Numerical simulation and wind tunnel tests investigation and validation of a morphing wing-tip demonstrator aerodynamic performance

Publisher’s version / Version de l'éditeur:

Aerospace Science and Technology, 2016-03-26

READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.

https://nrc-publications.canada.ca/eng/copyright

Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la

première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez

pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.

Questions? Contact the NRC Publications Archive team at

PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the

first page of the publication for their contact information.

Archives des publications du CNRC

This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. /

La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version

acceptée du manuscrit ou la version de l’éditeur.

For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien

DOI ci-dessous.

https://doi.org/10.1016/j.ast.2016.03.014

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at

Numerical simulation and wind tunnel tests investigation and

validation of a morphing wing-tip demonstrator aerodynamic

performance

Gabor, Oliviu Şugar; Koreanschi, Andreea; Botez, Ruxandra Mihaela;

Mamou, Mahmoud; Mebarki, Youssef

https://publications-cnrc.canada.ca/fra/droits

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site

LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

NRC Publications Record / Notice d'Archives des publications de CNRC:

https://nrc-publications.canada.ca/eng/view/object/?id=8ed03609-076e-4361-aaf3-778dc292de75

https://publications-cnrc.canada.ca/fra/voir/objet/?id=8ed03609-076e-4361-aaf3-778dc292de75

Contents lists available atScienceDirect

Aerospace

Science

and

Technology

www.elsevier.com/locate/aescte 1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Numerical

simulation

and

wind

tunnel

tests

investigation

and validation

of

a

morphing

wing-tip

demonstrator

aerodynamic

performance

Oliviu ¸Sugar Gabor

a

,

Andreea Koreanschi

a

,

Ruxandra Mihaela Botez

a

,

Mahmoud Mamou

b

_,

_{Youssef Mebarki}

b

a_{LARCASELaboratoryofAppliedResearchinActiveControl,AvionicsandAeroservoelasticity,EcoledeTechnologieSuperieure,Montreal,Que.,H3C1K3,Canada} b_{AerodynamicsLaboratory,NRCAerospace,NationalResearchCouncilCanada,Ottawa,Ont.,K1A0R6,Canada}

a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Articlehistory:

Received27November2015

Receivedinrevisedform20March2016 Accepted21March2016 Availableonlinexxxx Keywords: Morphingwing Morphingaileron CFD Infra-redtests Windtunneltests

Laminar-to-turbulentflowtransition

Thispaperpresentstheresultsobtainedfromthenumericalsimulationandexperimentalwindtunnel testing ofamorphingwingequipped with aflexible upper surfaceandcontrollable actuatedaileron. Thetechnologydemonstratorisrepresentativeofarealaircraftwingtipsection,anditwasdeveloped followingacomplex,multidisciplinarydesign process.Themodelwasfittedwithacompositematerial upper skin whose shape can be morphed, as a function of the flight condition, by four electrical actuatorsplacedinsidethewingstructure.Theoptimizationswereperformedwiththeaimofcontrolling the extent of the laminar flow region, and the resulting shapes were scanned using high-precision photogrammetry.ThenumericalsimulationswereperformedusingComputationalFluidDynamics(CFD) andincludedamodelforpredictingthelaminar-to-turbulentflowtransitionovertheentirewingsurface. The analysesincluded caseswith threeailerondeflection angles andangles ofattacksituated within fivedegreesrange.TheCFDresultswerecomparedwithinfraredthermographymeasurementsinterms oftransitionlocation,surfacepressuremeasurementsandbalanceloadsmeasurementsacquiredduring subsonicwindtunneltestsperformedattheNationalResearchCouncilCanada.

2016PublishedbyElsevierMassonSAS.

1. Introduction

The air transportation industry is a key contribution to eco-nomicdevelopmentaroundtheworld.Sincethebeginningofcivil aviation,therehasbeenasteadyincreaseinthenumberof passen-gersusingairplanesasafastandsafetransportationmethod,with airlinescarriedalmostthreebillionpassengersworldwidein2014 alone.Thishighlevelofdevelopment that hasbeenachievedhas alsotransformed theair transport industry into a non-negligible source of pollution. It is estimated that in 2014, over 2% of the worldwidecarbondioxide emissionswere attributedto commer-cialairlinecompanies[1].

The current growth rate is anticipated to accelerate over the next several decades. The International Civil Aviation Organiza-tion(ICAO)estimatesthatthenumberofflights willtriplebythe year2050[2].Thisgrowthrate,togetherwithgrowingglobal con-cern for the preservation of the environment and the reduction

E-mailaddress:ruxandra.botez@etsmtl.ca(R.M. Botez).

of greenhousegas emissionsis forcing the aerospaceindustry to searchforsolutionstoimproveaircraftfuelburnefficiency.

One possibility of achieving this desired efficiencyis through thenew-generation technologiesofmorphing variousaircraft lift-ing surface, that canbe activated anddeformedaccordingtothe flight conditions, thus allowing a multi-point design of the air-craftandimprovingaerodynamicsperformance.Amorphingwing could allow the aircraft to fly at optimal lift-to-drag ratios for anycondition encounteredduring flight bychanging someof the wing’scharacteristics.Researchershaveproposeddifferent techno-logical solutions for obtaining the desired wingadaptability, and some concepts have achieved important theoretical performance improvementscomparedtothebaselinedesign.However,the tech-nologyisstillintheearlystagesofdevelopment,itstechnological readiness level is still low, and only a few concepts have suffi-ciently progressed to reach wind tunnel testing, and even fewer haveactuallybeenflighttested[3].

Wing morphing techniques can be classifiedinto three major types: plan-form transformations (sweepangle, spanand chord), out-of-plane transformations (twisting, dihedral and spanwise bending) and airfoil transformations (camber andthickness) [3].

http://dx.doi.org/10.1016/j.ast.2016.03.014

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Morphing wings were used to adapt the wing span and airfoil camber [4,5] or the winglet’s cant and toe angles [6], and to replace conventional high-lift devices [7,8], or the conventional controlsurfaces[9].

LockheedMartindevelopedtheAgileHunterUnmannedAerial Vehicle(UAV)[10–12],capable offoldingthe innerregion ofthe wingoverthefuselageinordertoachievedrag reductionsduring transoniccruiseatloweraltitudes.Amorphingwindtunnel proto-typewasbuiltandtesteduptoaMachnumberof0.6.Themodel demonstrated a successful and accurate actuation under aerody-namicloads,achievingthedesiredwingshapechangein approx-imatelyone minute.An importantprojectforthedevelopment of morphing wings was the NexGen Aeronautics MFX1 UAV, show-ing wingsweep andchordchanges [13,14]. The wingwas based onamorphingtrussstructurethatcouldbecontrolledusing elec-tricalactuators.AprototypeoftheUAVwasbuiltandsuccessfully flighttested.Themorphingwingsustainedsweepanglevariations of 20◦ _{and area changes of 40% under aerodynamic loading, for}

flightspeedsupto100knots.

A detailed computational and experimental analysis was per-formedbySmithetal.[15]onthewingofaconventionalaircraft equippedwithtwooutboardmorphingpartitionscapableof vary-ing thetwist andthe dihedralangles. The morphing systemwas capableofprovidingtwistvariationsofupto3◦_,anddihedral

vari-ationsofupto90◦_{.ResearchersfromNASADrydenFlightResearch}

Center conducted several flight tests with a UAV equipped with inflatable wings whose spancould be modified by adjusting the pressureinput [16].The wingswere madefromseveralspanwise inflatabletubes,surroundedbyspongeandaflexiblenylonskinin orderto maintain theairfoil shapeduring flight. A variable wing plan-form UAV was designed andtested by Neal etal. [17]. The systemused pneumatic actuators to drive the telescopicand ro-tatingwing,capableofachievingsignificantwingspanandsweep anglechanges.Windtunneltestswereperformedandshowedthat onlythree morphing wingconfigurations were neededto signifi-cantlyincreasethe lift-to-dragratiofortheentireflight envelope oftheUAV.

