See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/267683178Bio-inspired design of flapping-wing micro airvehiclesARTICLE AUGUST 2005CITATIONS23DOWNLOADS61VIEWS384 AUTHORS, INCLUDING:Kevin JonesNaval Postgraduate School77 PUBLICATIONS 864 CITATIONS SEE PROFILEMax PlatzerNaval Postgraduate School189 PUBLICATIONS 1,495 CITATIONS SEE PROFILEAvailable from: Max PlatzerRetrieved on: 10 August 2015THE AERONAUTICAL JOURNAL AUGUST 2005 385Paper No. 2950. Manuscript received 11 January 2005, accepted 18 July 2005.Bio-inspired design of flapping-wing micro air vehiclesK. D. JonesC. J. Bradshaw J. PapadopoulosM. F. PlatzerDepartment of Mechanical and Astronautical EngineeringNaval Postgraduate SchoolMonterey, CAUSAABSTRACTInthispaperthedevelopmentandflighttestingofflapping-wingpropelled,radio-controlledmicroairvehiclesaredescribed.Theunconventionalvehiclesconsistofalowaspectratiofixed-wingwithatrailingpairofhigheraspectratioflappingwingswhichflapincounterphase.Thesymmetricflapping-wingpairprovidesamechanicallyandaerodynamicallybalancedplatform,increasesefficiencybyemulatingflightingroundeffect,andsuppressesstallover the main wing by entraining flow. The models weigh as little as11g,witha23cmspanand18cmlengthandwillflyforabout20minutes on a rechargeable battery. Stable flight at speeds between 2and5ms1hasbeendemonstrated,andthemodelsareessentiallystall-proofwhileunderpower.Thestatic-thrustfigureofmeritforthedeviceis60%higherthanpropellerswithasimilarscaleanddisk loading.NOMENCLATUREc aerofoil chord lengthCpcoefficient of powerCtcoefficient of thrustf frequency in Hzk reduced frequency, 2fc/UL/D lift over drag ratioUfreestream velocityx streamwise coordinatez cross-stream coordinate angle-of-attack in degrees Propulsive efficiency, Ct /Cp Phase angle in degrees1.0 INTRODUCTIONThebrilliantsuccessoforganismswhichachieveflightusingflapping-wingpropulsionhasbeenaninspirationtohumankindforhundredsifnotthousandsofyears.Thishasledtothe,perhapserroneous,beliefthatflapping-wingsaresuperiortootherformsofpropulsion. The justification for this belief is that nature has chosenflapping-wings through natural selection. Of course this argument israther naive, as it ignores the initial conditions and constraints of theevolutionary process. In truth, one does not find many rotating partsin nature, and therefore a reasonable argument is that nature did notselectflappingwingsoverpropellers,butratherpropellerswereexcluded from the process entirely.However,thereareseveralargumentssupportingflapping-wingpropulsionwhichdoholdmerit.Forexample,rotarypropellersgenerate torque and create helical slipstreams, both of which degradevehicleperformanceandhandling,particularlyatlowspeeds.Additionally, flapping-wings may easily have a larger actuator area,and therefore a lower actuator loading, which may lead to increasedefficiencies.Ontheotherhand,flapping-wingshaveinherentmechanicallossesandincreasedstructuralloads.Thestructuralandinertial requirements likely limit flapping-wing propulsion to smallerscales, however, at these smaller scales low Reynolds number effectsfurther complicate the problem, for both conventional and flapping-wing propellers.Manyhavearguedabouttherelativemeritsoffixed,flappingorrotary-wingdesign,seeforexampleWoodsetal(1).Whilemostinthe micro air vehicle (MAV) community have concentrated on fixed-wingaircraft,thecapabilitiesofflapping-wingdesignsinnature,whichcanhover,maneuverintightspaces,andevenlandonaceiling,greatlyexceedanythinghumankindhasbeenabletoproduce,andthisaloneismotivationenoughtopursueflapping-wing flight.Thecreativity,ambitionandpersistenceofcountlessenthusiastshave led to a variety of successful flapping-wing flyers over at leastthepast130years.Thesedesignsareoftenclassifiedaseitherbiomimetic(adesignwhichmimicsnature)orbiomorphicorbio-inspired(adesignwhichisinspiredbynature,butisnotacopyofnature). The configuration developed in this investigation, shown inFig. 