluctuating waveenergy that was being propagated through thesystem. The location, strength and spectralstructure of any flow noise sources/sinks wasthen derived from the small gains/losses inpower flux at the relevant sites. Pilotexperiments [2, 4] revealed that the flowgenerated sound spectra were always closelyrelated to the system acoustic resonances. Earlier experimental studies of flow generatedsound [1, 5] have largely been concerned withmeasurements of the sound power radiated fromthe open termination into anechoic,semianechoic or reverberant enclosures, ratherthan at the sites where it is generated.However, the observed harmonic structure ofthe emitted sound was also closely associatedwith the system resonances, while the overallintensity was proportional to the mean flowMach number M raised to a power lyingbetween four and six for M < 0.4. Experimentswith flow excited expansion chambers [1, 4, 5]showed that the sound power was also a rathercomplex nonlinear function of x/d, 2 < x/d < 15,where x is the chamber length or that of the freeshear layer between inlet and exit pipes withdiameter d. Similarly complex nonlinearbehaviour was observed with other typicalexhaust system component geometries. Measurements were also made with stronglyexcited expansion chambers [2, 3], either on thetest bench, or in the exhaust line of a runningengine accelerating to full speed on a test bedwith wide open throttle. The observationsprovided clear evidence of reverberantamplification or attenuation of the incidentsound at specific frequencies, although therelationship with the associated boundarygeometry, acoustic climate and fluid motion hasnot been established yet. Otherwise, therelative observed spectral behaviourcorresponded to that predicted by onedimensional linear acoustic models. This wasin spite of the fact that the observed amplitudeof some spectral components of the cyclicexcitation exceeded 3.5 kPa, corresponding to165 dB. With purely flow excited systems onthe test bench [2, 4] the maximum soundpressure levels recorded were some 20 dBlower than this, although the correspondingoverall fluctuating pressure levels were close to

160 dB. Otherwise, acoustic characteristics calculatedwith linear acoustic models were found to be in goodagreement with measurements, indicating that wavesteepening was not yet a significant factor at theconditions of these experiments.

DISCUSSION

Pilot studies [2-4] have established and validatedrobust and sufficiently precise experimentaltechnology for establishing the position strength andspectral characteristics of flow noise sources in relationto the boundary geometry, the mass flow and theassociated acoustic climate. This includes theinfluence of regenerative amplification by acousticfeedback [2, 3] on source characteristics, with somebasic examples [4] of source modelling. The recordshows that problems arising from poor signal to noiseratio associated with the presence of standing wavesand signal contamination by turbulent pressurefluctuations can be overcome [2, 3] by adopting sweptsine or swept periodic excitation combined withappropriate selective averaging. With uncontrolledexcitation by flow, coherent power flux measurements[2, 4] were closely identified with the acoustic powerflux. Further progress in predictive modelling ofaeroacoustic sources depends on establishingappropriate details [4] of the associated fluid motion.This includes the mapping of the measured orcalculated distribution of the fluctuating potential andvortical velocity components with the correspondingCoriolis accelerations. To accomplish suchmeasurements without introducing further sourcesrepresents a challenge that is a subject of currentstudies at the ISVR.

REFERENCES

1. Davies, P.O.A.L., and Holland, K.R., Journal of Soundand Vibration 223, 425-444 (1999).

2. Holland, K.R., Davies, P.O.A.L., and van der Walt, D.C.,ISVR Technical Report No. 291, University ofSouthampton (2000).

3. van der Walt, D.C., Journal of Sound and Vibration 243,797-821 (2001).

4. Davies, P.O.A.L., and Holland, K.R., Journal of Soundand Vibration 239, 695-708 (2001).

5. Desantes, J.M., Torregrosa, A.J., and Broach, A., Acustica87, 46-55 (2001).

Flow Induced Noise from a Simple Expansion Cavity TypeMuffler with Flow Pulsation

N. Kojimaa, M. Mikamia and T. Esakib

a Department of Mechanical Engg., Yamaguchi University, Tokiwadai Ube, 755-8611 Japanb Graduate School of Yamaguchi University, Tokiwadai Ube, 755-8611 Japan

The high flow rate in a muffler induces flow noise. This noise sacrifices the noise reduction effect of the muffler. The flowinduced noise from a simple cavity type muffler with pulsating flow and steady flow was comparatively studied. Thepredominant components of flow induced noise were related to the resonance of the cavity and the tail pipe of the muffler. Theflow induced noise was extracted from the total noise radiating from the open end of the muffler by introducing a noisegeneration model. The sound power level of the flow induced noise in the case with pulsating flow was different from the casewith steady flow, especially for the cavity resonance components. Since the pressure fluctuation on the solid boundary is themain source of flow induced noise for a low Mach number flow, the pressure fluctuation spectrum was compared between thecases with pulsating flow and steady flow.

INTRODUCTION

According to the improvement of automobile engine toprovide high speed and high power, the amount ofworking gas increases. When a high flow rate of gasexists, a large flow induced noise radiates from theopen end of the muffler. This flow induced noisesometimes sacrifices the noise reduction effect ofmuffler predicted through the acoustic theory. In the previous papers[1,2], the flow induced noisegenerated from the simple cavity type and the insertedtype mufflers with steady flow was experimentallystudied. In this paper, the characteristics of flow noisegenerated from simple expansion cavity type mufflerwith pulsating flow were studied. The noise sourcesinside the muffler were also discussed in comparisonof the case with steady flow.

EXPERIMENT AND DISCUSSION

The experimental set up is illustrated in Fig.1. The airflow from a blower was supplied to the pre-mufflerwhich had the sound absorbing structure inside itswalls. In the experiments for pulsating flow, thepulsation generator was inserted at the position of3070mm up-stream side of the entrance of the muffler.The probe microphone was employed to detect thepressure fluctuation on the inner wall of the mufflercavity. The mean velocity of the air flow supplied tothe muffler was 33m/s both for steady and pulsatingflow experiments.

Extraction of Flow Induced Noise from TotalRadiated Noise of Muffler

In the experiments for the pulsating flow inducednoise, the radiated noise contains not only the flow

noise but also the mechanical noise from pulsationgenerator. It is necessary to extract the noise generatedby pulsating flow from the total noise of muffler.Then, the noise generation model of muffler wasintroduced. The noise power radiated from the open end of themuffler Wm is recognized as a sum of two kinds of noisepower. The noise carried into a muffler from theupstream side is attenuated by the filter effect ofmuffler and carried out. This noise power can beevaluated as WpF, where Wp is the noise powerradiated from the open end of the straight pipe(without muffler), and F is the insertion loss of themuffler. The other one is the noise power induced bythe air flow Wf inside the muffler, then FWWW pmf (1)

Figure 2 shows the flow induced noise for pulsatingflow calculated by this method in comparison of thecase with steady flow. In the case with pulsating flow,the noise power of tail pipe resonance is larger 10-12dB than in the case with steady flow. For thecomponents of cavity resonance in axial direction, thisdifference becomes larger as 15-25dB. These resultsstrongly suggest that the generation mechanism offlow induced noise in the pulsating flow is not thesame as that in the steady flow.

