Experimental study of realistic low–noise technologies applied to a full–scale nose landing gear

(1)

Experimental study of realistic low–noise technologies applied to a full–scale nose landing

gear

Merino-Martínez, Roberto; Kennedy, John; Bennett, Gareth J.

DOI

10.1016/j.ast.2021.106705

Publication date

2021

Document Version

Final published version

Published in

Aerospace Science and Technology

Citation (APA)

Merino-Martínez, R., Kennedy, J., & Bennett, G. J. (2021). Experimental study of realistic low–noise

technologies applied to a full–scale nose landing gear. Aerospace Science and Technology, 113, [106705].

https://doi.org/10.1016/j.ast.2021.106705

Important note

To cite this publication, please use the final published version (if applicable).

Please check the document version above.

Copyright

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons. Takedown policy

Please contact us and provide details if you believe this document breaches copyrights. We will remove access to the work immediately and investigate your claim.

This work is downloaded from Delft University of Technology.

(2)

Roberto Merino-Martínez

a

,

∗

,

John Kennedy

b

,

Gareth

J. Bennett

b

a_Faculty_of_Aerospace_Engineering,_Delft_University_of_Technology,_Kluyverweg_1,₂₆₂₉_HS_Delft,_the_Netherlands b_Department_of_Mechanical_and_{Manufacturing}_Engineering,_Trinity_College_Dublin,_Ireland

a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Articlehistory:

Received22September2020

Receivedinrevisedform26March2021 Accepted28March2021

Availableonline31March2021 CommunicatedbyGursulIsmet

Keywords:

Landinggearnoise Lownoisetechnologies Acousticimaging Wind–tunnelmeasurements Aeroacoustics

Thelanding gear systemis thedominant airframe noise sourcefor mostmodern commercialaircraft

duringapproach.ThismanuscriptreportstheresultsfromtheALLEGRA(Advanced LowNoiseLanding

(Main and Nose)Gear for RegionalAircraft)project. Thisproject assessedthe performanceof several

highlyrealisticlow–noise technologies(LNTs)appliedto adetailed full–scalenose landinggear (NLG)

model in aeroacoustic wind–tunnel experiments. Four individual low–noise concepts tested, namely

a ramp door spoiler, a solid wheel axle fairing, wheel hub caps, and multiple perforated fairings.

CombinationsandsmallvariationsofsomeoftheseLNTswerealsoevaluated.Theuseofmultipleplanar

microphonearraysallowedfortheapplicationof2Dand3Dacousticimagingalgorithmstoassessthe

locationand strengthofthe noise sourceswithinthe NLGsystemindifferentemission directionsfor

eachconﬁguration.Thewheelaxle,theinnerwheelhubs,thesteeringpinionsandthetorquelinkwere

identiﬁedasthenoisiest NLGelements.Thesolidwheel axlefairingwas themosteffectiveindividual

LNT, and it improvedits performance whenapplied in combinationwith the rampdoor spoiler and

wheelhubcaps,reachingoverallnoisereductionsofmorethan4 dBA.

license(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Noise emissionsgenerated by aircraftcause severe annoyance to tensofmillionsofpeople livinginthevicinity ofairportsand pose environmental constraints for airport operations, with the consequenceofrevenueloss.Thesuccessfuldevelopmentoflow– noisepropulsiontechnologieshasincreasedthesigniﬁcanceof air-framenoiseinmodern commercialaircraft[1].Hence,inorderto fulﬁl the ambitious aircraft noise reduction requirements set by governmental organisations, such asACARE[2] (Advisory Council forAviationResearchandInnovationinEurope),andprojectssuch as Flight Path 2050 [3], airframe noise levels need to be further reduced, as they set the thresholdto aircraftnoise in the future [4,5].

Withrespecttoenvironmentalnoise,thelandinggear(LG) sys-temisoftenthedominantairframenoisesourceduring approach and landing [6–8] depending on the aircraft. A typical LG sys-temconsistsofcomplicatedstructuresofbluffbodies(struts,links, wheels,tires,fairings,etc.)ofconsiderablydifferentsizes. Because of its criticalimportance for theaircraft’s safety,this systemhas littleaerodynamic andaeroacousticreﬁnementtoeaseits

inspec-*

Correspondingauthor.

E-mailaddress:r.merinomartinez@tudelft.nl(R. Merino-Martínez).

tionandmaintenance,whichareessential tothereliabilityofthe LG [9]. Assuch, aerodynamic noise isproduced wheninteracting withthesurroundingairflow[9].Thebluffbodycomponentsofthe landinggearradiatenoisedirectly[10] andnoiseisalsogenerated bythesubsequentinteractionofturbulent wakeflowwith down-stream located gear elements [11]. In addition, research projects continuetoexaminetheseemingly everpresentproblemoftonal cavitynoise in landing gears of commercialaircraft [12–14], ini-tiallyidentifiedasanissueseveraldecadesago[10,15].

Dependingontheirsize,LGelementsarenormallydividedinto threecategories[16,17]:

1. Large–scalestructures suchasthewheelsandthewheelbay, whichcontributetothelow–frequencynoise.

2. Mid–scalestructures such as the main struts, which con-tributetothemid–frequencynoise.

3. Small–scalestructures suchasthehydrauliclines,wires,and LGdressings,whichcontributetothehigh–frequencynoise. Research to date hasattempted to measure the contributions ofsome ofthese componentsto forma basis forsemi–empirical airframenoise prediction models[6,16–19] and the designof LG low–noise technologies(LNTs)[20].Because oftheir nature, such semi–empirical prediction models have limited reliability when https://doi.org/10.1016/j.ast.2021.106705

(3)

estimating the performance of LNTs or unconventional LG de-signs [5]. Therefore, dedicated experimental research is required for evaluating and validating the potential noise reductions that theseLNTscanprovide.Thefindingsfromsuchstudieswould pro-videvaluableinformationonthenoisegenerationmechanismsand low–noise design criteriafor futureandimproved LNTs. Previous studies[21,22] showedthatthehigh–frequencynoiseemissionsof LGsareasignificantcontributortotheoverallsoundpressurelevel (Lp,overall) and,therefore,inordertoproperlyrepresentthem,itis essentialtoemployhighlydetailedandideallyfullscalemodelsin wind–tunnelexperimentsandincomputationalsimulations[5,20]. The current study employs the full–scale nose landing gear (NLG) model from the ALLEGRA (Advanced Low Noise Landing (MainandNose) Gear forRegional Aircraft)project[23–26]. This included a full representation of the NLG details and associated structures(e.g.,wheelbaycavity,baydoors,fairings,nosefuselage andhydraulic dressings).Oneoftheadditional benefitsof manu-facturingtheaccurate nosefuselage inthe testcampaign,aswell asto house the wheelbay, was to providethe correct boundary layer and aerodynamic flow field, resulting from the associated curvatures,to impingeupon theNLG.Thisunusuallevel ofdetail was intended tobring thewind–tunnel testsclosertothe condi-tionsexperiencedinrealflight.

SeveralLNTsfordecreasingthesoundlevelsemittedbytheLG systemhavebeeninvestigatedinthelastdecades.Ingeneral,LNTs fortheLGsystemaimatreducingtheincomingflowvelocity, sup-presscavityresonance,shieldsmallcomponents,orprevent wake interactions[9].Theycanbegroupedintofourcategories depend-ing on how they function: component enhancement, component smoothing, flow enhancement, and flow deflection. Whereas the designoptimizationofcertainLG components(suchasthewheel spacing, boogie angle, leg–door structure, and brakes) has been evaluated in projects such as SILENCER [27] and TIMPAN (Tech-nologies toImproveAirframeNoise)[28],theuseofnoise abate-ment treatments as passive add–ons (such as fairings, spoilers, caps, and optimized components) is considered as the simplest retrofit technology approach to noise reduction for existing gear designs[29].Considerableworkhasbeendevotedinthelastyears to the study of active noise reduction devices as well, such as air curtains[30,31] orplasma actuators [9],but their technology readinesslevel(TRL)isrelativelylow todate.Agoodoverviewof mostLNTsdevelopedtodateispresentedin[9].

The objective of the present research is to assess the perfor-manceoffourrealistic LNTsapplied, individually,incertain com-binations, and with variations, to a NLG model in aeroacoustic wind–tunnel measurements: a ramp door spoiler, a wheel axle windshield,wheelhubcaps,andaperforatedfairing.These tech-nologies were selected based on criteria, such as their potential noisereduction,weight,cost,easeofimplementation,TRL,and op-erationalandmaintenance impact.TheseLNTshavea mediumto highTRLandaresuitableforﬂighttestingandcommercial imple-mentationintheneartermintheearly2020s.

