Nemanjakomatinovic 2006

M.Sc.
PrograminWindEnergy TechnicalUniversityofDenmark
AshikagaInstituteofTechnology CollaborativeResearchCenter

InvestigationoftheSavoniustypeMagnus WindTurbine
MasterThesisProject
NemanjaKomatinovic,s040570 MEK,TechnicalUniversityofDenmark Kgs.Lyngby,DK2800 Supervisors:JensNrkrSrensen(TechnicalUniversityofDenmark) IzumiUshiyama(AshikagaInstituteofTechnology) October2006
InvestigationoftheSavoniustypeMagnusWindTurbine________________
Acknowledgements
ThisstudyisapartofaMasterThesisprojecttakenonTechnicalUniversityofDenmark during the period AprilOctober 2006. Project was done in Japan as a part of the exchangebetweenTechnicalUniversityofDenmarkandCollaborativeResearchCenter at Ashikaga Institute of Technology, under the supervision of professor Jens Nrkr SrensenandprofessorIzumiUshiyama. Iowespecialthankstoboth,professorSrensenforallowingmetoparticipateinsuch remoteexchangeprogramandtoprofessorUshiyamaforacceptingmeonthisproject. Also I am thankful for their professional guidance during preparation of my thesis. I alsowouldliketothankNishizawaYoshifumiandKimishimaYoshinobufornecessary technicalsupportandSatoshiKawashimaandAkiraItoforassistanceduringthewind tunnelexperiments.
Abstract
ThispapersdealswiththeconceptofthesocalledSavoniustypeMagnuswindturbine. Theturbinesupportstheideaofclassicalhorizontalaxiswindturbinewhereinsteadof the airfoil blades, Savonius rotors were used to create a lift forces. Blade simplicity allowsproductionofthewindturbinesatlowercostandpossibilityofspreadingsuch technologyforwindpowerproductionatmoreaffordablerates.After2DCFDanalysis, static torque was found for the various Savonius configurations, and the four most promising, were chosen for the practical blade design. The wind tunnel tests were performed where the torque and rpm measurements of the main rotor took place. In addition,themeasurementsoftheSavoniusbladerpmwererecordedforthepurposeof finding the relationship between the rotational lift and a power output for different generator loading cases. A modal analysis of the Savonius blade was found to be necessarysincealargevibrationsoccurredduringthetestingroughlyaround1000rpm. Also a centrifugal buckling analysis was made with identification of the highstress locations. ASavoniustypeMagnuswindturbinewasfoundtobeafeasibledeviceasastandalone low power production electromechanical system, and some aspects of possible prototypedesignwerepresentedasapartofthecurrentstudy.
Contents
ListofSymbols ............................................................................................................................ 6 1.Introduction............................................................................................................................. 8 1.1MagnusEffectTheory ..................................................................................................... 8 1.1.1Overview ..................................................................................................................... 8 1.1.2ShortHistoricalNote ............................................................................................. 12 1.2SavoniusRotorOutline ................................................................................................ 13 1.2.1SummaryofInvestigationsonSavoniusRotor ................................................ 13 1.2.2SomeCharacteristicsoftheSavoniusRotor...................................................... 14 1.2.3OperatingPrinciple ................................................................................................ 15 1.3IdeaBehindSavoniustypeMagnusWindTurbine ............................................... 17 1.4PlanofInvestigations.................................................................................................... 21 2.DeterminationoftheStaticTorque ................................................................................... 23 2.1Definition ......................................................................................................................... 23 2.2NumericalStudy.............................................................................................................. 23 2.3GeometryDescription .................................................................................................... 23 2.4FlowDescription ............................................................................................................. 25 2.5NumericalMethod .......................................................................................................... 25 2.6TurbulenceModel........................................................................................................... 26 2.7DomainSizeandMeshConfiguration ....................................................................... 28 2.8BoundaryConditions.......................................................................................................... 33 2.9SolverAlgorithm............................................................................................................. 33 2.10Postprocessing .............................................................................................................. 35 2.11Results ............................................................................................................................. 35 2.11.1ResultsValidation ................................................................................................. 35 2.11.2TorqueAnalysis ..................................................................................................... 38 3.BladeDesignandPrototype ................................................................................................ 44 3.1DesignObjectives ........................................................................................................... 44 4
InvestigationoftheSavoniustypeMagnusWindTurbine________________ 3.2PrototypeDescription .................................................................................................... 45 4.ExperimentalInvestigation ................................................................................................ 52 4.1ExperimentSetupandFacility...................................................................................... 52 4.2SavoniusRotationMeasurement ................................................................................. 54 4.3FullModelTestProcedure ............................................................................................ 55 5.Results .................................................................................................................................... 56 5.1RotorBladeExperimentalResults ............................................................................... 56 5.2MagnusExperimentalResults ...................................................................................... 62 6.ModalAnalysisoftheBlades ............................................................................................ 80 7.StressandDeformationAnalysis....................................................................................... 88 8.FutureInvestigations ............................................................................................................ 93 9.ConclusionsandRecommendations.................................................................................. 94 10.Bibliography......................................................................................................................... 95 AppendixA:ListofFigures..................................................................................................... 98 AppendixB:ListofTables .................................................................................................... 102
ListofSymbols
Note:Listisnotthorough,andomitssomesymbolsthatcould beuniquetoparticular chapters. a Savoniusrotoroverlap AM SweptareacoveredbytheMagnusrotor AR Aspectratio b Savoniusrotorseparationgap CD Totaldragforcecoefficient cD CL cL CP CQ D DC DM DS Fn Ft GP HS L nM nS OL Pmech Qmech R Rcyl Re RM RS St Vfree Vrel Vwind Local2Ddragcoefficient Totalliftforcecoefficient Local2Dliftcoefficient Powercoefficient Torquecoefficient Dragforce DiameteroftheSavoniusrotorsemicylinders(paddles) DiameteroftheMagnusturbinerotor DiameteroftheSavoniusrotor Normalcomponentofthetotalforcevector Tangentialcomponentofthetotalforcevector Savoniusrotorgapratio HeightoftheSavoniusrotor Liftforce RotationalspeedoftheMagnusturbinerotor RotationalspeedoftheSavoniusrotor Savoniusrotoroverlapratio MechanicalpoweronthemainshaftoftheMagnusturbinerotor MechanicaltorqueonthemainshaftoftheMagnusturbinerotor Totalforce Radiusofthecylinder Reynoldsnumber RadiusoftheMagnusturbinerotor RadiusoftheSavoniusrotor Strouhalnumber Velocityofthefreestreamflow RelativevelocityseenbytheSavoniusrotorcrosssection Undisturbedwindspeed AzimuthalangleofthebladeintheMagnusturbinerotorplane Tipspeedratio/rotationalrate TipspeedratiooftheMagnusturbinerotor
blade
M
InvestigationoftheSavoniustypeMagnusWindTurbine________________ S free cyl M S TipspeedratiooftheSavoniusrotor Densityofthefreestreamflow Angularvelocityofthecylinder AngularvelocityoftheMagnusturbinerotor AngularvelocityoftheSavoniusrotor
1.Introduction
Beforepresentingtheprojecttogetherwiththeresultsofinvestigation,somenotesfrom a Magnus effect theory and principles behind the Savonius turbine will be presented first. Reason lies in more complex application of aerodynamic lift theory and different approachthatcombinesdragdrivendeviceinordertocreatetherotationallift.
1.1MagnusEffectTheory
1.1.1Overview
Sideforce effect of the rotating bodies was noticed first time by an eminent English scientist Benjamin Roberts in 1742 during his investigations of spinning artillery projectilesusingtheswirlingarmdevice.RegardlessoftheadmirationforRobertswork, Euler disputed his results and ascribed transversal force of the rotating body to manufacturingirregularities[33].
Figure1.Inviscidirrotationalflowpastacylindera)zerorotation,b)subcriticalrotation,c) criticalrotation,d)supercriticalrotation
About the century after, German scientist Gustav Magnus explained this phenomenon asanaerodynamiceffect.FurthercontributioncamefromPrandtlandhismodification ofKuttaJukowskitheoremforbodiesofrotation.Assumingtheinviscidandirrotational flow,anddefiningarotationrate=R/Vwind,whereRistheradiusoftherotatingbody, itispossibletodefinealiftforceLasafunctionof.Onecandistinguishbetweenfour general flowcasesdependingon the positions ofthestagnation points onthecylinder
InvestigationoftheSavoniustypeMagnusWindTurbine________________ (seeFigure1).Fromthepotentialflowtheoryandintroducingthecirculation,weyield theliftcreatedbycylinderrotationas:
L = freeV free = freeV free Vrot dl = freeV free Rcyl cyl 2Rcyl L = freeV free 2Rcyl cyl
2
[1.1]
where dl is the integration of the full circumferential length of the cylinder wall and indexfreeassignsfreestreamparameters.Thereforealiftcoefficientcanbeexpressed:
CL =
L 1 freeV free 2 D 2
2 Rcyl cyl V free
= 2 ;
[1.2]
Utilizing1.2, Prandtlpresumedmaximumnondimensional lift obtainable could be no biggerthenCLmax=2=4(for=2case)sincesocalledahalfsaddlepoint(Figure1c) would move to the flow below creating a two zones of the flow with the formed vorticityatthewallsofthecylinder.However,recentinvestigationsonrotatingcylinder phenomena done by Tokumaru and Dimotakis [38] showed that for a large enough cylinder aspect ratio AR=18.7, Reynolds numbers Re=3800 and a rotational rate =5, mentioned vorticity is convected and diffused from the cylinder in disproportionate manner, therefore increase of the lift force could be observed above Prandtls limit. Glauert also found possibility of exceeding Prandtls limit for high values of , under circumstancesthatflowseparationbehindthecylinderissuppressed[26].Otherauthors, such as Kang and Choi [21] report strong 3D effects for Re > 200 which negatively influencesliftcharacteristicofthecylinderandInghamandTang[36]defineacylinder laminarregimeforRe<47andforsmallrotationalrates(<3).
Figure2.ComputedwakebehindtherotatingcylinderforRe=100and=1[35]
Stojkovic, Breuer and Durst [35] prove a logarithmic dependence between a critical rotationalratecritical,whereavortexsheddingdisappearsandReynoldsnumberstating theindependenceofaStrouhalnumberSt,forreasonablerangeofrotationalratevalues belowcritical.Alsocomputationsconfirmahighlyasymmetricwakebehindthecylinder (see Figure 2). Supercritical rotation rates reveal that the threedimensionality of the flow is largely suppressed due the Coriolis forces being predominant comparing to
InvestigationoftheSavoniustypeMagnusWindTurbine________________ convectionandviscousdiffusion[33]butthecertaincentrifugalinstabilitiescouldoccur (seeFigure4)[26]. Conclusions emerged from most of the papers agree about at least two important aspectsabouttherotatingcylinderflowmodeling: Specialattentionshouldbepaidondefinitionoftheouterboundaryconditions, since they are very hard to model adequately. This comes due the Reynolds number dependent length scale ratio between cylinder diameter and viscous diffusionlength[26],thereforestreamfunctionneedssomekindofenhanced boundarycondition. Forces acting upon the cylinder suffer from temporal instabilities in flow evolutionin earlyflowdevelopment stages,andcyclical behaviorindeveloped stageoftheflow.Therefore,solvingoftheunsteady,timedependent2DNavier Stokesequationsisnecessarywithsufficientlysmalltimestepstocatchtemporal instabilities(seeFigure3).
Figure3.Timehistoriesofliftanddragcoefficientfor2Dand3DflowforRe=200and=5[26]
10
InvestigationoftheSavoniustypeMagnusWindTurbine________________ In spite of the fact that hundreds of papers have covered problematics of flow around rotatingcylinder,previousbriefoverviewwasnecessaryonlyinordertogivebasicidea aboutthisextremecomplexityofthephenomena. Mostrelevant conclusionsforthe running studyofMagnus wind turbine are summed upintheexperimentalworkofTokumaruandDimotakis[38]andnumericalapproach by Mittal [26]. Studies proved a strong dependence between aspect ratio of cylinders (ratioofspanwiselengthtodiameter)andliftcoefficient.Thelargertheaspectratio,the liftvaluesareclosertothetheoreticalonesfora2Dflow(seeFigure3).Alsotheslip wall case which corresponds to the cylinder with end plates (Figure 4 top), shows reduction of flow separation behind the cylinder comparing to the noslip wall case (Figure4,secondfromthetop)forthesamevaluesofAR.Asaresult,drasticdropofthe dragandincreaseoftheliftcoefficientwasnoticed(seeFigure3).
Figure4.IsosurfacesofthespanwisevelocityforvariousARofcylinderinthe3Dflowfor Re=200and=5[26]
An analysis undertaken for investigation of Savoniustype Magnus wind turbine includesrotationalliftstudy,butitsrathersimpler.Oneofthereasonsliesinafactthat rotationalrate(laterintroducedasatipspeedratiooftheSavoniusrotor)isverycloseto
11
InvestigationoftheSavoniustypeMagnusWindTurbine________________ unity;thereforesupercriticalrotationscanneverbereachedundernormalcircumstances. Thisisinasametime,amaindisadvantageoftheconceptpresentedinthisstudyand willbeexplainedmoredetailedinthefollowingchapters. However,recentconclusionsaretolargeextentapplicableforlowerrotationalrates.The contemporarydesignoftheMagnuscylindersandSavoniusrotorsismainlyinfluenced bytheresultsofmentionedstudies.
1.1.2ShortHistoricalNote
One of the first and most famous use of the Magnus effect happened in early 1920s, whenaGermanengineerAntonFlettnerpatentedwhatbecameknownsubsequentlyas the Flettner rotor for a ship propulsion system instead of the ships fitted with sails. Vessels named Buckau (later renamed to BadenBaden) and Barbara (see Figure 5) were yielding considerable dynamic stability due the low center of gravity caused by the positionofthedieselenginesthatwerepoweringthecylinders.However,Magnusforce createdwasinsufficienttoputthisconceptintothecommonuse.
Figure5.ShipBuckau:rotatingcylidersutilizedtocreateMagnuspropulsionforce[37].9thof May1926,aftersailingacrossAtlantic,shipenteredNewYorkharbor.
Nowadays, rotating cylinders are playing important role in industrial fluid dynamics investigationsasanactivevortex suppression devices against flowinduced vibrations. Theyactbydisruptingtheformationofanorganized2Dvortexsheddingstructuresin ordertopreventresonancethatcouldoccurifthesheddingratecorrespondstonatural frequencyofthestructureinaflow[14].
12
1.2SavoniusRotorOutline
1.2.1SummaryofInvestigationsonSavoniusRotor