1, clearly falls into the second category.1.1 Biomimetic designsFlapping-wingmodelaircraftdatebackatleastto1874,whenAlphonsePenaudbuiltarubber-bandpoweredornithopter.Sincethen, with just a few exceptions, almost all successful flapping-wingaircrafthavebeenbiomimetic,atleastfortheflapping-wings.ProbablythemostwellknownbiomimeticflapperintheaerospacecommunityisAeroVironmentsMicrobat.Theirphilosophywastotakeafunctionalrubberbandpowereddesignandsubstituteanelectricalmotordrive-trainfortherubberbandandaddaradioforcontrol(2).Theymakeaninterestingnote,statingthatthespecificenergyoftherubberbandiscomparabletothatofthemotor/gearbox/battery/DC-DC-converterassembly.In2003,aversionoftheMicrobatusingatwo-cellLithium-polymer(Li-poly)battery,threechannelradio,andwitha23cmspanand14gtotalweight, reportedly made a flight of about 25 minutes.1.2 Biomorphic designsProjectsliketheMicrobatclearlydemonstratethatbiomimicrycanleadtosuccess,butoneislefttoponderwhetherornotthereisabetterway.Forexample,modelairplaneenthusiastshavebeenbuildingrubberbandpoweredornithoptersformorethanacentury,and to date, the greatest endurance has been achieved by somethingthatdidnotlookatalllikeabird,usinginsteadabiplaneflappingmechanism,asshowninFig.2.ThebiplaneorX-wingcanardflapper,builtin1985byFrankKieser,setanewrecordforindoorfreeflightendurance(3).Variationsofthisdesignstillholdtherecord.Whilethemodeldoesnotappeartolooklikeanythinginnature,infact,theflappingofitswing-pairsismuchlikeabirdflying in ground effect. Followingasimilarpath,acollaborationbetweenSRIandtheUniversityofTorontoproducedtheMentor,reportedlythefirstflapping-wingaircrafttohoverunderitsownpower(4).LikeFrankKiesersdesign,theMentorisdesignedtocapitaliseontheWeis-Fogh or clap-and-fling effect, and seems to be designed primarily forhoveringflightaflapping-winghelicopterofsorts,althoughforwardflighthasalsobeendemonstrated.Neitherdesignlooksanythinglikeabirdorinsect,butmanybirdsandinsectsmakeuseof the Weis-Fogh effect and were an inspiration for these designs.Wehavefounditinstructivetotrytodiscernwhichaspectsofanimalmorphologyevolvedasaresultoforganicand/orenviron-mentallimitations,andwhichaspectsareperformanceenhancing.While this is not always simple, one approach we have found usefulistolookathowanimalbehaviorisusedtocompensatefordesignlimitations.Forexample,birdshaveevolvedwithasinglepairofwings,soconfigurationssuchasthemodelshowninFig.2arenotpossible.However,thefactthatbirdsflylowovertheoceanindicatesthatbirdshaveindeeddiscoveredthebenefitsofflappingingroundeffect,abehaviourthatmakesthemaerodynamicallysimilartotheX-wingconfiguration.Theexploitationofgroundeffectisoneofthebehavioraladaptationswhichareutilisedinthisinvestigation.For the present design, much of the development process has beendocumented over the years in AIAA papers and other publications. Afairly detailed summary of our past work can be found in Jones et al(5)and Jones and Platzer(6), with references to most of our other publica-tions. In this paper the history is briefly summarised, with more detailsprovidedonprogressmadesincetheearlierpublications,andsomesummarising remarks on what the future might bring.2.0 DESIGN PHILOSOPHYAs previously mentioned, much can be learned from observations ofanimalflight,butcaremustbetakentodeterminewhichfeatureshave evolved to enhance performance, and which were forced due toorganicorenvironmentalconstraints.Forexample,abirdflapsits386 THE AERONAUTICAL JOURNAL AUGUST 2005Figure 1. NPS flapping-wing MAV. Figure 2. Frank Kiesers winning X-wing flapper.wingssuchthattheflap-amplitudevariesalongthespan.Thisdoesnot appear to be an optimal arrangement, but the bird does not haveanalternative.Wewouldprefertoflapthewingwithconstantamplitude to produce thrust from the root section of the wing, and toprovide for a more efficient span loading. Intheearlyyearsofourresearch,significanteffortwasdirectedtowardsimulationsoffairlybasicflapping-wingmechanicsinanefforttobetterunderstandthecomplicatedunsteadyflowphenomena, and with the hope of developing numerical tools to aidin a design optimisation methodology. Unsteady panel methods weredevelopedinconjunctionwithflowvisualisationandlaserDopplervelocimetry(LDV)measurementsinwindandwatertunnels(7-10),and several significant conclusions were made.First, the benefit of