Effects of Cavity Walls on Flow Induced NoiseTwo types of noise source can be considered for theflow noise generated inside the muffler. One is theturbulence of the flow and the other is the pressurefluctuation at the solid boundaries. Curle[3] proposedthe relationship between these two types of noisesource and generated sound pressure. Introducing theadiabatic process, the sound pressure formed at a point

apart r from these noise sources is derived as;

S iV ij ydP

trx

cydT

trxx

cp iji 2

223

2

2

32 41

41

(2)

The first term of this formula expresses the soundpressure component generated by the disturbance ofthe flow, and the second term expresses the componentgenerated by the fluctuation of the pressure at the solidboundaries such as inner walls of muffler. From theresults of order estimation of these two terms, the firstterm was two orders smaller than the second term in aslow Mach number flow as employed in thisexperiment. Then, the flow induced noise generatedinside the muffler has to be considered in therelationship with the second term of Eq(2). Figure 3 shows the typical example of pressurefluctuation on the inner walls of the muffler cavity incomparison between the case with steady flow andpulsating flow. As shown in Figs.3(a) and (b), thepressure fluctuations on the inlet wall and the side wallof the cavity are larger in the case with pulsating flowthan in the case with steady flow as much as 2.0-3.0 inlog(∂P⁄∂t). This difference will correspond to 20-30dBin the sound power level of noise sources. One of thereasons for this increase in pressure fluctuation ispossible that the disturbance of flow in radial directionbecomes strong by introducing the pulsating jet flowinto the cavity. When we introduce the steady flow to the muffler,the pressure fluctuation on the side wall becomes smallin high frequency range as shown in Fig.3(b). Whileon the exit wall of the cavity, the pressure fluctuationis larger than on the side wall as much as 1.5-2.0 inlog(∂p⁄∂t) as shown in Fig.3(c). This must come fromthe situation that the disturbance in jetting flow isincreased by introducing the flow and is transportedmostly in axial direction, then impinges on the exitwall of the cavity. From these results of the order estimation and theexperiments, the following two matters were clarified.(1) The pressure fluctuation affecting the inner walls

of muffler cavity has a superior power as the noise

source of flow induced noise in comparison with theshare of jetting flow. (2) In the case with pulsatingflow, the power of flow noise source was predominanton the inlet and side walls of the muffler cavity, whileon the exit wall in the case with steady flow.

REFERENCES

1. N. Kojima, B. Liu and H. Zhou, Relation between thePredominance of Acoustic Resonance Noise and AirFlow in Muffler, SAE Paper No.951262, P-291, pp.217-225(1995).

2. B. Liu, N. Kojima and M. Mikami, Generation of FlowInduced Noise from Inserted Pipe Type Muffler, Proc.ASME, Paper No.95-WA/NCA-22, 7p. (1995).

3. N. Curle, The Influence of Solid Boundaries onAerodynamic Sound, Proc. Roy. Soc., (A)231-1187,pp.505-514(1955).

FIGURE 1. Experimental set up

FIGURE 2. Power level of flow induced noise

FIGURE 3. Pressure fluctuation spectrum

Codenser

mi crophone

Bl ower Anechoi c room

Muff l er

Sound absorbi ng materi al

Tank ƒÓ600~900

2480

300

180

Pul se generator

ƒÓ32ƒÓ55ƒÓ100ƒÓ145

uni t : mm

3070

45‹

815 675 770

30

40

50

60

70

80

0 1000 2000 3000 4000 5000

Steady flowPulsating flow

Frequency Hz

tail pipe resonance

axial direction resonance of the cavity

Power level dB

8

10

12

14

16

0 1000 2000 3000 4000 5000

(a) On the inlet wall of the cavity (x=0,r=45)

Steady flow

Pulsating flow

log(ÝP/Ýt)

Frequency Hz

8

10

12

14

16

0 1000 2000 3000 4000 5000

(b) On the side wall of the cavity (x=50,r=75)

Steady flowPulsating flow

log(ÝP/Ýt)

Frequency Hz

8

10

12

14

16

0 1000 2000 3000 4000 5000

(c) On the exit wall of the cavity (x=150,r=35)

Steady flow

Pulsating flow

log(ÝP/Ýt)

Frequency Hz

Effect of Tabs on Reduction of Feed-back ResonanceGenerated in a Short Enlargement Muffler with High Speed

FlowY. Nakazonoa and T. Tanisugia

aDepartment of Mechanical Systems Engineering, Kyushu Tokai University, Kumamoto, 8628652 Japan

Effect of the tabs on the acoustic characteristics of resonance generated in short enlargement muffler (cavity) was examined of0.3 and 0.5. As the results, tabs are very effective to decrease feedback resonance generated in the muffler. The reduction dependson de-correlate of the azimuth coherence vortices. For an area blockage of tabs to the nozzle exit greater than 3 %, feedbackresonance is reduced considerably.

INTRODUCTION

When the jet vortex hits directly against the trainingedge of a short enlargement muffler, then the pressurewave is generated, and the wave returns to the leadingedge of the muffler. This phenomena is repeated itselfmany times, then feed-back resonance occurs. Todecrease feedback resonance, many techniquesinclining the downstream wall of the muffler have beentried 1. But the problem of these methods is that themuffler structure becomes too long. Tabs have been shown to produce stream-wisevortices on both sides of the tabs. This device has beenapplied for reducing jet noise. Therefore, it isconsidered that the application of tabs to the reductionof feedback resonance generated in the enlargementmuffler is effective. The present paper examines the effect of tabs on thereduction of feedback resonance.