Phasedmicrophonearraysandacousticimagingtechniquesare normallyemployedforestimatingthelocationandstrengthofthe soundsources[32–34] andto isolatetheircontributions. This ap-proach has been previously used in studying landing gear noise in flyover measurements [35–38] and wind–tunnel experiments [25,26,39,40]. In the ALLEGRA wind–tunnel experimental cam-paign,severalmicrophonearrayswereusedsynchronouslyto mea-suretheNLGnoiseemissions,whichtheauthorshaveidentifiedas an opportunity for 3D acoustic sourcemapping [41]. Henceforth, thisapproachisemployed forstudyingthethree–dimensional lo-cation and strength of the noise sources within the NLG model, andhowtheLNTsinfluencethese.

The paperisstructured asfollows.Section2describesthe ex-perimentalfacilityutilised,aswellastheNLGmodelandtheLNTs

Fig. 1. SchemeshowingtheALLEGRANLGmodelplacedinsidethePininfarinawind tunnel,thecoordinatesystem,thefourmicrophonearrays(denotedasblack,blue, red,andmagentadots,respectively),thewind–tunnel’snozzleexitplane(incyan) andthelimitsofthescangridselected(ingreen).Theﬂowmovesinthepositivex

direction.(Forinterpretationofthecoloursintheﬁgure(s),thereaderisreferredto thewebversionofthisarticle.)

considered.Theacousticimagingmethodsemployedforthethree– dimensional source mapping and the quantiﬁcation of the noise levels arebrieﬂy explained insection 3. Theresults obtainedare discussedinsection4,whereasthemainconclusionsaregathered insection5.

2. Experimentalsetup 2.1. Wind–tunnelfacility

The experimental measurements were performed in the Pin-infarina open–jet, semi–cylindrical wind–tunnel facility in Turin, Italy, whichhasa test section of8 m(length)

×

9.60 m (width)

×

4.20 m (height). The semi–circular nozzle has a diameter of 5.64m[42].Thefacilityisequippedwithalow–noise,high–speed fan–drivesystemconsistingof13fans,whichprovidesflow veloc-itiesupto72.2 m/s.Forthisstudy,flowvelocities U_∞ of40 m/s, 50 m/s,and60 m/swereconsidered,providingamaximumMach numberofabout0.18.ThebackgroundnoisefortheALLEGRANLG testswasaconsideredasacombinationofthewind–tunnelnoise andthe noise produced by the fuselage itself with the bay cav-ity sealed. Both are mostly low–frequency noise sources (below 100 Hz), outsideofthefrequency rangeofinterest inthispaper. Thereaderisdirectedtoa separatestudy whichfocusesonlow– frequencywheel bay noise [13]. The addition of the NLG causes an increase in the noise levels between5 and 12 dBA across a verywidefrequencyrange[24].Theflowvelocityproducedbythe windtunnelisveryuniform,sinceitvariesbyonly0.5%overthe test area. The turbulence intensity level hada value of approxi-mately0.3%intheseexperiments.

Fig.1depictstherelativepositionoftheNLGmodelinthewind tunnel,aswell asthe coordinatesystememployed,where thexz planeisthesymmetry planeofthetestmodel,the yz plane cor-respondstothewind–tunnelnozzleexitandtheoriginissituated onthefloorofthetestingplatform(xy plane).TheNLGcomplete model,withthepartialfuselage,wasplacedinthewindtunnelso thatthedistancebetweenthewind–tunnel’snozzleexitplaneand theNLGwheelaxiswas2.8 m.Thewind–tunnel’sshearlayerwas measuredanditwasconfirmedthatitdidnotimpingeonthetest modelforanyoftheflowvelocitiesconsidered.Thecoordinatesof themiddlepointofthewheelaxiswere:

(4)

Fig. 2. LocationofthesidelineararrayandemissionangleswithrespecttothemiddlepointofthewheelaxisoftheNLGmodel.Theﬂowmovesinthepositivex direction.

x =2.8 m; y =0 m;z =2.175 m.

Henceforth,twodifferentanglesare employedfordeﬁning the directionofthenoiseemissionsoftheNLGmodelwithrespectto thecoordinatesofthemiddlepointofthewheelaxis:(1)thepolar emissionangle

θ

withrespecttotheexpectedﬂightdirection,with

θ

=

≈

90◦) the side array was used (see blue dots in Fig. 1). The array was positioned in the y

=

−

4

.

22 mplane, i.e.parallel tothe modelplane ofsymmetry andconsistedof66microphonesarrangedinahalf–wheel dis-tributionwithadiameterofapproximately3 m.

3. A spiralfront arrayconsistingof 15 microphones was placed upstreamtheNLGforminganangleof10◦ withthe yz plane (seereddotsinFig.1).

4. Lastly,asidelineararrayof13microphonesonthesameside andatthesamehorizontaldistancefromthemodelaxisasthe side array.Thesemicrophonesaredepictedasmagentadotsin Fig. 1.Fig. 2 illustrates the emission angles fromthe middle pointoftheNLGwheelaxlethatcanbemeasuredbytheside linear array,including a single azimuthal sideline angle

φ

of 81◦ andpolaremission angles (withrespect tothe expected ﬂight direction)

θ

coveringarangefromapproximately 6◦ to 171◦. The shear layer caused by the ﬂow motioninside the open–jetwindtunnelrequiresvelocity–dependentgeometrical corrections,thatlimit theeffectivemeasurablerangeof emis-sionangles[32].

The dataofall microphoneswas acquiredsimultaneouslyata samplingfrequency of32,768 Hz for10 sper measurement. The time–averagedcross–spectral matrices(CSMs) [43] wereobtained byusingfrequencyspectraprocessedwithablocklengthof8192

samples(0.25 s),Hanning windowing and50%data overlap, pro-viding a converged solution with a frequency resolution

f of 4 Hz.

Thefrequencyrangeofinterestselectedforpostprocessing ex-tendsfrom200 Hzto4000 Hz.Thelowerlimitwasdeﬁnedbythe backgroundnoiseandthe spatial resolutionofthearray inorder toproperlyseparatethesoundcomingfromthe NLGmodelfrom other noise sources. The higher frequency limit was imposed by theminimum distancebetween microphonesto prevent aliasing, theamountofsidelobes,andthesignal–to–noiseratio.

Foradditionaldetailsabouttheexperimental setup,thereader isreferredto[44].

2.2.Noselandinggear(NLG)model

Thefull–scaleNLGmodelusedintheALLEGRAprojectincluded afull representationof theNLG details andassociated structures (e.g.,baycavity,baydoors,nosefuselageandhydraulic dressings). ThisLG geometryis that of an advanced regional turboprop air-craftdesign andwasprovided tothe authorsby membersofthe Clean–Sky consortium [23]. The diameter and width of the NLG tires were0.577 m and0.221 m, respectively,andtherim diam-eterwas 0.286 m. The separation betweenwheels was 0.159 m. Themain struthadalength of1.295 mandanaverage diameter of0.11 m.Additionaldetailsonthegeometrycanbefoundinthe thesisofNeri[44].

Fig.3showsapictureofthemodelinsidethewindtunnel.The nosefuselagehadastreamwiselengthof6.697 mandamaximum widthinthey directionof3.417 m.Theheightofthemodel, com-biningthefuselageandNLGmodel,was2.463 m.

Thewind–tunnelmodelhadafixed,built–inangleofattackof 4◦.Eachmodelconfigurationwastestedforarangeofflowspeeds andyaw angles (

±

5◦ and

±

10◦), allowing the analysisof condi-tions equivalent to landing with crosswind [44]. For this paper, onlytheyawangleof0◦ (nocrosswind)wasconsidered.

Fig. 4 identiﬁes the principal components that can generate noiseonaschematicoftheALLEGRANLG.Thebaseline conﬁgura-tionisconsideredtobetheextendedlanding gearwithhydraulic dressingappliedandisreferredtoastestcaseNLG.

Henceforth, the NLG baseline configuration is referred to as NLG–BASE,see Table 1.The configuration withonly the fuselage (withthe bay cavity sealedandwithout the NLG present)is de-noted as NLF and it was used as a reference to calculate the wind–tunnel’sbackgroundnoiselevels.Anadditionalconfiguration withoutthehydraulicdressingswastestedandreferredtoasNLU. Acomponentnoise assessmentonthebaseline configurationwas performedbyBennettetal.[24] andreportedhigh–amplitudebay cavityresonancemodesatfrequenciesbelow100 Hz,especiallyin theflyover direction[13].