flapping in ground effect was quantified, bothnumerically and experimentally, as shown in Figs 3 and 4. For thesesimulations,anairfoilwasplungedsinusoidallyatzeroangleofattack,withthespecifiedfrequency,andwithplungeamplitude0.4c.Theground-effectsimulationsweremodeledusinganimageairfoil below ground, with a mean separation of 14c, correspondingtoameandistancefromthegroundof07c.Theexperimentalmeasurementswereperformedonabiplanemodel,wherethetwowingsflappedincounterphase,essentiallyduplicatingthepanel-codemodelemulatinggroundeffectthroughsymmetry.A5to10% increase in both the thrust coefficient and propulsive efficiencywere predicted. Additionally, the agreement between experiment andthe panel code were quite good, particularly for the low experimentalReynolds numbers (less than 5 104).Asecondarybenefitofthebiplanearrangementwasthatthemodel, as a whole, was dynamically balanced, minimising vibrationsoftheapparatus.Whiledynamicbalancingwasconvenientfortestinginawind-tunnel,itappearedtobeevenmoreadvantageousforaflyingmodel.Flappingasinglewinginflightwouldresultinoscillationsofthefuselageworkdonewithnogaininperfor-mance.However,byflappingtwowingsincounterphase,thefuselagewouldremainsteady,andalltheworkwouldbeusedtomove components which directly produce thrust.Itislikelythatbirdshaveevolvedtocompensateforthisimbalance.Theirneckishighlyarticulated,suchthattheirheadisinertiallystableforimprovedvision,andtheirbodyandtailmostlikelycreateadditionalthrustastheyoscillateinoppositiontothewingmotion.However,itisdoubtfulthatanyman-madeornithoptersbenefitfromfuselageoscillations,asthisrequiresalevel of sophistication beyond our current capabilities.Anotheraerodynamicphenomenonwewishedtoexploitwastheabilitytosuppressflowseparationusingtheflappingwings.Dynamic stall delay due to oscillatory pitching is a fairly well knownphenomenon,ofparticularinteresttorotary-wingengineers(11,12).However,alesserknownapplicationistheuseofaflappingwingdownstream of a larger airfoil, using the favorable pressure gradientahead of the flapping wing to suppress flow separation on the largerwing.Water-tunnelexperimentsdemonstratedthisphenomenonforflowoverabackward-facingstep(8),andforseparationcontrolofflow over several blunt trailing edge airfoils(9,13), as shown in Fig. 5.In the upper image the general experimental setup is shown, illus-trating a large, stationary airfoil with a cylindrical leading edge and arather abrupt cusped trailing edge. The airfoil thickens slightly alongthechordtokeeptheboundarylayerthin.Dyeisemittedfromasmallholejustupstreamofthecuspedtrailingedgeregion,wherethe flow is expected to be attached. Following the stationary wing isa small wing which may be flapped. The lower images show experi-mentalsnapshots.Intheleftimagethetrailingwingisstationary,JONES ET AL BIO-INSPIRED DESIGN OF FLAPPING-WING MICRO AIR VEHICLES 387Figure 3. Predicted benefits of ground effect.Experimental schematic.Flapping dye indicates eattachment Static dye indicates separationFigure 5. Flow reattachment due to flow entrainment.Figure 4. Thrust measurements in ground effect.and the flow completely separates at the onset of the recompression.In the right image, with the trailing wing flapping, the dye indicatesthat the flow has reattached.Itwashopedthatasimilarsuccessmightberealisedonaflapping-wing MAV. This facet is of particular interest for low speedMAVs,astheflightReynoldsnumbersmaybequitelow,ontheorder of 2 104, where the flow is laminar, and separation is likely.TheseconsiderationsledustotheconfigurationshowninFig.1,with a biplane-pair of trailing wings, flapping in counterphase, coupledwith a large fixed wing located just upstream of the flapping wings.3.0 DESIGN METHODOLOGYAspreviouslymentioned,theearlyyearsofourresearchwerefocusedondevelopingsimulationsoftwarewhichmightbeusedinanoptimisationalgorithmtodesignanefficientflapping-wingMAV.Unfortunately,whatourresearchdemonstratedwasthatthelowReynoldsnumbersledtoflowphysicswhichcouldnotbeadequatelypredictedusinginexpensivemethodssuchasourpanelcode.Simulationswithatwo-dimensionalunsteadyNavier-Stokessolverprovidedmuchbetteragreementwithexperimentalmeasure-ments(10,14),butwerefartoocostlytobeembeddedaspartof