Results and Discussion

Schematic diagram of experimental apparatus isshown in Fig.1. High speed air flow is exhausted froma nozzle diameter of 20mm attached to a plenumchamber in which 5 screens are set. The aperture of thetrailing edge is the same with the nozzle diameter.Cavity configuration is L(0 to 80mm)H60mmW60mm and its section is made of clearPlexiglas to visualize the flow. Triangular tabs are fittedto a circular nozzle being the inlet shape of theenlargement muffler. The heights of the tabs are 2 or4mm, and tabs number is 1, 2, 3, 4 and 5. The areablockage of each tab is about 0.6% of the nozzle exitarea for the tab height of 2mm, and is about 2.5% of thenozzle exit area for the tab height of 4mm. The cavitylength L was varied from 0 to 80mm by

moving the trailing edge. The sound pressure spectra of the noise generated

aerodynamically in the muffler was measured at aposition of R=15cm and =60º by a 1/4 inchmicrophone. The sound pressure spectra without a tab showed a lotof the complicated discrete frequencies. The dominantfrequency is called cavity tone frequency in the presentexperiment. The cavity tone frequencies for Machnumber 0.5 are shown in Fig.2 as a function of cavitylength L. In the same figure, the calculation values ofboth resonance frequencies for cavity length directionand lateral direction are shown. The frequency modesfor the lateral direction fH and for the cavity length fLare defined as na0/(2H) and na0/(2L) respectively,where n is integer, a0 sound speed, H cavity height andL cavity length. The frequency mode of fH is seen ataround L=30mm and 55mm. The frequency mode of fLis seen at around L=15, 40 and 65mm. These frequencymodes were also produced at a Mach number of 0.3.The acoustic feedback mode (Rossiter’s equation2) andthe cavity length mode shows a similar frequency,therefore it is difficult to separate these frequencies.Tam & Ahuja3 stated that the feedback doesn’t occurbelow Mach number of 0.4. But in the presentexperiment the feedback was produced at Machnumber 0.3. These phenomena might depend on thedifference of experimental apparatus whether the box isinstalled or not. Varying the shear layer by fitting tabs to the nozzle,jet vortex structure changes. As the vortex becomessmaller than before fitting the tabs, the hit of the vortexis alleviated. The pressure wave produced by the hit isdecreased, and then noise generation is reduced. The effect of tabs on the overall sound pressure level(SPL) is shown in Fig.3 as a function of the cavitylength. With no tabs, an intensive sound pressure levelof 140dB is generated. For tabs of h=2mm and nt=5which corresponds to blockage area ratio 3.1%, thesound pressure level is 115dB, that is, a noise reductionof 25dB is found. According to photographs of the flowvisualization4), the hit of the vortices against the trailingedge generates the discrete frequency tone. When the

100

110

120

130

140

150

0 10 20 30 40 50 60 70 80

L(mm)

OASPL(dB

Normalh=2mm nt=2h=2mm nt=3h=2mm nt=4h=2mm nt=5h=4mm nt=2h=4mm nt=3

vortex collapses before hitting against the trailing edge,the discrete tone does not appear. In the case of h=2mmand nt=3, it is found that the vortices hit against edgethrough the photographs of the flow visualization.Therefore, the lateral frequency mode is observed alittle, but the frequency mode of length direction is notseen. The tabs decrease the acoustic resonance ofcavity length direction more effectivethan that of lateral direction. By increasing tab numbersand tab height, the cavity tone is decreased. Theappropriate tab numbers and tab height should beselected. In the present experiment, for the tab height of2mm, tab numbers not less than 4 are proper, and forthe tab height of 4mm, 2 tabs is good. The blockage oftheir tabs area to the nozzle area is 2.5% and 5%,respectively. Moreover, an additional experiment forthe tab height of 3mm from 2 to 4 tabs was conducted.Noise reduction of 15dB (overall) for 2 tabs (blockage2.8%) was obtained. Then, the blockage above 3% iseffective to decrease the cavity tone. To examine the reason reducing the cavity tone, thecoherence function of the velocity fluctuation in theshear layer was measured using 2 hot wires. In thiscase, the side of the enlargement muffler was removedto do the measurement of coherence easily. As theresults, the values of coherence function of the velocityfluctuation with tabs were lower than that without tab.It is found that tabs de-correlate the azimuth coherencevortices.

Conclusions

Tabs are very effective to decrease the cavity tonegenerated in the short enlargement muffler. Thereduction depends on de-correlation of the azimuthcoherence vortices.

REFERENCES

1. Liu B. & Kojima N., Noise Control 20-5,55(1997) (in Japanese).

2. Rossiter J.E., Ministry of Aviation RM 3438,(1964).

3. Tam C.K.W. & Ahuja K.K., J. Fluid Mech. 214,67 (1990).

4. Nakazono Y. & Sano Y., AIAA Paper 99-1828,

(1999).

FIGURE 1. Experimental apparatus.

FIGURE 2. Cavity tone frequency vs. cavity length LM=0.5, Rossiter’s Eq.L mode,: H mode.

FIGURE 3. Overall sound pressure level as a functionof Cavity length L, M=0.5.

A Dynamical Countermeasure Method for the Non-stationaryWind-induced Noise based on Generalized Regression Model

and Its Field ExperimentK. Hatakeyama a and M. Ohta b

aKinki University, Takaya-Umenobe 1, Higashi-Hiroshima, 739-2116 Japan. hatake@hiro.kindai.ca.jpbEmeritus Prof. Hiroshima University, Matoba 1-7-10, Minamiku, Hiroshima, 738-0824 Japan. ohta-

32222@hiro,kindai.ac.jp

In this paper, the problem of detecting the acoustic signals contaminated by random background and wind-induced noises is discussed fromthe statistical viewpoint. For evaluating the effects of wind-induced noise, we will pay our special attention to the measurement of windvelocities which cause the wind-induced noise. By grasping the statistical relationship between wind velocities and wind-induced noise inthe form of generalized regression model, a new type wide-sense digital filter for detecting the acoustic signals under wind-induced noisesis proposed from Bayesian viewpoint theoretically and experimentally.

INTRODUCTION A wind-induced noise influences the acoustic mea-surements in the outdoor. The wind-induced noisefluctuates randomly owing to the temporal changes ofwind velocity at the observation point and shows arbitraryprobability distribution forms of non-Gaussian type. It isnecessary to establish some systematic countermeasuremethods for the wind-induced noise especially whenmeasuring the low frequency acoustic signals, because thewind-induced noise includes components in a lowfrequency range which can not be reduced with use of theusual methods ( such as the usage of wind screen andothers). This is due to the fact that the physical mechanismof wind-induced noise is not known and the stationaryproperty of wind-induced noise can not be assumed. Here, we will pay our attention to the fact that the wind-induced noise is generated from the temporal change ofwind velocity, which can be measured separately whenobserving the acoustic data. The wind velocity changesrelatively slower than the changes of the wind-inducednoise and, in addition to the amplitude of wind velocities,the directions and the frequency components of them arealso important. Then, for predicting the wholeprobability form of wind-induced noise, we shouldobserve the velocities in different times and/or positions,and then consider the nonlinear multivariate correlationinformations between the wind-induced noise and thewind velocities. In this paper, from the above viewpoint, a new trial ofdynamical countermeasure method for the wind-inducednoise will be proposed by considering how to useeffectively the multivariate information of wind velocities.More concretely, we will expand first the multivariatemoment generation function hierarchical-ly and derive themultivariate form of generalized regression model of thewind-induced noise on the set of wind velocities. Next,by predicting various statistics of the wind-inducedrecursively, we evaluate the true acoustic signals

dynamically under the wind-induced noise from Bayesianviewpoint. Finally, the experimental confirmation of the proposedmethod has been confirmed by applying it to the actualdata of field measurement.