(5)

Fig. 3. ALLEGRANLGmodelinsidethewindtunnelwiththetoparrayvisible.The viewdirectionisfromupstreamtowardsdownstream[44].

Fig. 4. Principalcomponents ontheNLGmodelasseenfromupstream(left)and downstream(right).

2.3. Low–noisetechnologies(LNTs)

ThispaperreportsonfourdifferentLNTs,whichincluded mul-tiple separate approaches to noise reduction. These technologies are applied individually and in certain combinations and varia-tionstotheNLGmodelwithhydraulicdressing(i.e.theNLG–BASE conﬁguration).ThemaincharacteristicsoftheLNTsconsideredare summarisedinTable1.SomephotographsoftheNLGmodelwith thefourLNTsarepresentedinFig.5.

Abriefdescription ofeach oftheLNTs consideredis provided below:

1. The firstLNT considered was a retractableramp doorspoiler (denoted as NL1) placed upstream of the bay cavity, see Fig. 5a, which causesa deflectionof theflow onto theouter regions of the extended gear. Its main objective is to miti-gatethebaycavitynoisegenerationmechanismsandtoshield the steering pinion, drag stay, lower arm, upper strut, and baydoorsfromtheincomingflow. However,thepotential de-flected flowinteractionwiththefront edgeofthewheelbay doors may generate new noise sources, as well as the flow separationfromthetrailingedgeoftherampdoorspoiler[9]

Table 1

ALLEGRAlow–noisetechnologies.

Test ID Fuselage Landing gear Low–noise technology NLF Sealed None None

NLU Open Bay Undressed None NLG–BASE Open Bay Dressed None

NL1 Open Bay Dressed Ramp door spoiler NL2 Open Bay Dressed Solid wheel axle fairing NL3 Open Bay Dressed Wheel hub caps

NL4 Open Bay Dressed Multiple perforated fairings NL5 Open Bay Dressed NL2 + NL3 + NL4 NL6 Open Bay Dressed NL1 + NL2 + NL3

NL7 Open Bay Dressed NL4 with alternative material NL8 Open Bay Dressed NL4 only applied to the wheel axle

andtheinteractionofitswakewiththeleg.Theprimary bene-ﬁtsofthisLNTareexpectedtobeobservedforlowfrequencies (below300 Hz).

2. ThesecondLNTemployed(NL2)wasawheelaxlewindshield solid fairing,see Fig.5b,whose function ismainly to deflect flowandproduceamoreaerodynamicand,hence,quieter ge-ometry.ThisLNThasprobablythehighestTRLduetoseveral previousexperiments,eveninrealflightscenarios[9,20].This LNT is expected to effectively mitigate the relatively strong noise source between the wheels which has been identified inpreviousstudies [20,24,45].In addition,thisfairingdesign, was designed to allow for wheel deformation during touch-down.Duetothedesignoftheaircraft,therewerenobrakes intheNLG,whichallowsforthelocationofsuchalarge fair-ing. Such a solution would typically not be possible for an MLG, however, as its presence would compromise the con-vective cooling required for its brakes. With regards to the NLG,apotentialpenaltyforthisLNTisthegenerationoflow– frequencynoise (below300 Hz)duetotheflowdeflectionto adjacentanddownstreamuncoveredgearcomponents[9].In addition,therelativelylargefairingsizemayalsocausevortex shedding.

3. The NL3 treatment consisted of both inner and outer wheel hubcaps(seeFig.5ctocoverwheelhubvoidsand,hence, re-duce the noise caused by the interaction of the air andthe wheel[46,47].Theinnerhubcapsarepossibleduetothelack ofbrakes in the NLG design, butthis LNT is expected to be easilyinstalledandtohaveanegligibleimpactinthelanding gearoperation [9]. Tire deﬂection was takeninto account in thedetailoftheirdesign.

4. Thefourthcase,NL4,wasaperforatedfairingdesigncovering multiple regions of the NLG (lower arm, the steering pinion and wheel axle in Fig. 4), see elements highlighted in blue colourinFig.5d.Comparedwithsolidfairings (suchasNL2), porous fairings with correctly designed porosity, reduce the velocity ofthe deflectedflow towards the fairing’ssides and thevorticity generated [9]. In addition,they are expectedto havea lower weight and benefits on brake cooling [48]. On the other hand, they are expected to cause high–frequency noiseduetotheshearingflowpasttheperforations,although correctdesign can push thesefrequencies above the thresh-old of hearing. It should be noted that the wheel axle solid fairingofNL2was largerthanthat ofNL4. Moreinformation aboutthepropertiesoftheperforatedmaterialisprovidedin section2.3.1.

ThesefourLNTsaffecta rangeofareasandcomponentsofthe NLG system and, as such, may be applied either in isolation or in combination. Two combinationtechnologies which included a

(6)

Fig. 5. Photographs of the NLG model ﬁtted with LNTs inside of the wind–tunnel.

subsetofthreeoftheindividualtechnologiesweretested,referred toas:

5. NL5, whichwas a combinationof NL2, NL3, andNL4. In or-der to allowforthe combinationofthese technologiestobe tested,thewheelaxlefairingofNL4wasexcludedduetothe identicalplacement ofthe componentwiththeNL2 technol-ogy,i.e.onlytheelementshighlightedby thelowerblueoval inFig.5dwerepresent.

6. NL6,whichwasacombinationofNL1,NL2,andNL3.ThisLNT addressesthenoisesourcesonthesteeringpinions,leg struc-ture,wheels,andbaydoors.

Furthermore, two additional LNTs were tested with different perforatedfairingcharacteristics:

7. ThetechnologyNL7consistedofthesamefairinglocationsand dimensionsofNL4butwithanalternativeperforatedmaterial, seesection2.3.1andFig.6b.

8. Since nodirectcomparison couldbe madebetweenNL2and NL4, becauseNL4alsoincludedfairings inother locationson the gear, aconﬁguration which includedjust thewheel axle component(theelements highlightedbythehigherblueoval in Fig. 5d) of the perforated fairing with the same mate-rial asNL4was includedinthe testcampaign, referred toas NL8.

Whereas all of the selected LNTs were considered to be ata mediumtohighTRLthisdoesnotmeanthatalltheirdesign con-siderations are complete, and certain considerations can still be optimized, such as theconsequent addedweight, their inﬂuence on the aerodynamic performance (i.e. a potential drag increase), brake cooling,and structuralstability of theNLG [49]. Moreover,

theseLNTs shouldbe compatiblewith thesafetyregulations and not hinder the inspection andmaintenance procedures. Lastly, it should be noted that, although some of these LNTs have been relatively successful in wind tunnel measurements [21,27,28,50], carefulconsiderationsneedtobetakenforscaling–upwind–tunnel resultstoﬂighttestswithactualaircraft[20,51–54],suchas instal-lationeffects.

2.3.1. Perforatedfairings

Oneofthekey parameters whendesigning perforatedfairings is the porosity of the material. The noise reductions achievable byporousmaterials greatlydepend ontheirporosity [48,55].The percentageporosity

σ

ofaperforatedplateofcircularholes with triangularpitchdependsonthediameterd thepitchp,seeFig.6a, andisdeﬁnedas[56]:

σ

=

100

√

3

π

6

d p

2 (1) The aerodynamic and acoustic characteristics of a perforated materialstronglydepend onthe geometryoftheperforate. Thus, itispossibleto designandoptimizeaperforatedmaterialforthe desiredaeroacousticperformance.Boorsmaetal.[56,57] defineda setofguidelinestobefollowedwhenapplyingperforations toLG fairings.Itwas reportedthat perforatedfairingshadthepotential tobreakdownlarge–scaleflowstructures suchasvortices,which are a negative consequenceof usingsolid fairings(such asNL2), henceconsiderablyreducingthebroadbandnoiselevels,including thefairingself–noise.However,valuesofporosity whichweretoo highwereshowntoallowtoomuchfluidtobebledthrough,thus recreatinga noise source at theLG strut itself.The self–noise of aperforatedmaterial isinfluenced bythe massflow throughthe

(7)

Fig. 6. (a) Explanation of the parameters deﬁning the perforated meshes considered. (b) Detail of the perforated fairing around the drag stay for NL7 (withσ=40.31%).

Table 2

Perforatedmaterialpropertiesforthe NL4andNL7testcases.