an iterative design process. Therefore, in the end we had to resort toa more hands-on or heuristic approach, using failures and successesto drive the evolutionary design process.3.1 Mechanics of thrust generationOne of the first lessons to be learned was simplicity in design. On alargemodelcomplicatedmechanismscouldbedevelopedtoaddseveraldegreesoffreedomtoaflappingmechanism.However,working on a MAV-scale, size and weight issues dominated, and theflappingmechanismhadtobereducedtoasingleactivedegreeoffreedomplunge.Testmodels,liketheoneshowninFig.6wereassembled,allowingforthedirectmeasurementofthrust,anditquicklybecameapparentthatapitchdegreeoffreedomwasnecessary.Tokeepthingsmechanicallysimple,thiswasdonepassively by attaching the flapping wings to the flapping mechanismwithaflexiblejointsothattheywereabletopitchaeroelastically.Typical performance from the models is shown in Fig. 7.Inspired by Cylinder et al(15), a second passive degree of freedomwasattemptedcamber.ThemodelisshowninactioninFig.8,with the flash-frozen wings shown at mid-stroke as they pulled awayfrom each other. An elastic angle of attack of about 30 is visible, aswellasthefullycamberedwing.ThemeasuredperformanceofthecamberingwingsisshowninFig.9.Whilethecamberprovidedexcellentandpredictableperformanceforlowerfrequencies,athigherfrequenciesthewingsdidnothaveenoughtimetochange388 THE AERONAUTICAL JOURNAL AUGUST 2005Figure 6. 15cm length/span MAV propulsion model.Figure 7. Performance of the propulsion model.Figure 8. Aeroelastic cambering wings.Figure 9. Static performance of cambering wings.camber, and the performance degraded to that of the flat-plate wings.Withsomeadditionalwork,perhapsthistoocouldberesolved,butdue to the added weight and complexity of the cambering wings, forthe time being they were eliminated from the design.Alloftheseearlytestmodelsusedtiny,gearedsteppingmotors,which required large external power supplies and controllers. Whilethesedrivesystemsweredesirableforwind-tunnelmodels,theywerenotsuitableforflyingmodels,andtheywerereplacedbybrushedDCmotorswithcustomgearboxesforthefollowingmodels.3.2 Lift generationThenextstepwastheintegrationoftheflapping-wingpairwithafixedleadingwing.Thefirsttestmodel,showninFig.10,stillconformedwithDARPAs15cmsquaresizecriteria.Thismodelwastestedinthewind-tunnel,andusingflowvisualisationwithasmokewire,providedthefirstevidenceofseparationcontrolusingthe flapping wings. This is illustrated in Fig. 11, where the wings arenotflappingontheupperimage,withtheflowseparatingattheleadingedge,andonthelowerimage,withthewingsflapping,theflow appears to have reattached.Tethered flight on a radial-arm test stand indicated that speeds onthe order of 2 to 3ms1were possible while generating enough lift tosupportthe7gweight.Atthattimecommercial-off-the-shelf(COTS)batteriesandradiogearaddeduptoaminimumofabout20g, so untethered flight was not feasible.3.3 Energy storageWearefortunatethatsomeofthecriticalcomponentsweusearealsocriticaltothecell-phoneindustry,andhenceperformanceimprovementshavebeentremendous;specifically,batteries,motors(themotorsthatvibratepagersandcell-phones),andDC-DCconverters to step-up a low voltage.Batterytechnologyhasbeendrivenbythecell-phoneindustryatanextremepace.Fiveyearsago,aboutthebestsolutionwasa50mAhnickelcadmiumcell,providing006Whina35gcellanenergy density of about 17Wh/kg. The rechargeable Li-poly cells weusenowhaveacapacityof135mAhandcanprovide05Wh.Weighingamere3g,theyhaveanenergydensityofabout170Wh/kg, or ten times what the NiCd cell provided.3.4 Motor/gear assemblyThemotorsarecurrentlyoneoftheweakestlinks.Whilethecell-phoneindustryhaspushedhardtomakesmall,lightandcheapmotors,theyhavedonelittletoimprovetheefficiency.Themotorsweused(DidelMK06L-10orMK06-04.5)weighedabout1.3to15g, and used a two-stage gear system. For the first models we hadtofabricateourowngearsystems,butnowweusegearassembliesbuiltfortheindoormodelaircrafthobbyist(Didel6R21).Usingaminiaturedynamometerwemeasuredtheefficiencyofthemotor/gear assembly, and at flight loads it was