Not that much literature has been published about an experimental and numerical analysisoftheSavoniusrotoranditsutilization.However,adevelopmentofSavonius rotor is closely related to the investigation and utilization of Magnus effect. Finnish engineer, Sigurd. J. Savonius noticed that it was possible to harness the wind so as to maintaintheFlettnercylinder(Chapter 1.1.2)rotating and in that way toeliminate the needforaDieselengineusedforthecylinderrotation[34].Ideawastosimplysplitthe rotorinhalfanddisplacesidewaystwosemicylindricalsurfaces(paddles)alongthecut plane. AftertheSrotorwasofficiallyintroducedinthe1928[32],relativelysmallnumberof papersoccurredsincethattime.Bach[4]investigatedpossibilityforimplementationofa newpossiblebladeshapes,anduntilmid1960swhenamoreseriousexperimentalwork in terms of utilization took place, no serious effort has been done in improving the design of the Savonius rotor. As a dragdriven type machine with low efficiency, this conceptdidntsucceedindrawingtoomuchattention. In the 1970s, following the renewed interest in wind energy Sandia National Laboratories launched a series of experiments, Blackwell, Sheldahl, Feltz [5], with the purpose of investigating a torque characteristics of 2 and 3 bucketed Savonius configurations.Afunctionofthesetestswastomakeacorrectionforasevereblockage factorsthatmostofthepreviouswindtunneltestswerenotimmuneto. During the 1980s detailed analysis of the Savonius rotor experimental aerodynamics wasdonebyUshiyamaandNagai[39]wheredynamicandstatictorquemeasurements weretakentogetherwith thestarting and power characteristics for a various Savonius geometrical parameters. The unsteady flow field around the Savonius rotor at the maximum power performance was also studied by Fujisawa [15] using a smokewire flowvisualizationandahotwireanemometermeasurements. Inapassedtwodecadesnumericalinvestigationfounditsplaceininvestigationofthe Savonius rotor aerodynamics. Modi and Fernando [11] used a discrete vortex method forpredictionoftheperformanceofaSavoniusturbineforbothstationaryandrotating cases.SimilarinvestigationfollowedbyFujisawa[16]andrecentlyverycomprehensive 3D flow analysis around a Savonius and Bachtype turbine was taken by Ishimatsu, KageandOkubayashi[19]. Also,effortwasmadebyCochran,BanksandTaylor[9]tocreateacorrelationbetween computational,reducedscaledandtherealsizemodelinordertodetermineaSavonius turbinepowerperformancecharacteristicandefficiencyinacosteffectiveway.
13
1.2.2SomeCharacteristicsoftheSavoniusRotor
Not particularly big, but a constant interest in Savonius wind turbine is kept until nowadays.AreasonforthisliesinmiscellaneousadvantagesthatSavoniusrotoroffers: Simple,easytobuildandalowcostdesign, Veryhighstartingtorquethatgivesthemadvantageoflowcutinwindspeeds, Properdesignenablesoperationevenatthehighwindspeeds(whenmostofthe highspeedHAWTmustbestoppedforsafetyreasons), UsingarecentlydevelopedLcriterion[25]itwasshownthatSavoniusrotoris more resistant to mechanical stress then any high speed HAWT, i.e. for a same stressvalue,powerperunitsurfaceishigherthenforthehighspeedHAWT, Abilitytooperateregardlessofthewinddirection, Lownoiselevel. On the other hand, Savonius rotor suffers from at least several serious disadvantages: Slowrotationalrateinatermsthatrotationalspeedoftherotoristhesameorder ofmagnitudeasawindvelocity, Wind and load fluctuation cause changes in the output voltage and the frequency,hencetryingtofollowalowcostadvantage,thissystemsarenotto beconnectedtothepublicelectricalgrid, Savoniusrotoriflargeinsize,mustbeinstalledclosetotheground, Greatermaterialexpenditurepersquaremeterofsurfaceusedforpoweringthe VAWTincomparisontoHAWTinstallations[30], LowefficiencyandpowercoefficientCPduethepooraerodynamicalproperties ofthepaddleswhicharemostlydrivenbythedifferenceofthedragforces(see Figure6). Indeed, Savonius original prediction of 31% efficiency in a wind tunnel and 37% efficiency on the open field was overestimated. As Savonius himself stated, Betzs prediction of roughly 20% efficiency was more likely to be correct. Some sources and recent researches on that field indicate that former high efficiency of 30% or more can actually be obtained with carefully chosen design, but no additional or a certainly widespreadconformationwasfoundonthisissue[20]. Mostofthepresenteddisadvantagesarepointingclearlytoimplausibilityofdeveloping large VAWT systems and wind farms. However, most of these disadvantages are vanishing if we consider use of the small Savonius wind rotors instead. In fact, until nowadays,smallSavoniusrotorshavefoundtheirplaceintheindustry.Theyarewidely used as a centrifugal ventilators, airturbines, flowmeters or a turbines for harnessing the tidal power. In wind power industry for reasons given above, their utilization is
14
InvestigationoftheSavoniustypeMagnusWindTurbine________________ limited,sotheyareusedaslowcostwaterpumpingdevices,irrigationdevicesorfora localhouseholdelectricitygenerationi.e.abatterycharging.Also,ahighstartingtorque of the Savonius rotor is used as a compensation for poor starting performance of the Darrieuswindturbine,soitisacommonpracticetoinstallSavoniusturbineonthemain shaftoftheeggbeaterrotor.
Figure6.Performanceoftheconventionalwindconversionsystemsgivenasefficiencyvs.tip speedratio[24]
1.2.3OperatingPrinciple
OperatingprincipleoftheSavoniusorsocalledSrotor(seeFigure7)israthersimple andisverysimilartotheoneobservedonsimplecupanemometers.Dragforcecreated bycuporsemicylinderlikesurfaces(paddles)producesthetorqueonthemainholding shaftthuscreatingpowerthatcouldbeutilizedformultiplepurposes. In the case of a Savonius rotor, it is shown that a geometrical characteristics such as a separationgapbetweenpaddles, overlap ratioand aspect ratio are found to be a most importantforoptimumperformanceoftherotor[39]. Wethereforedefinethoseparametersthroughthefollowingrelations: Aspectratio: AR =
HS ; DS
Overlapratio: OL =
a ; DC
Separationgap: GP =
b ; DC
15
a) b) Figure 7. Various Savonius rotors: a) Chowchilla, California [2], b) EMAT Ltd, England commercialmodel[10]
Figure8.BasicgeometricalfeaturesofSavoniusrotor
In the equations, DC is a diameter of the semicylinder (paddle) and DS is the overall diameter of the Savonius rotor. After adopting some convention we say that positive value of overlap ratio OL indicates that there is an overlap between contours of semi cylinders. On the other hand, as can be see from numerical definition above, positive separation gap GP means that contour of one paddles surface doesnt penetrate a diameter crosssectional plane of the other semicylinder. Therefore, a rotor presented on Figure 8.a can be described with a positive overlap and a zero gap and the one on Figure8.bwithanegativegap.
16
InvestigationoftheSavoniustypeMagnusWindTurbine________________ The relevant data important for a Savonius performance evaluation are a rotor static torque, power and torque coefficients and starting characteristics function i.e. a rotational speed of the rotor. Numerous experiments showed that the optimum performanceintermsofthepowerefficiencyisobtainedforAR4,OL=2025%and GP=0(5)%.
0.25 No Plates Experiment Fitted With Plates Experiment Fitted 0.2
0.15
Cp 0.1 0.05
0.5 Tip Speed Ratio s
1.5
Figure9.TipplateeffectonSavoniusrotorperformance[39]
Also,confirmingresultsvalidforarotatingcylinders(Chapter1.1.1)useofatipplates have shown to improve an overall characteristic of rotor significantly (see Figure 8.c). EffectofatipplatesnotsodistinguishedforlowtipspeedratiosoftheSavoniusrotor (lowrotationalspeedregimes)wasfoundtobeindispensableforatipspeedratiosS> 0.5. Increase of about 25% in a peak power efficiency was noticed for a range of a tip speedratiosthatwasfoundtobesignificantlyexpandedinafirstplace(seeFigure9). Diameterofthetipplate DP isabout10%largerthendiameteroftheSavoniusrotorD incaseofoptimumperformance,i.e. D P DS 1.1 .
1.3IdeaBehindSavoniustypeMagnusWindTurbine
Evolution from a classical HAWT concept to the one which is the subject of our investigationisrepresentedonFigure 10.Ideadeveloped by Kozlov and Bychkov [22] wastoreplaceanairfoils(seeFigure10.a)witharotatingcylinders(seeFigure10.b)in ordercreatetheliftforceutilizingaMagnuseffect.Asixbladedprototypedevelopedat InstituteofTheoreticalandAppliedMechanicsfromNovosibirsk,Russiaand5bladed
17
InvestigationoftheSavoniustypeMagnusWindTurbine________________ 12kW rated power commercial model developed by Akita Magnus Association [1], Japan were both proved to be operational. A certain amount of electrical power is constantly needed to facilitate the cylinders. Indeed, at least two sources indicate relationbetweenapowerlossestorunthecylinderandthecylinderdrag.
a)
b)
c)
Figure 10. Lift generating device on the wind turbines: a) airfoil, b) rotating cylinder, c) Savoniusrotor
18
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Goldstein[13]suggeststhatpowertoruncylindercouldbeexpressedasanequivalent of 20% increase of the drag for stationary cylinder. However, in more explicit way, regarding the lift of the blunt bodies, Hoerner [18] states that power to rotate the cylinderin2Dflowduetotheaerodynamicforcescanbeestimatedfromtheequation: [1.1] Paero = 4.762 10 5 C D U 2DN ;[W] where CD is the skin friction drag of the rotating cylinder for corresponding Reynolds number case, U is the tangential rotational speed of the cylinder, D is the cylinder diameterandNisthecylinderrpm. Thevalueobtainedpresentsapowerneededforcylinderofunitylength,soarescaling is necessary to get the power to run the cylinder of an arbitrary length. However this expression was found to substantially reduce power requirements since electromotor losses and unavoidable friction in bearings and transmission system should also be accountedfor.Asaresult,neededpowertorunthecylinderscanbeconsiderablybigger. Thereforewehavegeneralexpressionforpowerneededfromelectromotors:
Pmotor =
Paero
motor bearings transmission
[1.2]
Thisexpressionwilllargelydependnotonlyonthemotorchoiceandtype,butalsofrom typeandconditionofbearingsandtransmissionsystem. Forthepurposesofillustration,Table1containsapproximatedandextrapolatedvalues for the power needed to run the cylinders at 500 rpm, for the model made by Kozlov and Bychkov at SBRAS, Akita turbine and the rotating cylinder model tested by Reid [29].ValuesaregivenasthepowerinWatts,scaledperonecylinderandunitylength. Also,twoReynoldsnumbervaluesweregivenbasedoncylinderdiameterandtheblade lengthtogetherwiththeirratioRe, Case Red ReL Re P[W] SBRAS[22] Reid[29] AkitaTurbine[1] 3.5x104 4.4x104 7.8x104 1.2x105 5.9x105 1.8x106 3.4 13.4 23.1 17 6 ~20
Table1.Powerrequirementsforrunningthecylinderat500rpm
WhatshouldbenotedhereisthatdiscrepancyforSBRASresultsshouldbeaccounted foratransmissionlossesforsixrotatingcylindersthatarepoweredbyamotor.Incase ofaReidexperiment,onlyone,adoublybearingsupportedcylinderwastested.
19
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Alsowhatcomesasanimportantconclusionisthataspectratiooftherotatingcylinders should be high enough in order that turbine could support itself with the sufficient power for the cylinders. Indeed, only Akita model that is holding a very high aspect ratio of the cylinder blades is capable of running in an autonomous regime. For the reasons of confidentiality only can be said that the overall power requirements for runningthecylindersareroughlybetween1015%ofthepowergeneratedbythewind turbine. Also should be taken into account that cylinders are equipped with the spiral superstructuresandtipplatessopowerefficiencyofsucharrangementisbiggerthenit wouldbeforabarecylindercase[22],[1]. ThepurposeofthisstudyistobringanattempttocreateMagnuseffectpoweredwind turbine that would utilize Savonius rotors instead of the cylinders (see Figure 10.c). Majoradvantagesofthissetupwouldbe: NoexternalpowerisneededtorunSavoniuscylinders, Simplicityofconstructionduethelackofelectromotorsandtransmissionsystem, Costandweightreduction, Furthersimplifiedmaintenance.
Figure11.LocalvelocitiesandforcesonSavoniusrotor
ExploitingselfstartingcharacteristicoftheSavoniusrotor,thecentralideaistoutilizea dragforceofthesemicylindersforcreatingaMagnusliftovertherotatingsurface.As can be seen from a Figure 11, created lift due the rotation is normal to the relative
20
InvestigationoftheSavoniustypeMagnusWindTurbine________________ velocity vector Vrel, seen by the Savonius rotor section which depends on both, wind velocityandrotationalspeedoftherotor. Vectorsumisdefinedthroughtheflowanglethatinthecaseofneglectinginduction factors can be defined over an inverse tangent function of the Magnus rotor tip speed ratio:
= cot 1
Vrot Vwind
R = cot 1 M M V wind
1 = cot 1 M = ; 1 tan M
[1.3]
Tangential component of the created force Ft, would generate the torque on the main shaftoftheturbine.SinceMagnusliftdependsontherotationalspeedofthecylinderi.e. tipspeedratio(seeChapter1.1)whatcomesasaconclusionisthatweareinterestedin exploiting the speed characteristic of Savonius rotor, not its torque and power characteristic. However,previousparametersarenotalwaysinaperfectcorrelation.Theexperiments run by Ushiyama and Nagai [39] showed that for certain separation gap values of the Savonius rotor and an overlap ratio of the Bachtype rotor, best rpm performance in idlingregimeisnotcorrespondingwiththehighestCQandCPvalues.Asanexample,for the later Bachtype rotor, overlap of 50% between semistructures yields the superior rpm performance comparing to the other, smaller overlap configurations, but power efficiencyisabout16%smallerthenforthecasewith10%overlap.Higherrpmcanbe explained simply by applying the conservation of momentum theory. Bigger overlap holdssmallermomentofinertiavaluethereforehigherrpmcanbeobtained.
1.4PlanofInvestigations
Investigations of the Savoniustype Magnus rotor includes several studies of different aspectsoftherotor. First, static torque of the different Savonius rotors as a measure of selfstarting characteristic is investigated numerically using commercially available CFD tools. On the basis of those results, most successful configurations were used for creating the practicalrotordesign.Abriefanalysiswasmaderegardingthechoiceofmaterialsand propertiesoftheprototypestobetested. An experimental analysis included a torque and power measurements of the Magnus rotorpoweredbySavoniusbladesinthewindtunnelforthevariouswindspeedsand numberofblades.Aresultswerecomparedinordertofindthemostpromisingdesign intermsofthehighestpowerandtorquecoefficient.
21
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Power characteristic and noticed behavior of the turbine, imposed a supplementary investigationofthemechanicalandmodalpropertiesoftheSavoniusbladessinceitwas foundthattheyyieldasignificantimpactontheMagnusrotorperformance.
22
2.DeterminationoftheStaticTorque
2.1Definition
Beforestartinganyexperimentalinvestigation,oneshouldatleastintuitivelydetermine basicparametersthatwoulddefinethedesignoftheMagnusturbine.Asalreadyseen, most important of them is a crosssection design of the Savonius blade that plays an important role in defining the behavior of the rotor, therefore contributes to the lift characteristicofthedevice. One of the criteria needed to establish a good choice among different profile configurationsusedtopowerupaSavoniusrotorwouldcertainlybeanassessmentof the static or starting torque. For a Savonius rotor, static torque corresponds to a maximum value of the torque when device is blocked i.e. without ability to rotate. All other torque values for full operating range of rotor speeds or tip speed ratios will be lowerthenthisvalue.Thereforewecanstatethatstatictorquerepresentsaproperty,an abilityoftheexternallypoweredmachine(inthiscaseawindturbineorrotor)tostart itself.ASavoniusrotorasstatedinpreviouschapterhasanadvantagethatyieldsavery highvalueofstatictorque.
2.2NumericalStudy
Tofindthebeststartingcharacteristic,finiteelementbasedNavierStokescodesuchas ANSYSFLOTRANwasutilizedforsimulatingtheflowaroundthebladerotor.Problem was treated as a stationary, twodimensional; the rotor crosssection was placed into computationaldomainandpressuredistributionandresultingmomentswerefound.
2.3GeometryDescription
For the purpose of finding the optimum design, some ofthe most representative rotor crosssections from previously reviewed literature (Chapter 1.2.1) were taken into consideration for rotor design. All rotors are yielding the same DS=0.06 m diameter measuredasthedistancebetweentherotorouteredgeswhichcorrespondstothespan oftherotoralongxaxis(seeFigure12). A first profile (see Figure 12.a) proposed by Bach [4], was taken as one of the most successful solutions for high torque and rpm performance, which was confirmed later by Ushiyama [39] and Ishimatsu [19]. Geometrical features for Bach profile were influencedbythesetupsuggestedbythelaterauthor.
23
a)
b)