NL4 NL7 σ 29.61% 40.31%

p 3.5 mm 6 mm

d 2 mm 4 mm

holes. Whilethisis dependenton theﬂow speed,there isalso a dependency on the hole size [55]. It was suggested that a pore diameteroflessthan3mmwouldgeneratenoiseoutsidethe au-diblerangeforapproachconditions[56,57].Boorsmaetal.[56,57] applieda fairingwitha porosityof40% toaLG modelforwind– tunneltests.InthecontextofALLEGRA,thetwomaterialsavailable metone,butnotbothoftheseconditionsofsmallholesizes and relatively high porosity.The NL4case consistedof the smaller 2 mmholesize,butwithalowerporosityof29

.

6% whereastheNL7 case consistedofa larger hole size, 4mm, witha higher poros-ityof40%.Bothperforatedmesheshadathicknessof1mmanda pitchangleof60◦,seeFig.6a.Thematerialpropertiesaregivenin Table2.

Thesetwoporousmaterialswereselectedbasedoninputfrom theCleanSkyGRAmembers,manufacturingconsiderations,a liter-aturereview [56,57] andsmall–scaletestscompleted priorto the wind–tunneltestcampaign.Theliteraturereviewsuggestedan im-proved performance fora greater porosity witha highfrequency penalty whichisa functionof theholesize.Forpractical aircraft applications, an appropriate material selection is requireddue to thearduousworkingenvironment[9].

3. Acousticimagingmethod

Thedatarecordedbythemicrophonearrayswasemployedfor theapplicationofbothtwo–andthree–dimensionalacoustic map-ping algorithms. In all cases, the main diagonal ofthe CSM was removed in order to eliminate the inﬂuence of noise incoherent forallthemicrophones[36],suchasthewindnoise.The convec-tion ofthesoundwavesdueto theﬂow velocity wasconsidered andastandardshearlayercorrection,asdescribedbyAmiet[58], wasapplied.Inaddition,aweightingfunctionwasappliedto min-imizecoherenceloss,especiallyforoutermicrophonesandathigh frequencies[33].

3.1. Two–dimensionalacousticmapping

Conventionalfrequencydomainbeamforming(CFDBF)[33] was applied tothe acousticdata acquiredby thetop andside micro-phonearrays,separately,inordertoobtainthe2Dacousticsource

maps in the xy and yz planes,respectively. Foreach point in a deﬁned scan grid, CFDBF estimates the agreement between the recordedpressuresbythemicrophonearrayandthepressuresfor amodeledsoundsourceaccordingtotheradiationmodelselected by the user (typically an omnidirectional monopole). This tech-nique iswidely used for sourcemapping, since it is robust, fast, andintuitive[59].

6 m.These ROIswereselectedto coverthenoisesources ontheNLG model, seeFigs.7and8.

Additionally, the enhanced high–resolution deconvolution methodEHR–CLEAN–SC[62–64] wasalsoappliedtothedatafrom botharraysinordertoobtaina betterdynamic range(fewerand lower sidelobes,i.e.spurious sources) andtoinvestigate whether one or multiple sound sources were present (even beyond the Rayleighresolutionlimit [62,63]).Theloopgain[62] selectedwas 0.9and the number of sound sources considered was estimated basedonthenumberofdominanteigenvaluesoftheCSMforeach frequencycase[59]. The widthofthe syntheticclean main lobes hastobeinputbytheuserand,inthiscase,alargeenoughwidth was selected to ensure a clear visualization. For the frequency range considered (200 Hz to 4 kHz), the differences between the obtained spectra by the SPI technique and EHR–CLEAN–SC wererelativelysmall.Henceforth,thespectraobtainedwithEHR– CLEAN–SCarepresentedinthispaper.

(8)

Fig. 7. EHR–CLEAN–SCsourcemapsfortheflyover directionforaone–third–octave–bandcentred at1250 HzandU∞=50 m/s.ThereferencecaseNLG–BASEisshownin subfigure(a)andsubfigures(b–i)correspondtothecasesNL1–NL8,respectively.TheROIisdenotedasadashedgreenrectangle.

3.2. Three–dimensionalacousticmapping

701 grid points. As forthe2D case, thewidthsofthe clean main lobesof EHR–CLEAN–SCwereselectedtoberelativelylargetoeasethe vi-sualization.

Theadvancedexperimentalsetup,withseveralmicrophone ar-raysondifferentplanes(i.e.withdifferentpointsofview)is ben-eﬁcialforusing3D acousticmapping [41,65]. Insteadofusingall the microphones synchronously as a larger array, the 3D source mapsobtained,respectively,bythetop,side,andfrontarrays sep-aratelywerecalculatedandthencombinedintoatotal3D source map usinga multiplicative approach.Foreach scanpoint the ge-ometricaverageofthe sourceautopowersinthe3D sourcemaps obtained by each of the three arrayswas calculated (i.e. the cu-bicrootoftheproduct)[41,65].Inthatway,thedifferentsidelobe patterns ofeach array are averaged out, sincethese strongly de-pend onthe relative position betweenthesound source andthe microphonearray. Porteouset al.[41] showedthat better results wereobtainedwhenusingthistechniquecomparedtosimply

us-ing allthe available microphonessimultaneously. The beamform-ing formulation has to be adapted accordinglyfor obtaining the correctsource locations in3D [66]. Battista etal.[67] employed inverse methods and CLEAN–SC for three–dimensional acoustic sourcemapping ofan aircraftmodelusing thesame microphone arrays in the same wind tunnel, and suggested that the use of multiple arraysis preferred when source localizationaccuracy is crucial. However, it was argued that excessive source directivity orpotential shieldingofthesoundemissionsincertaindirections maydeterioratethelocalization,butspeciallythequantificationof noise sources using this approach [67]. In the currentsetup, for example, a noise source located between the wheels would not bedirectlyvisibletotheside array.Therefore,3D acousticsource maps are difficult to directly interpret to obtain quantitative re-sults.Hence,forthepresentresearch,3Dacousticsourcemapping isonlyemployedforsourcelocalization,andthequantificationof thenoiseemissionsintheflyover andside directionsisperformed byintegratingthe2DROIsdefinedinsection3.1.

The3D acousticsourcemapspresented henceforthdepict iso-contoursofthegridpointsthathaveaLp belowacertain thresh-oldindBbelowthepeakvaluewithinthemap.

4. Results

Asdiscussed,the study presentedinthisarticlecontinuesthe analysis of data from the ALLEGRA test campaign. With respect to the NLG results,the outcome of the work published by

(9)

Ben-Fig. 8. EHR–CLEAN–SCsourcemapsfortheside directionforaone–third–octave–bandcentred at1250 Hzand U∞=50 m/s.ThereferencecaseNLG–BASEisshownin subﬁgure(a)andsubﬁgures(b–i)correspondtothecasesNL1–NL8,respectively.TheROIisdenotedasadashedgreenrectangle.

nettetal.[24] andNerietal.[13] resultedincertainconclusions. Thefirstwasthatthepresenceofthebay,doors,anddressedNLG increasedthe noise levelsover the NLFconfiguration(closed bay fuselagewithnoleg)byupto15 dBasmeasuredbymicrophones in the linear array.The majority of thenoise increase was mea-sured in the 200 Hz to 1 kHz frequency range with the most significant increases centred around 200 Hz and 350 Hz with a biasindirectivitytowardstoforwardarc.Asecondkeyconclusion was that significantnoise canbe generatedinthewheelbay be-low 200 Hzduetotheshear layerexcitationover thebay cavity, butthattheseHelmholtz–resonanceandductmodenoisesources tendtobe attenuatedby thepresenceoftheleg anddoors.With specific focuson thenoise generated bythe NLG wheels,a third set of conclusions found that the wheels generated noise inthe 315 Hz,630 Hzand1.25 kHzone–third–octavebands,withthe lat-tertwofrequencyrangesbeingshowntoresultfrominter–wheel noisesources.ItwasalsoshowninBennettetal.[24],thatadding hub caps (internal andexternal)was aseffectiveat reducing the noiselevelassimplyremovingthewheelsentirely.

4.1. Two–dimensionalresults 4.1.1. Two–dimensionalsourcemaps

Thissectioncontainssomeillustrativeexamplesof2Dacoustic sourcemapsobtainedbythetopandsidearrays,usedseparately. Thesesourcemapsareemployed toobtaininformationaboutthe location ofthe mainnoise sources within theNLG modelandto

quantifythenoise levelsemitted by integratingthemovera ROI, asexplainedinsection 3.1.Only theresultswiththe ﬂow veloc-ityU_∞

=

50 m/sare reportedforreasonsofbrevityandtoallow for comparisons with the results to be found in previous publi-cations[13,24–26].Resultscorrespondingto theone–third–octave frequencyband centred at1250 Hz are presented,as LNTs were foundtodramaticallyreduce noiseinthisfrequencyband,aswill be seen in Figs. 10and 12. The source maps are plotted over a cutout of the NLG 3D model at the location of the scan plane (see greycontours in Figs. 7 and8). Unfortunately,the 3D mod-els of the LNTs were not available, so the 3D NLG model cor-responds to the baseline case (NLG–BASE). Lastly, the Lp values shown are relative to the maximum value of the reference case (NLG–BASE) and all 2D source maps have a dynamic range of 12 dB.