below 25%.Ontheotherhand,themotor/gearassemblieswereverylight,about22to25g,andtheywereabletodelivermorethan0.4Wofshaftpower,apowerdensityof180W/kg,whichisquitegoodforthis scale.3.5 Power conversionAvarietyofmotorswereavailable,withnominalvoltagesbetween08 and 24V. The first flying models built used motors designed for12V with a rated power output of about 0.14W. In order to achieveJONES ET AL BIO-INSPIRED DESIGN OF FLAPPING-WING MICRO AIR VEHICLES 389Figure 10. First fixed/flapping-wing configuration.StaticFigure 11. Separation control at high AOA.FlappingthepowerrequiredtoflytheMAV,weneededtoincreasethepoweroutputconsiderably,andthiswasdonebyrunningthemotoratamuchhighervoltage.AminiatureDC-DCstep-upcircuit was assembled to bump the 30 to 42V from the battery uptoaregulated5V.ThecircuitwasbasedonanICdesignedforcell-phone type devices, but was miniaturised as much as possible,withsomesacrificeinefficiency.Thecircuitweighedabout0.35g, and produced a regulated 5V supply up to about 05A.Withsomeofthelowerresistancemotorsnowavailable,thepowerconverterisnotrequired;althoughtheflightperformancedegrades somewhat as the battery life is exhausted and the voltagedrops.3.6 AvionicsTheradiogearisCOTShobbyistequipment,designedforpark-flyersandindoormodelaircraft.Thethreechannelreceiver(SlowflyRFFS-100)weighsabout2g,slightlylessifyouremovethebulkyconnectorpinsitcomeswith,anditincludesanelectronicspeedcontrolanddriversfortwomagneticactuators.Thereceiverisnotnarrowband,andislimitedtoabout100mrange.Themagneticactuatorsarecomprisedofsmallcoilsandamagnetpair.Thecoilisgluedtotheaircraftframe,andthemagnets are glued to the control surface, such that they are insidethe coil. When the coil is energised, the magnets try to align withthe coil field, deflecting the control surface. If the control surfaceislightlyspringloadedtoreturntocenter,thenproportionalcontrol can be achieved by varying the duty cycle of the coil.Theactuatorsaretheotherweakestlink.Theyarenotlight,typically07to1geach,andtheyarenotefficient,drawingasmuchas05Weach,buttheyarerelativelyinexpensive,easytouse and quite durable.3.7 Vehicle sizingThenextstepinthedesignprocedurewastosizetheconfigurationfor the required payload, in this case, just the propulsion and controlsystem.Thecomponentsweusedonourfirstmodelwereslightlyheavierthanthoselistedabove,addinguptoabout925gforabattery,converter,drivesystemandreceiver.Frompastbuildingexperience, we estimated the weight of the structure to be about 5g,yielding a total weight of just over 14g.Whilelittleexperimentaldatawasavailableforlowaspect-ratiowingsinthe2104to5104Reynoldsnumberrange,papersbyLaitone(16)and Torres and Mueller(17)suggested that lift coefficientsontheorderofabout06werepossiblewithanL/Dmaxofaround5onanaspect-ratio2wing.Thisindicatedthatawingareaofabout006m2was needed.4.0 DESIGN EVOLUTIONThe first model, shown in Fig. 12, flew in December of 2002. It had amainwingwitha30cmspanand145cmchord,andflappingwingswitha25cmspanand4cmchord.Itusedasingle-channelcontrol,throttle-only, with fixed rudder trim for a shallow turn, and fixed pitchtrim to give it a constant nose-up attitude. The flying weight was about144g,andthemodelmadeanumberofflights,thelongestlastingabout three minutes before coming to rest high up in a tree.Flight speed was only about 2ms1, and the ability of the model toflyatveryhighanglesofattackwithoutstallingsuggestedthatseparationcontrolwasineffect.Withthepoweroff,themodelwouldstallquiteeasilyinresponsetogusts,butunderpower,itwouldmerelysettlebackintolevelflightwithoutloosinganyaltitude.Severalmonthslaterasecondmodelwasbuilt;slightlysmaller,witha27cmspan,andincludingaruddercontrol.Theweightwasreduced to 134g, and the model could now sustain longer flights, astreesandbuildingscouldbeavoided.DuringanAIAAtechnicalseminar at NASA Ames on 12 February 2003, the model was flowninthetestsectionofthe80by120ftNFACtunneloneofthesmallest aircraft flying the worlds largest wind-tunnel.Thesecondmodelwasstillwithoutelevatorcontrol,soitwastrimmedwithanose-upattitude,andthrottlewasusedtocontrolrate-of-climb.Bychangingthepitchtrim,theflightspeedcouldbeadjusted.Themodelflewwellfromspeedsaslowas2ms1uptoabout5ms1.Higherspeedswouldbepossiblewithactivepitchcontrol,butwereriskywithapresetpitchtrim,asthemodelmighteasily