c) d)
e)
Figure 12. Various rotor configurations: a) Bach, b) Benesh, c) Modi, d) Savonius OL=0.21, e) SavoniusOL=0.67
Second profile presented by Rahai [28] is based on an investigation of A.H. Benesh where profile that combines a lift and a drag characteristics, is introduced in order to improve a rotor characteristics. However, according to the author, lift contribution to overall performanceis limitedonly ina certain range ofangles ofattack from 0 20 and 180 200. As Figure 12.b shows, Benesh profile camber is approximated by the followingsetofequations:
m 2 px x 2 ; p2 m (1 2 p ) + 2 px x 2 ; y= 2 (1 p ) y=
for for
0 x 0.3 ; 0.3 x 1 ;
[2.1] [2.2]
24
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Profileappliedinourcaseyieldssmallthicknessbutforx0.3,constantinsteadofthe taperedthicknesswasapplied. AthirdprofileproposedbyModi,FernandoandRoth[27]wasyieldinganoutstanding power performance, however the authors contributed such results to the too high blockage factor during the wind tunnel testing (B = 16.4%). A major geometrical parametersareconsideredtobeprofilearcangle,andtheratiobetweenlinearpartof profileandthearcradiusp/q(seeFigure12.c).Theoptimumdesignisconsideredtobe for=135andp/q=0.2. The fourth and the fifth profile are typical Savonius configurations with semicircle displaced crosssections described more thoroughly in works of Ushiyama [39] and Blackwell [5]. They are yielding overlap of 21% (see Figure 12.d) and 67% (see Figure 12.e)respectively.
2.4FlowDescription
The flow was solved for 6m/s wind velocity case and the standard atmosphere parameters for pressure, temperature and viscosity. In order to investigate torque for different flow angles, velocity is given through its components in x and y directions applied on the domain boundaries. Since the all models are yielding rotational symmetry, the flow was modeled for the angles in range between 0 180. Reynolds numberbasedontherotordiameterforflowspeedof6m/sis:
Re =
V free DS
6 m s 0.06m 24000 ; 1.5x10 -5 m 2 s
[2.3]
TakingintoaccountthecomplexgeometryandthevalueofReynoldsnumber,itcomes asaconclusionthatturbulentflowinsteadoflaminaronehastobeconsidered.
2.5NumericalMethod
FlowwillbetreatedasviscousandsolvedusingtheNavierStokessetofequations[3]:
V x ( V xV x ) (V xV y ) P V V + + = g x + Rx + e x + e x t x y x x x y y
+ Tx ;[2.4]
V y t
(V yV y ) (V xV y ) P V y + = g y + Ry + e y x y x x
V y + e y y
+ T y ;[2.5]
25
InvestigationoftheSavoniustypeMagnusWindTurbine________________ alsofromthelawofmassconservationcomesthecontinuityequation:
(V x ) (V y ) + + = 0 ; t x y
[2.6]
ThegtermsaregravityaccelerationsandRtermsrepresentanysourceterms.Both aretobeneglectedinfollowingsimulations.TheTtermsareviscouslossterms,which are eliminated in the incompressible, constant property case, but in compressible flow case,arenot.
2.6TurbulenceModel
Turbulence model implemented for all simulations is RNG (ReNormalized Group) model,whichisconsideredtobeconvergencestableandeffectivewherethegeometry has a strong curvature [3]. Savonius rotor crosssection computational domain leaves spaceforsuchconsiderations. Bydefinition,RNGmodelisaktypeofmodelderivedfromtheinstantaneousNavier Stokesequations,thatusestechniquecalledrenormalizationgroupmethodstoderive theequationsfortheturbulencekineticenergyandtheturbulencedissipationrate[12]. Thebasicmethodisasimpleiterativeproceduretoeliminatethesmallereddiesandthe replacement of their mean effect on the remaining larger eddies by increasing the viscosity. Similar to other techniques, this is another way to damp out the smaller eddies. The resulting equation is rescaled through an iterative procedure until two successiveiterationsmatchclosely. Standardkmodelyieldsturbulentkineticanddissipationequations:
k ( V x k ) (V y k ) t k t k + + t + + + = t x y x k x y k y C 4 t T T gx + + gy t x y
and
[2.7]
( V x ) (V y ) t t 2 + + C1 t C 2 + + = + t x k k y x k x y k y [2.8] C (1 C3 ) t T T gx + + gy x y t
26
InvestigationoftheSavoniustypeMagnusWindTurbine________________ where k is a turbulent energy term and is a viscous dissipation term. C terms representconstantsandtisaturbulentPrandtl(Schmidt)Number.Turbulentviscosity tiscalculatedas:
t = C
whereCisalsoaconstant.
k2
[2.9]
RNGmodeldealswiththedifferentialequationforturbulentviscosity[12]:
2k d
= 1.72 d 3 1 + C
[2.10]
where = e / ,andC100.
Fromequations[2.9]and[2.10]usingarenormalizationprocedureweobtainC=0.0845. Also,C1constantisgivenasafunctionofoneoftheinvariants:
C1 = 1.42 1 + 3 1
[2.11]
ValuesoftheconstantsusedforRNGmodelaregiveninaTable2: Value C1 C2 C k Default 0.12 1.42 1.68 0.085 0.72 0.72 4.38
Table2.ValuesofconstantsforRNGturbulencemodel
27
2.7DomainSizeandMeshConfiguration
Figure13.Exampleofcomputationaldomainofthe2Dflowaroundmodeledrotors
A mesh was formed on an operational 2D domain taking into account distances from modeledrotorsthatareembeddedinthemiddleofthedomain.Coordinatecenterofthe domaincorrespondswiththecenterofmassandrotationforallprofiles,intermstoease findingthestatic(pitching)moments.AscanbeseenfromtheFigure13,onabasisof modelingtheflowaroundcylinder[35],circularcomputationaldomainwaschosenwith its diameter 100 times bigger then diameter of the rotor models. In another words, domain yields a distance of fifty rotor diameters around the rotor. A circular shape makes a good candidate to represent the far field boundary since it has no discontinuitiesinslope,enablingthepartialcontinuityofasmoothmeshintheinterior ofthedomain.Oneshouldnoticethatoutsidedomainwasformedusingthesemifree quadrilateralmeshwithforcedcellsizegradients(decreasingspacingratio)indiameter direction. Since aerodynamic properties of the rotor should cover functional range of the flow directions (0 180) using the velocity components of the flow as a part of the outer boundary conditions, a grid was formed in a way that gradual change in a cell size should remain almost same regardless of the flow direction. All discontinuities on the domain are due the complexity of the rotor geometries and special attention such as a forcedmeshingtechniquehastobeappliedintheclosevicinityoftherotors(seeFigure
28
InvestigationoftheSavoniustypeMagnusWindTurbine________________ 14).Quadrilateralgridcellswerealsousedformodelingtheforcedmeshbecausethey canbestretchedeasilytoaccountfordifferentsizegradientsindifferentdirections[3]. Some characteristics of the mesh are given in Table 3, with the pressure convergence terminationcriteriaasoneofthemainparametersforevaluatingthegridquality.
Rotortype BachType BeneshType ModiType Savonius OL=0.67 Savonius OL=0.21 No.of elements 9765 7638 7761 6043 8853 No.of nodes 9961 7792 7917 6189 9012 No.ofnodes ontherotor wall 81 121 121 65 65 Pressure termination criteria 1x107 1x107 1x107 1x107 1x106
Table3.Gridparametersandterminationcriteria
The gradients normal to the rotor wall are greater than ones tangent to the surface. Consequently, the cells near the surface have been modeled with the high grid aspect ratios (see Figure 15). Such biasing of the grid is useful, so the boundary layer can be properlymodeled.
a)
29
b)
c)
d)
30
e) Figure14.Meshingofthevariousrotorconfigurations:a)Bach,b)Benesh,c)SavoniusOL=0.67, d)Modi,e)SavoniusOL=0.21
Figure15.BoundarylayeradequategridonBachtyperotorwall
31
Bach Profile 7 Back Back Front Front Profile - Windward Side Profile - Leeward Side Profile - Windward Side Profile - Leeward Side
Benesh Profile 4.5 Leeward Side Windward Side 4
3.5
5
4 Y+
2.5 Y+
1.5
2
1
1
0.5
0.1
0.2
0.3
0.4 0.5 0.6 Distance along the profile (x/c)
0.7
0.8
0.9
0.1
0.2
0.3
0.7
0.8
0.9
a)
Modi Profile 3.5
Leeward Side Windward Side
b)
Savonius OL=0.67 Profile Back Back Front Front Profile - Windward Side Profile - Leeward Side Profile - Windward Side Profile - Leeward Side
2.5
2 Y+ Y+ 1.5
0.5
0.1
0.2
0.3
0.7
0.8
0.9
0.1
0.2
0.3
0.7
0.8
0.9
c)
7
Savonius OL=0.21 Profile
Back Back Front Front
Profile - Windward Side Profile - Leeward Side Profile - Windward Side Profile - Leeward Side
d)
4 Y+ 3 2 1 0 0
0.1
0.2
0.3
0.7
0.8
0.9
e) Figure 16. Y+ distribution along the walls of various profiles for Re = 2.4 x 104 : a) Bach, b) Benesh,c)Modi,d)SavoniusOL=0.67,e)SavoniusOL=0.21
Inordertomodeltheboundarylayerintheapprovedmanner,validrangeofY+values was taken into account so that the node on the first celllayer could be put on
32
InvestigationoftheSavoniustypeMagnusWindTurbine________________ appropriate distance away from the wall. The needed distance is practically independentfromthemesh,andbelongsonlytothenatureoftheflow(infirstinstants thevelocity). Followingtheprocedurefrom[7],thevalueofsmallestnodedistancefromthewallfor the adjacent cells is given as function of the operating Reynolds number, reference lengthanddesiredvalueofY+.ForY+=3whichisconsideredtobereasonablysmall value,wecomeupwiththevalueofdwall0.12mmwhichisinanagreementwithnode distancestakenformeshingthefirstcelllayer.Also,aruleofthumbwasused,andthe growthratioofcellsizeintheboundarylayerissetuptobeabout1.21.25(seeFigure 15). ValuesofY+forthefirstmeshcellaregivenofFigure16.Theyrepresentthereadings fromallflowadjacentnodesonboth,windwardandleewardsideoftheprofilewalls, andinthecaseofBachandSavoniustyperotors,theyaregivenforbothsemistructure profiles.
2.8BoundaryConditions
Domains outer boundary conditions were defined as the Dirichlettype conditions where velocity and pressure loads are given. In this case, coupling of pressure and velocity values for a domains outer boundaries gives a so called pressure far field boundaryconditionthatsimulatesafreestreamconditionsatinfinity.Forthisreason,to effectivelyapproximatetrueinfiniteextentconditions,weplacedthefarfieldboundary farenoughfromtherotormodels. Anoslipboundaryconditionwasappliedontheallcontourlinesoftherotorstested. It represents a stationary wall case, where a fluid layer adjacent to the wall suffers no motion,thereforehasbothvelocitycomponentssettozero.
2.9SolverAlgorithm
FLOTRANdistinguishesbetweentwodifferentalgorithmsettings.Theybothbelongto theclassoftheSemiImplicitMethodforPressureLinkedEquations(SIMPLE). SIMPLEF (segregated pressurevelocity coupled algorithm) uses TriDiagonal Matrix Algorithm(TDMA)solver.Neverthelessthemaindisadvantageisthatevenwhenexact solutions are obtained on those individual equations for pressure, velocity, energy, turbulence and momentum, the overall rate of convergence will not improve. This is consequenceoftheweakcouplingbetweenthepressureandthemomentumequations
33
InvestigationoftheSavoniustypeMagnusWindTurbine________________ [3].Otherproblemisconnectedtotherelaxationfactorsthatcaninthiscase,causelarge instabilitiesinthesolutioniftheybecametolargesovaluesabove0.5forpressure are notrecommended.Thereforearateofconvergenceisslower. On the other hand, the overall rate of convergence of the SIMPLEN (enhanced segregated algorithm), can be improved considerably when more exact solutions are obtained for each individual equation. More stabile solution causes faster convergence and possibility to increase relaxation factors up to 0.8 for momentum equation for instance. new=(1RELX)previous+RELXcalculated [2.12] Thereforeinordertoobtainfasterconvergencerate,SIMPLENalgorithmwasusedfor all rotor crosssections. Due the fact that SIMPLEN algorithm is very sensitive on any grid inconsistency, one can be used as a test for grid quality. If we disregard initial instabilities,convergenceofthepressureequationsshowedmonotonedecreasebehavior. Therefore it is once more confirmed that mesh around the crosssection was properly modeled. Table4.containsnumberofiterationsbasedforreachingthe106convergenceinvicinity ofthemajorflowangles. Numberofiterationsfor106pressureterminationcriteria Rotortype 0 45 90 135 180 BachType BeneshType ModiType Savonius OL=0.67 Savonius OL=0.21 1750 1620 1310 510 1290 980 1550 2600 480 1910 1520 1970 1270 520 2200 2500 1640 1370 730 2630 2640 1650 1390 500 2870
Table4.Numberofiterationsforvariousrotorsandflowangles
Ascanbeseen,SavoniusrotorwithOL=0.67yieldsmuchbetterconvergencethenrestof configurations mainly due the successive grid adjustment and refinement procedure. However the results for the improved grid held significant change neither in the flow picturenorintheresultsforthetorque. A solver permits choice between the steady and transient state analysis, but due the natureofourworkandalsoprocessorandcomputationaltimerestrictions,steadytype ofanalysiswasdeployed.
34
2.10Postprocessing
The postprocessor tool computes a force quantities integrated for chosen nodal results overadefinedsurface.Thetotalforcesaresimplythexandycomponentsofsumofall thepressureandviscousforces.
r r FX TOTAL = FPRESS + FVISC i
[2.13] [2.14]
r r FY TOTAL = FPRESS + FVISC j
A resulting moment Mz around the center of mass of the profile is simply calculated fromthesumofmomentscreatedbytheviscousandpressureforcestakenforadomain contour.ByapplyingtheStokestheoremweyield:
r r r r M TOTAL = x FPRESS dl + x FVISC dl