Pecora et al. [18] demonstrated the effectiveness of replac-ing the conventional segmentedflap witha morphing compliant high-lift device, in the case of a regional transport aircraft. Bil-genetal.[19,20]alsopresentedtheconceptofreplacingthewing trailing-edgedeviceswitha morphing surface, capableof achiev-ingcontinuous camber variationsinstead ofrigiddeflections.The morphing systemwas designed toreplace the aileronsofa UAV, andthus used rapid,electrical actuationmechanisms. Both wind tunnel experiments and preliminary flight test were performed, and demonstrated the effectiveness of the concept at providing accuraterollcontrol. PankonienandInman[21] presenteda con-cept formorphing ailerons designed to replace the conventional wingcontrol surfacesofa UAV.The aerodynamic performance of the system was evaluated using wing tunnel testing, with mea-surementsfocusedonthedrag coefficientpenaltyassociatedwith classic control surface deflections at off-design flight conditions. Themorphingtrailingedgeachieveddrag reductionsofupto20% comparedtotheoriginaldesign,thusjustifyingitsincreasedmass andcomplexity.

The CRIAQ 7.1 project, which took place between 2006 and 2009,was realizedfollowingacollaborativeeffortbetweenteams from the École de Technologie Supérieure (ÉTS), École Polytech-niquede Montréal,BombardierAerospace,ThalesCanada andthe National Research Council Canada (CNRC). The objective of the projectwastoimproveandcontrolthelaminarityoftheflowpast a morphing wingin orderto obtain a substantial drag reduction

[22].

Inthatproject,theactivestructureofthemorphingwing con-sistedofthreemainsubsystems:a flexibleskin;acomposite

ma-terial upper surface spanning between3% and70% of the airfoil chord, a rigid inner surface and a Shape Memory Alloy (SMA) actuator group located inside the wing box, which could morph the flexible skin at two actuation points, located respectively at 25.3%and47.6%ofthechord[23].Thereferenceairfoilchosenfor the projectwas the WTEA laminarairfoilandthe morphing sys-temwas designedforlow subsonicflowconditions.A theoretical studyofthemorphingwingsystemwasperformed[24],andvery promising results were obtained: the morphing systemwas able to delay the transitionlocation downstream by up to30% of the chord,andtoreducetheairfoildragbyupto22%.

Two control approaches were used for providing the optimal SMA actuatordisplacementsforeach differentflightcondition. In the open-loop configuration, the desired displacements were di-rectlyimposedonthesystem[25],whileanovel,adaptive, neuro-fuzzy approach which was used to predict and control the mor-phingwingperformance[26].Intheclosed-loopconfiguration,the displacementswereautomaticallydeterminedasafunctionofthe pressurereadingsfromthewinguppersurface[27,28].Inaddition, twonewcontrollersweredeveloped;thefirstonewasbasedonan optimalcombinationofthebi-positionalandProportional-Integral (PI) laws[29,30],whilethe second onewas ahybrid fuzzy logic-PID controller [31,32]. The wind tunnel tests were performed in the2m

×

3m atmosphericclosedcircuitsubsonicwindtunnelat NRC.The windtunnelmeasurementswere analyzedtoassessthe validityofthenumericalwingoptimizations[33]andthedesigned controltechniques[34].

2. DescriptionoftheCRIAQMDO505project

2.1. Projectinformation

The CRIAQ MDO 505 project is performed as a continuation of the CRIAQ 7.1 project on adaptive upper-surface wing con-cept. In this multidisciplinary project a real industrial wing and aileron (classicalandmorphing)structurewasconsideredand de-signed following structuraland materials optimizations based on newaerodynamicoptimizationconstraintsandnewmorphingskin controlchallenges,usinganelectricalactuationsystemalongwith classicalandadaptiveailerons.

Theresearchpresentedinthispaperwasperformedwithinthe framework of the MDO 505 project, a multiple partners project involving an international collaboration between Canadian and Italian industries, universities and research centers (Bombardier Aerospace, Thales Canada and Alenia Aeronautica, on the indus-try side,École de TéchnologieSupérieure,ÉcolePolytechnique de Montreal andthe UniversityofNaples,ontheacademicside,and NRC andtheItalian Institute forAerospaceResearch CIRAon the researchcenters side).

The purposeoftheCRIAQ MDO505project istodemonstrate thestructural, aerodynamic andcontrol abilitiesofawind tunnel wingmodelequippedwithan adaptiveuppersurfaceandboth a rigid and an adaptive aileron, designedfor low-speed (subsonic) windtunneltests.Thenoveltyofthisprojectconsistsinthe mul-tidisciplinarydesign,analysisandmanufacturingofawindtunnel representativemodelthatrespectsthestructuralandaerodynamic properties ofa realaircraftwing-tip section.The morphing wing model structure was designed to closely follow the real aircraft wing-tip section dimensions, while allowing the insertion of the actuation system. Reynolds number similitude of the morphing wingmodelwiththerealaircraftwasachievedforthelow-speed aerodynamic testsbyusingthe samespanwisechorddistribution astheonerealwing-tipsection,andmaintainingthesamemean aerodynamic chord. Fig. 1presents theposition of themorphing upper skin on a typical aircraft wing, while Fig. 2 presents the structuralelementsofthemorphingwingmodel.

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 1.The layout of the morphing skin on the aircraft wing.

Fig. 2.ThestructuralelementsoftheCRIAQMDO505morphingwingconcept(the morphingskinisnotshowninthefigure).

2.2.Generaldetailsaboutthemorphingwingmodel

Thefull-scalemorphingwingmodelisastructurewitha1.5 m span and a 1.5 m root chord, a taper ratio of 0.72, and leading andtrailing edges sweep angles of 8◦_{. The chorddistribution of}

thewingmodelfollowsthechorddistributionoftherealwing-tip section,whilethesweepangleandthespanwisetwistdistribution were modified in orderto reduce 3D flow effects. The wingbox andits internal structure(spars, ribs,andlowerskin) were man-ufacturedfromaluminumalloymaterial,whiletheadaptiveupper surface was positioned between20% and65% ofthe wingchord. Theadaptiveuppersurfaceskinwasspecificallydesignedand op-timized fortheproject,andwas manufacturedusingcarbonfiber compositematerials[35].

Thedeformation ofskin shape,driven by actuators placed in-side the wingbox structure, isa function of the flight condition (defined in terms of Mach number, Reynolds number and angle ofattack). Theseactuators were specifically designedand manu-facturedto meettheprojectrequirements.Fourelectricactuators wereinstalled ontwo actuationlines;two actuatorseach, placed at37%and75%ofthewingspan,fixedtotheribsandtothe com-positeskin.Eachactuatorhastheabilitytooperateindependently fromtheothers,andhasadisplacementrangebetween

±

3.5 mm. Oneachactuationline, theactuators werepositioned at32% and 48%ofthelocalwingchord.

Theaileron articulation was locatedat 72% ofthechord. Two aileronswere designedandmanufactured.One aileronwas struc-turally rigid, while the other one represented a new morphing aileronconcept.Bothaileronsweredesignedtobeattachedtothe samehingeaxisofthe wingbox,andbothare abletoundergoa

Fig. 3.CRIAQ MDO 505 morphing wing concept.

controlleddeflectionbetween

−

7◦ and

+

7◦.Thisintervalismore restrictedthanthenormaldeflectionrangeofanaileron,butitwas considered sufficientto demonstratethe proof ofconcept forthe morphingaileron.Thisrestrictionwasdeterminedbytheavailable spaceinsidetheNRCwindtunnelandbythemechanicaldesignof themorphing mechanism. Fig. 3presentsa sketchofthe morph-ingwingmodelconceptasitwouldbemountedandtestedinthe NRCsubsonicwindtunnel.

2.3. Thestructuraldesignofthemorphingwingmodel

TheMDO505morphingwingwasdesignedtohaveastructural rigidity similar to a realaircraft wing-tip. This means that when the wingmodel was subjected toin-flight 1g loads, thebending moment,thetorsionalmomentandtheshearstresses atthe ribs were similar to thoseobtained forthe realaircraftwing-tip. The uppersurfacemorphingskinwascreatednotonlytobeanactive structuralelement, rigidlyfixed around its perimeterandableto withstandrealflight loads,butalso toallowthe obtainingofthe required aerodynamic shape changes andactuator displacements whileremainingstructurallyloaded.