dive into the ground in response to a gust.A Watt-meter was attached to the model on the bench top, and atfullpower15Wwasdrawnfromthebatterywhilethemodelproduced about 10g of static thrust. Considering only the motor/gearefficiency(ignoringlossesthroughtheDC-DCconverter,receiver/speed-controller,andcrankshaftassembly),thissuggeststhat about 0375W shaft power was delivered, resulting in a figure ofmerit(FOM)ofabout30g/Wataneffectivedisk-loadingofabout6N/m2.Thisisabout60%higherthancomparablerotarywingvehicles(18).390 THE AERONAUTICAL JOURNAL AUGUST 2005Table 1Model detailsModel Span Length Avionics BatteryMotor/gear Structural Totalweight weight weight weight weight1 30cm 20cm 27g 38g 27g 52g 144g2 27cm 19cm 31g 32g 25g 45g 134g3 25cm 18cm 27g 32g 23g 42g 124g4 23cm 17cm 27g 23g 23g 38g 110gFigure 12. First flying model.A third model, shown previously in Fig. 1, is slightly smaller yet,witha25cmspanand124gflyingweight.Itusesalowervoltagemotor,eliminatingtheneedfortheDC-DCconverter,andithasinterchangeablepartsallowingformoresystematicoptimisation.The main wings on this model incorporate about 5 of leading-edgesweepandareduceddihedral-angle.Theyaw-rollcouplingonthethirdmodelishigherthanthepreviousmodelsallowingforaverytight turn radius, even without active pitch control.A fourth model was assembled to compete in the 8th InternationalMAVcompetition,inApril2004.Ithadanevensmallerspan(aswell as the longest linear dimension) of 23cm, and weighed just 11g.A summary of model dimensions and weight breakdown is shown inTable 1.5.0 DESIGN ANALYSISWith flying models in hand, we now went back to the wind-tunnel inordertogainabetterunderstandingoftheflowphysics;hopefullyallowingustooptimisethedesign.Sincethepagermotorsusedintheflyingmodelshadarelativelyshortlifespan,amodelwasbuiltwith the same fixed and flapping-wing geometry as the second radiocontrolled model, but with a larger fuselage to house a bigger motorandrotaryencoderandwithinterchangeableparts.Thenewmodel,shown in Fig. 13, was attached to a two-component force balance tomeasureliftandthrust,andflowvisualisationandunsteadyLDVexperiments were run.5.1 Flow visualisationStreaklinesweregeneratedbyasmokewirewhichwasconstructedfrom025mmdiameterNiCrbeadedwire,heatedbypassingacurrentthroughit,andusingRoscoFogJuiceasthesmokeagent.Imagerywasrecordedusingeitheradigitalstillcameraoradigitalvideocamerawithahighshutterspeedtofreezethemotionofthewings and streaklines. Details of the methods can be found in Jonesand Platzer(6)and Papadopoulos(19).Flowvisualisationexperimentswereperformedwiththemodelmountedata15angleofattack,ataflowspeedofabout2ms1,approximating the low-speed flight conditions. Initially the flappingwingswereatrest,andtheywerethenquicklyacceleratedtoaflapping frequency of about 30Hz. The results are shown in Fig. 14,viewing the model from the left rear corner forward; an angle whichprovidesagoodviewoftheflowovertheuppersurfaceoftheleftwing.Intheupperimage,withoutwingflapping,itisclearlyseenthattheflowseparatesattheleadingedge,andthewingisfullystalled. In the upper image, after just four flapping strokes, the flowisalreadyreattached.Whiletheboundarylayerappearstobeverythick and unsteady, the outer flow remains parallel to the upper wingsurfaceandreattachesatthetrailingedge.Notonlyistheflowentrainment sufficient to reattach the flow, but it requires only aboutatenthofasecondtotransition.TheReynoldsnumberisabout2104for the main wing, and just 5 103for the flapping wings.5.1 Laser-doppler flow measurementsA TSI two-channel LDV system with a single probe was used, withflowseedingprovidedbyaRoscofoggenerator.Forunsteadymeasurements,thesignalfromtherotaryencoderwasfedintoarotarymotionresolver(RMR),whichallowedtheLDVsystemtorecordperiodicdata,synchronisedwiththewingflapping.Furtherdetails about the setup and measurement procedures can be found inBradshaw(20).UseoftheRMRprovidedforhigherfidelityvelocitymeasure-ments,removingtheeffectsofphase-biasedflowseeding.SinceLDV