dS dS
[2.15]
wherexisthepositionvectorofthenodeelementrelativetocenterpoint,anddlisaline elementalongtheboundaryS,ofmagnitudedl.Torquecoefficientthereforecorresponds tomomentcoefficientforcenterpointi.e.pointofrotation:
CQ =
M TOTAL 1 U 2 DS 2 2
[2.16]
2.11Results
2.11.1ResultsValidation
Beforeafurtheranalysis,notewillbemadeaboutvalidityofprocedure.Resultsofthe CFDanalysiswereconfirmedbycomparingthemwiththeexperimentalandnumerical resultsfromModiandFernando[27]&[11]forwhatisinthisstudycalledModiprofile (see Figure 17). Following figures show the results for a pressure distribution and a torquerespectivelyforastaticcasemeasurementsandtheflowangleof=30.Pressure valuesonthefrontandthebacksideoftheModiFernandoprofilewereobtainedusing 46pressuregaugesmountedontheprofile.In current simulation free stream velocity wassettobeVwind=6m/s. What easily can be noticed from Figure 17 is slight difference in designs due the existence of the overlap shift between leading and trailing paddle, where such gap is filledwithcentralshaft.Forcurrentstudy,themajorgeometricalparameters,suchasa
35
InvestigationoftheSavoniustypeMagnusWindTurbine________________ profile arc angle , and ratio between linear part of profile and the arc radius p/q are yieldingoptimumvaluesassuggested.
Figure17.ModiFernandoprofilewiththepressuretapssetup[27]
Figure18.Modiprofilepressuredistributionfor=30andVwind=6m/s
Ontheotherhand,thesegeometricalfeatureswillcausethedifferenceswhentheresults are compared. Since the actual dimensions of the ModiFernando profile and shaft are notknown,onlygeneralanalysiswillbemadehere.
36
Figure19.FlowfieldpressuredistributionoverthetheoreticalModiprofilefor=30andVwind =6m/s
Figure18showsthatnumericallypredictedpressuredistributionfromthecurrentstudy formostofthedomainisyieldingareasonableaccuracywiththeexperimentalresults. The difference is noticeable for the pressure coefficient drops and also for point of separationlocationofthetrailingpaddle. Numerical result from FLOTRAN is presented on Figure 19, and clearly states low pressure zone located closer to the trailing edge then the one stated by Modi and Fernando which is according to the measurements positioned around 32nd pressure gauge. However,mismatchinlocationandmagnitudeoflowpressurezone,canbeattributed tothegeometricaldifferencesbetweentwomodels.Thisdifferenceisexpressednotonly in terms of the Reynolds number disparity, but due the fact that for the equivalent profiletotallengthandrestofparameters(,p/q)leftthesame,abuildinpositionangle ofthebothpaddlesduetheexistenceoftheshaftisnotsame.Bothpaddleedgescloseto thecenterofrotationwillbeslightlyshiftedupduetheexistenceofthecentralstructure. Thiscausesshiftinthelowpressurezoneinmoredownwinddirectionontheleading paddleandmoreupwindonthetrailingpaddle.Also,ahighblockagefactorreported duringModiFernandoexperimentsshouldbealsotakenintoaccount. Figure20showsstatictorquevaluesobtainednumericallyandexperimentallybyModi and Fernando and compared with the current study simulation for a range of a flow angles. The effect of the central shaft is pronounced once again through the shift phenomenaofthecomputedcurves.
37
Figure20.StatictorquevariationwithflowangleforModiprofile
However, the current simulation obtains better agreement with experimental data in termsofthetorqueminimum,sinceforthehighflowangleseffectoftheshaftisnotso distinctiveduethetrailingpaddleshadoweffect.Ifwecomparestatictorquevaluesfor previously examined =30 flow angle case, we come up with the reasonable match betweenthosedata.Arelativeerrorcomparingnumericalresultswiththeexperimentis given in a Table 5 below. A value of 6.6% represents a good agreement with the experimental data taking into account the differences between these two groups of measurements. StaticTorque [%] ModiFernandoExperiment ModiFernandoNumerical CurrentAnalysisNumerical 0.48 0.495 0.51 / 3.3 6.6
Table5.Statictorqueresultsrelativeerror
2.11.2TorqueAnalysis
A summed results are presented on the Figure 21, where static torque is given as a functionoftheflowangle.Itisnoticeablethattorquevaluesareyieldingthesymmetry for flow angles higher then 180. All profiles are characterized by rise of the positive valuesforthestatictorqueinarangeofanglesfrom0upto2245wheremaximum
38
InvestigationoftheSavoniustypeMagnusWindTurbine________________ value is reached. Benesh profile holds the lowest maximum of nearly CQ 0.4 for approximately27comparingtotheBachprofilewhichreachesCQ0.58foraflowangle of45.ModiandSavoniusOL=0.21areyieldingmaximumvaluesofastatictorquefor approximatelysameflowanglesasaBeneshprofile,whileSavoniusOL=0.67profileis obtainingmaximumtorqueforroughly40. Characterizedbysimilarslope(k7.3x103),statictorqueforallprofilesisdeclining, reachingtherangeofnegativetorquevaluesbetween130and175.Asidefromthefact that both Savonius rotors are yielding moderate values of negative torque (CQ 0.05), Modi profile is represented by distinguishable negative torque of CQ 0.2, while the Bachprofileshowedtendencyofholdingthechangeablebutconstantlypositivevalue, with CQmin 0.1. A torque for Benesh profile seems to vary around the zero, therefore showinghighlystabilebehaviorforhighflowangleswhichisnotofourinterest.
0.6 Modi Savonius OL=0.67 Benesh Savonius OL=0.21 Bach
0.5
0.4 Static Torque Coefficient C Q
0.3
0.2
0.1
-0.1
-0.2
20
40
60
80
100
120
140
160
180
Flow Angle [ ]
Figure21.Statictorqueofvariousbladeprofilesasafunctionofflowangle
HighertorqueforBachandSavoniustyperotorsisobtainedbecauseoftheexistenceof separationgapbetweenpaddles.Inthisway,theairflowfromtheadvancingpaddlecan enter the wake zone ofthe returning paddle and diminishor completely eliminate the existence ofthe vortex structure which is responsible for decreasingpressure therefore creatingawakesuctionzone.
39
a)
b)
c)
Figure22.ComputedvectorflowfieldaroundSavoniusOL=0.21rotor;a)0vortexstructures inside both buckets, b) 16 pressure increase on the walls of the returning bucket, c) 32 centralvortexstructureoutsideofthebucket,withsecondvortexdownwashed.
40
InvestigationoftheSavoniustypeMagnusWindTurbine________________ AnexampleofsuchbehaviorisgivenonFigure22above,foracaseofSavoniusrotor withOL=0.21.Avectorflowfieldisgivenforanglesof0,16and32. Examinationoftheflowrevealsfor0caseatleastthreelargevortexstructuresaround therotor.Oneincenteroftheforwardingbucketandtwointhereturningbucketzone, where the upper one is created due the separation of the flow from the forwarding bucket concave surface. However, the pressure inside of the forwarding bucket is still biggertheninthereturningone.Theairpassingthegapisyieldingasmalldirectional angle,thereforeisunabletocreateabiggermomentaroundtherotationaxis. When the flow angle is increased, the air flows at higher angle, pushing the vortex structuremoreoutsidethebucket. Also, secondary vortex due theseparation from the bucketisalsoshiftedmoredownstream. Foracaseofflowangleclosetotheoptimumone(correspondingtoour=32case)the flow fully penetrates inside the paddle leaving no space for large vortices in the advancingpaddle.Separatedstructureissufficientlyfarconvectedfromtherotor,while the bucket vortex is outside of the returning bucket. The highest impact on rotor behaviorcomeshoweverfromthehighpressurefielddistributedalongtheinnersideof thereturningbucket.
Figure23.VelocityplotaroundSavoniusOL=0.21rotorfor=164
In a same way, a negative starting characteristic is caused when the pressure on the trailingtipofthereturningpaddleexceedsthevalueinsidethebucketwhichispossible for high flow angles. Maximum negative torque is calculated for flow angle of =164 andvelocityplotaroundtherotorisgivenonFigure23.Astagnationpointcorresponds
41
InvestigationoftheSavoniustypeMagnusWindTurbine________________ with the maximum value of the pressure coefficient and the small perpendicular distancefromstagnationpointtocenterofrotationexplainsmoderatevaluesofnegative torque coefficient. Also an attention should be paid on two counterrotating vortex structuresinsidethebothbucketareas.Allcommentsmadeherearemoreorlessvalid foralltheotherprofiles,intermsoftheexplainingthedrivingmechanism.
Figure24.PressureplotaroundBachtyperotorfor=143
Figure25.PressureplotaroundBachtyperotorfor=161
A note will be given as well regarding the Bach profile that showed the outstanding behaviorintermsofpositivestatictorqueforfullrangeofflowangles.Figure24gives
42
InvestigationoftheSavoniustypeMagnusWindTurbine________________ the pressure distribution for =143 and explains the mechanism that enables Bach profilenottofailintheperformanceforhighflowangleslikealltheothermodels. Duetheslimoverlapareaplacedunderacertainangle(inthecurrentcase30),airflow isabletoreachtheconvexcurvatureoftheadvancingpaddleforhighflowangles.As the figure shows, inner side of the paddle is yielding still a significant pressure force comparing to the pressure on the outer side of the returning paddle. With further increase of the flow angle, a pressure slightly drops but redistributes along the inner convexsurfaceincreasingthetorque.Itssecondmaximumvalueisreachedfor=161 and the pressure plot around the profile and Cp values along the inner and outer surfacesofthereturningandadvancingpaddlesaregiveninFigures25and26.Onecan see that in spite the fact that highest value of the pressure coefficient is found on the outer wall of the returning paddle, a total area under that pressure curve for an advancing paddle is bigger. The same is valid for the opposite walls of the both structures.Innerpressureonthereturningpaddleisbiggerthenthepressureinawake behindtheadvancingone,thereforeadditionallycontributingtothetorqueperformance. Alsoitshouldnotbeneglectedthatduethecurvedgeometry,aprofileforthehighflow angles yields tangential component of the pressure force on outer surface of the returning paddle. In this way, regardless of the pressure force loss on the advancing surfaceandtheintensitydrop,profilestillholdsapositivetorque.
1.5 Advancing In Advancing Out Returning Out Returning In 1
0.5
CP
-0.5
-1
-1.5
0.1
0.2
0.3
0.7
0.8
0.9
Figure26.PressuredistributionaroundBachtyperotorfor=161
43
3.BladeDesignandPrototype
3.1DesignObjectives
Afterthenumericalinvestigationofthetorque,acriterionwasestablishedforachoiceof the appropriate design of the rotor blades in order to exploit combination of good performance and reasonable price. Following design objectives have been considered and they represent the datum for design evaluation and directions in selecting the suitableconfiguration. Good mechanical performance in the terms of starting characteristics and high revolutionnumberofblades.Thisparametersiffulfilled,guaranteeselfstarting capability of the blade for low wind speeds, and satisfying torque and power performanceoftheturbineaswell. Lowmanufacturingcostofthebladesisacrucialprerequisiteforthesuccessful turbine implementation. Since the current research is focused on a smalltype Magnusturbineperformance,acomparisonshouldbemadeontheappropriate scale. Situation check reveals that a market price for 28 (0.71m) carbon reinforcedbladeofthefamousSouthwestAirXtypeisinarangeofUSD1520 per blade. This price can be easily explained by decreased expenditure due the massproduction.Thisalsomeansthatourproductioncostsshouldbealsokept toaminimum. Simplicity of construction is also a very important feature. Savonius rotors are sufferingfromimpossibilityofchangingtheirrpmnumberofrotationtherefore they should be treated as a passive mechanisms. This attribute corresponds to thelackofpitchmechanismonmostoftheconventionalsmalllowpowerwind turbines. On the other hand this enables the much simpler construction and thereforeisfavorableintermsofthecostsandmaintenance. Reliability and durability of the turbine are yielding high importance in the futuredesign.Onceinstalled,thepurposeofthewindsystemistorunforyears without any major repairs aside from the annual checking and the regular maintenance procedure. A life expectancy of the system should be at least 10 yearsandduringthattime,itwouldbehighlyadvisablethatnoseriousfailure or incident occurs. A maintenance costs should not exceed the benefit from installingtheturbine. Useofrecycledpartsshouldbealsoseriouslyconsideredasastrategyforcost reductionpolicy.SimplicityoftheSavoniusrotoranddragbasedfeaturesdonot
44
InvestigationoftheSavoniustypeMagnusWindTurbine________________ demand special construction and precise machining like in a case of the conventionalwindturbineblades.Thereforescrapedplasticoraluminumpipes, steelarmatureorcaralternatorscouldbeutilizedfordesigningtherotor. Taking into account stated objectives, a two most promising designs have been taken intoconsiderationforpreliminarydesignoftheblade;aclassicSavoniusprofilebecause ofthesimplicityofthepaddledesignandBachtypeprofileforitssuperiortorqueand rpmcharacteristics.Theywillbetestedforpowerperformance. Choice of materials used for a blade manufacturing is based on an investigation and decisional matrix proposed by Menet [23], where design objectives above were implementedinasimilarway. Outside Conditions Temp. Sensitivity Assembly Easiness 0 0 5 3 2 2 1 3 Rigidity Weight Total 20 18 25 20 16 13 13 23 45 Price 4 4 5 2 4 2 2 4
Material
PolyethyleneHighDensity Polypropylene PolyvinylChloride(PVC) Plexiglas ShockPolystyrene AcrylonitrilButadeneStyrene AcrylonitrilStyrene Aluminum
4 3 5 4 3 3 3 5
5 5 2 3 4 4 4 1
2 2 4 3 2 2 3 5
5 4 4 5 1 0 0 5
Table6.MaterialdecisionmatrixbasedonSavoniusrotorresearchbyMenet[23]
Adecisionmatrixis basedona gradesystemfrom 0 to 5where 0isa grade given for extremelybadperformance/behaviorand5forexcellentone.AmatrixgiveninaTable6 takes into account some important features that are necessary for a successful implementation of theturbine in the outdoor environment. Table shows that PVC and aluminumareshowingthebestoutdoorperformancewiththehighestpriceandweight to be taken into account. However this seems to be the best tradeoff at the moment especiallyonthisstageoftheexperimentalevaluation.
3.2PrototypeDescription
Duringinvestigation,atotaloffourprototypesweremade;threeSavoniusandoneBach bladeweretested.Theywereyieldinganapproximatelythesamelengthsofthepaddle blades, with the different overlap and rotorblade diameters. A sketches of
InvestigationoftheSavoniustypeMagnusWindTurbine________________ manufactured rotor blades with most important geometrical parameters are presented on Figures 27 30 and most important blade features are given in a Table 6. As a supplement,Table7containsbasicmaterialpropertiesusedforfurtheranalysis. An examination of all cases reveals that two different conceptions were used for designingarotationalblade.AfirstthreeprototypesmadeofthePVCplastic,yieldthe centralshaftandthetwobearingsmountedontheendplatesoftheSavoniusrotor.On theotherhand,aBachprototypelacksacentralshaftduetheverysmalloverlapspace inbetweenthebuckets. Paddle Number Central Profile DS Paddle Name OL Length of Axis Type [mm] Material [mm] bearings Shaft Prototype01 Prototype02 Prototype03 Prototype04 Savonius Savonius Savonius Bach 0.22 0.63 0.47 0.5 41 60 90 60 2 2 2 1 yes yes yes no PVC PVC PVC Al
351
Table7.Characteristicsoftestedblades
Material PVC Aluminum Steel
Density [kg/m3] 1400 2710 7860
YoungModulus [N/m2] 3x109 7x1010 2x1011
PoissonRatio 0.4 0.346 0.266
Table8.Materialproperties
46
Figure27.Prototype01isometricsketchwiththecrosssectiondetail
47
48
49
Figure30.Prototype04(Bachtypeblade)isometricsketchwiththecrosssectiondetail
50
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Placing the shaft in theslot would mean blockage of the airflow and loss of the major advantages of a Bach rotor. Since steelmade shaft also adds additional stiffness to the blade, for Bach blade case PVC paddles wouldnt be appropriate, therefore, aluminum whichisyieldingahigherYoungmodulusofelasticitywasused.Also,toensureproper staticallydeterminedsupportoftheblade,double row bearing pair was placed on the hub(seeFigure31). In this way, taking into account the lack of the shaft, weight of the rotor/blade was furtherreduced.Also,aplacementofthebearingmechanismclosertothecenterofthe rotationreducespotentialperipheralmassproblemsandvibrationsthatcouldbecaused duethemissbalancedassembly. Ingeneral,overallmassoftheBachblade(Prototype04)wasreducedcomparingtothe massoftheSavoniusbladeofthecorresponding diameter(Prototype 02).Comparison betweentotalmassesofthebladesmanufacturedisgivenintheTable8.Ascanbeseen, areductionof54%inmasswasobtained.
Figure31.Hub/bladeconnectionforSavoniusblade(left)andBachblade(right)
Prototype Mass [kg] 01 0.390 02 0.480 03 0.575 04 0.220
Table9.Masspropertiesoftestedblades
51
4.ExperimentalInvestigation
4.1ExperimentSetupandFacility
ExperimentstookplaceintheEiffeltype,opencircuit,lowspeedwindtunnelattheAIT (seeFigure32).Setoftheoperatingwindvelocitiesfrom223m/sisobtainedbyuseof the3mdiametercentrifugalblowerpoweredbythevariablethreephase30kWmotor. Theflowiscontrolledbychangingtheblowerrotationfrom2250rpmandmonitored using the Betztype manometer placed in the undisturbed field, right before the test section.Duethesizeofthemodelandextremelyhighblockageratio,openedtestsection setupwasusedwiththeleveledsupportstructure(seeFigure33).Testapparatuswith mounted wind turbine model was placed approximately 1m distance from 1050mm x 1050mmcrosssection(seeFigure34).