TwoFiniteElementModels(FEMs)werecreatedforthedesign process: a simplified, thus general model (GFEM) that was used todesignandoptimizethecarbonfiber uppersurfaceskin,anda detailedmodel(DFEM) thatwas usedforthedesignand numeri-cal analysisoftherigidstructure(lower skin,spars,ribs,internal actuators). The FEM’s were created using the Altair Hyperworks software package, while NASTRAN was used as the FEM solver. Theaerodynamic loadsusedfordimensioningandcalculatingthe structuralelementsofthemorphingwingmodelcorrespondtothe limitloadfactorsof

+

2.5gand

−

1gtypicalofciviltransport avia-tion,multipliedbytheultimatesecuritycoefficientof1.5[35].

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 4.Overview of the morphing wing control system.

Fortheleadingedge,lowerskinandribs,theGFEMwas gener-atedusing 2D CQUAD4 andCTRIA3elements, having the PSHELL property. The flexible upper skin 2D elements had the PCOMP property attributed. The spars and stringers were modeled with 1D BEAM elements, having the PBEAML property. The complete GFEMmodelhadapproximately180000elements.TheDFEM com-prisedalmost4millionelements,allowing toperformamore de-tailednumericalsimulation.A mixtureof1D,2Dand3Delements (CTETRA havingthePSOLIDproperty)wasused,whilebothupper andlowerskinsweremodeledusingonly2DCQUAD4andCTRIA3 elements,butwithmuchhighermeshdensitythantheGFEM[35]. The flexible skindesignand optimizationwere performed us-ingtheAltairOptistructsolver,whiletrying tomatchstructurally ascloseaspossiblethe aerodynamicallyoptimized uppersurface shapes[35].Theoptimizationapproachwas done inthreestages. Thefree-sizeoptimizationfocusedondeterminingtheglobal thick-nessoftheskinandofthecomposite plies.Thedimensional opti-mizationrefinedthepliesthicknessesasfunctionoftheir orienta-tionandtopography. The finalstage was the optimizationsofthe plieslayoutasfunctionofmanufacturingconstraints.

An erroranalysis performedfor a numberof optimizedcases showedthat theaverage shape error betweenthe skin FEM and the spline target shapes was 0.25 mm, or 7% of the maximum actuator displacement. Given the multidisciplinary nature of the MDO505projectandthehighnumberofstructuralrequirements onthecarbonfiber skin, agoodreproductionofthedesired opti-mizedshapeswasnumericallyobtainedusingtheFEManalysis. 2.4. Thewingmodelcontrolsystem

The core of the wing control system was the embedded real timecontrollerPXI-e8135ofNationalInstruments.Thiscontroller ran on a real time operating system, and was connected to all the system hardware peripherals through several input and out-putmodules.Allfourupperskinactuators(BLDCmotors),therigid aileronactuator,the LVDTsensorsforprovidingtheactuators po-sitionsfeedback,aswellastheupperskinKulitepressuresensors wereconnectedtothePXI-e8135system.Thecontrollerwas mon-itored by the host PC via an Ethernet network using the TCP/IP

communication protocol, which hada staticIP address that was personalized andfixed by the systemoperator. The Windows OS machine(thehostPC)servedforthecontrolprogramdeployment, system state control, and data monitoring inreal time. All com-municationtasks,controlanddataloggingwereentirelyoperated bythePXI-e8135,whichranindependentlyofthehostPC.Fig. 4

presentsan overviewofthe integratedcontrolleranddata moni-toringsystemofthemorphingwingmodel.

3. Flowequations,turbulenceandtransitionmodels

CFD simulations were performed to simulate the flow past the wing under the wind tunnel test flow conditions and set up. The dynamics of fluid flow are governed by the Navier– Stokes equations, which represent the fundamental principles of mass, momentum and energy conservation. For turbulent flows, theReynoldsAveragingtechniqueisusedtodecomposethe instan-taneousflowvariablesintotheiraveragevaluesandturbulent fluc-tuations,whiletheBoussinesqeddy-viscosityhypothesisisusedto relate theReynoldsstresstensorandturbulentheat fluxtermsto the averageflow variables.Withtheseassumptions, theReynolds Averaged Navier–Stokes(RANS) equations(withReynolds average flowvariables)canbewrittenasfollows:

∂

ρ

∂

t

+

∂

∂

xj

(

ρ

Uj

) =

0 (1)

∂

∂

t

(

ρ

Ui

) +

∂

∂

xj

(

ρ

UjUi

)

= −

∂

P

∂

xi

+

∂

∂

xj



μ

eff

 ∂

Ui

∂

xj

+

∂

Uj

∂

xi



−

2 3

μ

eff

∂

Uk

∂

xk

δ

_{i j}



(2)

∂

∂

t

(

ρ

H

) −

∂

P

∂

t

+

∂

∂

xj

(

ρ

UjH

)

=

∂

∂

xj



λ

∂

T

∂

xj

+

μ

t Prt

∂

h

∂

xj



+

∂

∂

xj



Ui



μ

eff

 ∂

Ui

∂

xj

+

∂

Uj

∂

xi



−

2 3

μ

eff

∂

Uk

∂

xk

δ

i j



+

μ

∂

k

∂

xj



(3)

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

where

ρ

is the fluid density, Ui are the velocity components, P isthesum ofthestaticpressure andthe

(

2

ρ

δ

i jk

)/

3 term result-ingfromtheBoussinesqassumption,

μ

eff istheeffectiveviscosity, givenbythe sumofthemolecular viscosity

μ

andtheturbulent viscosity

μ

t,

δ

i j istheKroneckerdeltafunction, Histhetotal en-thalpy, T is the fluid temperature,

λ

is the thermalconductivity, Prt istheturbulentPrandtlnumber,histhestaticenthalpy andk istheturbulentkineticenergy.

Theturbulent viscosityandthekinetic energyare determined using the k–

ω

Shear Stress Transport (SST) model [36]. The SST modelrepresentsablendofthek–

ω

model,usedinthenearwall region, and the k–

ε

model, used for the rest of the flow. Thus, itachievesbothaccurateboundary layerrepresentationuptothe viscoussub-layerandinsensitivitytoboundaryconditionsat free-stream flow. The SST turbulence model equations are presented below:

∂

∂

t

(

ρ

k

) +

∂

∂

xj

(

ρ

Ujk

) =

Pk

−

Dk

+

∂

∂

xj



(

μ

+

σ

k

μ

t

)

∂

k

∂

xj



(4)

∂

∂

t

(

ρω

) +

∂

∂

xj

(

ρ

Uj

ω

)

=

γ

μ

Pk

−

β

ρω

2

₊

∂

∂

xj



(

μ

+

σ

ω

μ

t

)

∂

ω

∂

xj



+

2

(

1

−

F1

)

ρσ

ω21

ω

∂

k

∂

xj

∂

ω

∂

xj (5) where k is the turbulent kinetic energy, Pk is the turbulent ki-netic energyproductionterm, Dk is theturbulent kinetic energy destructionterm,

ω

isthespecific turbulencedissipation rate, F1

isablendingfunctionrelatedtotheSSTmodel,and

γ

,

β

,

σ

k,

σ

ω and

σ

ω2 aretheconstantsofthemodel.Theturbulentviscosityis

calculatedas:

μ

t

=

min



ρ

k

ω

,

a1

ρ

k S F2



(6) wherea1 is adampingcoefficient, S isthestrain ratemagnitude andF2isasecondblendingfunctionrelatedtotheSSTmodel.