THEORETICAL CONSIDERATIONSThe Generalized Regression Model of Wind-

induced Noise on Wind Velocities It is well-known that there is the averaged relation-shipbetween a wind-induced noise vk and a wind velocity

uk2 (power) :vk uk

2, (1)where n: a priori known integer, and : known pro-portional constant. But the wind-induced noise can not beexplained only with use of the averaged relationship inEq.(1). Then, we will consider the deviations from Eq.(1): k vk uk

2 . (2)Then, we will detect and utilise the correlation informationembedded under k as much as possible, by using theinformation on the wind velocities powers. Here, weconsider the m observations of wind velocities

u1k2 , u2k

2 ,, umk2 ( u1k

2 is the present vel-ocity

u1k2

uk2 ) at the different time or positions. For grasping

the whole probability form of the deviations of a wind-induced noise without minimum information losses, thejoint probability function P(k, u1k

2 , u2k2 ,

, umk2 )

should be considered.By introducing the joint characteristic function

M(, 1, 2,, m) ( expk u1k21 umk

2m )

associated with the joint probability function on wind-induced noise and wind velocities, the hierarchicallyexpanded joint characteristic function can be obtained asfollows :

M(, 1, 2,, m) exp0,,1,,0 jj1m

0,,2,,0 j2 / 2 gi1,,im ()i1im 0

1i1m

im / (i1!i m!), (3)

w

here

i,i1,,i m denotes the

(i, i1,, im) -th jointcummulant of the wind noise and the wind velocities. Thecoefficient of each expansion term is defined as :

gi1,,im () ( j

)j1m

i j

expi1im2

i,i1,i2 ,,im 1i12

i2m

i m / i1!i2! im! | j0 . (4)

The correlation information among the wind noise and thewind velocities can be reflected in each expansioncoefficient hierarchically. By using the inverse Laplacetransformation of the moment generating function, thestatistical prediction of arbitrary moments of a deviationterm k can be derived in the form of function on theobservation of wind velocities :

kN | u1k

2 , u2k2 ,, umk

2

i1i m0

gi1,,im(N) (0) Hi j (

uj2 0,,1,,0 0,,2,,0

)j1

m

/( 0,,2,,0 )i j / gi1,,i m (0)

i1im 0

Hi j (uj

2 0,,1,,00,,2,,0

)j1

m / ( 0,,0,2,0,,0 )i j (5)

where

gi1,,im(N) (0) ( )N gi1,,im () |0 .

That is, the generalized multivariate regression model canbe realized for predicting the arbitrary moments of thewind noise with use of the wind velocities.

A Wide-Sense Digital Filter for Detecting AcousticSignals Embedded under Wind-induced noise

In the same analytical viewpoints, a dynami-calsignal detection method under the background and thewind noises can be derived. Here, based on the analysisbased on the additive property of energy quantities and thephysical mechanism, the system and observation equationsfor the acoustic systems are formulated as:

xk1 Fk(xk ,uk ), yk Hk(xk , vk ), (6, 7)

where xk denotes the unknown acoustic signal at a timestage k, and yk denotes the contaminated obser-vation.vk and uk are the background noise and the wind noise. For detecting the acoustic signal xk with use of theobservation recursively, the conditioned joint mo-mentgenerating function M*(, 1) ( expxk

yk1 | Yk ) should be considered. Here Yk is theset of past observations

y1, y2, , yk. Then, by considering Bayes theorem, the unifiedalgorithm of estimating the unknown acoustic signalsunder these random noises can be obtained :

xkN | Yk gj

*(N)(0)j0

H j(yk 0,1

*

0,2*

) / ( 0,2* )j

/ 1 gj*(0)

j1

H j(yk 0,1

*

0,2*

) / ( 0,2* )j

, (8)

where i, j* denotes the (i,j)-th predicted joint cummulant

of xk and yk . And the coefficients are defined as :

g j*(N)(0) ( )Ngj

*() |0, (9)

gj*() (

1)j exp 0m

*1

m / m!m3

im*

i1

m / i!m!m0

i1

|10 . (10)

Finally, by combining it with the unified predictionalgorithm: xk1N | Yk Fk

N(xk ,uk ) | Yk , thedynamical state estimation algorithm in a recursive formcan be realized.

EXPERIMENTAL CONFIRMATIONS The effectiveness of proposed estimation theory hasbeen experimentally confirmed by applying it to the actualacoustic data in the outdoor field under the wind-inducednoises (omitted owing to page-limit).

CONCLUSIONS In this paper, for the acoustic system in an outdoor

environment, a new derivation of the systematiccountermeasure method for the wind noise was proposedwith use of the wind velocities.

ACKNOWLEDGMENTSWe will thank to Mr.M. Kubo, Mr.K.Furukawa andMr.H.Kondo.

REFERENCES

1) K.Hatakeyama & M.Ohta, Research Report on Random Phenomenaand Stochastic Process(1995.11) . 2) T.Katayama:Kalman Filter andApplications (Asakura, 1983). 3) Anderson,B.D.O. &Moore,J.B.:Optimal Filtering ( Prentice-Hall,1979 ).

Low Frequency Noise in Automotive Interiorsdue to Flow through Window Openings

P. Lenzuni a,b, R. Sisto b, A. Pieroni b

a Italian National Institute for Occupational Prevention and Safety (ISPESL), Department of Florence, 50121Florence, Italy. E-mail paolo_lenzuni@hotmail.com

b Italian National Institute for Occupational Prevention and Safety (ISPESL), Department of OccupationalHygiene, 00040 Monteporzio Catone (Rome), Italy. E-mail familyname@ispesl.castelliromani.it

We investigate how the acoustic field created by air inflowing through the side windows of automotives of various sizesdepends on the vehicle speed. Extremely large sound pressure levels are found near the frequency of the fundamental mode of thevehicle’s Helmholtz resonance (about 20 Hz). Much smaller peaks are also found at the frequencies corresponding to higherharmonics. A positive, tight correlation exists between the peak frequency and the vehicle speed. The peak sound pressure levelsincreases with the vehicle speed, up to a critical speed, and declines afterwards. In absolute terms, much larger sound pressurelevels were found inside a sub-compact car, as compared to a four-door sedan and to a station wagon.

INTRODUCTION

Interior noise in the low frequency region is usuallyof minor concern inside contemporary cars. However,the presence of open sunroofs and/or side windows caneasily result in very high amplitude, low frequencynoise, which strongly interferes with the driver’sconcentration and wakefulness, causing considerablediscomfort and even possibly impairing his ability toprovide safe driving [1]. This contribution reports the first results of anongoing investigation aimed at characterizing the lowfrequency sound pressure field in vehicle interiors, as afunction of vehicle type and speed, as well as of theopening sizes, number and locations. Here we focus onthe relation between the peak sound pressure level andfrequency and the vehicle speed.