Figure32.SketchoftheAIT1.05x1.05mopencircuitlowspeedwindtunnel
Wind turbine is coupled with the 400W/200V 50Hz pole induction motor/generator andforpurposesofinvestigationofthedifferentloadconditionsandthetipspeedratios oftherotor;aloadiscontrolledusingthe200W/100ratedbreakingresistorconnected withtheinverter.
52
Figure33.Opened1.05x1.05mtestsection
Figure34.Experimentalapparatus
The torque is measured using the torque transducer and converter coupled with the mainshaft.Torquecapacityofthemeasuringsystemis10Nmandiscapableofhandling the range of rotations up to 8000rpm. The converter is also connected with the RPM detectorunitthatmeasuresrotationofthemainshaftoftheMagnusturbine.
53
4.2SavoniusRotationMeasurement
InordertobuildapropermodelfordescribingthebehaviorofSavoniustypeMagnus windturbine,moredetailanalysisisneededabouttheloadsontheSavoniusandBach blades used in these experiments. Since the lift and the drag forces that act upon the blade rotor are function of the angular velocity of the blade itself, measurement of the bladerpmwasnecessary. However the angular velocity of the Savonius and Bach blades were suspected to be differentdependingonthepositionoftherotorbladeintheMagnusturbinerotational plane,thereforecontributingtothefluctuationoftheloadsaswell.Forthatpurpose,a measurementsoftherpmwereperformedforeachoftherotationalbladesforMagnus turbineazimuthalanglesbladeof0,45,90,135and180(seeFigure35).
Figure35.Valuesofmeasurementazimuthalanglesblade
Ascanbeseenfromthefigure,theuppermostpositionistakenasazero,northvalue. A portable turn detector was used for the rpm measurements. After the blade was positioned at the desired angle, the rotation of the Magnus wheel was set to still conditionandthemeasurementtookplaceaftersomesettlingperiodoftimesoconstant rotationalspeedofthebladecouldbereached.Aprocedurewasrepeatedforeachofthe fiveazimuthalangles,forallthebladesfortheeveryprototypetested.Thetestsforother halfoftherotorplaneangleswerenottakenforthereasonsofsymmetry.
54
4.3FullModelTestProcedure
Testing the each prototype of Magnus wind turbine is done in order to obtain torque andpowercoefficients,CQandCPasafunctionoftheMagnusturbinetipspeedratioM. After reaching the desired wind speed, turbine was released from the previous still condition and set into motion. The reasons for this hold/start procedure will be given later (Chapter 6). After constant rpm of the Magnus rotor was reached in the idling mode,generatorloadwasactivatedusingtheinverter/breakingresistorcircuitinaway thatloadwasgraduallyincreasedfromzeroloadcondition,equivalenttoidlingmode, up to the highest possible load which corresponded to approximately 0.7 Hz rotating frequency of the rotor. The set of the high loads due the enormous resistive heat dissipationwasnotyieldinganypracticalsignificance,butpurposewassolelytomodel the efficiency in as large range of tip speed ratios as possible. The procedure was repeatedforvariouswindspeedsforcasesof2,3,4and5bladesthatwereconsequently mountedonthehubplate. Since not all rotational blades were yielding the same performance for the particular windspeeds,selectioncriteriawastotrytomatchthebladeswithrathersimilarrpmif possible for reasons of balanced load. If the difference in rpm between 2 blades for instance,islarge,discrepancyinloadswouldcausetheadditionalstressesonthemain shaftoftheturbineandthatwouldinfluenceatorquecharacteristicaswell.Inacaseof increasing number of blades, that effect is a partially diminished by the fact that contribution of one blade on rotor performance is reduced, therefore shaft torsion stressesarelower.
Figure36.Threebladestestsetup
55
5.Results
5.1RotorBladeExperimentalResults
Inthischapter,theresultswillbegivenforbladenumberofrevolutionperminuteasa functionoftheazimuthalpositionintherotorplane,forarangeoftestedwindspeeds. Theresultswillcoverallprototypesthatweretested. Afirsttwobladeprototypeswerefullytestedforall5blades,andPrototypes03and04 werepresentedonbasisofthreemeasurementsinaformerandoneblademeasurement inalatercase.Reasonforlimitedsetofdatacomesfromthedestructionofsomeblades duringthetestingproceduresincethemeasurementsofthebladerpmtookplaceafter the Magnus torque tests and certain workout period in order to obtain better bearing performance.Areferencerpmdataaretakenfor4,6,8and10m/sregardlessofthefact thatsomeofthebladeprototypesoperatesatisfactoryforhigherwindspeedsandsome showedmuchbettercutincharacteristicthentheothers.
12m/s Wind Speed 3000 Blade 11 Blade 12 Blade 13 Blade 14 Blade 15
2500 10m/s Wind Speed 2000 1500 Savonius Revolutions [rpm] 1000 1500 8m/s Wind Speed
1000 6m/s Wind Speed 1000 800 600 600 500 400 300 0 20 40 60 80 100 Azimuthal Position [] 120 140 160 180 4m/s Wind Speed
Figure37.Prototype01bladerotationasafunctionofthewindspeedandazimuthalangle
56
InvestigationoftheSavoniustypeMagnusWindTurbine________________ AFigure37representstherotationaldatatakenforthesmallest,Prototype01case,and firstthingthatcomesintoattentionisextremelybadstartingcharacteristicoftheblades, therefore the Magnus rotor itself. Only one blade was able to start at the 4 m/s wind speed and in order to start all five blades, a wind speed of approximately 6 m/s was necessary. It is properly suspected that this bad characteristic came from the fact that centralshaftoccupiesthesignificantoverlapspace(seeFigure27),thereforeneutralizing effect of the air flow extending to the returning paddle. The shaft influence will be examinedlaterasaseparateproblem.Onecannoticethediscrepancybetweentherpm ofthebladesdependingonthebladeitselfandthepositionofthebladeintermsofthe azimuthalanglewithbestrevolutionratefortheazimuthnorthpositionforallblades. A very bad aerodynamical performance of the Prototype 01, makes a blade more sensitive on the fluctuating bearing loading that is caused by the weight of the blade itself. APrototype02bladebringsdrasticincreaseintherpmrate(seeFigure38),notonlydue theincreasedsizeofthemodel,butalsoduethegreateroverlapratio(63%comparingto the22%oftheprototype1)andthereforeasmallercontributionofthecentralshaftina flowblockage.
10m/s Wind Speed 3500 3000 2500 Savonius Revolutions [rpm] 2000 8m/s Wind Speed 2600 2400 2200 2000 1800 1600 6m/s Wind Speed 1600 1400 1200 1000 800 4m/s Wind Speed 900 800 700 600 500 0 20 40 60 80 100 120 140 160 180 Blade 21 Blade 22 Blade 23 Blade 24 Blade 25
Savonius Position []
Figure38.Prototype02bladerotationasafunctionofthewindspeedandazimuthalangle
57
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Also it was observed that 4 from 5 blades tested, started revolving at approximately 3 m/s, therefore moving cutin characteristics toward more expected values for a small windturbines.Therpmperformancedependsonanangularpositionofthebladeagain duethebearingloadissue. Prototype 03 tested for a largest Savonius blade was made in a manner which was considered to be a reasonable match between an approximate optimum in overlap values(50%)andaspectratio(4).Asmallervaluesoftheoverlapratioaccordingto Ushiyama were yielding no significant RPM improvement of the rotor, and on a contrary could reduce it due the increase of the inertia moment. Also smaller overlap wouldincreaseacentralshaftblockageeffect. Starting characteristic of the blades was found to be similar as for a Prototype 02, and thebladerevolutionratesmallerthenforpreviousprototypecases(seeFigure39). Bach rotor results were presented on Figure 40, and are based on results of measurements from one blade. Due the manufacturing problems, other blades were givingverytediousresultsthatwerenotrealisticintermsofexplainingtherotationallift phenomena.
10m/s Wind Speed 2400 2200 2000 1800 Savonius Revolutions [rpm] 8m/s Wind Speed 2000 1800 1600 1400 6m/s Wind Speed 1400 1200 1000 800 600 4m/s Wind Speed 600 400 200 Blade 32 Blade 33 Blade 35
20
40
60
80 100 Azimuthal Position [ ]
120
140
160
180
Figure39.Prototype03bladerotationasafunctionofthewindspeedandazimuthalangle (resultsfrom3blades)
58
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Thereasonsforthiswillbegiveninthelaterchapters.Oneofthenoticeableadvantages of the Bach blade was very low cutin speed needed to start the rotor. All blades smoothly started spinning at approximately 2 m/s. Also for 4 m/s case, Bach blade showedsuperbresultscomparingtotheotherprototypes.Howeverduethementioned manufacturingproblems,riseofthebladerpmforhigherwindspeedswasfoundtobe prettymuchunsatisfactory.Thiswascreditedtotheveryhighlateralvibrationsofthe blade due the misbalanced mass problems that would simply influence the rpm characteristicofthebladesinanegativeway. In order to better understand and compare various models, a rotation results for all prototypes were compared and given on Figure 41 directly as an rpm values and on Figure42as tipspeedratiosofthe tested prototypes as afunction of thewindspeed. TipspeedratiooftheSavonius(Bach)bladeisdefinedas:
blade
Dblade 2 nblade R = 2 = blade blade Vwind Vwind
[5.1]
wherebladeisthetipspeedratioofthebladeandDbladeisthebladediameter.
10m/s Wind Speed 1400
1300
1200 Bach Revolutions [rpm] 8m/s Wind Speed 1300 1200 1100 1000 6m/s Wind Speed 1200 1000 800 4m/s Wind Speed 1100 1000 900 800 700 0 20 40 60 80 100 120 140 160 180
Azimuthal Position [ ]
Figure40.Prototype04bladerotationasafunctionofthewindspeedandazimuthalangle (resultsfromoneblade)
59
3000 Prototype 01 Prototype 02 Prototype 03 Prototype 04 2500
k 360
2000 Blade Revolutions [rpm]
1500
1000
500
7 Wind Speed [m/s]
10
Figure41.RPMcharacteristicoftestedprototypes
1 Prototype 01 Prototype 02 Prototype 03 Prototype 04
0.9
k 0.081
0.8
0.7
blade
0.6 0.5 0.4 0.3 4
7 Wind Speed [m/s]
10
Figure42.Tipspeedratioofthebladesasafunctionofthewindspeed
60
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Asinglerpmvaluefortheonewindspeedisobtainedasthemeanoftheallazimuthal rpmvaluesandthenaveragedforthetestednumberofblades:
nblade _ i =
1 N
n
j =1
[5.2]
where N = 5 is a number of the azimuthal positions and nblade _ i mean for one blade. And:
nblade =
1 N blades
N blades i =1
blade _ i
[5.3]
whereNbladesisanumberofbladestested(15)and nblade isameanrpmforallblades. A both figures reveal a trend in increase of the rotational speed for the all prototypes. Redlinerepresentsafoundlinearfitonlyintermsofslope,whilethestartingrotation value for cutin speed (yaxis slice) depends on the geometrical characteristics of the prototypes. A deviation from the fitted line practically characterizes the various set of problemsinthebladefunctioningthatarereducingperformance,vibrationsandbearing conditiononafirstplace. IfweadoptfittedlineasafunctionthatexplainsthetheoreticalbehavioroftheSavonius rotor, we see that the Bach profile characterized by the superb starting performance actuallyfailstoholdsuchbehaviorduethemanufacturingproblems. Plots confirm the theory that tip speed ratio greater then one can not be obtained for Savoniustypeofrotor,thereforethemeasureofsuccessofcertainbladeconceptisthe consideredintermshowsoonabladewillreachmaximumtipspeedratioofone.Again Prototypes 02 and 03 seems to hold the best performance with Prototype 02 showing dropinthetrendforwindspeedshigherthen8m/s.Ontheotherhand,theBachblade (Prototype 04) again shows the greatest potential for reaching a maximum tip speed ratioforwindspeedsofroughly7m/s,butagaintheresultsarenotasexpectedduethe manufacturingandstructuralproblems. Thisactuallyconfirmstheideaofusingtheelectricallypoweredcylindersthatwouldbe abletoachieveahigherrotationalrates(greaterthenunity)thusbeingabletocreatea higherrotationallift. Sincethebladeprototypeperformancedirectlyinfluencesthepowerperformanceofthe Magnusturbine,resultsforpowerandtorquecoefficientwillbepresentedandgrouped inawaythatemphasizesimportantrelations.
61
5.2MagnusExperimentalResults
In this chapter, the results are given for torque and power coefficients as a function of the Magnus turbine tip speed ratio. Since mechanical torque and rotor rpm were measured,otherquantitieswerenumericallyderivedfromtheminthefollowingway:
DM 2 nM R M = 2 = M M Vwind Vwind
[5.4]
Pmech = Qmech rot
[5.5]
where M is the tip speed ratio of the rotor and DM is the Magnus rotor diameter. Replacingthebasicdefinitionsofthewindturbinepowerandtorque[17],inexpression 5.5,weyield:
1 1 3 2 C P Vwind AM = CQ Vwind AM RM M 2 2
therefore:
[5.5] [5.6]
C P = CQ M
400 Prototype 01 Prototype 02 Prototype 03 Prototype 04
350
300 Magnus Rotation [rpm]
250
200
150
100
50
500
1000
1500 Blade Rotation [rpm]
2000
2500
3000
Figure43.Relationbetweenbladeandturbinerpmforidlingcase(2blades).Markingpoints onlinesrepresentwindspeedsof4,6,8and10m/s
62
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Even without the generator load, it is possible to anticipate the performance of the manufacturedprototypesjustbyobservingtheidlingcharacteristicoftheMagnusrotor. Regardlessofthefactthatrotorthensufferszerotorque,arelationbetweentheMagnus rotations clearly reflects the lift capabilities of the driving blades. Figure 43 shows the Magnus rotor revolutions as a function of rotation of the all blade prototypes for 2 blades case. Plot doesnt differentiate much for a case of 3, 4 and 5 blades, and it was found that the Magnus rotor suffers a maximum drop in the revolutions of approximately914%forthe5bladescasecomparingtothecurrentone. Clearly the best performance is shown for a Prototype 03 blade that is capable of developingsameorhigherMagnusidlingrateforasmallerrpmandsamewindspeed. This clearly comes from the larger diameter of the Prototype 03 and the increased circulation around the blade which allows such behavior. A graph again confirms superiorperformanceoftheBachbladebutonlyfor4m/scase;TheBachbladewiththe samediameterasaPrototype02,develops25%higherrpmrateandinduces50%higher idlingrotationalspeedoftheMagnusrotor. For this reasons, power and torque characteristic of all prototypes will be given in comparisontotheresultsforaPrototype03model.Itspowerandtorqueperformance arepresentedasafunctionofthewindspeedandnumberofbladesonFigures4451. Subsequently power and torque results of the Prototype 03 are compared with other prototypes for a various wind speed cases, given in a manner that shows additional dependenceofthenumberofblades.TheyarepresentedonFigures5273. All figures show that apparently the tip speed ratio of the Prototype 03 powered MagnusturbinecannotoverexceedM1.4andthelimitisreducingwhenthenumber of rotor blades is increased. Also the maximum values of the power coefficient are obtainedforthetipspeedratiosofapproximatelyM=0.65~0.8(movingtowardlower valueforincreaseinthenumberoftheblades)withexceptionof4m/scasethatyields maximumatM0.6.Additionalgeneratorloaddecreasespowercoefficientasexpected. Theriseinatorqueandpowercoefficientisdirectlyproportionaltothewindspeedand the number of blades. A former one is a solely characteristic of this particular turbine and doesnt yield any similarity with classical HAWT systems. Power coefficient dependence on the wind speed can be attributed actually to the dependence of the MagnusrotorperformanceonthetipspeedratiooftheSavoniusblades.Thereforewe havearelationbetweentipspeedratiosoftheSavoniusandtheMagnuswindturbine. Inthatmanner,itisalsopossibletopresentthepowerandthetorquecurvesonFigures 4473asafunctionoftheSavoniustipspeedratioinsteadofthewindspeed(usingthe dependencefromFigure42).
63
0.035 4 m/s 6 m/s 8 m/s 10 m/s
0.03
0.025
Power Coefficient Cp
0.02
0.015
0.01
0.005
0 0.2
0.4
0.6
0.8 1 Tip Speed Ratio M
1.2
1.4
1.6
Figure44.PowercoefficientofthePrototype03modelasafunctionofMagnustipspeedratio for2bladescase
0.07 4 m/s 6 m/s 8 m/s 10 m/s
0.06
0.05 Torque Coefficient Cq
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
Figure 45. Torque coefficient of the Prototype 03 model as a function of Magnus tipspeed ratiofor2bladescase
64
0.05 4 m/s 6 m/s 8 m/s 10 m/s
0.045
0.04
0.035 Power Coefficient Cp
0.03
0.025
0.02
0.015
0.01
0.005
0 0.2
0.4
0.6
1.2
1.4
1.6
Huge Savonius - 3 Blades 0.12 4 m/s 6 m/s 8 m/s 10 m/s 0.1
0.06
0.04
0.02
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
65
0.07 4 m/s 6 m/s 8 m/s 10 m/s
0.06
0.05
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
0.12 4 m/s 6 m/s 8 m/s 10 m/s 0.1
0.06
0.04
0.02
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
66
0.08 4 m/s 6 m/s 8 m/s 10 m/s
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
1.2
1.4
1.6
0.16 4 m/s 6 m/s 8 m/s 10 m/s
0.14
0.12
Torque Coefficient Cq
0.1
0.08
0.06
0.04
0.02
0 0.2
0.4
0.6
1.2
1.4
1.6
67
0.06 P03 - 2 Blades P03 - 3 Blades P03 - 4 Blades P03 - 5 Blades P02 - 2 Blades P02 - 3 Blades P02 - 4 Blades P02 - 5 Blades
0.05
0.03
0.02
0.01
0.2
0.4
0.6 0.8 Tip Speed Ratio M
1.2
1.4
Figure52.Comparisonofpowercoefficientsbetweenprototypes03and02for4m/scase
0.1
0.06
0.04
0.02
0.2
0.4
1.2
1.4
Figure53.Comparisonoftorquecoefficientsbetweenprototypes03and02for4m/scase
68
0.06 P03 - 2 Blades P03 - 3 Blades P03 - 4 Blades P03 - 5 Blades P04 - 2 Blades P04 - 4 Blades
0.05
0.03
0.02
0.01
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
0.1
0.06
0.04
0.02
0.2
0.4
1.2
1.4
69
0.06
0.05
0.04
0.03
0.02
0.01
0.2
0.4
0.6
0.8
1.2
1.4
Tip Speed Ratio M
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0.2
0.4
1.2
1.4
70
0.06
0.05
0.04
0.03
0.02
0.01
0.2
0.4
1.2
1.4
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0.2
0.4
1.2
1.4
71
0.06
0.05
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
1.2
1.4
1.6
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0 0.2
0.4
0.6
1.2
1.4
1.6
72
0.06
0.05
0.04
0.03
0.02
0.01
0.5 Tip Speed Ratio M
1.5
0.14
0.12
0.1
0.08
0.06
0.04
0.02
1.5
73
0.06
0.05
0.04
0.03
0.02
0.01
1.5
0.14
0.12
0.1
0.08
0.06
0.04
0.02
1.5
74
0.06
0.05
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
0.8
1.2
1.4
1.6
Tip Speed Ratio M
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0 0.2
0.4
0.6
1.2
1.4
1.6
75
0.07
0.06
0.05
0.04
0.03
0.02
0.01
1.5
0.12
0.08
0.06
0.04
0.02
1.5
76
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.2
0.4
0.6
1.2
1.4
1.6
1.8
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0.2
0.4
0.6
1.2
1.4
1.6
1.8
77
0.08 P03 - 2 Blades P03 - 3 Blades P03 - 4 Blades P03 - 5 Blades P04 - 2 Blades
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0 0.2
0.4
0.6
1.2
1.4
1.6
0.14 P03 - 2 Blades P03 - 3 Blades P03 - 4 Blades P03 - 5 Blades P04 - 2 Blades
0.12
0.08
0.06
0.04
0.02
0 0.2
0.4
0.6
1.2
1.4
1.6
78
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Itcanbenoticedthattorqueandpowerdonotyieldalinearrisewithwindspeedand set of optimal values for the power coefficient at certain wind speed reduces since the parabola curve elongates and thus reduces the optimum performance zone. Hence, choiceofthepropergeneratorisveryimportant.Followingthepracticeforasmallwind turbines, nominal wind velocity will be set to V = 10 m/s, and it can be used as a referencevalueforsettingthepossibleload. Further plots reveal obvious inferiority of the Magnus rotors powered by other blade prototypescomparingtothePrototype03blade. Prototype01poweredMagnus rotor againshowedthepoor qualityduethecombined roleofgapshaftblockageandsmallbladediameter,whilePrototype02poweredmodel isyieldingsuperiorperformancecomparingtoPrototype03onlyfora10m/scaseina rangeofthetipspeedratioshigherthenM0.82(seeFigures70and71).Inaddition, thetipspeedratiosforPrototype02basedMagnusrotorarehigherthenforanyother prototype and they reach up to M = 1.8. Again, the tip speed ratio was found to be decreasingwithnumberofbladesandincreasingforhigherwindspeeds. Prototype04obtainshigherpowerefficiencythenPrototype03modelfora4m/scasein arangeoftipspeedratiosaboveapproximatelyM0.6forboth,twoandfourblades cases(seeFigures54and55). Overall,apowerperformanceoftheSavoniusbasedMagnusrotorislow.Forthebest case of Prototype 03 powered rotor with 5 blades, the power coefficient was approximatelyCP=0.075whichisabout3timessmallerCPthenforatypicalSavonius rotor and about 6 times smaller then the power coefficient of the properly designed 3 bladed HAWT system. On the other hand, this study showed that such system is possibletobuildandthatitsinferiorperformanceactuallycorrespondswiththeoverall low expenses for such device. In order to further reduce the costs, constant load generator and fixed rotor speed system could be used instead of the one with the controller for a variable speed regulation. In such case, generator should be optimized forsomewindspeedthatwasfoundtobeappropriateforthedesiredlocation.Table9 contains optimum tip speed ratio values (in terms of power coefficient) for maximum poweroutputfortestedrangeofwindspeeds. Wind Optimum RotorRotation Mechanical Speed[m/s] TSR [rpm] Power[W] 4 6 8 10 0.62 0.74 0.74 0.69 50 90 120 140 1.3 6.3 14 29.9
Table10.Magnusturbineshaftpowerforoptimumvaluesoftipspeedratio
79
6.ModalAnalysisoftheBlades
During the wind tunnel tests, it was noticed that in a range of certain wind speeds, a power performance of the rotor was drastically dropping. The measured torque was showing highly unstable behavior together with the number of revolutions of the Magnusrotor.Asignificantlyloweraveragevaluesofthetorqueandsometimesalow values of the tip speed ratio were reason for concern and further investigation. The powercurveingeneralwasexpectedtobesubstantiallyhigherforagivenwindspeed comparing to the other wind speed measurements, and some tests were proved to be unsuccessful.TheexamplesofsuchbehavioraregivenonFigure43below.
Prototype 01 - 4 Blades 0.012 8 m/s Valid Test 8 m/s Invalid Test
0.04 0.045 10 m/s Valid Test 10 m/s Invalid Test Prototype 02 - 3 Blades
0.01
0.035