Inorderto includethe effectsof laminarflow andmodelthe laminar-to-turbulenttransitionprocess,the

γ

–Reθt modelisused

[37]. Thismodelincludes two more equations,one forthe inter-mittencyandoneforthetransitionmomentumthicknessReynolds number:

∂

∂

t

(

ργ

) +

∂

∂

xj

(

ρ

Uj

γ

) =

Pγ

−

Eγ

+

∂

∂

xj



μ

+

μ

t

σ

f

 ∂

γ

∂

xj



(7)

∂

∂

t

(

ρ

Reθt

) +

∂

∂

xj

(

ρ

UjReθt

) =

Pθt

+

∂

∂

xj



σ

θt

(

μ

+

μ

t

)

∂

Reθt

∂

xj



(8) where

γ

istheintermittency, Pγ isthe intermittencyproduction term, Eγ is the intermittency destruction/relaminarization term, Reθt is thetransitionmomentumthicknessReynoldsnumber, Pθt isthe transition momentum thickness Reynolds number produc-tionterm,and

σ

f and

σ

θt aremodelconstants.

The transition onset is controlled by an empirical correlation between Reθc, the critical Reynolds number where the intermit-tencystarts to increasein theboundary layer,andReθt [37].The modelcontainscorrectiontermstoaccountforlaminar separation-inducedtransitionandstrongpressure-gradientflows.Couplingof the

γ

–Reθt transitionmodel withthek–

ω

SST turbulencemodel is done by modifying Pk and Dk, the turbulent kinetic energy productionand destructionterms, andthus deactivating the tur-bulencemodelforthelaminarboundarylayerregion.

The numericalcomputations were performed withthe ANSYS FLUENTsolver [38]. The steady-stateflow equationswere solved

usinga projection method,achievingtheconstraintofmass con-servation by solving the pressure equation, with the pressure-velocity couplingaccomplishedby usinga high-order Rhie–Chow scheme.Thecell-facevaluesofthepressurewere interpolated us-ingasecond-ordercentraldifferencingscheme,whileforallother variables,includingtheturbulenceandtransitionmodelequations, a second-order upwindscheme was used. The discrete nonlinear equationsweresolvedinafullyimplicit,coupledmanner. Conver-gence acceleration was achieved witha coupled algebraic multi-grid (AMG) approach, using a block-method Incomplete Lower– Upper(ILU)factorizationschemeasthelinearsystemsmoother.

4. Morphedgeometriesandmeshgeneration

4.1. Thetheoreticaloptimizeduppersurfaceshapes

The core concept of an active morphing of the wing upper surfaceistoprovideanoptimizedairfoilshapeforeachflight con-dition. A single point optimization must be performed for each combinationofMach number, Reynoldsnumberandangle of at-tack.Thisprocedureincreasestheaerodynamicperformanceofthe shape-changing airfoil (with respect to the desired optimization objective)comparedtothemulti-pointdesignedbaselineairfoil.

Aerodynamic optimizations were performed to determine the actuator-driven displacements required to improve the perfor-mance of the morphing wing with respect to the original wing. Inorderto reducegreatly calculationtimes,theaerodynamic op-timizationswere performedundertwo-dimensionalflow assump-tionusingtheXFOILsolver[39]andanin-housegeneticalgorithm optimizer [40], forlocal flow conditions (local Reynolds number andangleofattack)correspondingtothemeanaerodynamicchord ofthewingmodel[41].

For the numerical optimizations, the upper skin shapes were approximatedusingcubicsplines, asfunction oftheactuator dis-placements. This mathematicalmodel was chosen because it en-forces the tangency condition with the rigid part of the airfoil (up to the curvature continuity given by the second derivative), it provides an iso-arc-length condition and it shares mathemati-calpropertieswithabeambendingunderanappliedload.Dueto constraintsrelatedtostructuralrigidityofthecompositeskin,the actuatordisplacementswerelimitedto

±

3.5 mm,whilethe max-imum difference between the two displacements was limitedto 6 mm.

4.2. Measurementoftherealuppersurfaceshapes

Due to the high degree of multidisciplinary involved in the development of the MDO 505 morphing wing project, the con-tradictory requirements that the morphing upper surface had to satisfy (rigid in orderto withstandflight loads, but at the same timeflexibleenoughtoallowpropercontrolleddeformations)and theveryhighprecisionrequiredfortheaerodynamicoptimization, it was decided to scan the shapesobtained afterthe completion ofthemanufacturingprocess.Thisway,thesimulations wouldbe performedon geometries that were practically achieved, andnot onsurfacesreconstructedusingmathematicalmodeling.

To construct the geometries required for the 3D calculations, the real shapes of the morphing skin surface for all flight cases were scanned using a highprecision photogrammetry procedure, utilizingthree3D-trackingcameras(Prime41fromNaturalPoint)

[42]. Circular retro-reflective markers were applied on the wing upper surface, and their positions were recorded for each skin shape. Fig. 5 presents the marker positions for the un-deformed skin, as measured with the scanning procedure. The estimated maximum position error with this procedure is 0.07 mm, using theknownpositionsofthefouractuatorsaxes.

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 5.Marker positions for the un-deformed upper skin.

Toincrease the resolution ofthe scanned data,a bi-harmonic splineinterpolationprocedurewasperformedbetweenthemarker positions,andthenumberofpoints wasincreasedto100inboth chord-wiseandspan-wisedirections.Fig. 6presentsthe100

×

100 grid points obtained by interpolation for the un-deformed skin (upperimage),andthedeformedshapeoftheflexibleskinforthe C41configuration(lowerimage).

Theprocedureofdeterminingthemarkerpositionsthroughthe photogrammetrytechniqueandtheninterpolatingusingsplinesto increase thedensityofsurface points was repeatedforall ofthe morphing cases. The data was further used to construct the ge-ometries requiredfor the 3D calculationsby patching the upper surfaceskinshapesontherigidgeometryrepresentingtherestof thewingmodel.Anaccuraterepresentationoftherealskinshapes wasthereforeavailableforperformingthenumericalsimulations. 4.3. Comparisonbetweentargetandobtainedshapes

Followingthe reconstructionof themorphedwinggeometries using the scan data, comparisons were performed between the target skin shapes (obtained after the aerodynamic optimization process)andtherealskinshapes. Fig. 7presentsthe comparison betweenthetargetandobtainedshapesforCase75.

The comparison was made at four stations along the wing span, located at 0.55 m, 0.75 m, 0.95 m and 1.15 m, as mea-sured from the wing root section. The first and the last section correspond to the positions of the two ribs where the electrical actuatorswereinstalled.From Fig. 7itcanbe seenthatforthese two sections thereis an excellent agreement betweenthe target shapesobtainedfromthenumericaloptimizationandtheobtained skin shapes. The other two sections correspond to positions be-tweenthetworibs.Forthesesections,thereisamoresignificant difference betweenthe desired and the scanned shapes. At cer-tainpoints,thisdifferencecanreachmagnitudesofapproximately 2 mm, thus an important percentageof the maximum displace-mentrangeoftheelectricalactuators(

±

3.5 mm).

Sincethe three-dimensionalnumericalstudyis performed us-ing the scanned upper surface shapes, this difference does not have an impact on the numerical versus experimental compar-isonsintermsoflaminar-to-turbulenttransitionlocation.However, theperformanceofthemorphedgeometrieswithreferencetothe original, un-deformed wing is expected to be smaller than pre-dicted,sincethetargetedoptimalshapesarenotobtainedforthe entirewingspan.

4.4. Gridconvergencestudy

Thestructuredmeshesusedforthenumericalsimulationwere generatedusingtheICEM-CFDsoftware.Agridconvergencestudy wasperformedinordertoevaluatethemeshdensityrequiredfor

Fig. 6.Interpolatedpointgridconstructedfromthescannedmarkerpositionsforthe un-deformedupperskin(upper)andtheCase41morpheduppersurface(lower). (Forinterpretationofthereferencestocolorinthisfigure,thereaderisreferredto thewebversionofthisarticle.)