METHODS

Three vehicles were tested, including one sub-compact city car (SC), one sedan (S), and one full sizestation wagon (SW). Measures were taken positioninga microphone with extended sensitivity to lowfrequency waves (Brüel & Kjær type 4193) at adistance of about 10 cm from the right ear of thedriver. Experiments were carried out between Marchand December 2000, with different vehicle speeds,ranging from 80 to 140 Km/h, and a fully open rearright window. Typical test durations were set at around90 seconds. Narrow band FFT spectra were calculated withbandwidths 0 – 100 or 0 – 200 Hz and resolutions 0.5Hz. One third octave band spectra were synthesizedfrom narrow band spectra. Data analysis was

performed using the software Kyplot (version 2.0) tocalculate best fitting models.

RESULTS

Spectral shape

Figure 1 shows the FFT spectrum measured insidethe SC vehicle at 100 Km/h with fully open right rearwindow. Despite its specific character, this spectrumcan be taken as generally representative of frequencyspectra collected in this study, since they all share thesame basic features.

FIGURE 1. Frequency spectrum of vehicle SC

After an initial area of low signal at very lowfrequencies (up to 8 - 12 Hz), the sound pressure levelshows a very fast rise, culminating with a sharp peak atfrequencies f1 in the range 16 – 22 Hz depending onthe various parameters affecting the system. This isfollowed by an equally sharp decline. Beyond 30 Hz, amore shallow downward trend sets in, which extends toabout 80 – 100 Hz. Superposed on this “noise floor”are the various peaks due to the higher harmonics, at

60

80

100

120

140

0 20 40 60 80 100Frequency [Hz]

SPL

[dB]

frequencies nf1. Their amplitudes are much smaller(at least 25 – 30 dB) than that of the first peak. The exciting mechanism is provided by theaerodynamical noise due to flow instabilities in theboundary layer between the vehicle interior and theouter flow [2]. These instabilities results in vortexshedding and the convection of discrete vortexes overthe opening determines the acoustic excitation of thecavity. Acoustic emission is triggered by theinteraction of the vortexes with the downstream edgeof the opening, and shows a characteristic frequency

where NS is the Strouhal number, v is the vehiclespeed and d is the opening streamwise length. Theactual emission takes place at a frequency dictated bythe acoustical response of the vehicle interior. Thelatter is very well approximated by a Helmholtzresonator [3]. Under the circumstances found invehicles, the fundamental proper mode has a frequency

where cs is the sound speed, A is the window openingarea, and V is the interior volume. For the typicalvalues assumed by these parameters in tested vehicles,fH is of order 20 Hz.

Peak frequency and velocity

Table 1 reports the peak frequencies and the peaknarrow band sound pressure levels in vehicles SC andSW. The very large difference between the soundlevels is likely due to a combination of a reducedinflow and a broader emission spectrum is vehicle SW.The non-monotonic trend of the SPL with velocity ispossibly related to the matching of the instabilitycharacteristic wavelength and the opening length.

velocity Vehicle SC Vehicle SWf-Peak SPL f-Peak SPL

80 16 117.4 16.5 110.390 17.5 126.3 17 113.8

100 18 126.9 17.5 111.3110 19 126.8 18 109.2120 19.5 127.5 18.5 106.2130 20 128.2 19.5 105.5

Table 1. Peak sound pressure levels and peak frequencies

Figure 2 shows the peak frequency as a function ofvelocity for all three tested vehicles. A very clean,systematic rising trend can be observed as a function of

vehicle speed, in agreement with the results of aprevious study [4].

Figure 2. Peak frequencies at various speeds

This trend is very robust, as it appears to be presentin all vehicles of different shapes and sizes. A powerlaw f v has been found to provide good fitting todata. Results indicate similar trends ( 0.5) for theSC and the S vehicles. A matched pair t-test showshowever that SPL values in the S vehicles aresignificantly larger (t = 6.54, P < 0.05) than in the SCvehicle. A more shallow slope characterizes the SWvehicle. The same behavior is replicated by the higherharmonics, although they prove at times hard to locatedue to unfavourable signal to noise ratios. Finally, thespectrum becomes more and more spread out asvelocity increases, with lower resonance peaks andhigher high frequency "noise floors", pointing to arising trend of damping in the system.

ACKNOWLEDGEMENTS

The assistance of Dr. Renato Gurin during themeasurements is gratefully acknowledged.

REFERENCES

1. Landström U., “Human exposure to infrasound”,in the Encyclopedia of Environmental controltechnology – Vol. 7 – Chapt. 20, edited by P.N.Cheresiminoff, 1987.

2. Pierce A.D., Acoustics – An introduction to itsphysical principles and applications, AcousticalSociety of America, New York, 1989.

3. Mongeau L., Brown D., Kook H., Zorea S., “Apredictive model for the interior pressureoscillations from flow over vehicle openings”, inProceedings of the 1997 SAE Noise and Vibrationconference.

4. Backteman O., Köhler J., Sjöberg L., Journal ofLow Frequency Noise and Vibration 3, 28 – 66(1984).

dvNf S

A

2/12/1

2/1

22

VAcf S

H

= 0,51 = 0,49

= 0,33

1,20

1,22

1,24

1,26

1,28

1,30

1,32

1,34

1,3 1,35 1,4 1,45 1,5 1,55 1,6

Log (velocity [m/s])

Log

(fre

quen

cy [H

z]) SC

SSW

Prediction of airplane interior noise due to flow over small stepsPart 2: Non-resonant sound transmission

B.M. Efimtsova, A.Ya. Zvereva

A.O. Anderssonb, S.V. KravchenkobaCentral Aerohydrodynamics Institute, Moscow, Russia

bThe Boeing Company, PO Box 3707, Seattle WA 98124, USA

Flow over small steps on the exterior of an airplane causes aerodynamic pressure fluctuations that are up to 30 dB greater thanthose under a turbulent boundary layer on a surface without pressure gradients. Although of small extent, regions with suchexcitation may contribute substantially to the interior noise in airplanes, particularly if well-radiating structures like windows areexcited. The aim of this investigation is the prediction of interior noise of airplanes resulting from flow over such steps, commonon airplane exteriors. This requires derivation of the relations governing the vibration and acoustic radiation of thin-walledstructures in the case of excitation by such pressure-fluctuation fields. Existing interior-noise prediction procedures in theframework of statistical-energy analysis are modified to reflect the peculiarities of these fields, which differ from those under aturbulent boundary layer in terms of correlation scales and non-uniformity scale. This paper is limited to the transmission due tonon-resonant excitation of the structure. A paper delivered at Inter-noise 2001, The Hague, Netherlands describes the somewhatgreater resonant contribution to the radiated field. This work is the result of cooperation between TsAGI, Moscow and Boeing,Seattle.

