FatigueFailureAccidentofwindturbinetowerinTaikoyamaWindPowerPlantYinLIU1,TakeshiISHIHARA21,2DepartmentofCivilEngineering,SchoolofEngineering,theUniversityofTokyo,Tokyo,Japan1.IntroductionIn2013thenacelleofNo.3windturbineinTaikoyamawindpowerplantcollapsedduetofatiguefailure[1].(a)Collapsednacelle(b)Fracturesection(c)VerticalcrosssectionFig.1AccidentsceneandschematicdiagramByobservingthefracturefacesoftheturbinetowertube,evidenceoffatiguecrackpropagationwasfoundattheinnersideofthetube.Furthermore,brokenboltswerefoundduringthefiledinvestigation.Bycomparingthetwoaspects,thisfailureisconsideredtobeprecededbyacertaindegreeoffatiguedamagecausedbythedecreaseofboltspre-tensionforce.Thiswindturbinecollapsedonly12yearsafteritwasbuilt,comparingtothelifetimeof20years.Furthermore,theperiodicalinspectionwascarriedoutonlythreemonthsbeforetheaccident.Additionally,therearemorethan120windturbinesinserviceofthesametype.Therefore,itisnecessaryandurgenttounderstandthecauseofthisaccident,sothatsimilarfailurecanbepreventedinthefuture.2.Approach(1)AerodynamicmodelisdevelopedtosimulatethedynamicperformancebyGLGarradHassanBladed[2].Thetowerpartreferencestheconstructiondrawings，thebladesinformationisbasedonGuidelinesforDesignofWindTurbineSupportStructuresandFoundations,JSCE[3],andthecontrolparametersarealsobasedonreference[3]butmodifiedtovalidatethedynamicsimulationresultswithmeasuredresults.Thecomparisonofthemeasuredresultsandsimulationresultsiscarriedouttoverifythemodel.(2)Stressdistributionatfracturesectioniscalculatedbyaerodynamicmodelinordertounderstandthecauseofthefatiguefracture;(3)A3DimensionalFEmodelisdevelopedtocalculatethelocalstressatthefracturesectionbeforeandaftertheboltsdamaged;(4)Eventually,byincorporatingthetimeseriesstressdistributionfromaerodynamicmodelandtherelationshipbetweenlocalstressandnominalstressfromFEmodel,thefatiguelifeofthefracturesectioncanbeevaluatedbyusingrain-flowcountingmethod,GoodmanformulaandMiner’sprinciple.1Presentingandcorrespondingauthor,PhDcandidate,E-mail:liuyin@bridge.t.u-tokyo.ac.jp50m45.96m0mFracturesection3.Mainbodyofabstract3.1WindloadcharacteristicThetowerbasemoment(height12.6m)wasmeasuredbyusingstraingaugesshowninFig.2tovalidatetheaerodynamicmodelfromFeb.2nd2015toFeb.28th2015.Themeasuredresultsarebasedonthe10minutesSCADAdataduringthesametime.Fig.2showsthebinaveragedturbulenceintensityderivesfromthemeasuredresult.Indynamicsimulation,measuredturbulenceintensitywasusedforwindspeedlowerthan15m/s.ForhigherwindspeedtheturbulenceintensityisextrapolatedassumingthenormalturbulenceintensityinIEC61400-1[4].Theredlineshowstheturbulenceintensityusedindynamicsimulation.00.10.20.30.40.50.60510152025measurementIECAdoptedcurveTurbulenceintensityWindspeed(m/s)Fig.2SchematicdiagramofstraingaugesinstallmentFig.3Turbulenceintensityvs.windspeedFig.4showsthecomparisonofbinaveragedpoweroutput,rotorspeedandtowerbasemomentbymeasurementanddynamicsimulation.Thefiguresshowgoodagreementwiththemeasurementresults.01002003004005006007000510152025measurementsimulationPoweroutput(kW)Windspeed(m/s)10152025300510152025measurementsimulationRotorspeed(rpm)Windspeed(m/s)-500050010001500200025000510152025measurementsimulationAveragemoment(kNm)Windspeed(m/s)(a)Poweroutput(b)Rotorspeed(c)TowerbasemomentFig.4Comparisonofthemeasurementandsimulationresult3.2Characteristicsofthetowertopsectio...
优质VIP