Table 1

Detailsaboutthefourgeneratedmeshes. Mesh type Chord-wisecells

onwall Span-wisecells onwall Maximum y+ Coarse 100 40 2.66 Medium 200 80 1.33 Fine 400 160 0.66 Extra fine 800 320 0.33

grid-independentaerodynamic coefficientsvalues.Fourmeshesof increasingcelldensityweregenerated,andeachonewasanalyzed at a Mach number of 0.15, a Reynolds number of 4.53E

+

06 (as calculated withthe wingmeanaerodynamic chord)andan angle ofattack of0◦_{.Thedetails regardingthe wallcelldensityforthe}

generatedmeshesarepresentedinTable 1.

The wingaerodynamic coefficientsvalues(lift, dragand pitch-ingmomentcoefficientabouttherootsectionquarterchordpoint) andthetransitionpointlocationsontheuppersurface,at37%and 75% of thespan stations arepresented inTable 2.The transition point locationswere determined usingthe intermittencyvariable

γ

distribution.Thetableshowsthatthedifferenceinaerodynamic coefficientvaluesbetweentheFinemeshlevelandtheRichardson extrapolation ofthe convergence studyis lessthan 1%, andthus the Finemeshprovidessufficiently accurateresults.Itcan be ob-served that the

γ

–Reθt model requiresa goodstream-wise mesh refinementlevelbeforethegridconvergenceofthetransitionpoint locationisachieved(thusalsoaffectingthegridconvergenceofthe drag coefficient, through thevariation ofthe laminarflow region length).

The characteristics of the meshesused to perform the simu-lations were determined basedon theresultsof thegrid conver-gencestudy.Inordertoensurethatthesamemeshingparameters

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 7.Comparisonsbetweentheuppersurfaceskinshapesofthetargetairfoilandofthescannedairfoil,atfourstationsalongthewingspan:Y=0.55m (a),Y=0.75m

(b),Y=0.95m (c)andY=1.15m (d).

Table 2

Resultsobtainedforthegridconvergencestudy.

Meshtype CL CD Cm Transitionat37%of

span(%oflocalchord) Transitionspan(%ofatlocal75%chord)of

Coarse 1.531E–01 1.308E–02 −9.235E–02 13.4% 3.4%

Medium 1.587E–01 9.855E–03 −9.264E–02 48.2% 32.8%

Fine 1.593E–01 9.621E–03 −9.273E–02 57.5% 36.9%

Extrafine 1.596E–01 9.609E–03 −9.274E–02 58.0% 37.1%

Richardson extrapolation

1.597E–01 9.605E–03 −9.276E–02 58.2% 37.1%

wereusedforallthemorphedwingcases,anautomaticmesh gen-erationprocedurewasimplementedbycreatingascripttobeused for the ICEM-CFD code. The automatic procedure can also han-dlerigidailerondeflections between

±

7◦. Themesheswere con-structedbasedontheFinemeshlevelcreatedfortheconvergence study,andinclude400cellsaroundthewingsection(200 cellson both the lower and upper surfaces), and 160 cells in the direc-tionofthespan(80 cells onboththe loweranduppersurfaces). The wall normal spacing was set to 3.0E–06 m, refined enough to provide the required y

+

<

1 condition. Figs. 8 and 9 present two cross-sectionviews ofthe meshconstructed around the

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 9.Spanwise cross-section view of the mesh.

Fig. 10.MDO 505 wing model setup in the wind tunnel test section.

5. Experimentaltestinganddataacquisition

The wind tunnel testswere performedat the 2 m

×

3 m at-mospheric closed circuit subsonic wind tunnel of the National ResearchCouncil Canada. The tunnel hasa maximum speed cor-respondingtoaMachnumberof0.40,atatmosphericpressure.

Theuppersurfaceflexibleskinwasequippedwith32high pre-cisionKulitepiezoelectric-typetransducers[43] forpressure mea-surement on the flexible skin and then processed to determine the laminar-to-turbulent transition location. These sensors were installed in two staggered lines (with 16 Kulite sensors on each line),situatedrespectivelyat0.600mand0.625 mfromthewing rootsection. InadditiontotheKulitepiezoelectric sensors,atthe sametwospanwisestations,60staticpressuretapswereinstalled (30 taps on each line), on the wing leading edge, lower surface andaileron,thusprovidingcompleteexperimentalpressure distri-butionaroundthewingcrosssectionat40%ofthewingspan.

The experimental measurements also included the use of a wakerakepressureacquisitionsystem,tomeasurethewingprofile drag atdifferentspan-wisepositions,andtheuse ofawind tun-nel balancefor measuring theaerodynamic forcesandmoments.

Fig. 10 presentstheMDO 505morphing wingmodel installedin the tunnel test section, viewed fromboth the leading edge (left figure)andthetrailingedge(rightfigure).

Infra-red (IR) thermography camera visualizations were per-formed for capturing the transition region over the entire wing model surface. The wing leading edge, its upper surface flexible skin and the aileron interface were coated with high emissivity

black paint to improve the quality of the IR photographs. The span-wise stationswherethetwopressuresensors lineswere in-stalled were not painted, in order to not influence the pressure measurements.A JenoptikVariocamcamera[44],witharesolution of 640by 480pixels, was used tomeasure the surface tempera-tures.Thiscamerawasequippedwith60◦ _{lensinordertocapture}

theflowtransitionontheentireuppersurfaceofthewing.A cus-tomwoodenwindowwasinstalledonthewindtunneltestsection wall,throughwhichtheIRcameraoperated.

The IR thermography visualization allows the identificationof the laminar-to-turbulent transitionregion basedon the tempera-turegradientbetweenthetwoflowregimes,whichisdetermined bythedifferentconvectiveheattransfercoefficientsandheatflux dissipationexistinginthetworegimeswhenthesurfaceisheated to a fixedtemperature. Fig. 11 presentsan example oftheIR vi-sualization of the wing model upper surface transition, for one flight condition (Mach numberof0.15, angle ofattack of 1◦ _and

no aileron deflection) andfor both un-morphed (left figure)and morphed(rightfigure)skinshapes.

Theblacklinerepresentstheaveragetransitionlineonthe up-persurface,anditsvariationasfunctionofthespan-wiseposition canclearlybeobserved.Thetwodashedwhitelinesrepresentthe estimated extentofthe transitionregion, determinedas function ofthe chord-wisetemperaturegradientexisting betweenlaminar and turbulent regimes. The transition from laminar to turbulent flowoccursoveranarrowregionanditwasautomaticallydetected forthewinguppersurfaceusingaMATLABcodethatwas specif-ically developed for the IR images post-processing [44]. The red dot correspondstotheestimatedtransitioninthespan-wise sec-tion situatedat0.612 mfromtherootsection(40%ofthemodel span), thatishalf-waybetweenthetwo Kulitepiezoelectric pres-suresensorslines.Theaccuracyofthetransitiondetectionforthis section was estimated to

±

2% of the local chord, based on the knownKulitepositionsandtheirthermalsignaturesintheimages.

6. Resultsanddiscussion

6.1. Thetestcases

The two-dimensional aerodynamic optimizations that deter-minedtheelectricalactuatorsdisplacementswereperformedwith theobjectiveofcontrollingtheextentoflaminarflowonthe up-persurfaceofthewingmodel.

These optimizations were performed for several flight condi-tions (expressedintermsofMachnumber,Reynoldsnumberand angle of attack) and several rigid aileron deflection angles. The casesthatwereoptimized,analyzed andexperimentallytestedfor laminarflowincreasearepresentedinTable 3.TheReynolds num-bers that correspond to the two Mach numbers are 4

.

28

×

106 and 5

.

27

×

106. A downwards aileron deflection was considered positive, while an upwards aileron deflection was considered as negative.

Foreachcase,thetransitionpointlocationonthepressure sen-sors linewasdetermined fromthe numericalsimulation andwas compared totheexperimentallymeasured transitionlocation, de-terminedusingtheIRthermography.Thetransitionpointlocation in the numerical results was determined by plotting the turbu-lence intermittency

γ

versus the local chord, forthe upper and lowerwingsurfaces.Inordertoconsistentlyextractthetransition location, the first derivative of the intermittency plot was used. Since theintermittencyisapproximatelyconstantforthelaminar boundary layerand itsvalue significantly increasesin the transi-tionregion,thefirstderivativecanbe usedtoidentifythisregion of high gradient. The transition point was considered to be the most upstream point wherethe derivative becomes non-zero. As an example,Fig. 12showsthe intermittencydistribution at0.612

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 11.IRvisualization ofthelaminar-to-turbulenttransitionregionontheuppersurfaceforbothun-morphed(left)andmorphed(right)skinshapes.(Forinterpretationof thereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)

Table 3

Testcasesoptimizedforlaminarflowimprovement.