is a statistical average of a large number of recorded events, inan unsteady, periodic flow, the seeding density may fluctuate period-ically,biasingthevelocitypredictiontowardthemoreheavilyseeded parts of the cycle. This biasing was removed using the RMR,resulting in somewhat higher velocity peaks.The flow-entrainment effect is illustrated in Figs. 15 and 16. In Fig.15thetime-averagedvelocityprofilejustinfrontoftheflappingwings (x = 0) is shown for three cases. In the first case, the main wingis removed, and the wings are flapped at 32Hz. In the second case theJONES ET AL BIO-INSPIRED DESIGN OF FLAPPING-WING MICRO AIR VEHICLES 391Figure 13. Wind-tunnel variant of the MAV.StaticFlappingFigure 14. Separation control at high AOA.mainwingisincluded,butthewingsarenotflapped.Inthethirdcase, the main wing is included and the wings are flapped at 32Hz. Inall three cases the freestream speed is 275ms1, and the model is setat a 15 angle of attack. Unfortunately, the dihedral of the main wingmaskedalargeareaabovethesymmetryplane,roughlywherethefigurelegendisplaced,suchthattheeffectoftheupperflappingwing is partially obscured.Comparing the flapping cases with and without the main wing, theentrainmenteffectisclearlyseenwithabouta30percentover-velocityatthecenterlineofthelowerflappingwing.Notethatthevelocityprofileisnearlyunaffectedbytheinclusionofthemainwing.Withoutflappingthewings,alargevelocitydeficitisseennearthestagnationpointontheleadingedgeofthelowerflappingwing.Alsonotethatwithoutflappingthewings,avelocitydeficitappearsmorethanachordlengthabovethemainwing,illustratingthe severity of the separated flow.InFig.16,velocityprofilesachordlengthupstreamoftheflappingwings(x =c)areshownforthesamethreecases,anditcan be seen that the entrainment effect has diminished considerably,indicating that the flapping wings must be quite close to the trailingedge of the main wing to capitalise on this phenomenon.6.0 SUMMARY & PROSPECTIVEAlmostadecadeofresearchinflapping-wingpropulsionandlowReynoldsnumberunsteadyaerodynamicshasculminatedinthedevelopmentofanunusualflapping-wingpropelledmicroairvehicle.Theconfigurationconsistsofabiplanepairofflappingwings,whichmoveincounterphase,locatedjustdownstreamofafixed wing. The relatively large fixed wing provides most of the lift,whiletheflapping-wingsprovidethethrustandsuppressflowseparationonthemainwingduetotheirflowentrainment.Thesymmetryoftheflapping-wingpairprovidesmechanicalandaerodynamicbalancing,andproducesthrustmoreefficientlythanconventional flapping-wing systems.Thepresentmodel,usingcommercialofftheshelfpowerandavionicsequipment,hasa25cmspan,18cmlengthandweighsapaltry124g,withtwo-channelcontrolandenoughbatterycapacityfor 15 to 20 minutes of flight. The model does not have active pitchcontrol,butthepitchmaybemanuallytrimmedtosettheflightspeed.Speedsbetweenabout2and5ms1havebeenachieved,andthemodelisvirtuallystall-proofwhileunderpower,duetotheseparation suppression characteristic of the design.Wind-tunnelmeasurementsofliftandthrust,aswellasflowvisualisationandunsteadyLDVmeasurementsofthesurroundingflowfield have verified the flow separation suppression phenomenon,andnowprovideameansforthesystematicanalysisofthedesign,whichwilleventuallyallowfordesignoptimisation.Whilethecurrentdesignfliesextremelywell,itisbynomeansoptimised.However,operatingwithastaticfigureofmeritofalmost30g/W,itexceeds low Reynolds number (5 103to 2 104) propeller perfor-manceatanequivalentdiskloadingbyatleast60percent.Furthermore,thedesignshouldscaledownwell,allowingformuchsmaller vehicles as power and avionics technology improves.While there are several natural directions for the project to evolve,itslowspeedperformancemakesithighlysuitableformissionswhereflightinconfinedareasisrequired.Itislikelythattheflightspeedcanbereducedmuchfurther,perhapseventuallyachievinghovering flight. Along with the low flight speed comes the ability topark the vehicle in strategic places, allowing it to gather and transmitdatalongpastitsusefulflightlifetime.Therearecertainlymanyotherpossibilities,limitedonlybeourcreativityandperseverance.Researchiscurrentlyunderwaytoinvestigatethecomplicatedinteraction