INTRODUCTIONRelations are found for prediction of structural-

acoustic radiation and airplane interior noise due toexcitation by pressure fluctuations in flow over steps.In reference [1] the relations associated with theresonant behavior of the structure were obtained. Theaim of the present investigation is to obtain therelations associated with non-resonant behavior of thestructure and are derived by studying a model task.The results were obtained in the course of scientificresearch contracts between the Boeing Company(USA) and the Central Aerohydrodynamics Institute(Russia).

TASK STATEMENT AND SOLUTIONMETHOD

We examine a thin infinite panel, with mass but nostiffness, separating two half spaces ( 1Ω and 2Ω ).We use an orthogonal system of coordinates

21 x,xx =ρ

in the plane of the plate, completed by acoordinate 3x positive in half-space 2Ω . The non-uniform pressure-fluctuation field excites the platefrom half-space 1Ω . The perturbation of the air in 1Ωand 2Ω is assumed small and non-vortical. The speedof sound ( 0c ) is taken to be equal in 1Ω and 2Ω butthe air density in the general case different ( 1ρ and

2ρ , respectively).The task is prediction of the spectral density of the

sound pressure in half-space 2Ω . This task is solvedby techniques analogous to those used for acoustic

radiation from a plate excited by a randompressure-fluctuation field [2].

From the multiplicative presentation of the space-correlation spectra [1] we will obtain relations for thefrequency/wavenumber spectra of the non-uniformpressure-fluctuation field:

( ) ( ) ( ) ( )ωϕω′ϕωΦ=ω′Φ ,k,k,k,k,k,k 221110211q (1)

( )( )

( ) ( )( )( ) ( )[ ]( ) ( )[ ]

( )( ) ( )( )( ) ( )[ ]( ) ( )[ ]

( ) ( )( )( ) ( )[ ]( ) ( )[ ]

( ) ( )( )( ) ( )[ ]( ) ( )[ ]

′++++++

′++−++

+′++−−′+

′+′−+−−

−′++−+++

′+−−−++

+

′+++−′+

′+′−++

π=ω′ϕ

21q

221q

21q1q

2

21q

2211

21q11

21q

221q

21q1q

21q

2211

21q11

2111

kkcakkca

kkkkca

kkcakka2

kkkkcaa2kkcakkca

kkkkcaca

kkcakka2

kkkkcaa2

22,k,k

( ) ( )22

22

222

k1,k

Λ+π

Λ=ωϕ L21a =

11c Λ= phq U/ω=κ

Here ( )ωΦ0 is the spectral density of the pressure-fluctuation field in the region of maximal intensity,L the scale of non uniformity of the pressure-fluctuation field, 1Λ the longitudinal correlation scale,

2Λ the transversal correlation scale, phU the phasevelocity of the pressure-fluctuation spectralcomponents, f2π=ω , and f the frequency in Hz.

In this case the following relation for prediction ofthe sound pressure spectral density at ( 1x , 3x ) in

2Ω is obtained:

( ) ( ) 211

122

21

322

21110

11 1

22

21

322

21110

211q30

231p ddd

i1

x1xikexp

i1

x1xikexp,,,k

4,x,x

22

21

22

21

2λλ′λ

τ+λ−λ′−

λ−λ′−+′λ′−×

τ−λ−λ−

λ−λ−+λ+ωλλ′λΦτ=ωΦ ∫∫∫

≤λ+λ′≤λ+λ

(2)

Here011 kk=λ ; 011 kk ′=λ′ ; 022 kk=λ ;

00 ck ω= ; ( )[ ]22101 mc ωρ+ρ=τ ; 21

12 ρ=ρτ=τ

The three-dimensional integral can be calculatednumerically. However, when the distance betweenobservation point and step (R) is large in comparisonwith the sound wavelength ( 1Rk0 >> ), it is possible togive an asymptotic analytical relation:

( ) ( ) ( )( ) ( )

×τ+Λ+

ΛωϕωΦ

πτ=ωΦ

=λ′=λ1

220

220

0102

p1k1

kR2 112

( )

τ+τ−Λ+×

1

1220 1

k1 (3)

When the component ( )ω′ϕ ,k,k 111 of a frequency-wavenumber spectrum in (2) varies slowly in the rangeof integration (which is true in almost all cases forfields at small steps), it is possible to give anotheranalytical relation for small distances betweenobservation point and step ( 1Rk0 << ). This relation isnot presented here due to space limitations.

The increase in sound pressure level related to theinertial behaviour of the structure, when excited byinteraction of the turbulent boundary layer with a step,is predicted by the following relation:

( ) ( ) ( ) ( )[ ]ωΦωΦ+⋅=∆ BLP

stP 22

1lg10L (4)

Here ( ) ( )ωΦ stP2

is the sound pressure spectral density

predicted according to (2) for the case of flow oversmall steps and ( )( )ωΦ BL

P2 is the same quantity for the

case of non-gradient turbulent boundary-layer flow ona smooth surface. This last quantity is predictedaccording to the well known relation (see [2]):

( )( ) ( ) ( )αωΦ

τ+

πτ

=ωΦ F11ln4

)bl(q

1

2BLP2

(5)

Here ( )ωΦ )bl(q is the spectral density of turbulent

boundary layer pressure fluctuations [3] and ( )αF is afunction of dimensionless parameters [2].

As an example, function L∆ presented in Fig. 1 waspredicted for the case of excitation of a flat flight-deckwindow pane at different values of the step height andat an observation-point distance of m1.0R = . Theinitial parameters of the calculation were the same as incase of resonant energy transmission [1].

Comparison of figure 1 [1] with figure 1 of thispaper shows that the increase in sound pressure levelsinside the airplane, due to the inertial behaviour of thestructure, are smaller than those due to its resonantbehaviour.

0

5

10

10 100 1000 10000Frequency, Hz

dB

h mm5

3

1

L∆

FIGURE 1. Increase of interior noise due to flow over astep considering only the inertial behaviour of the structure

CONCLUSIONS1. An analytical expression is obtained describing

the influence on airplane interior noise, due to theinertial behaviour of the structure, of flow over a stepof small height (in comparison with the thickness ofthe boundary layer). This expression is presented in theform of a three-dimensional integral which can beevaluated numerically.

2. Asymptotic relations are obtained from theanalytical expression for the cases when the distancefrom the observation point to the step is much greateror much smaller than the sound wavelength.

3. The predictions indicate that interior noise dueto inertial response of the structure to flow over stepsin the external surface is smaller than the noise due toresonant response.