Mach Delta [◦_] Angle of attack [◦_] 0 0.50 0.75 1.00 1.25 1.50 2.00 2.50 3.00 4.00 5.00 0.15 0 – – C39 C40 C41 C42 C43 C44 C45 – – 0.20 4 C68 C69 – C70 – C71 C72 C73 – – – 0.20 −4 C74 C75 – C76 – C77 C78 C79 C80 C81 C82

m span-wise section, for case C39 un-morphed. The laminar-to-turbulenttransitioncorrespondstotheregionofhighgradient. 6.2.Uppersurfacetransitionlocation

Fig. 13 presentsa comparison betweenthe predictedandthe measured transition location for the un-morphed and morphed wingupper surface skin,at a spanwise station corresponding to 40%ofthewingspan.Thecomparisonshowsbothnumericaland IRexperimentalresultsforcasesC39toC45(Machnumberof0.15, noailerondeflectionandanglesofattackbetween0.75and3◦_).

Fig. 14 displays the experimental transitionlocation measure-mentcomparedtothenumericalpredictionsforcasesC68toC73 (Machnumberof 0.20, 4◦ _{downwards aileron deflectionand}

an-glesofattack between0and2.5◦_{). No IRexperimental data was}

availableforcaseC68(0◦_{angleofattack).}

InFig. 15,the experimental and numericaltransition location detectionforcasesC74toC82(Machnumberof0.20,4◦_upwards

ailerondeflection andangles ofattack between0and5◦_{) is}

pre-sentedforbothun-morphedandmorphedwinggeometries.NoIR experimentaldatawas availableforcasesC74(0◦ _{angleofattack)}

andC80(3◦_{angleofattack).}

InFig. 13,itcanbeseenthatareasonableagreementexists be-tweentheexperimentalandthenumericallydeterminedtransition pointlocationatthepressuresensorssectionfortheun-morphed wing.Forthesecases(C39toC45,withnoailerondeflection),the

Fig. 12.TransitiondetectionforCase39un-morphedusingtheturbulence intermit-tencydistribution.

un-morphed wing error is around 5% of the local chord (corre-spondingto0.05Cinthefigure).TheIRexperimentalresultsshow asuccessfulimprovementoflaminarflowforthesectionof inter-est. The transitionis delayed towards the trailing edge by 3–5% of the chord (equivalent to 0.03–0.05C in the figures). For the

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 13.ComparisonbetweennumericalandIRexperimentaltransitiondetectionfor thestationlocatedat40%ofthespanfortheun-morphedandmorphedwingsof casesC39–C45.

Fig. 14.ComparisonbetweennumericalandIRexperimentaltransitiondetectionfor thestationlocatedat40%ofthespan,fortheun-morphedandmorphedwingsof cases C68–C73.

morphed geometries results presented in Fig. 13, the agreement betweenthe numerical andIR transitionpositions is better than fortheun-morphedwing,eventhoughthepredictedperformance improvementissmaller.

InFig. 14(casesC68toC73,havinga4◦_{ailerondeflection),for}

angles ofattack smallerthan 1◦_{,there isa very goodagreement}

between numerical versus experimental results obtained for the un-morphedwing.Thediscrepancyisseentoincreaseforanglesof attack higherthan 1.5◦_{, astheexperimental measurements show}

anearlyshiftofthetransitionoccurrencetowardsthewingleading edge.Again,asuccessfulimprovementoflaminarflowisobserved, withdelaysof9% ofthe chordobtainedfortwo angles ofattack values(1.5◦ _and2◦_{).ThemorphedgeometriespresentedinFig. 14}

show a very good level of agreement betweennumerical andIR experimental results,witherrorsofmaximum 2–3%of thechord (equivalentto0.02–0.03Cinthefigure).

For cases C74 to C82 (

−

4◦ aileron deflection), presented in

Fig. 15,thereisarathergoodagreementbetweentheIRdataand thenumericalresultsfortheun-morphedwing(transitionposition

Fig. 15.ComparisonbetweennumericalandIRexperimentaltransitiondetectionfor thestationlocatedat40%ofthespan,fortheun-morphedandmorphedwingsof casesC74–C82.

errorsofaround5%ofthechord),butthedifferencesarehigherfor themorphedwinggeometries.Fortheanglesofattackbetween1◦

and3◦ _{the laminarflow delay predictedby thenumericalresults}

isnotobservedintheIRmeasurements.

Since allof theabove presentedresultswere obtainedforthe sectionlocatedat40%ofthespan,theyonlyofferlocalinformation aboutthe performance ofthemorphing upperskin.Toprovide a qualitativeassessmentoftheskin’sinfluenceonthetransition re-gionfortheentireuppersurface,2DplotsarepresentedinFigs. 16 to 18, for cases C40 (Fig. 16), C72 (Fig. 17) and C77 (Fig. 18), respectively.Thesecaseswerechosen amongthoseforwhich im-portant transition location delayswere observed on the pressure sensors span-wise section (asshown inFigs. 13to 15),andthey cover all three aileron deflection angles. This choice verification whethertheextensionoflaminarflowwasaphenomenonpresent ontheentireuppersurface,orifitwas limitedtoacertain span-wiseinterval.

Inthenumericalresults,thedisturbancesintransitionposition appearing nearthe wingrootsection were givenby the6.5 mm gapbetweenthewingrootribandthesymmetryplane.Thisgap was presentintheexperimentalsetup andincludedinthe simu-lations.ItseffectwasnotcapturedwiththeIRmeasurementsdue tothedecreaseindataqualityintheregionclosetothewingroot section.

For the wing tip region, the precision of the numerical sim-ulations breaks down and an unrealisticlaminar flow appears in all theresults. Thiscanbe explainedby the factthat the

γ

–Reθt model contains one empirical correlation forthetransition onset that was calibrated especially for naturaltransition (stream-wise Tollmien–Schlichtinginstabilities) andlaminarseparationbubbles, while in reality the wingtip region isstrongly contaminated by complex, cross-flow instabilities induced by the presence of the wingtipvortex.

An analysisof Figs. 16 to 18 shows that the behavior of the laminar flow region(under the actuationof theupper skin)that wasobservedfromthepressuresensorsline(indicatedbythered dot inthe experimental IRdata andby theblackline inthe nu-mericalresults)canalsobeobservedforotherspan-wisesections. Thus,whenasuccessfultransitiondelaywasobtainedforthe pres-suresensorsline,thisdelaycanbeseenoccurringnotonlylocally butalsooverahighpercentageofthewing’sspan,indicatingthe effectivenessoftheuppersurfacemorphingskin.

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 16.Comparisonbetweenexperimentalandnumericaltransitionlocationon thewinguppersurfaceforcaseC40,forbothun-morphed(left)andmorphed (right) geometries.(Forinterpretationofthereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)

6.3.Pressurecoefficientdistributioncomparisons

Acomparison between the experimental and numerical pres-surecoefficient distributionsforthesection locatedat40%ofthe wingspanispresentedinFigs. 19to22,forthefollowing4 cases: C40(Mach number of0.15, angle of attack of1◦ _{and no aileron}

deflection),C68 (Machnumberof0.20, angleofattack of0◦ _and

4◦_{ailerondeflection)andforC79andC82(Machnumberof0.20,}

anglesofattackof2.5◦ _and5◦_and

₋

₄◦ _{ailerondeflection).}

Goodagreement exits betweennumerical predictions andthe wind tunnel measurements for the two sets of results given by caseC40 and C68 (Figs. 19 and 20). The influence of the upper skinshapechangecan beobserved fromthe differencesbetween the un-morphed (left) and morphed (right) pressure coefficient

distributions, forthechordwise intervalbetween25% and60% of the chord. The skin morphing extends the region where the air accelerates over the upper surface, thus creating more favorable conditionsforlaminarflow, thiseffectbeingclearly visibleinthe twofigures.