0.03
0.025
0.006
0.02
0.004
0.015
0.01
0.002
0.005
0 0.1
0.2
0.3
0.6
0.7
0.8
0.2
0.4
0.6
1.2
1.4
1.6
1.8
0.06
a)
Prototype 02 - 4 Blades
10 m/s Valid Test 10 m/s Invalid Test
0.08 0.07
b)
Prototype 02 - 5 Blades 10 m/s Valid Test 10 m/s Invalid Test
0.05
0.06
0.4 0.6 0.8 1 1.2 Tip Speed Ratio M 1.4 1.6 1.8
0.05
0.03
0.04
0.03
0.02
0.02
0.01
0.01
0 0.2
0 0.2
0.4
0.6
0.8
1 1.2 Tip Speed Ratio M
1.4
1.6
1.8
c)
d)
Figure74.Variouscasesofvibrationinfluencedpowerperformance
For this reason, like mentioned in Ch. 4.3, hold and start procedure was applied for every wind speed in order to determine the condition of the rotor before setting the rotor to idling state. A simple eye examination of the blocked rotor in some wind regimes, revealed a large amplitude lateral vibrations on the rotating Savonius blades thatweresuspectedtocorrespondtothenaturalfrequenciesofthebladesthemselves.
80
InvestigationoftheSavoniustypeMagnusWindTurbine________________ Duethelackoftheequipment,suchasaFFTanalyzeranddatalogger,itwasimpossible to doany experimental modalanalysis ofthe prototype blades. Instead, a simpler and rather inaccurate eyecheck method analysis was done together with FEM modal analysisusingANSYSsoftware.
Figure75.Prototype01FEMmodelusedformodalanalysis
A former one consisted of the simple observation of the blade on a blocked Magnus rotorwhileitsrotationalspeedwasdecreasingafterthesuddenstopofthewindtunnel
81
InvestigationoftheSavoniustypeMagnusWindTurbine________________ fan.Aportableturndetectorwasusedtorecordrpmofthebladesduringthecomplete seriesofvibrationevent,frominitializationoveritspeakvaluetoasettlingstateagain. Beyondthedoubt,suchtestsdontholdsufficientaccuracyforanyseriousanalysisbut its value is in the determination of the approximate range where resonance event can occur.
82
InvestigationoftheSavoniustypeMagnusWindTurbine________________ A conformation of such undesired events lies in results that should be treated from a qualitative,notquantitativepointofview.Regardlessofthesolutionsobtainedforthis particularsetofmodels,thepossiblefuturedesignandimprovementsmayvaryenough in choice of the materials, bearings and rotor size, therefore a new analysis would be necessary. The other method consists of ANSYS FEM analysis of the prototypes. All the blades were modeled with the high accuracy using the 3D 20node structural solid SOLID95 elements and the material properties given in a Table 7. A satisfying precision was reachedintermsofthegeometricalandinertial(momentofinertiaandmass)terms(see Figures7578)andthefreeformtetrahedralmesheswasimplementedforeachofthe elements. A contact between modeled bearings, shaft holder, shaft and aerodynamic paddlesweremodeledusingthesurfacetosurfacecontactpairfeaturewithCONTA174 elements for the contact surface and TARGE170 elements for the target surface. A frictionoftheballbearingswassetuptof=0.0015andfixed,bondedpairsweremade between shaft holders and the end plates of the paddles. Paddles and end plates were modeled and meshed as a single volume regardless of the bonding method used for their interconnection (adhesiveforP01, P02and P03and spotweldingforP04) which actually added additional stiffness to the model. Therefore slightly higher natural frequencieswereexpectedfromsimulation. The Block Lanczos default eigenvalue solver was used for finding the first 6 modes of thebladeswhileloadsweresetinformsofthezerodisplacementsonnodeswherethe centralshaftwasfastenedtothehub. The results are presented in the Table 10 below, together with the corresponding revolutionratesoftheprototypesthatmightbeaffected.Naturalfrequenciesoftherotor systemsarechangingwith therespect to theangular velocity ofthesystem (Campbell diagram) [31]. However, such changes in eigenvalues were found to be very small for currentoperatingrangeofrotationalspeedsofSavoniusrotor,soCampbellanalysiswas omitted. Mode Prototype01 [Hz]/[rpm] 1 2 3 4 5 6 22.186/1331 22.735/1364 74.568/4474 116.00/6960 118.19/7091 137.79/8267 Prototype02 18.586/1115 19.242/1154 43.660/2620 108.88/6533 117.57/7054 147.97/8878 Prototype03 17.562/1054 17.637/1059 25.448/1526 107.36/6441 111.65/6699 122.56/7354 Prototype04 38.260/2296 53.061/3184 111.67/6700 148.89/8933 155.99/9359 325.68/19541
Table11.NaturalfrequenciesusingANSYSFEMsolver
83
InvestigationoftheSavoniustypeMagnusWindTurbine________________ It can be seen from the table that due the high similarity between two longitudinal symmetries in perpendicular directions, first two modes are yielding the close frequencies. Prototype04holdsdifferencebetweenfirsttwomodesalonglocalprincipalaxesduethe elongated shape of blade. However aluminum paddles give rise in modes, so natural frequenciesareshiftedtowardhigherrpmvalues. An approximate eyecheck experimental values are lower then the simulated once as expectedandtheyarepresentedfor1stmodeinTable11below.Oneofthereasonsfor discrepancy lies like stated above in ideal connection between blades and endplates used in FEM modeling that is not very realistic. Second reason may lie in the possible delay between readout and true rpm at the moment when the vibrations occur since wholesystemishandportableandinthiscasedependsonskillofthepersonincharge tocatchamoment.Aformerreasonisresponsibleforthetoohighmodalvalue,and lateronecouldexplainthepossiblereadingundershooting. Experimental Prototype Prototype Prototype Prototype readoutofrpm 01 02 03 04 duringresonance 1stmode (upperlowerlimit) 1280 (1450850) 970 (1050800) 840 (920740) /
Figure79.Experimentalreadoutofthe1steigenmodefortestedblades
2200 Prototype 03 2000
1800
1600 Blade Revolutions [rpm]
1400
Calculated 1st mode Experim. 1st mode
Calculated 3rd mode
1200
1000
800
600
400
7 Wind Speed [m/s]
10
Figure80.RepresentationoftheeigenmodesonrpmwindvelocityplotforPrototype03
84
InvestigationoftheSavoniustypeMagnusWindTurbine________________ The fourth prototype wasnt tested for vibration analysis for the reason of large vibrations that were already occurring due the bad blade/tipplates mounting that causedtheunbalancedload. Figure 80 shows the approximate zones of the Prototype 03 rotor modal frequencies influence on a rpm vs. wind speed plot. Both, experimental and calculated values are presented. Results for the blade revolutions confirm the validity of the simple experimental analysis since upper limit of blade vibrations actually well corresponds withtheriseinaslopeoftherpmtrendline.Fora3rdmode,thenumericalprocedure based on a same criteria gives good prediction since around 1600rpm drop in blade revolutions was again observed which is reflected in decline of the rpm characteristic slope. Also should be emphasized that performance of each rotor of the same prototype is different and depends on individual settings and conditions of the blade itself (i.e. see Figure 39) therefore upper plot is presented for the averaged revolution rates of all measured blades. However, the results strongly imply that first and second eigenfrequency are characteristic for wind speeds around 6 m/s while the third mode correspondstothe8m/scase. Figures 81 86 show eigenshapes of the blade for all six modes. First and the second eigenmodearethebendingmodescorrespondingtotheflapwiseandedgewisemodes of the classic blade and main principle axes are not x and y axis. Third eigenmode is corresponding to the twisting of the Savonius blade while fourth and fifth modes are equivalenttophasedandoutofphasebulkingofthepaddles.Sixthmodeisyieldinga rathercomplexbendingofthepaddleswithlocalbuckingonthereturningpaddle.
85
Figure81.Firstmode,f1st=17.562Hz
Figure82.Secondmode,f =17.637Hz
2nd
Figure83.Thirdmode,f3rd=22.548Hz
86
Figure84.Fourthmode,f =107.36Hz
4th
5th
Figure85.Fifthmode,f =111.65Hz
Figure86.Sixthmode,f =122.56Hz
6th