Fig.7depictsthesourcemapsobtainedbythetoparrayforthe 1250Hzcaseforthebaselinecase(NLG–BASE)andtheeightLNTs. As is usually the case withﬂyover angle beamforming[68], this pointofviewdoesnotprovidepreciseinformationonthelocation ofthestrongestnoisesources.Thebaselinecase(Fig.7a)showsa mainnoise sourcelocatedatthe middlepointofthe wheelaxle, thatcouldjustaseasilybelocatedalongthemainstrut.Asimilar sourcedistributionwasobservedbyYokokawaetal.[20] ina com-parableexperiment.Thelargestnoisereductionforthisexampleis achievedbyNL2(solidwheelaxlefairing,seeFig.7c),showinga reduction of the peak value of approximately 6 dB, followed by NL3, NL5, and NL6 (Figs. 7d, f, and g, respectively), which offer

(10)

Fig. 9. (a)Narrowbandspectra(f=4 Hz)forallLNTsandthebaseline(NLG–BASE)intheﬂyover directionforU∞=60 m/s.(b)Correspondingreductioninthesound pressurelevelLpwithrespecttothebaseline(NLG–BASE)providedbyeachLNT.

peak reductionsbetween 4 to 6 dB. The fact that the combined LNTs NL5 and NL6 provide slightly lower noise reductions than the individual measure NL2 (even ifboth combined technologies have NL2equipped) is remarkableand highlightsthe complexity ofthephysicalmechanismsinvolvedinLGnoise.However,aswill be seen insection 4.1.2, NL6hasapproximatelythe same perfor-manceasNL2andbothNL6andNL5performbetterthanNL3.This isamoresensibleresultandhighlightsthefactthatinterpretation of such source maps without integration can be erroneous. The rest of the LNTs reach lower noise reductions forthis frequency band. The cases NL2, NL3, andNL5 presenta slight downstream shift of the main source position away from the main strut lo-cation, whereas NL4, NL6, NL7andNL8seem to shiftthe source locationintheupstreamdirection.Thecaseequippedwitharamp doorspoiler(NL1)slightlyincreasesthenoiseemissionsatthis fre-quencyband,perhapsduetotheinteractionofthedeﬂectedﬂow withcertainelementsoftheNLG.

The side array provides a clearerpoint of view forseparating differentnoisesources withinthe acousticsourcemaps,although itshouldbenotedthatthewheelandthebaydoorclosest tothe array (in the y

<

0 subspace) may shield the soundpropagation of any potential noise sources behind them. Fig. 8 contains the analogousresultsasforFig. 7(i.e.a frequencyof1250 Hz anda flowvelocityof50 m/s)butfromtheside direction.Inthisfigure, thebaselinecase(Fig.8a)showswhatappearstobeadistributed noise sourcealong the mainstrut, especiallynear thewheeland around thetowfittingandthesteeringpinions.Thissource loca-tion is possibly not correct, because the 3D source map forthis case, see Fig. 15a, seems to indicate that the noise source is in fact located between the wheels.Hence, the apparent source lo-cation observed in these2D source maps is perhaps due to the shielding effect of the wheelor other soundpropagation effects. The bestperformingLNTs areinlinewiththose observedforthe flyover direction in Fig. 7: NL2, NL3, NL5, and NL6, managing to significantlymitigatethenoisesourcealongthemainstrut.Asper theflyover case,thesefourLNTsareprimarilylocatedat/withinthe wheelsandsoitseemsthatthenoisesourceseeninthebaseline isattenuatedbythesewheelsLNTs mostprobablyasaresultofa changeinairflowatthislocation.ThecaseforNL6(Fig.8g), how-ever,presents anadditionalnoise sourceatthe frontedge ofthe wheelbaydoor(orthetrailingedgeoftherampdoorspoiler).This ismostlikelyduetothepresenceoftherampdoorspoiler,sincea similar(althoughweaker)noisesourceisalsoobservedinthecase ofNL1(Fig.8b).Theeffectoftheflowdeflectionbytherampdoor spoiler seems to alsoshorten the extent ofthe distributednoise source along the mainstrut and cluster thenoise sources at the outerpartofit.ThoseLNTsequipped(only)withporousmaterials

(NL4, NL7, andNL8) do provide some noise mitigation, but con-siderablylower,eventhoughcomponentsofthemarealsolocated betweenthewheels.

4.1.2. Frequencyspectra

Thissection discusses thefrequencyspectra obtainedby inte-gratingtheEHR–CLEAN–SC2DsourcemapswithintheROIsshown insection 4.1.1fortheﬂyover andtheside directions,respectively. The noise reductions are presented as relative values with re-spect to the baseline case(NLG–BASE). Therefore, the parameter

Lp

=

Lp,baseline

−

Lp,NLi is used, where Lp,baseline represents the noise levels ofthe baseline case, and Lp,NLi those of the ith LNT, with i

=

1

. . .

8. In this manner,

40 m/s.The firstperforatedfairing(NL4) achieves a poorer performance than the solid fairing case (NL2) withaveragenoisereductionsaround2 dB forthethreeflow ve-locities,forfrequencieshigherthan1 kHz.Thesevalues agreewith thefindingsofMurayamaetal.intheircomputationalstudy[48]. The fairing with an alternative perforated material with higher percentage porosity (NL7)performs considerably worse than the less porous caseof NL4, andeven increases the noise emissions at 400 Hz for U_∞

=

40 m/s and 50 m/s, and at 1600 Hz for U_∞

=

60 m/s.NL8(perforated fairingonlyatthewheelaxle) es-sentially offers the same results as NL4 (fairing with the same perforatedmaterialbutcoveringmultipleregionsoftheNLG),but still considerablypoorerperformance thanthesolidfairing(NL2). This indicates that the perforated fairingis most effectiveat the wheelaxleandthattherestofthefairingscouldberemovedwith the consequent save in weight and complexity. Lastly, the other LNT thatcombinesseveraltechnologies(NL5=NL2+NL3+NL4) providesacceptableresults,butcomparablylowernoisereductions thanNL2whenusedalone.

Sidedirection Thenoise emissionsintheside directionforeach NLG conﬁguration for U_∞

=

60 m/s are depicted in Fig. 11a as

narrowbandspectra.In general,the noiselevels measuredin this directionarehigherthanthoseintheﬂyover directionformostof thefrequencyrange.Similar low–frequencytonesareobservedin thiscase.ThenoisereductionsachievedbyeachLNTarepresented inFig. 11b,where the low–frequency tonesare again greatly at-tenuated by NL1 and NL6. Overall, considerably lower noise re-ductions are reported in this direction compared to the ﬂyover one.

(12)

Fig. 11. (a)Narrowbandspectra(f=4 Hz)forallLNTsandthebaseline(NLG–BASE)intheside directionforU∞=60 m/s.(b)Correspondingreductioninthesound pressurelevelLpwithrespecttothebaseline(NLG–BASE)providedbyeachLNT.

Fig. 12. ReductioninthesoundpressurelevelLpwithrespecttothebaseline(NLG–BASE)providedbyeachLNTintheside directionasone–third–octavebandspectrafor (a)U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s.

directionthe noise reduction decreases from6.9dB to 6.8 dB to 5 dB with increasing velocity (40 m/s, 50 m/s, and60 m/s). For theside directionthenoisereductiondecreasesfrom3.5dBat40 m/s to 2.2 dB at 60 m/s with a slight increase to 3.8 dB at 50 m/s. In contrast, NL1increases inefficiency withincreasing flow velocity with 4.1 dB, 4.3 dB and 6.7 dB being the noise reduc-tionattainedwithincreasing velocityfortheflyover direction,for example.A similar trendis observedin theside direction, which seems toindicate thatthose noisesources increase inmagnitude withincreasing flowvelocity and alsothat their behaviour is al-mostomnidirectionalinnature,beingequallymeasurableinboth directions.