ForcasesC79andC82(showninFigs. 21 and22),asmall dif-ference existsinthe uppersurfacepressure coefficientup to50% ofthechord,andverygoodagreementexistsbetweenthe numer-icalandexperimental resultsfortheaileron, rigidlowerskinand theuppersurfacedownstreamof50%ofthechord.Again,the in-fluence of the morphing skin is clearly observable by comparing theleft (un-morphed)andright(morphed) pressuredistributions forthetwofigures.

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 17.Comparisonbetweenexperimentaland numericaltransitionlocationonthe winguppersurfacefor caseC72,for bothun-morphed(left)and morphed(right) geometries.(Forinterpretationofthereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)

6.4. Aerodynamiccoefficientscomparison

InTables 4 and 5,acomparisonis madebetweenthe liftand dragcoefficientsfortheun-morphedandmorphedgeometries, ob-tained through the numerical simulations and the experimental tests. The comparison is presented for cases C38 to C45, which were analyzed ata Machnumberof0.15andhadnoaileron de-flection.

The numericallift coefficient valuesin Table 4were found to be in good agreement with the experimental values included in

Table 5,a small underestimation being observed forthe 2.5 and 3◦ _{anglesofattack.Concerningthedragcoefficient,thenumerical}

valuesare always under-predicted comparedto the experimental ones, theaverageerrorbeingaround 25%. Theexperimental drag

coefficientdatashowsthatthemorphingoftheuppersurfaceskin caused a reduction of the wing model drag coefficient, with re-ductions between 0.20% and0.60%, for all analyzed cases, while the numericalsimulationsdid notcapturethisreduction.Table 6

shows the detailed differences obtained between the numerical andexperimentalcoefficients.

Another comparison was done betweenthe lift anddrag co-efficients obtainedforthe un-morphedandmorphed geometries, throughthenumericalsimulationsandtheexperimentaltests,and is presented in Tables 7 and 8. Cases C68 to C73, analyzed at a Mach numberof 0.20 andwith a 4◦ _{aileron deflection were}

in-cluded in the comparison. Details aboutthe differencesobtained betweenthenumericalandexperimentalresultsareshownin Ta-ble 9.

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 18.Comparisonbetweenexperimentalandnumerical transitionlocationonthewinguppersurfaceforcaseC77,for bothun-morphed(left)andmorphed (right) geometries.(Forinterpretationofthereferencestocolorinthisfigure,thereaderisreferredtothewebversionofthisarticle.)

Thequalitative behavior oftheresultsremains thesameasfor casesC39toC45.Agoodagreementbetweentheexperimentaland numericallift coefficients, anda 20–25% under-estimationof the numericallycalculateddrag,comparedtotheexperimentalvalues. Theupperskinmorphing determines0.15–0.40% reductionofthe wingdragcoefficient,asconfirmedbytheresultsshowninTable 7. Thenumericallycalculateddragcoefficient forthemorphedwing was higher than the value calculated for the un-morphed wing, thusnotpredictingthereductioneffectobservedexperimentally.

Tables 10 and 11 show the comparison betweenthe lift and drag coefficientsforthe un-morphed andmorphedwing geome-tries,using theresultsthat were obtainedthrough thenumerical simulations and the experimental test. This comparison is pre-sentedforcasesC74 toC82,analyzed at a Machnumberof 0.20 andanailerondeflectionangleof

−

4◦.

Concerningthecomparisonbetweenthenumericaland exper-imental results, the remarks made in the paragraph above also applyforcasesC74toC82.Thereisanunder-estimationofthe cal-culateddragcoefficient,andthereisabetteragreementinthecase ofthelift.Theimpactoftheuppersurfaceskinmorphing onthe dragcoefficientwasnotuniform.Insomecasesaslightreductions of up to 0.67% were obtained, while for others,an increase was obtained. These drag variations were presentin both experimen-talandnumericalresults.Table 12showsthedetaileddifferences obtainedbetweenthenumericalandexperimentalresults. 6.5. Generalobservationsaboutthemorphingconcept

The drag coefficient reductions obtained following the wing tunnelexperimental testingarerelatively small.Itmustbe noted

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132

Fig. 19.Comparisonofexperimentalversusnumericalpressurecoefficient distribu-tionforcaseC40correspondingtoun-morphed(left)andmorphed(right)wing.

Fig. 20.Comparisonofexperimentalversusnumericalpressurecoefficient distribu-tionforcaseC68correspondingtoun-morphed(left)andmorphed(right)wing.

that theaerodynamic loads balancethat was used during testing wasaveryhighprecisioninstrument,withdragmeasurement un-certaintiesthatwereestimatedtobeoftheorderof0.10–0.15%of thedrag values.Thus,forthecaseswherethedrag reduction be-tweenthemorphedandun-morphedgeometrieswas higherthan themeasurementerror,themorphinguppersurfacehasshownthe abilityofgeneratingasmalllevelofquantifiableimprovements.

Upper skin morphing reduces the friction drag coefficient throughtheextension ofthelaminarflowregion.Dueto itsvery low aspect ratio, the wing model gives a poor performance in terms of the lift-induced drag, which has a much higher contri-butionto thetotal drag thaninthe caseofa typicalhighaspect ratiowing(thecompletewingofanaircraft).Thus,whenthe fric-tiondrags’percentagecontributiontothetotaldrag ishigher(as, forexample,ahighaspectratiowingduringcruiseflight),thedrag reductionobtainedbytheconceptcouldbehigher.

Duringthedesignphaseoftheflexiblecompositeskin,special attentionwas givento thereduction ofthe weightofupper sur-face,in comparisonwiththe originalaluminum design.This was

Fig. 21.Comparisonofexperimentalversusnumericalpressurecoefficient distribu-tionforcaseC79correspondingtoun-morphed(left)andmorphed(right)wing.

Fig. 22.Comparisonofexperimentalversusnumericalpressurecoefficient distribu-tionforcaseC82correspondingtoun-morphed(left)andmorphed(right)wing.

achieved by varying the skin thickness as function of the local stress and by using composite material stringers (instead of the originalaluminumstringers).Thecompositeskinweightwasmore than2kglighterthantheoriginalaluminumskin,andthisweight reduction perfectlycompensatedthe introductionofthe 4 actua-torsinsidethewingbox,whoseoverallweightwasapproximately 2 kg. Thus, the difference in terms of weight between the non-morphingandmorphingstructurewasnegligible.

InthepresentmultidisciplinaryCRIAQMDO505project,an in-dustrial wing-tipwas designedandmorphed,andforthisreason, severalstrictstructuralconstraintsandupperskinbehavior uncer-taintiesexisted, whichwereaddedtotheerrorsexistingbetween numerical models/simulations and experimental results. Detailed testsmustbeperformedinordertoquantifytheimpactoftheskin shapevariations (differencesbetweentargetandobtainedshapes) onthequality ofthelaminarflowoptimizations.Inorderto con-tinueperformingothertypesofresearchstudiesonuppersurface morphing wings and aero-structuraloptimizations, it will be

in-1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132 Table 4

Comparisonbetweenthenumericalun-morphedandmorphedwingliftanddragcoefficientsfor casesC39toC45.

Numerical results

Case Angleof

attack[◦_]

Un-morphed wing Morphed wing Drag

variation[%] CL CD CL CD C39 0.75 0.2058 0.0118 0.2059 0.0118 −0.01% C40 1 0.2191 0.0126 0.2196 0.0126 0.24% C41 1.25 0.2325 0.0134 0.2330 0.0134 0.18% C42 1.50 0.2460 0.0142 0.2464 0.0142 0.10% C43 2 0.2729 0.0161 0.2736 0.0161 0.19% C44 2.50 0.3002 0.0183 0.3009 0.0183 0.13% C45 3 0.3276 0.0206 0.3278 0.0206 0.18% Table 5

Comparisonbetweentheexperimentalun-morphedandmorphedwingliftanddragcoefficientsfor casesC39toC45.