between the three wings. Low Reynolds number Navier-Stokes analysis of the aeroelastic pitching mechanism is in progress,aswellassimulationsofthebi-planeinteractionoftheflappingwings.Futureworkwillincludeallthreewingsectionsinordertosimulate the flow-entrainment and reattachment phenomenon.ACKNOWLEDGMENTSWe are grateful for the support received from Richard Foch, head ofthe Vehicle Research Section of the Naval Research Laboratory, andprojectmonitorsKevinAilinger,JillDahlburgandJamesKellogg.REFERENCES1. WOODS,M.I.,HENDERSON,J.F.and LOCK,G.D.Energyrequirementsfortheflightofmicroairvehicles,2001, AeroJ,105,(1045),pp135-149.2. KEENNON,M.T.and GRASMEYER,J.M. DevelopmentoftheBlackWidow and Microbat MAVs and a vision of the future of MAV design,2003, AIAA Paper No 2003-3327.392 THE AERONAUTICAL JOURNAL AUGUST 2005Figure 15. Time-averaged velocity at x = 0. Figure 16. Time-averaged velocity at x = c.3. CHRONISTER,N. Ornithoptermediacollection,2001,http://www.ornothopter.org/rubber.html.4. BRIDGES,A. Flyingrobotscreateabuzz,MontereyCountyHerald,28July, 2002.5. JONES,K.D.,LUND,T.C.and PLATZER,M.F. ExperimentalandComputationalInvestigationofFlapping-WingPropulsionforMicroAirVehicles,2001,Chapter16,FixedandFlappingWingAerodynamicsforMicroAirVehicleApplications,AIAAProgressinAerospace Sciences Series, MUELLER, T. (Ed), 195.6. JONES,K.D.and PLATZER,M.F. Experimentalinvestigationoftheaerodynamic characteristics of flapping-wing micro air vehicles, AIAAPaper No 2003-0418, 2003.7. JONES,K.D.,DOHRING,C.M.and PLATZER,M.F. AnexperimentalandcomputationalinvestigationoftheKnoller-Betzeffect,AIAAJ,1998,36, (7), pp 1240-1246.8. LAI, J.C.S., YUE, J. and PLATZER, M.F. Control of backward facing stepflow using a flapping airfoil, Experiments in Fluids, 2002, 32, pp 44-54.9. DOHRING,C.M.,FOTTNER,L.and PLATZER,M.F. Experimentalandnumerical investigation of flapping wing propulsion and its applicationforboundarylayercontrol,1998,ASMEInternationalGasTurbineCongress, 98-GT-046.10. JONES,K.D.,CASTRO,B.M.,MAHMOUD,O.and PLATZER,M.F. Anumericalandexperimentalinvestigationofflapping-wingpropulsionin ground effect, 2002, AIAA Paper No 2002-0866.11. CARR,L.W.and CHANDRASEKHARA,M.S.Compressibilityeffectsondynamicstall,ProgressinAerospaceSciences,,199632,pp523-573.12. MCCROSKEY, W.J. Unsteady airfoils, Ann Rev Fluid Mech, 1982, 14, pp285-311.13. TUNCER, I.H., LAI, J.C.S. and PLATZER, M.F. A computational study offlowreattachmentoverastationary/flappingairfoilcombinationintandem, 1998, AIAA Paper No 98-0109.14. CASTRO,B.M. Multi-blockParallelNavier-StokesSimulationsofUnsteadyWindTunnelandGroundInterferenceEffects,2001,PhDthesis,DepartmentofAeronautics&Astronautics,NavalPostgraduateSchool.15. CYLINDER,D.,SRULL,D.and KELLOGG,J. Biomorphicapproachestomicro air vehicles, 2003, UAV Asia-Pacific 2003 Conference.16. LAITONE,E.V. WindtunneltestsofwingsandringsatlowReynoldsnumbers, Chapter 5, Fixed and Flapping Wing Aerodynamics for MicroAir Vehicle Applications, AIAA Progress in Aerospace Sciences Series,MUELLER, T. (Ed), 195, pp 83-90.17. TORRES,G.E.and MUELLER,T.J. AerodynamiccharacteristicsoflowaspectratiowingsatlowReynoldsnumbers,2001,Chapter7,FixedandFlappingWingAerodynamicsforMicroAirVehicleApplications,AIAAProgressinAerospaceSciencesSeries,MUELLER,T.(Ed),195,pp 115-142.18. KORNBLUH, R. D., LOW, T. P., STANFORD, S. E., VINANDE, E., BONWIT,N.,HOLEMAN,D.,DELAURIER,J.D.,LOEWEN,D.,ZDUNICH,P.,MACMASTER,M.and BILYK,D. Flapping-wingpropulsionusingelectroactivepolymerartificialmuscleactuators,Phase2:radio-controlled flapping-wing testbed, 2002, SRI International Report ITAD-3470-FR-03-009.19. PAPADOPOULOS,J.P. AnExperimentalInvestigationoftheGeometricCharacteristicsofFlapping-wingPropulsionforaMicroAirVehicle,2003,Mastersthesis,DepartmentofAeronautics&Astronautics,Naval Postgraduate School.20. BRADSHAW,C.J.AnExperimentalInvestigationofFlappingWingAerodynamicsinMicroAirVehicles,2003,Mastersthesis,Department of Aeronautics & Astronautics, Naval Postgraduate School.JONES ET AL BIO-INSPIRED DESIGN OF FLAPPING-WING MICRO AIR VEHICLES 393