At ﬁrst inspection, this may seem to be counter–intuitive as if it is assumed that aerodynamic noise increases with U6

∞, or

sometimeswithU7

∞ athigherfrequenciesduetosmallfeaturesas

discussedbyGuoetal.[69] andBennettetal.[24],thenonemight expecttheefficiencyofthe LNTstoincrease withvelocitydueto anincreasingnoisesource.However,asdiscussedbyCasalinoetal. [70] significantNLG noiseisgeneratedasa resultofnarrowband acousticmodesbetweenthewheelsrelatedtotherimsbeing ex-citedbytheflow.Thisisconsideredtobeasignificantnoisesource inourcaseasit was examinedin aprevious paperon thesame experimentalrig[24] andinthisworkitissupportedby thefact thatthesolidwheelaxlefairing(NL2)andthehubcaps(NL3)are themostefficientLNTsforthe wheels.Thus,itwouldnot be ex-pectedthatthesenoisesources would, increasesignificantly with increasingflow speed andthedecreasingLNT efficiencycould be duetothenoiseoverthetopandbottomofthewheelsincreasing asU6

(13)

Fig. 13. Three–dimensionalEHR–CLEAN–SCsourcemapsfortheone–third–octave–bandcentred at200 HzandU∞=50 m/s.ThereferencecaseNLG–BASEisshownin subﬁgure(a)andsubﬁgures(b–i)correspondtothecasesNL1–NL8,respectively.Thevaluesshownarevisualizedasisocontours3 dB(inred)and6 dB(inyellow)below thepeakvalueofthereferencecase(NLG–BASE).

4.2. Three–dimensionalresults

The two–dimensional source maps presented in section 4.1.1

provide some insighton thelocation ofthe noisesources within the NLG model, but it is challenging to identify the exact NLG elements generatingthe noise in each case, since both points of viewhavetheirownlimitations.Inaddition,itisconceivablethat bothofthesearraysmaybeonthesamedipolarnodalplaneand thus will not be sensitive to certain noise sources with such ra-diationpatterns.The3D acousticmappingtechnique explainedin section 3.2aimsatsolving thisissueby combiningthe resultsof thetop, side,andfrontmicrophonearraysina multiplicative ap-proach, beneﬁting from their different points of view. This way, thetypicallypoorerresolutioninthedirectionsnormaltoeach ar-ray’s planeispartiallymitigated.Basedontheobservationsmade on the integrated spectra of Figs. 10 and 12 the examples pre-sented below correspond to a ﬂow velocity of50 m/s and one– third–octave–bands centred at 200 Hz, 315 Hz and1250 Hz

be-causethose frequencies showedthe highest noise reductions for most of the LNTs considered. However, it was noticed that per-formance is better for different LNTs, highlighting the fact that the LNTs address differentnoise sources atdifferent frequencies. Acousticimagingatthesefrequencybandsthereforehelpsto iden-tify exactly which sources are being addressed. The 50 m/s test pointwaschosensothatresultscanbecomparedwiththosefrom earlierpublications[13,24].Therelative Lp valuesshownare visu-alizedasisocontours3 dB(inred)and6 dB(inyellow)belowthe peakvalueofthereferencecase(NLG–BASE).

Fig.13 illustrates this approach forthe 200 Hz case. The ref-erence case(Fig. 13a) shows that the noise is mostly generated around the lower jointarm and thejunction between the steer-ingpinionsandthemainstrut,i.e.notbetweenthewheels,which explainswhythewheel–locatedLNTsarenot effectiveatthis fre-quency.TheweakersourcesawayfromtheNLGmodelpresentin someofthesource mapsare duetothepoorerspatialresolution of the arraysat low frequencies. Consistentwith the ﬁndings of

(14)

section 4.1.2, the cases equipped with a ramp door spoiler(NL1 and NL6) completely remove that noise source (in this dynamic range),mostlikelyduetotheﬂowdeﬂection.Other LNTssuchas NL2, NL3, NL5 andNL7provide smallernoise reductions. Twoof the technologies equipped with porous materials (NL4 and NL8) offerapoorerperformanceandevenincreasethenoiseemissions slightly.

Forthe frequencyof315 Hz (see Fig.14) thebaseline config-uration shows a cluster of noise sources along the outer part of themainstrutuntilthejunctionwiththewheelaxle,seeFig.14a. In thiscase, thoseLNTs equippedof solid fairings(NL2, NL5and NL6)andwheelhubcaps(NL3)considerablydecreasethestrength oftheaforementionednoisesource.Thetechnologiesconsistingof porousfairings(NL4,NL7andNL8),ontheotherhand,donot pro-vide agoodperformance forthisfrequencyband.Lastly,NL1also achieves a notable noise reduction (perhaps dueto the flow de-flection bytherampdoorspoiler)butsmallerthanthosebyNL2, NL5,andNL6.

Lastly, the source maps corresponding to the frequency of 1250 Hz are depicted in Fig. 15. For the baseline configuration (Fig. 15a) the main noise source is now located at the middle point of the wheel axle (between both wheels). Once again the LNTsequippedofsolidfairings(NL2,NL5andNL6)andwheelhub caps (NL3) manage to successfully reduce the noise levels mea-sured.NL4 andNL8 providean almost identicalperformance for thisfrequencyband, confirmingagainthat theperforatedfairings aremosteffectivewhenapplied atthewheelaxle. Thecasewith a fairing made of a more porous material (NL7) performs con-siderably worse than the other material option. Lastly, the ramp door spoiler of NL1 barely modifies the baseline’s source distri-bution, since this frequencyis out of the intended range ofthis LNT.

In general, 3D source maps have the added value of offering amore preciseandglobalassessment ofthelocation ofcomplex noise sources within a system by combining different points of view (in this case top, side and front). An example of this was

(15)

the caseofthe 3D source mapsof Figs. 15aand b,compared to themisleading2D sourcemapsofFigs. 8aandb.Thus,3D source maps should be employed in combination with the 2D source maps obtained by the individual planar arrays, since those pro-vide an easierwayto determinethenoise levels inthe emission directionsofinterest,asexplainedsection4.1.2.

4.3. Overallnoisereduction

The acoustic imagingresults of sections4.1.1 and4.2 demon-strate that the NLG is a complex arrangement of sound sources, andthatwhereastheLNTstargetanindividualsourceandmay re-duce its level,thismaynot stronglyaffectthe overallnoise level as there are potentially multiple other sources of similar magni-tudeondifferentsectionofthelandinggear.

Innoiseassessmentapplications,itiscommontoemploysound metrics that consider the overall frequency spectra. Given the strong low–frequency content of the noise signature of the NLG

model tested [71], it was decided to apply A–weighting to the integratedspectra, inorder to highlightthe relevance of the ob-tained noise reductions by the LNTs. Therefore, the

Lp,A,overall metricwasused,inasimilarwayasinsection4.1.2:

Lp,A,overall

=

Lp,A,overall,baseline

−

Lp,A,overall,NLi. As before,

Lp,A,overall

>

0 rep-resents a noise reduction. Additionally, the D–weighted results (

Lp,D,overall)arealsoexamined,sincethisweightingﬁlterwas es-pecially developed for aircraft noise measurements and is more representativeoftheeffectiveperceived noiselevel(EPNL)metric usedforaircraftcertiﬁcation[1].Theoverallvalueswereobtained byintegratingthespectrapresentedinFigs.10and12withinthe frequencyrangeofinterest(200 Hzto4 kHz).

4.3.1. Flyoverdirection

TheoverallA–weightednoisereductionsobtainedby eachLNT inthe ﬂyover directionare presented asbar plotsin Fig.16. Not surprisingly,similar ﬁndings are observed aswhen analyzingthe frequencyspectra insection 4.1.2: NL6 (NL1+NL2 +NL3) isthe

(16)

Fig. 16. ReductionintheA–weightedoverallsoundpressurelevelLp,A,overallwithrespecttothebaseline(NLG–BASE)providedbyeachLNTintheﬂyover directionfor(a)

U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s.

best performing LNT, achievingoverall noise reductionsof6 dBA for U_∞

=

40 m/s and50 m/s, and 4 dBAfor 60 m/s.NL2 (solid wheel axle fairing) is the best performing individual LNT (with noise reductionsup to4 dBA)followedby NL3(wheel hubcaps), whichachievessimilarnoisereductionsasthecombined technol-ogy NL5 (NL2 + NL3 + NL4). Overall, this indicates that simply addingasmanyLNTsaspossiblemightbecounterproductive,even whenthoseLNTshaveproventohavefavourableindividualeffects. SinceNL1(ramp doorspoiler)mostlyreducesthenoiseemissions atlow frequencies onlyandthose are heavilyreduced bythe A– weighting,itonlyachieves

Lp,A,overallvaluesofaround1 dBA.As mentioned in section 4.1.2,the additional treatedregions of NL4 comparedtotheperforatedfairingonlyappliedtothewheelaxle (NL8) do not provide any beneﬁt. Lastly, the fairing with higher porosity (NL7)performs slightlyworse thanthelessporouscases (NL4 andNL8). Ingenerallower noise reductionsare reachedby allLNTsasthevelocityincreases,whichcanconditionthe perfor-manceachievedin casehigherapproachvelocities are employed. Thus,wind–tunnelexperimentsathigherﬂowvelocitieswouldbe ofhighinterest.