Experimental results (loads balance measurements)

Case Angleof

attack[◦_]

Un-morphed wing Morphed wing Drag

variation[%]

CL CD CL CD

C39 0.75 N/A N/A N/A N/A N/A

C40 1 0.2150 0.0156 0.2165 0.0156 −0.20% C41 1.25 0.2324 0.0168 0.2329 0.0167 −0.47% C42 1.50 0.2483 0.0180 0.2490 0.0178 −0.51% C43 2 0.2794 0.0206 0.2788 0.0204 −0.60% C44 2.50 0.3102 0.0235 0.3109 0.0234 −0.40% C45 3 0.3434 0.0267 0.3424 0.0266 −0.23% Table 6

DeltasbetweenthenumericalandexperimentalwingliftanddragcoefficientsforcasesC39toC45. Case Angleof

attack[deg.]

Un-morphed wing Morphed wing

CL[·10−2_{] CD[·10}−3_{] CL[·10}−2_{] CD[·10}−3_]

C39 0.75 N/A N/A N/A N/A

C40 1 −0.409 3.030 −0.305 2.968 C41 1.25 −0.015 3.412 −0.007 3.308 C42 1.50 0.228 3.753 0.257 3.648 C43 2 0.652 4.444 0.524 4.291 C44 2.50 0.999 5.220 1.001 5.101 C45 3 1.578 6.110 1.457 6.013 Table 7

Comparisonbetweenthenumericalun-morphedandmorphedwingliftanddragcoefficientsfor casesC68toC73.

Numerical results Case Angleof

attack[deg.]

Un-morphed wing Morphed wing Drag

variation[%] CL CD CL CD C68 0 0.2990 0.0191 0.2994 0.0191 0.03% C69 0.5 0.3254 0.0212 0.3260 0.0212 0.07% C70 1 0.3519 0.0236 0.3527 0.0237 0.19% C71 1.5 0.3783 0.0263 0.3780 0.0263 0.13% C72 2 0.4047 0.0292 0.4057 0.0292 0.15% C73 2.5 0.4318 0.0323 0.4300 0.0322 −0.09% Table 8

Comparisonbetweentheexperimentalun-morphedandmorphedwingliftanddragcoefficientsfor casesC68toC73.

Experimental results (loads balance measurements) Case Angleof

attack[deg.]

Un-morphed wing Morphed wing Drag

variation[%] CL CD CL CD C68 0 0.3023 0.0231 0.3034 0.0230 −0.15% C69 0.5 0.3350 0.0261 0.3358 0.0260 −0.36% C70 1 0.3671 0.0295 0.3671 0.0294 −0.41% C71 1.5 0.3996 0.0333 0.3999 0.0332 −0.15% C72 2 0.4318 0.0373 0.4329 0.0372 −0.26% C73 2.5 0.4660 0.0417 0.4634 0.0416 −0.25%

1 67 2 68 3 69 4 70 5 71 6 72 7 73 8 74 9 75 10 76 11 77 12 78 13 79 14 80 15 81 16 82 17 83 18 84 19 85 20 86 21 87 22 88 23 89 24 90 25 91 26 92 27 93 28 94 29 95 30 96 31 97 32 98 33 99 34 100 35 101 36 102 37 103 38 104 39 105 40 106 41 107 42 108 43 109 44 110 45 111 46 112 47 113 48 114 49 115 50 116 51 117 52 118 53 119 54 120 55 121 56 122 57 123 58 124 59 125 60 126 61 127 62 128 63 129 64 130 65 131 66 132 Table 9

DeltasbetweenthenumericalandexperimentalwingliftanddragcoefficientsforcasesC68toC73. Case Angleof

attack[deg.]

Un-morphed wing Morphed wing

CL[·10−2_{] CD[·10}−3_{] CL[·10}−2_{] CD[·10}−3_] C68 0 0.329 3.954 0.397 3.914 C69 0.5 0.963 4.922 0.980 4.813 C70 1 1.516 5.892 1.437 5.726 C71 1.5 2.132 6.960 2.093 6.878 C72 2 2.708 8.117 2.718 7.979 C73 2.5 3.412 9.447 3.335 9.373 Table 10

Comparisonbetweenthenumericalun-morphedandmorphedwingliftanddragcoefficientsfor casesC74toC82.

Numerical results Case Angleof

attack[deg.]

Un-morphed wing Morphed wing Drag

variation[%] CL CD CL CD C74 0 0.0206 0.0061 0.0208 0.0061 −0.13% C75 0.5 0.0461 0.0063 0.0461 0.0062 −0.28% C76 1 0.0716 0.0064 0.0718 0.0065 0.67% C77 1.5 0.0967 0.0069 0.0963 0.0070 0.20% C78 2 0.1222 0.0077 0.1227 0.0076 −0.51% C79 2.5 0.1477 0.0086 0.1470 0.0085 −0.34% C80 3 0.1733 0.0098 0.1738 0.0097 −0.44% C81 4 0.2250 0.0128 0.2243 0.0128 0.01% C82 5 0.2765 0.0168 0.2766 0.0168 −0.05% Table 11

Comparisonbetweentheexperimentalun-morphedandmorphedwingliftanddragcoefficientsfor casesC74toC82.

Experimental results (loads balance measurements) Case Angleof

attack[deg.]

Un-morphed wing Morphed wing Drag

variation[%] CL CD CL CD C74 0 0.0082 0.0083 0.0082 0.0083 0.04% C75 0.5 0.0383 0.0084 0.0382 0.0084 −0.09% C76 1 0.0679 0.0088 0.0680 0.0088 0.33% C77 1.5 0.0983 0.0094 0.0992 0.0095 0.76% C78 2 0.1294 0.0105 0.1230 0.0105 0.09% C79 2.5 0.1602 0.0119 0.1560 0.0119 −0.14% C80 3 0.1917 0.0137 0.1912 0.0136 −0.67% C81 4 0.2531 0.0182 0.2541 0.0183 0.59% C82 5 0.3175 0.0241 0.3171 0.02401 −0.03% Table 12

ErrorsbetweenthenumericalandexperimentalwingliftanddragcoefficientsforcasesC74toC82. Case Angleof

attack[deg.]

Un-morphed wing Morphed wing

CL[·10−2_{] CD[·10}−3_{] CL[·10}−2_{] CD[·10}−3_] C74 0 −1.241 2.078 −1.252 2.089 C75 0.5 −0.774 2.132 −0.788 2.142 C76 1 −0.369 2.349 −0.381 2.334 C77 1.5 0.158 2.452 0.284 2.510 C78 2 0.726 2.784 0.730 2.832 C79 2.5 1.252 3.267 1.274 3.279 C80 3 1.845 3.876 1.737 3.828 C81 4 2.813 5.362 2.981 5.467 C82 5 4.102 7.307 4.047 7.308

terestingtostudyothermultidisciplinary approaches, asthe ones definedin[45].

Theexperimentaltestsandresultsobtaineduptothisdayand presented in thispaper included only the wing model equipped withtherigidaileron. The CRIAQMDO 505projectwill continue byperformingthenumericalstudiesandexperimentaltestsonthe morphingaileron.Theaimwillbetorealizetheproofofconcept, wherethenewmorphing ailerondesign wouldrequirelower de-flectionanglesinordertoprovidethesamedegreeofrollauthority astherigidaileron.

7. Conclusions

TheresultsobtainedusingCFDnumericalsimulationand exper-imental wind tunnel testingfor a morphing wingequippedwith a flexible upper surface and controllable rigid aileron were pre-sented. Themorphingwingtipwas manufacturedandfittedwith a composite material upper skin.Two-dimensional optimizations wereperformedwiththeaimofcontrollingtheextentofthe lam-inarflowregion,andtheresultingskinshapeswerescannedusing high-precisionphotogrammetry.Agridconvergencestudywas