REFERENCES1. Efimtsov B.M., Zverev A.Ya., Andersson A.O., Kravchenko

S.V. “Prediction of airplane interior noise due to flow over smallstep - Part 1. Resonant sound transmission”, paper 141, inter-noise 2001, The Hague, Netherlands, 2001.

2. Efimtsov B.M. ‘’Influence of the correlation space scales ofrandom pressure fluctuations on acoustic radiation from aplate’’, Sov.Phys.Acoust. 26(4),1980, pp.307-312.

3. Efimtsov B.M. “Similarity criteria of wall pressure fluctuationspectra of turbulent boundary layer”. Sov.Phys.Acoust.30(1),1984.

Determination of Source Impedance of Fluid Machines byLumped Element Model

J. Lavrentjev

Department of Machinery, Tallinn Technical University, Ehitajate tee 5, 19086, Tallinn Estonia,lavren@edu.ttu.ee

Fluid machines (internal combustion engines, pumps, fans) with inlet and outlet systems form complicated acousticalsystems. To understand sound generation and to design properly the pipe or duct system for the fluid machine, it isnecessary to know the source impedance of the machines. In most cases there is no general theory to calculate it andthe impedance has to be determined experimentally. However, this is time consuming and costly and mathematicalmodels for the source impedance would be a help. In the paper as an example, a complete lumped-impedance modelfor a centrifugal fan is derived and validated by experiments. The model can give satisfactory results at lowfrequencies, where the wave propagation effects can be neglected. However most of the acoustical energy of fluidmachines is generated at low frequency region. In order to verify the derived lumped-impedance model, measurementswere carried out on centrifugal fans, By comparison with measured curves the predicted source impedance were woundto agree well up to approximately 400 Hz.

INRODUCTION

Fluid machines such as pumps, fans etc., cangenerate high sound levels in connected pipe and ductsystems. The sound power generated by the sourcedepends both on the impedance of the duct system andthe source. The impedance of the duct system canusually be easily calculated but very few investigationscan be found concerning the calculation of theimpedance of the ducted fans. Yeow [1,2] used alumped impedance-element approach to study theinfluence of the fan’s structural parts impedances onthe radiated sound power. Lavrentjev [3] used similarmodel and scaling laws to determine the 1-port sourcedata for fans. The aim of this paper is to develop acomplete source impedance model for centrifugal fans.

MATHEMATICAL MODEL

The lumped model of the fan consists of fourdifferent basic types of elements. In the first type ofelement the kinetic energy of the gas dominates overthe potential energy (due to compression) and theelement is characterised by the inertia of the gas (masstype element, ZL). In the second type of element thepotential energy of the gas dominates over the kineticenergy of the flow (spring type element, ZC). The thirdtype of element is the resistive element, which causes a

pressure drop when a gas flow passes (resistiveelement, ZR). The fourth element needed is atermination element, which can describe the radiationfrom the fan inlet opening (Zterm). Taking into account the elements described, a typicalcentrifugal fan (Figure 1) can be split into lumped-elements as illustrated in Figure 2 and 3.

3

4

2

1

FIGURE 1. Geometry of the tested centrifugal fans. 1-outlet neck, 2 - volume of the casing; 3 - channels formed bythe blades of the impeller; 4 - inlet necks.

The inlet and outlet necks and the channels formedby the blade rows of the impeller can be modelledeither as mass type element (Figure 2) or spring typeelement (Figure 3).

The choice depends on the relative lengths of the bladerows.

Zterm

ZL1 ZR1 ZL3 ZR3 ZL4

ZC2

FIGURE 2. Lumped-element model for the centrifugal fanin Figure 1, analogue circuit representation. Rotor bladesmodeled as mass elements.

Zterm

ZL1 ZR1 ZR31 ZR32 ZL4

ZC2 ZC3

FIGURE 3. Lumped-element model for the centrifugal fan.Rotor blades modeled as spring elements.

Since blade rows can also be expected to introducecertain losses, they will include also resistive elements.Quantities applied in the elements of models andcharacterizing fans are purely geometrical (volumesand lengths) except resistive elements which have tobe determined by theoretical experiments.

EXPERIMENTS

Models described here were used successfully fordifferent fans. As an example, a model, presented inFigure 3 was applied for the centrifugal fan (13backward curved blades, n=2800 1/min). Sourceimpedance was determined experimentally for twoflow rate cases (100% and 20%) by using 2-microphone method and 1-port model approach. Thecomparison between model and measurement results isshown in Figure 4 and Figure 5. As one can see thediscrepancy between predicted and measured curvesare small.

CONCLUSIONS

By using the lumped impedance model the centrifugalfans impedance can be determined. For different typesof fans different models of impeller may be applied.

50 100 150 200 250 300 350 4000

0.5

1

1.5

2

2.5

3 x 104

Zs(r

eal)

Frequency(Hz)

FIGURE 4. The real part of the source impedance, _____

measured, flow 100%; _ _ _ measured, flow 20%; _ . _ . _

calculated, flow 100%; _ . . _ . . _ calculated, flow 20%.

50 100 150 200 250 300 350 400-1

-0.5

0

0.5

1

1.5

2

2.5 x 104

Zs(im

ag)

Frequency(Hz)

FIGURE 5. The imaginary part of the source impedance,_____ measured, flow 100%; _ _ _ measured, flow 20%; _ . _ . _

calculated, flow 100%; _ . . _ . . _ calculated, flow 20%..

REFERENCES

1. K. W. Yeow, Journal of Sound and Vibration, (1974),32(1), pp. 143-152.

2. K. W. Yeow, Journal of Sound and Vibration, (1974),32(2), pp. 203-226.

3 J. Lavrentjev, INTER-NOISE 98, CD-ROM.

On reduction of acoustic disturbances in supersonic separatedflow by means of passive flow control

R. Khakimova, A. D. Khon’kina,b

aCentral Aerohydrodynamic Institute (TsAGI), Zhukovsky street, 1, Zhukovsky, 140180, RussiabMoscow Institute of Physics and Technology, Gagarin street, 16, Zhukovsky, 140180, Russia

Effect of passive control on acoustic disturbances caused by flow unsteadiness in a shock wave/turbulent boundary layer interactionregion at supersonic speeds was studied experimentally. The shock wave was generated by blunted fin mounted on a wind tunnelsidewall. The cavity covered by perforated cover and located in the interaction region was used as a passive control device. Testswere conducted at Mach numbers from 2 to 4 and unit Reynolds numbers from 9 6 106 to 51 2 106 . Spectra and integral levels ofwall pressure fluctuations were measured during experiments as well as static pressure fields were measured. It was demonstrated thatpassive flow control can sufficiently decrease acoustic disturbances caused by flow unsteadiness in the interaction region.