Ingeneral,theD–weightednoisereductionsachievedbyallthe LNTs (see Fig. 17) are slightlylower values than theA–weighted ones, exceptforNL1andNL6whichachieveslightlyhigherlevels because their reductionsat low frequencies are morerelevant in thecomputationof

Lp,D,overall.

4.3.2. Sidedirection

Similar trends are observed for the case ofthe side direction see Fig.18, althoughwithconsiderably lower values.NL6in this caseonly reaches overall noise reductionsof3 dBA andthe rest oftheLNTsalsoprovideapoorerperformance.OnlyNL1seemsto maintainsimilarvaluesasintheﬂyover direction.

Asfortheﬂyover direction,all LNTsprovidelower D–weighted noise reductions than the A–weighted ones, except for NL1 and NL6,seeFig.19).

4.4.Directivityinthepolardirection

Thesidelineararrayallowsforasimplestudyofthedirectivity patternthattheLNTspresentfortheirnoisereductions.Thesame

Lp,A,overallmetricasforFigs.16and18isemployed,althoughthis timethedatausedfortheanalysisdoesnotcomefromthe inte-grationofacousticsourcemaps,butratherfromtheacousticsignal oftheindividualmicrophonesofthesidelineararray.Itshouldbe notedthat theuseofsingle microphonesismoresensitivetothe backgroundnoise inside ofthewind–tunnel facility,as the sepa-rationofnoise sourcesisnotpossibleaswithmicrophonearrays. Inaddition,thesidelineararrayhasasidelineangle

φ

of81◦,see Fig. 2b. These two aspects are the reason whythe noise reduc-tionvalues presentedin thissection donot precisely correspond tothoseobservedinFigs.16and18.

As aforementioned,the presence of theshear layer limitsthe effectivemeasurablerangeof polaremissionangles corrected for soundconvection[32].Therefore,theavailablerangefor

θ

extends fromapproximately35◦to145◦forU_∞

=

40 m/sand50 m/s,and from35◦to110◦forU_∞

=

60 m/s.Fig.20depictsthe

Lp,A,overall valuesforallﬂowvelocities.Theresultsobservedarequalitatively consistentwiththoseofFigs.16and18:NL6isthebest perform-ingLNT,followedbyNL2,andthenoisereductionsdecreasewhen theﬂowvelocityincreases.Ingeneral,alltheLNTstestedshowan almostomnidirectionalpatternintheirnoisereductions.

5. Conclusions

This manuscript hasinvestigated the acoustic performance of severallow–noise technologies(LNTs) appliedtoafull–scalenose landing gear (NLG) in open–jet wind–tunnel measurements. The LNTsanalyzedwere selectedwithin theEuropeanClean–Sky pro-gramme and the ALLEGRA project and were highly realistic and atamediumtohightechnologyreadinesslevel,allowingthemto be applied to actual aircraftin the near future. There were four

(17)

Fig. 17. ReductionintheD–weightedoverallsoundpressurelevelLp,D,overall withrespecttothebaseline(NLG–BASE)providedbyeachLNTintheﬂyover directionfor(a)

U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s.

Fig. 18. ReductionintheA–weightedoverallsoundpressurelevelLp,A,overall withrespecttothebaseline(NLG–BASE)providedbyeachLNTintheside directionfor(a)

U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s.

main low–noise concepts tested experimentally, namely a ramp door spoiler (NL1), a solid wheel axle fairing (NL2), wheel hub caps (NL3), and multiple perforated fairings (NL4). Combinations andsmallvariationsofsomeoftheseLNTswerealsoinvestigated. Theacousticsignature oftheNLGbaselinemodelandthe con-ﬁgurations equippedwith LNTs were measured by four different

microphone arrays, which allowed the study of the noise emis-sions inthe ﬂyover and side directions, aswell as the useof 3D acousticimagingtechniques.Thistechnique wasprovenvery use-fulfordeterminingthepreciselocationofthenoisesourceswithin theNLG,comparedtothetypical2Dsourcemapsasvisibility be-tweenthewheelsanddoorscouldbeobtainedaswellasincluding

(18)

Fig. 19. ReductionintheD–weightedoverallsoundpressure levelLp,D,overall withrespecttothebaseline(NLG–BASE)providedbyeachLNTintheside directionfor(a)

U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s.

Fig. 20. ReductionintheA–weightedoverallsoundpressurelevelLp,A,overallwithrespecttothebaseline(NLG–BASE)providedbyeachLNTfordifferentpolaremissionangles

θasmeasuredbythesidelineararrayfor(a)U∞=40 m/s,(b)U∞=50 m/s,and(c)U∞=60 m/s. extradatafromthefrontradiationdirection.Dependingonthe

fre-quency,itwasfoundthatforthebaselineNLG,thewheelaxle,the main strut,andthetow ﬁttingwerethe elements generatingthe highestnoiselevels.

TheLNTsconsideredaredesignedtomitigatenoisegeneration at different locationsof the NLG system. The ramp door spoiler (NL1) achieved large noise reductions (more than 6 dB) at low frequencies (around 200 Hz). The solid wheel axle fairing (NL2)

(19)

was thebestperformingLNT appliedindividually,offeringoverall noise reductionsbetween2and6 dBA, particularlyintheﬂyover direction andpeaking at frequencies which correspond to previ-ously determinedwheelnoise sourcefrequencies. Thewheelhub caps (NL3) presented similar but slightly lower noise reductions than NL2(2 to4 dBA). Regardingthe perforatedfairings(NL4), it wasfoundthatthewheelaxle(NL8)isthemosteffectivelocation for their application andthat fairings with a larger porosity and holesize(NL7)offeredslightlypoorerresults.Giventhehigh com-plexityofporousmaterialsforaeroacousticapplications,however, these ﬁndings are not enough to induce anydesign criteria, and moreresearchonthistopicisrecommended.

Since each LNT actson a different region ofthe NLG system, theyarewellsuitedtobeingappliedincombination.Itwasfound thatjoiningtherampdoorspoiler(NL1),thesolidwheelaxle fair-ing (NL2), and the wheel hub caps (NL3) into a combined LNT (denotedasNL6)offeredthebestglobalresults,withoverallnoise reductions between 4 and7 dBA, showing a mutuallybeneﬁcial union ofLNTs. The other combined technologyconsidered (NL5), consistingofthesolidwheelaxlefairing(NL2),thewheelhubcaps (NL3),andthemultipleperforatedfairings(NL4),however,didnot alwaysoutperform theindividualconstituentLNTs,indicatingthat certaincombinationsofLNTsmightbecounterproductive.Thisfact highlights againthe complexityinvolvedin thedesignofoptimal LNTs.

In general, all the LNTs considered showed a better perfor-manceintheflyover directionthanintheside emissiondirection, andslightlylowernoisereductionswereachievedwhenincreasing theflowspeedfrom40 m/sto60 m/s.NL1andNL6areexceptions to thesecomments.Overall, conductingwind–tunnel experiments athigherflowvelocitieswouldbeofhighinteresttoevaluatethis trendandmorerepresentative resultsforactual flightconditions. Asimpledirectivity studyinthepolaremissiondirectionshowed thatnoneofthenoisereductionsofferedbyanyoftheLNTshave apronouncedradiationpatternalongthisparticulararc.

Declarationofcompetinginterest

Theauthorsdeclarethattheyhavenoknowncompeting ﬁnan-cialinterestsorpersonalrelationshipsthatcouldhaveappearedto inﬂuencetheworkreportedinthispaper.

Acknowledgements

Theresearchleadingtotheseresultshasreceivedfundingfrom the European Union Seventh Framework Programme (FP7/2007– 2013) for the Clean Sky Joint Technology Initiative under grant agreements[308225](ALLEGRA)and[620188](ARTIC).

The authors would like to acknowledge the work of Dr. Francesco Amoroso of Eurotech who led the manufacture of the modelandMr.MarcoEspositoofTeknosudwholedthewind tun-nelmodeldesign.