87
7.StressandDeformationAnalysis
Ashortanalysiswillbegivenhereregardingthecentrifugalbuckingphenomenonthat was found to have a significant impact on the Savonius based Magnus turbine performance. In the numerous cases was found that misuse of the adhesive materials between tip plates and the PVC blades lead to the abrupt break of the blades around 25003000RPMatapprox.1012m/swindspeeds.Duethehighinertialloads,buckling ofthebladestressincreasesinthezoneswherebladeisconnectedtothetipplatesandif the local stress is bigger then yielding value of the bonded contact, connection simply brakes apart (see Figure 87). In order to investigate such phenomena numerical simulationwasinvokedoncemore.
Figure87.Prototype03bladedestroyedat12m/sduethecentrifugalbuckling
Again, ANSYS calculation was made on a Prototype 03 model using a large displacementtransientanalysisinordertoproperlymodelgeometricnonlinearityofthe buckets. Atypicalruleofthumbstatesthatiftheoutofplanedeflectionofsomeplateorshellis greater than half of its thickness, then membrane forces start to become significant in resisting the applied load [6]. In ANSYS, this calls for activating a large displacement solution (a.k.a. geometric nonlinearity). Since the wall thickness was 6 mm and it was largelysuspectedthatbladedeformationsmightbebiggerthen3mm. A transient dynamic analysis is usually used to determine the dynamic response of a structureundertheactionofanygeneraltimedependentloads.Thistypeoftheanalysis wasusedtodeterminethetimevaryingdisplacements,strains,stresses,andforcesina structuresinceitgoodcorrespondstocombinationofthestatic,transientandharmonic loadsandspeciallywheninertialloadsareimportantlikeinthecurrentstudy.
88
InvestigationoftheSavoniustypeMagnusWindTurbine________________ AbasicequationofmotionsolvedbyatransientdynamicanalysisinANSYSisgivenin aform: && & M {u} + D{u} + K {u} = F (t ) [7.1] && & whereM,D,andKaremass,dampingandstiffnessmatrixrespectivelyand u , u andu representnodalacceleration,nodalvelocityandnodaldisplacementvectors.Onaright && hand side force vector is given. Terms such as M {u} are used to account for inertial forceswedealwithinthiscase.Forthispurpose,aNewmarktimeintegrationmethod wasusedasadefaultone. A tested prototype was modeled as a solely PVC blade with tip plates without any bearings and central shaft. As a load vector angular velocity of 314.16 rad/s was set whichcorrespondstothe3000RPM.Deformationvectorsumofthebladedisplacement isgivenonFigure88below.
Figure88.DeformationvectorsumoftheP03bladefor3000RPM
As expected highest value of displacement is on the mid point of the edge of the advancingpaddleandlineardeformationinlocalxdirectioniscalculatedtobeux=4.72 mm which seems to be a good match with observed buckling. Experimental data lack becauseofthedifficultiestomeasuregeometrydeformationduringtherotationduethe lackofadequateequipment.Ahighestvalueofthetotaldisplacementismeasuredtobe usum=6.12mmanditslocatedontheedgeofreturningpaddle.Thisvalueincludesthe sumofallthreedirectionaldisplacements.Bucklingofthestructureisalsonoticeablein ydirection(seeFigure89)togetherwiththedeformationofthetipplates.
89
Figure89.DeformationofthebladeandtipplatesforP03model
Tipplatedeformationsarenotunexpectedeveninacasewithmountedbearings,since theconnectionbetweenplatesandbearingisusuallyestablishedwiththesmallgapthat enableshigherrpmofthebladescomparingtothecaseofdirectcontactbetweenthem. Alsostressdistributionofthebladeispresentedasacontourplotofthestressintensity (see Figure 90). Stress intensity I is defined over principal stresses as a largest of absolutevaluesofdifferencesinthecalculatednodesi.e.: I=max[(12),(31),(23)] [7.2] wheretheprincipalstressesaredefinedoverastressesinx,yandzdirectionas:
x i xy xz yx y i yz = 0 zx zy z i
[7.3]
fori=1,2,3. In order to establish criteria for predicting the onset of yield in ductile materials we derivevonMisesorequivalentstressfromthedistortionalenergydensitycriterion.This criterionstatesthatfailureoccurswhentheenergyofdistortionreachesthesameenergy for yield/failure in uniaxial tension (or compression). The distortional strain energy is the energy associated with a change in the shape of a body. The calculated von Mises
90
InvestigationoftheSavoniustypeMagnusWindTurbine________________ stressisacombinationofprincipalandyieldstressesanditisexpressedmathematically as:
M2
1 ( 1 2 )2 + ( 2 3 )2 + ( 3 1 )2 2
Figure90.MaximumvaluesofprincipalandvonMisesstressesontherotatingblade
Analysis confirms that the maximum values of the principal stresses and value of von Mises stress are located on the corner edges of the advancing blade (see Figure 90). StressvaluesaregiveninPascalunitswithmaximumvalueofthefirstprincipalstress
91
InvestigationoftheSavoniustypeMagnusWindTurbine________________ S1 = 66.7 Mpa, second one S2 = 27.3 MPa and third one S3 = 22 Mpa. Maximum equivalent stress value yields value of 46.2 Mpa. The principal stresses are ordered so thatS1isthemostpositive(tensile)andS3isthemostnegative(compressive). This clearly proves sensitivity of this spot on tension stress in cases of inadequate contactstrengthbetweentipplatesandtheSavoniuspaddles.Numerousdamagedcases andfailuresofhighrotatingSavoniusbladesinthisstudyconfirmedthisanalysis,since looseningofthebondedcontactevenbeforethedestructionofthebladeoccurredonthe spotshownbymeansofnumericalanalysis.
92
8.FutureInvestigations
RegardlessofthefactthattheexperimentalanalysisoftheSavoniustypeMagnuswind turbine was performed, current study leaves open space for the possible model improvementsandfurtherinvestigations. Model improvement could include solution for problems such as an overstress of the adhesive bondageforwindspeedshigherthen 12m/s andpossible substitution of the materials or support structure in order to avoid the resonance due the low natural frequenciesoftheblades.Othersolutionsotherthencentralshaftshouldbeinvestigated aswell,togetherwiththepossibilityofbiggerprototypeinvestigations. Also Bachtype blade should be given a special attention in terms that it beats the performance of the current configurations when properly manufactured. Its performance at 4 m/s when the misbalance wasnt so distinguished showed quite promising behavior with strong intuition that it could possibly reach power efficiency over0.1forwindspeedshigherthen10m/s.Inaddition,choiceofmaterialsanddelayed bladeresonancefor smallwindspeeds suggest the othersignificant advantages of this model.HoweverproductionoftheBachbladeismorecomplicatedandmorecostlythen exploitationofthePVCbasedSavoniusrotor.
Figure91.BachtypebasedMagnusrotor
AlsoliftanddragoftheSavoniusbladesshouldbeinvestigatedforvariouswindspeeds. Chauvin and Benghrib [8] suggest on basis of their investigations of aerodynamic coefficientoftheOL=0.43Savoniusrotorthatforsmallvaluesoftipspeedratio(S< 0.25)liftcoefficientisnegativeanddragisrelativelyconstantinwithslightlydecreasing towardS1.Suchanalysiscouldbeusefulforpossiblebuildupofthebladeelement momentumtheoryforMagnusturbine.
93
9.ConclusionsandRecommendations
Current study showed that regardless of the small output power and efficiency, a Savoniusbaseddesignisratherpossible.Cheapmanufacturingcostarewhatisthemain advantageofthisdesigntogetherwithitsobviousbenefitasasmallpowergenerating systemwhencomparedwiththeclassicMagnuswindturbinespoweredbytheelectro motors.Small,motorpoweredMagnusturbinewouldntbeabletogeneratetheoutput powerinaweakwindspeedregimes. Othermaindisadvantagesoftheturbinearerelatedtothelownaturalfrequenciesthat aresignificantlyinfluencingthepowerproductionandhighstressesinacontactregions thatarethreateningtostructuralintegrityofthemechanism. With improvements in material and the design of the blades such system could be implemented as a lowcost battery charger or direct water pumping device for remote locationsorThirdworldcountriesthatareeverydayfacingthedrinkingwaterproblem. Utilization of other lowcost components could make fully functional Magnus turbine designpossible.Asimplecaralternatorsarerepresentativesofcheapandhighlyreliable technologies that after minor modifications such as rewinding could be used as generators. Also,typical,largelyavailable12Vcarbatteryknownforitsslowdischargecapabilities is an excellent choice as a storage facility of electrical power. Other advantage of car battery is that allows large number of charges and discharges which is highly suitable forwindenergypurposes. SincetheSavoniustypeMagnuswindturbineconsistsofthelargenumberofrotational parts,specialcareshouldbetakenonbearingsthatareused.Windtunneltestsarenot enough,notonlyintermsofnaturalturbulenceandhighloadfluctuations,butalsoin terms of natural airpollution, dust and humidity that can significantly influence functioningoftheturbine.Bearingsshouldbeadditionallysealedandprotectedwhich mightreducebladerotorperformance. Therefore,choiceofthebearingsforoutdoorsystemshouldbedrivenbytheirdurability and long life criteria. They should be able to work for at least couple of hundreds of hours since maintenance should be brought to a minimum. Main load criteria should includebendingmomentsofthebladesduethedragoftherotorinhighspeedregimes.
94
10.Bibliography
[1] [2] [3] [4] AkitaMagnusAssociationOnlineResource:www.akita21.com/magnus/ AlternativeEnergyOnlineResource:www.southcom.com.au/~windmill/ ANSYSRelease9.0DocumentationandTutorial,SASIP,Inc.,2004. Bach, G. von, Untersuchen uber SavoniusRotoren und verwandte Stromungs maschinen, Forschung auf dem Gebiete des Ingenieurwesens, vol.2,pages218231,1931. Blackwell,B.,Sheldahl,R.E.,FeltzL.V.,WindTunnelPerformanceData for Two and ThreeBucket Savonius Rotors, Sandia Laboratories, SAND760131,1977. Budgell, P., ANSYS Tips and ANSYS Tricks, Burlington, Ontario, CA,
[5]
[6] [7]
1998,1999.
CFD Online Resource Network, Guidelines for Modeling Turbomachinery,http://www.cfdonline.com/Wiki/Best_practise_ guidelines_for_turbomachinery_CFD Chauvin,A.andBenghrib,D.,DragandLiftCoefficientsEvolutionofa SavoniusRotor,ExperimentsinFluids,vol.8,pages118120,1989. Cochran, B.C., Banks, D., Taylor, S.J., A ThreeTiered Approach for Designing and Evaluating Performance Characteristics of Novel WECS (WindEnergyConversionSystems),AIAA20041362Paper,2004.
[8] [9]
[10] Encyclopedia of Alternative Energy & Sustainable Living, a resource of DavidDarling: www.daviddarling.info/encyclopedia/S/AE_Savonius_turbine.html [11] Fernando, M.S.U.K. and Modi, V.J., A Numerical Analysis of the Unsteady Flow Past a Savonius Wind Turbine, Journal of Wind EngineeringandIndustrialAerodynamicsvol.32,pages303327,1989. FLUENT 5 Tutorial Guide, Chapter 9, Turbulence Models, Fluent Inc., 1998. Fox, R.W. , McDonald, A.T, Pritchard, P.J, Introduction to Fluid Mechanics,JohnWiley&Sons,Sixthedition,2003. Fujisawa, N., Kawaji, Y. and Ikemoto, K., Feedback Control of Vortex SheddingfromaCircularCylinderbyRotationalOscillations,Journalof FluidsandStructures,vol.15,pages2337,2001. Fujisawa,N.,ShiraiH.,ExperimentalInvestigationontheUnsteadyFlow Field Around a Savonius Rotor at the Maximum Power Performance, WindEngineering,vol.11,Issue4,pages195206,1987.
[12] [13] [14]
[15]
95
InvestigationoftheSavoniustypeMagnusWindTurbine________________ [16] Fujisawa,N.,VelocityMeasurementsandNumericalCalculationsofFlow FieldsinandaroundSavoniusRotors,JournalofWindEngineeringand IndustrialAerodynamicsvol.59,pages3950,1996. Hansen, Martin O.L., Aerodynamics of Wind Turbines, James & James, London,UK,2003. Hoerner, S.F., Borst, H.V., FluidDynamic Lift, Practical Information on Aerodynamic and Hydrodynamic Lift, Published by L.A. Hoerner, SecondEdition,1985. Ishimatsu, K., Kage, K. and Okubayashi, T., Simulation of the Flow Around Cross Flow Turbine With End Plates, Japan Society of ComputationalFluidDynamics,14thJSCFDSymposium,C062,2006. Johnson, G.L., Wind Energy Systems, PrenticeHall, Electronic Edition, 2001. Kang, Sangmo and Choi, Haecheon, Laminar Flow Past a Rotating Circular Cylinder, Physics of Fluids, vol. 11, No. 11, pages 33122231, November,1999. Kozlov,V.andBychkov,N.,MagnusWindTurbine,ExperimentalResults oftheModelTesting,InstituteofTheoreticalandAppliedMechanicsSB RAS,2005. Menet, J.L., A Doublestep Savonius Rotor for Local Production of Electricity:ADesignStudy, Renewable Energy vol. 29, pages 18431862, 2004. Menet,J.L.,Bourabaa,N.,IncreaseintheSavoniusRotorEfficiencyviaa Parametric Investigation, European Wind Energy Conference , Proceedings,London,2224November,2004. Menet, J.L., Valdes, L.C., Menart, B., A Comparative Calculation of the Wind Turbines Capacities on a Basis of the L Criterion, Renewable Energyvol.22,pages491506,2001. Mittal, S., ThreeDimensional Instabilities in Flow Past a Rotating Cylinder, Journal of Applied Mechanics, vol. 71, pages 8995, January, 2004. Modi, V.J., Fernando, M.S.U.K., and Roth, N.J., Aerodynamics of the Savonius Rotor: Experiments and Analysis, Energy Conversion EngineeringConference,IECEC90vol.5,pages213218,1990. Rahai, H., Development of Optimum Design Configuration and Performance for Vertical Axis Wind Turbine, Feasibility Analysis and FinalEnergyInnovations,SmallGrantProgramReport,CaliforniaState University,2005.
[17] [18]
[19]
[20] [21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
96
InvestigationoftheSavoniustypeMagnusWindTurbine________________ [29] Reid, E.G., Tests of Rotating Cylinders, Langley Memorial Aeronautical Laboratory,NACATN209Report,December,1924. [30] [31] Riegler, Hannes, HAWT versus VAWT, ROPATEC AG, Business InnovationCenter,REFOCUS,July/August,2003. Santos, Ferreira Ilmar, Vibrations in Rotating Machinery, Technical UniversityofDenmark,LectureNotes,2005.
[32] Savonius, S.J., The Wing Rotor in Theory and Practice, Savonius Co, Finland,1928. [33] [34] Sengupta, Tapan K. and Talla, Srikanth B., RobinsMagnus Effect: A ContinuingSaga,CurrentScience,vol.86,No.7,10April,2004. Shikha, Bhatti, T.S. and Kothari, D.P., Early Development of Modern Vertical and Horizontal Axis Wind Turbines: A Review, Wind Engineeringvol.29,No.3,pages287299,May2005. Stojkovic,D.,BreuerM.andDurst,F.,EffectofHighRotationRateson theLaminarFlowAroundaCircularCylinder,PhysicsofFluids,vol.14, No.9,September,2002. Tang, T and Ingham D.B., A Numerical Investigation Into the Steady Flow Past a Rotating Cylinder at Low and Intermediate Reynolds Numbers,JournalofComputationalPhysics,vol.87,No.91,1990. The Center for the Study of Technology and Society http://www.tecsoc.org Tokumaru, P.T. and Dimotakis, P.E., The Lift of a Cylinder Executing RotaryMotionsinaUniformFlow,JournalofFluidMechanics,vol.255, pages110,1993.
[35]
[36]
[37] [38]
[39] Ushiyama, I. and Nagai, H., Optimum Design Configurations and PerformanceofSavoniusRotors, WindEngineeringvol.12,No.1,pages 5975,1988.
97
AppendixA:ListofFigures