INTRODUCTION

The shock wave/boundary layer interactions occurs insupersonic flows around such elements of modern air-crafts as inlets, control surfaces, pylons etc. Flow in in-teraction region has a complex structure and can suffi-ciently effect on aircraft aerodynamics [1]. In additionto these, separation is a cause of flow unsteadiness, andits presence leads to a high level of aeroacoustic load-s on a vehicle elements [2]. Present paper concerns theflow around control surface-fuselage junction at super-sonic speeds. The main object was to determine the effec-tiveness of passive control at Mach numbers M∞

2 4for reduction of flow unsteadiness in interaction region.Flow was studied experimentally due to its complexity.The blunted fin was used as a model [3]. The shock gen-erated by fin interacts with a turbulent boundary layer ona flat plate (wing tunnel sidewall). Due to sufficient pres-sure overfall in a shock wave, the flow decelerates andmain separation zone occurs. Inside an interaction zonethe horseshoe vortex, developing downstream forms.

EXPERIMENT

Experimental facility and model

Experiments were carried out in supersonic low-turbulent blowdown wind tunnel with cross section0 21 0 20 m2.

Wind tunnel equipped with removable nozzles forfreestream Mach number variation. Unit Reynolds num-bers are varying by means of plenum chamber pressurevariation. Tests were conducted at Mach numbers M∞

2 3 4 and plenum chamber pressure from 2 105 Pato 2 105 Pa, while unit Reynolds numbers were from9 7 106 m 1 to 51 2 106 m 1. Dimensionless D

δ pa-

rameter (where D is a bluntness diameter of leading edge,and δ is a boundary layer thickness) was changed from0.287 to 0.417.

Model (Figure 1) was consisted of shock wave genera-tor and passive control device. As a shock wave generatorthe rectangular blunted fin was used. The fin was 120 mmlong, 10 mm wide and 10 mm thick. The bluntness diam-eter was 10 mm. Fin was mounted on a remote controlled

120 mm

10 mm

70 m

m

4.5 mm

Piezoresistive Pressure Transducer

Cover

155 mm Remote Controlled Drive

U

Wind Tunnel Sidewall

FIGURE 1. Scheme of a model.

drive, which allowed to move the fin parallel to wing tun-nel axis of symmetry with a minimal step of 5 mm bymeans of micrometer screw. Passive control device con-sisted of plug with cavity mounted in a port on a windtunnel sidewall. Cavity diameter was 155 mm, and cavitydepth was 114 mm. Cavity can be covered flush by solidor perforated cover. The thickness of cover was 4.5 mm.When model without control device was studied, the cav-ity was covered by solid cover. In case of passive controlthe perforated cover was used. Perforated cover have had4256 holes 0.8 millimeters in diameter. The perforationratio was 6 25%.

The pressure coefficient was calculated by static pres-sure measurements in a single orifice, relative to whichthe fin was shifted. The plug was turned in such a waythat measuring orifice was shifted relative to the axis ofsymmetry of the model by 9.89 and 14.0 mm. Due to this,the measurements of Cp in three sections of Y constplane were performed. Piezoresistive pressure transduc-er Endevco type 8514-10 with case diameter of 1.65 m-

m was used for studying the effect of passive control onwall pressure fluctuations. Transducer was mounted in aminiature plug placed in bypass cavity. Pressure fluctua-tions spectra were measured at Y 0 section with a stepof 5 mm. These was measured directly during the experi-ment in third-octave frequency bands by spectra analyzer.Analyzer was consisted of type 2607 measuring amplifi-er, type 1614 filter and type 2305 plotter by "Brüel &Kjær". In present experiments 100 Hz - 25 kHz frequen-cy band was studied. Using measured spectra the integrallevel of pressure fluctuations LΣ was determined, whichcharacterized the magnitude of acoustic disturbances.

RESULTS

Spectral characteristics of the staticpressure fluctuations

The flow in separation region is unsteady. It meansthat place and structure of shock waves system, in whichthe shock decomposes near the wall, is time-dependent.This unsteadiness produces acoustic disturbances. Pres-sure fluctuation spectra, measured along the model sym-metry axis in section Y 0 indicates the strong flow al-ternation in interaction region, while spectra measured inundisturbed flow are characterized by low amplitude inwhole frequency range. Example of the wall pressurefluctuations spectra measured in 100 Hz - 25 kHz fre-quency band in separation zone (x

D 2 5) are pre-

sented on Figure 2. The pressure fluctuations spectra are

120

125

130

135

140

145

150

155

160

5000 10000 15000 20000 25000

A, D

b

f, Hz

x/D=-2.5

solid surfaceperforated surface

FIGURE 2. Example of spectra of wall-pressure fluctuationsentire separation zone, M∞

2.

characterized by wide-band power spectra with high en-ergy concentration in low-frequency region, conditionedby unsteady shock wave displacement. Air bypass em-ployment leads to reduction of magnitude of disturbancesin interaction region. It should be noted that for plot p-resented, the maximal reduction of pulsation amplitudehappens in low frequency domain. This leads to redis-tribution of wall pressure fluctuations amplitudes by fre-quencies. Mentioned above indicates that unsteadinesscaused by shock displacements becomes lower.

Integral level of pressure fluctuations

Plot of integral level of pressure fluctuations on thewall surface in section Y const for solid and perforatedcover is presented on Figure 3. Integral level of wall pres-sure fluctuations is characterize the magnitude of acousticdisturbances. It should be pointed out that level of acous-tic disturbances caused by shock wave/boundary layer in-teractions is extremely high and achieves 180 decibels atMach 2.

140

145

150

155

160

165

170

175

180

0 -1 -2 -3 -4 -5

L ,

Db

x/D

Σ

solid surfaceperforated surface

FIGURE 3. Example of an integral level of wall-pressure fluc-tuations, M∞

2.

CONCLUSIONS

Experimentally studied interaction of shock wave,generated by blunted fin, with turbulent boundary layerat supersonic speeds.

For reduction of separation intensity the passive flowcontrol was used. Passive control device was consistedof cavity covered by perforated surface with perforationratio was 6 25%.

It was demonstrated that passive control allows to de-crease the length of separation zone, to relieve the pres-sure on interaction region, and to lower acoustic distur-bances in separation zone.

Maximal reduction of integral pressure fluctuationslevel was up to 5.7 decibels (at Mach 3).

Effectiveness of flow control device was demonstratedat all Mach and Reynolds numbers traced.

REFERENCES

1. Hemsh M.J., Nielsen J.N. (1986) Tactical missile aerody-namics. AIAA, New York

2. Delery J. (1999) Flow Physics Involved in Shock Wave-Boundary Layer Interaction Control. FLOWCON Book ofProceedings, KLUWER, Netherlands

3. Khakimov R.A., Khon’kin A.D. (1998) On effective pas-sive control for 3D shock wave/boundary layer interactions.FLOWCON Book of Abstracts, Goettingen, Germany