References

[1] G.Ruijgrok,ElementsofAviationAcoustics,secondedition,VSSD,Vereniging voorStudie- enStudentenbelangenteDelft,Delft,theNetherlands,ISBN 90-6275-899-1,2007,https://repository.tudelft.nl/islandora/object/uuid:e4589b31 -ffa3-4a5d-9400-1e2b446f60f3/?collection=research.

[2] ACARE–StrategicResearch &InnovationAgenda, Tech.Rep.,2012, https:// www.acare4europe.org/sites/acare4europe.org/ﬁles/document/ACARE-Strategic -Research-Innovation-Volume-1.pdf.

[3] Flightpath2050Europe’sVisionforAviation,Tech.Rep.,EuropeanCommission, ISBN 978-92-79-19724-6, 2012, http://ec.europa.eu/transport/sites/transport/ ﬁles/modes/air/doc/ﬂightpath2050.pdf.

[4] H.H. Heller, W.M. Dobrzynski, Sound radiation from aircraft wheel– well/landing–gearconﬁguration,J.Aircr.14 (8)(1977)768–774,https://doi.org/ 10.2514/3.58851,http://arc.aiaa.org/doi/abs/10.2514/3.58851.

[5] W. Dobrzynski, Almost 40 years of airframe noise research: what did we achieve?, J. Aircr. 47 (2) (2010) 353–367, https://doi.org/10.2514/1.44457,

http://arc.aiaa.org/doi/pdf/10.2514/1.44457.

[6] L. Bertsch,Noise Predictionwithin ConceptualAircraft Design, Ph.D.thesis, DLR, Bunsenstrasse 10, 37073 Göttingen, Germany, DLR Forschungsbericht, ISRNDLR–FB–2013–20,ISSN 1434-8454,2013,http://elib.dlr.de/84386/1/phd_ thesis_lothar_bertsch.pdf.

[7] R. Merino-Martinez, L. Bertsch, M. Snellen, D.G. Simons, Analysis of land-ing gearnoise during approach,in: 22nd AIAA/CEASAeroacoustics Confer-ence,Lyon,France,May30–June12016,2016,AIAApaper2016–2769,http:// arc.aiaa.org/doi/pdf/10.2514/6.2016-2769.

[8] R.Merino-Martinez,A.Vieira,M.Snellen,D.G.Simons,Soundqualitymetrics appliedtoaircraftcomponentsunderoperational conditionsusinga micro-phonearray,in:25th AIAA/CEASAeroacousticsConference,Delft,the Nether-lands,May20–242019,2019,AIAApaper2019–2513,http://arc.aiaa.org/doi/ pdf/10.2514/6.2019-2513.

[9] K. Zhao,P. Okolo, E. Neri, P.Chen, J. Kennedy, G.J. Bennett,Noise reduc-tion technologies for aircraft landing gear – a bibliographic review, Prog. Aerosp. Sci. (2019), https://doi.org/10.1016/j.paerosci.2019.100589, https:// www.sciencedirect.com/science/article/pii/S0376042119300338.

[10] D.B.Bliss,R.E.Hayden,LandingGearandCavityNoisePrediction,Tech.Rep. NASAContractorReportCR–2714,NASAContractorReport2714,1976,https:// ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19760021873.pdf.

[11] Y.Li,X.Wang,D.Zhang,Controlstrategiesforaircraftairframenoisereduction, Chin.J.Aeronaut.26 (2)(2013)249–260,https://doi.org/10.1016/j.cja.2013.02. 001.

[12] S.Khelil,P.Bardoux,J.Godard,T.LeGarrec,J.Kennedy,G.J.Bennett, Investiga-tionofthenoiseemissionofaregionalaircraftmainlandinggearbay,in:23rd AIAA/CEASAeroacousticsConference, Denver,Colorado,USA, June5–92017, 2017,AIAApaper2017–3012,http://arc.aiaa.org/doi/pdf/10.2514/6.2017-3012. [13] E.Neri,J.Kennedy,G.J.Bennett,Baycavitynoiseforfull–scalenoselanding

gear:acomparisonbetweenexperimentalandnumericalresults,Aerosp.Sci. Technol.72(2018)278–291,https://doi.org/10.1016/j.ast.2017.11.016. [14] R.Merino-Martinez,M.Snellen,Implementationoftonalcavitynoise

estima-tionsinlandinggearnoisepredictionmodels,in:26th AIAA/CEAS Aeroacous-ticsConference,June15–192020,VirtualEvent,2020,AIAApaper2020–2578,

http://arc.aiaa.org/doi/pdf/10.2514/6.2020-2578.

[15] D.G.Crighton,AirframeNoise.(Chapter7ofAeroacousticsofFlightVehicles: TheoryandPractice–Volume1:NoiseSources), Tech.Rep.NASATechnical Memorandum1258,NASAReferencePublication1258,1991,https://ntrs.nasa. gov/archive/nasa/casi.ntrs.nasa.gov/19920001380.pdf.

[16] Y.Guo,EmpiricalPredictionofAircraftLandingGearNoise,Tech.Rep.NASA TM–2005–213780,NASA,July2005,https://ntrs.nasa.gov/archive/nasa/casi.ntrs. nasa.gov/20050209966.pdf.

[17] Y.Guo, A semi–empirical model for aircraftlanding gear noise prediction, in:12th AIAA/CEASAeroacousticsConference,Cambridge,Massachusetts,USA, May8–102006, 2006,AIAApaper 2006–2627,http://arc.aiaa.org/doi/pdf/10. 2514/6.2006-2627.

[18] M.R.Fink,Noisecomponentmethodforairframenoise,in:4th AIAA Aeroa-cousticsConference, Atlanta,Georgia, USA,October3–51977,October1977, AIAApaper1977–1271,http://arc.aiaa.org/doi/pdf/10.2514/6.1977-1271,1977. [19] M.G.Smith,L.C.Chow,Validationofapredictionmodelforaerodynamicnoise

fromaircraftlandinggear,in:8th AIAA/CEASAeroacousticsConference, Breck-enridge, Co, USA, 17–19 June 2002, 2002, AIAA paper 2002–2581, http:// arc.aiaa.org/doi/abs/10.2514/6.2002-2581.

[20] Y.Yokokawa,T.Imamura,H.Ura,H.Kobayashi,H.Uchida,K.Yamamoto, Ex-perimentalstudyonnoisegenerationofatwo–wheelmainlandinggear,in: 16th AIAA/CEASAeroacousticsConference,Stockholm,Sweden,2010,AIAA pa-per2010–3973,http://arc.aiaa.org/doi/pdf/10.2514/6.2010-3973.

[21] W.Dobrzynski,H.Buchholz,Full–scalenoisetestingonAirbuslandinggearsin theGermanDutchWindTunnel,in:3rd AIAA/CEASAeroacousticsConference, Atlanta,GA,USA,May12–141997,1997,AIAApaper1997–1597,http://arc. aiaa.org/doi/pdf/10.2514/6.1997-1597.

[22] R.Merino-Martinez,E.Neri,M.Snellen,J.Kennedy,D.G.Simons,G.J.Bennett, Analysisofnoselandinggearnoisecomparingnumericalcomputations, predic-tionmodelsandﬂyoverandwind–tunnelmeasurements,in:24th AIAA/CEAS AeroacousticsConference,Atlanta,Georgia,USA,June25–292018,2018,AIAA paper2018–3299,http://arc.aiaa.org/doi/pdf/10.2514/6.2018-3299.

[23] Green Regional Aircraft (GRA), https://www.cleansky.eu/green-regional -aircraft-gra.(Accessed November2019).

[24] G.J.Bennett,E.Neri,J.Kennedy,Noisecharacterizationofafull-scalenose land-inggear,J.Aircr.55 (6)(2018)2476–2490,https://doi.org/10.2514/1.C034750. [25] E. Neri,J. Kennedy, G. Bennett,Characterization of low noise technologies

appliedtoafullscalefuselagemountednose landinggear,in: Proceedings ofthe Internoise2015/ASMENCADMeeting,SanFrancisco,CA,USA,August 9–122015,AmericanSocietyofMechanicalEngineers(ASME),TwoPark Av-enue,New York,USA, 2015, NCAD2015–5911,https://www.researchgate.net/ proﬁle/Eleonora_Neri/publication/283082549_Characterization_of_low_noise_ technologies_applied_to_a_full_scale_fuselage_mounted_nose_landing_gear/ links/56aa06a708aef6e05df43d32/Characterization-of-low-noise-technologies -applied-to-a-full-scale-fuselage-mounted-nose-landing-gear.pdf.