Figure1.Inviscidirrotationalflowpastacylindera)zerorotation,b)subcriticalrotation,c) criticalrotation,d)supercriticalrotation ...................................................................................... 8 Figure2.ComputedwakebehindtherotatingcylinderforRe=100and=1[35] ......................... 9 Figure3.Timehistoriesofliftanddragcoefficientfor2Dand3DflowforRe=200and=5[26] ............................................................................................................................................................ 10 Figure4.IsosurfacesofthespanwisevelocityforvariousARofcylinderinthe3Dflowfor Re=200and=5[26]......................................................................................................................... 11 Figure5.ShipBuckau:rotatingcylidersutilizedtocreateMagnuspropulsionforce[37].9thof May1926,aftersailingacrossAtlantic,shipenteredNewYorkharbor.............................. 12 Figure6.Performanceoftheconventionalwindconversionsystemsgivenasefficiencyvs.tip speedratio[24]................................................................................................................................. 15 Figure7.VariousSavoniusrotors:a)Chowchilla,California[2],b)EMATLtd,England commercialmodel[10] ................................................................................................................... 16 Figure8.BasicgeometricalfeaturesofSavoniusrotor..................................................................... 16 Figure9.TipplateeffectonSavoniusrotorperformance[39]......................................................... 17 Figure10.Liftgeneratingdeviceonthewindturbines:a)airfoil,b)rotatingcylinder,c) Savoniusrotor.................................................................................................................................. 18 Figure11.LocalvelocitiesandforcesonSavoniusrotor................................................................... 20 Figure12.Variousrotorconfigurations:a)Bach,b)Benesh,c)Modi,d)SavoniusOL=0.21,e) SavoniusOL=0.67............................................................................................................................ 24 Figure13.Exampleofcomputationaldomainofthe2Dflowaroundmodeledrotors ............... 28 Figure14.Meshingofthevariousrotorconfigurations:a)Bach,b)Benesh,c)SavoniusOL=0.67, d)Modi,e)SavoniusOL=0.21....................................................................................................... 31 Figure15.BoundarylayeradequategridonBachtyperotorwall .................................................. 31 Figure16.Y+distributionalongthewallsofvariousprofilesforRe=2.4x104:a)Bach,b) Benesh,c)Modi,d)SavoniusOL=0.67,e)SavoniusOL=0.21 ................................................. 32 Figure17.ModiFernandoprofilewiththepressuretapssetup[27].............................................. 36 Figure18.Modiprofilepressuredistributionfor=30andVwind=6m/s....................................... 36
98
Figure19.FlowfieldpressuredistributionoverthetheoreticalModiprofilefor=30andVwind =6m/s.................................................................................................................................................. 37 Figure20.StatictorquevariationwithflowangleforModiprofile............................................... 38 Figure21.Statictorqueofvariousbladeprofilesasafunctionofflowangle.............................. 39 Figure22.ComputedvectorflowfieldaroundSavoniusOL=0.21rotor;a)0vortexstructures insidebothbuckets,b)16pressureincreaseonthewallsofthereturningbucket,c)32 centralvortexstructureoutsideofthebucket,withsecondvortexdownwashed............ 40 Figure23.VelocityplotaroundSavoniusOL=0.21rotorfor=164............................................... 41 Figure24.PressureplotaroundBachtyperotorfor=143............................................................. 42 Figure25.PressureplotaroundBachtyperotorfor=161............................................................. 42 Figure26.PressuredistributionaroundBachtyperotorfor=161 .............................................. 43 Figure27.Prototype01isometricsketchwiththecrosssectiondetail........................................... 47 Figure28.Prototype02isometricsketchwiththecrosssectiondetail........................................... 48 Figure29.Prototype03isometricsketchwiththecrosssectiondetail........................................... 49 Figure30.Prototype04(Bachtypeblade)isometricsketchwiththecrosssectiondetail .......... 50 Figure31.Hub/bladeconnectionforSavoniusblade(left)andBachblade(right)..................... 51 Figure32.SketchoftheAIT1.05x1.05mopencircuitlowspeedwindtunnel ............................. 52 Figure33.Opened1.05x1.05mtestsection ........................................................................................... 53 Figure34.Experimentalapparatus ........................................................................................................ 53 Figure35.Valuesofmeasurementazimuthalanglesblade ............................................................... 54 Figure36.Threebladestestsetup ........................................................................................................ 55 Figure37.Prototype01bladerotationasafunctionofthewindspeedandazimuthalangle 56 Figure38.Prototype02bladerotationasafunctionofthewindspeedandazimuthalangle 57 Figure39.Prototype03bladerotationasafunctionofthewindspeedandazimuthalangle (resultsfrom3blades).................................................................................................................... 58 Figure40.Prototype04bladerotationasafunctionofthewindspeedandazimuthalangle (resultsfromoneblade)................................................................................................................. 59 Figure41.RPMcharacteristicoftestedprototypes ............................................................................ 60 Figure42.Tipspeedratioofthebladesasafunctionofthewindspeed...................................... 60 Figure43.Relationbetweenbladeandturbinerpmforidlingcase(2blades).Markingpoints onlinesrepresentwindspeedsof4,6,8and10m/s ................................................................. 62
99
Figure44.PowercoefficientofthePrototype03modelasafunctionofMagnustipspeedratio for2bladescase............................................................................................................................... 64 Figure45.TorquecoefficientofthePrototype03modelasafunctionofMagnustipspeed ratiofor2bladescase ..................................................................................................................... 64 Figure46.PowercoefficientofthePrototype03modelasafunctionofMagnustipspeedratio for3bladescase............................................................................................................................... 65 Figure47.TorquecoefficientofthePrototype03modelasafunctionofMagnustipspeed ratiofor3bladescase ..................................................................................................................... 65 Figure48.PowercoefficientofthePrototype03modelasafunctionofMagnustipspeedratio for4bladescase............................................................................................................................... 66 Figure49.TorquecoefficientofthePrototype03modelasafunctionofMagnustipspeed ratiofor4bladescase ..................................................................................................................... 66 Figure50.PowercoefficientofthePrototype03modelasafunctionofMagnustipspeedratio for5bladescase............................................................................................................................... 67 Figure51.TorquecoefficientofthePrototype03modelasafunctionofMagnustipspeed ratiofor5bladescase ..................................................................................................................... 67 Figure52.Comparisonofpowercoefficientsbetweenprototypes03and02for4m/scase ...... 68 Figure53.Comparisonoftorquecoefficientsbetweenprototypes03and02for4m/scase...... 68 Figure54.Comparisonofpowercoefficientsbetweenprototypes03and04for4m/scase ...... 69 Figure55.Comparisonoftorquecoefficientsbetweenprototypes03and04for4m/scase...... 69 Figure56.Comparisonofpowercoefficientsbetweenprototypes03and01for6m/scase ...... 70 Figure57.Comparisonoftorquecoefficientsbetweenprototypes03and01for6m/scase...... 70 Figure58.Comparisonofpowercoefficientsbetweenprototypes03and02for6m/scase ...... 71 Figure59.Comparisonoftorquecoefficientsbetweenprototypes03and02for6m/scase...... 71 Figure60.Comparisonofpowercoefficientsbetweenprototypes03and04for6m/scase ...... 72 Figure61.Comparisonoftorquecoefficientsbetweenprototypes03and04for6m/scase...... 72 Figure62.Comparisonofpowercoefficientsbetweenprototypes03and01for8m/scase ...... 73 Figure63.Comparisonoftorquecoefficientsbetweenprototypes03and01for8m/scase...... 73 Figure64.Comparisonofpowercoefficientsbetweenprototypes03and02for8m/scase ...... 74 Figure65.Comparisonoftorquecoefficientsbetweenprototypes03and02for8m/scase...... 74 Figure66.Comparisonofpowercoefficientsbetweenprototypes03and04for8m/scase ...... 75 Figure67.Comparisonoftorquecoefficientsbetweenprototypes03and04for8m/scase...... 75
100
Figure68.Comparisonofpowercoefficientsbetweenprototypes03and01for10m/scase .... 76 Figure69.Comparisonoftorquecoefficientsbetweenprototypes03and01for10m/scase.... 76 Figure70.Comparisonofpowercoefficientsbetweenprototypes03and02for10m/scase .... 77 Figure71.Comparisonoftorquecoefficientsbetweenprototypes03and02for10m/scase.... 77 Figure72.Comparisonofpowercoefficientsbetweenprototypes03and04for10m/scase .... 78 Figure73.Comparisonoftorquecoefficientsbetweenprototypes03and04for10m/scase.... 78 Figure74.Variouscasesofvibrationinfluencedpowerperformance ........................................... 80 Figure75.Prototype01FEMmodelusedformodalanalysis .......................................................... 81 Figure76.Prototype02FEMmodelusedformodalanalysis .......................................................... 81 Figure77.Prototype03FEMmodelusedformodalanalysis .......................................................... 82 Figure78.Prototype04FEMmodelusedformodalanalysis .......................................................... 82 Figure79.Experimentalreadoutofthe1steigenmodefortestedblades...................................... 84 Figure80.RepresentationoftheeigenmodesonrpmwindvelocityplotforPrototype03 ....... 84 Figure81.Firstmode,f1st=17.562Hz ................................................................................................... 86 Figure82.Secondmode,f2nd=17.637Hz .............................................................................................. 86 Figure83.Thirdmode,f3rd=22.548Hz.................................................................................................. 86 Figure84.Fourthmode,f4th=107.36Hz Figure85.Fifthmode,f5th=111.65Hz........................ 87
Figure86.Sixthmode,f6th=122.56Hz................................................................................................... 87 Figure87.Prototype03bladedestroyedat12m/sduethecentrifugalbuckling ......................... 88 Figure88.DeformationvectorsumoftheP03bladefor3000RPM................................................. 89 Figure89.DeformationofthebladeandtipplatesforP03model ................................................. 90 Figure90.MaximumvaluesofprincipalandvonMisesstressesontherotatingblade ............ 91 Figure91.BachtypebasedMagnusrotor ............................................................................................ 93
101
AppendixB:ListofTables
Table1.Powerrequirementsforrunningthecylinderat500rpm .................................................. 19 Table2.ValuesofconstantsforRNGturbulencemodel ................................................................. 27 Table3.Gridparametersandterminationcriteria............................................................................. 29 Table4.Numberofiterationsforvariousrotorsandflowangles .................................................. 34 Table5.Statictorqueresultsrelativeerror.......................................................................................... 38 Table6.MaterialdecisionmatrixbasedonSavoniusrotorresearchbyMenet[23].................... 45 Table7.Characteristicsoftestedblades .............................................................................................. 46 Table8.Materialproperties.................................................................................................................... 46 Table9.Masspropertiesoftestedblades ............................................................................................ 51 Table10.Magnusturbineshaftpowerforoptimumvaluesoftipspeedratio ............................ 79 Table11.NaturalfrequenciesusingANSYSFEMsolver................................................................. 83
102

Nemanjakomatinovic 2006

Enviado por

Dados do documento

Descrição original:

Título original

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Nemanjakomatinovic 2006

Enviado por

Direitos autorais:

Formatos disponíveis

M.Sc.

InvestigationoftheSavoniustypeMagnusWindTurbine________________ S free cyl M S TipspeedratiooftheSavoniusrotor Densityofthefreestreamflow Angularvelocityofthecylinder AngularvelocityoftheMagnusturbinerotor AngularvelocityoftheSavoniusrotor

InvestigationoftheSavoniustypeMagnusWindTurbine________________ (seeFigure1).Fromthepotentialflowtheoryandintroducingthecirculation,weyield theliftcreatedbycylinderrotationas:

2 Rcyl cyl V free

0.5 Tip Speed Ratio s

motor bearings transmission

WhatshouldbenotedhereisthatdiscrepancyforSBRASresultsshouldbeaccounted foratransmissionlossesforsixrotatingcylindersthatarepoweredbyamotor.Incase ofaReidexperiment,onlyone,adoublybearingsupportedcylinderwastested.

6 m s 0.06m 24000 ; 1.5x10 -5 m 2 s

Benesh Profile 4.5 Leeward Side Windward Side 4

0.4 0.5 0.6 Distance along the profile (x/c)

0.4 0.5 0.6 Distance along the profile (x/c)

Leeward Side Windward Side

0.4 0.5 0.6 Distance along the profile (x/c)

0.4 0.5 0.6 Distance along the profile (x/c)

Savonius OL=0.21 Profile

Back Back Front Front

0.4 0.5 0.6 Distance along the profile (x/c)

r r FX TOTAL = FPRESS + FVISC i

r r FY TOTAL = FPRESS + FVISC j

r r r r M TOTAL = x FPRESS dl + x FVISC dl

wherexisthepositionvectorofthenodeelementrelativetocenterpoint,anddlisaline elementalongtheboundaryS,ofmagnitudedl.Torquecoefficientthereforecorresponds tomomentcoefficientforcenterpointi.e.pointofrotation:

0.4 Static Torque Coefficient C Q

0.4 0.5 0.6 Distance along the profile (x/c)

PolyethyleneHighDensity Polypropylene PolyvinylChloride(PVC) Plexiglas ShockPolystyrene AcrylonitrilButadeneStyrene AcrylonitrilStyrene Aluminum

Material PVC Aluminum Steel

Density [kg/m3] 1400 2710 7860

YoungModulus [N/m2] 3x109 7x1010 2x1011

PoissonRatio 0.4 0.346 0.266

Prototype Mass [kg] 01 0.390 02 0.480 03 0.575 04 0.220

80 100 Azimuthal Position [ ]

Dblade 2 nblade R = 2 = blade blade Vwind Vwind

7 Wind Speed [m/s]

1 Prototype 01 Prototype 02 Prototype 03 Prototype 04

7 Wind Speed [m/s]

InvestigationoftheSavoniustypeMagnusWindTurbine________________ Asinglerpmvaluefortheonewindspeedisobtainedasthemeanoftheallazimuthal rpmvaluesandthenaveragedforthetestednumberofblades:

Pmech = Qmech rot

400 Prototype 01 Prototype 02 Prototype 03 Prototype 04

300 Magnus Rotation [rpm]

1500 Blade Rotation [rpm]

0.8 1 Tip Speed Ratio M

0.07 4 m/s 6 m/s 8 m/s 10 m/s

0.05 Torque Coefficient Cq

Tip Speed Ratio M

0.035 Power Coefficient Cp

0.8 1 Tip Speed Ratio M

0.08 Torque Coefficient Cq

Tip Speed Ratio M

Tip Speed Ratio M

0.12 4 m/s 6 m/s 8 m/s 10 m/s 0.1

0.08 Torque Coefficient Cq

Tip Speed Ratio M

0.8 1 Tip Speed Ratio M

0.16 4 m/s 6 m/s 8 m/s 10 m/s

0.8 1 Tip Speed Ratio M

0.04 Power Coefficient Cp

0.6 0.8 Tip Speed Ratio M

0.08 Torque Coefficient Cq

0.6 0.8 Tip Speed Ratio M

0.04 Power Coefficient Cp

Tip Speed Ratio M

0.08 Torque Coefficient Cq

0.6 0.8 Tip Speed Ratio M