西班牙可再生能源中心風(fēng)功率預(yù)測(cè)

PedroCorreia:LauraFrías:IvánMoya:WindResourceAssessmentandPrediction-EPRTheNumericalWeatherPredictionmodels(NWP),havebeentraditionallyusedtopredicttherealstateofearth'satmosphere.Theinitialstateoftheatmosphere(ANALYSIS)isreproducedusingmeasurementsfromsatellites,weatherstations,etc,andconvertedtoaregulargridthatcanbeusedtofeedthemesoescalemodel.Byresolvingtheprimitiveequationswiththatinputdata,theNWPmodelscanpredicttheweatherinthefuture.There'sdifferenttypesofatmosphericmodels:Global:GlobalForecastingsystem(GFS)fromNcep/NCARandtheECMWFGlobalmodel,...Regional:Skiron;EtaModel;WRF;MM5,Hirlam;Aladin,...NWP:MesoescaleModels–ShortDescriptionMesoescalemodelthatusesthebaseoftheETAmodel.RequiresanUNIXOperatingSystem;Itisabletousetheweatherinputdatafrom;GFS(GlobalForecastingSystem);NCEP/NCARReanalysis1;ECMWF(GlobalModel)?CENERworkswithSKIRONsinceOctober,2023.CENERworkswithSKIRONsinceOctober,2023.Itwasfirstconfiguredtorunreal-timeforecasts,allowingCENERtoobtaindailyweatherpredictions.Fromsometimenow,themodelisalsobeenusedtowindanddirectsolarradiationresourceassessmentinanwiderangeoflocationsthroughouttheglobe.Togeneratewind/solarradiationmaps,it'sdesirabletorunSKIRONaslongaspossible(morethan5years),inordertoobtainthelongtermbehaviorofthedifferentmeteorologicalvariables,suchaspressure,windvelocityanddirection(atseveralheightsabovegroundlevel),directsolarradiation,temperature,etc.NWP:MesoescaleModels–SKIRON

SKIRON:Real-timepredictionsIt'sexecutedin16processorswithanhorizonof180ph->5h30minHorizontalResolution:0.1ox0.1o(~10kmx10km)->341x281ptsVerticalResolution:38Etaverticallevels.NonestingTemporalresolution:outputfrequency=1h(180dailyfiles)Dailydownloadandstorageof:GFS12UTC,SST,SnowcoverandSnowdepthBackupsystem:Thesamemodelconfiguration,indifferentmachineswhicharelocatedinanotherarea.Thisallowsustoguaranteetheclientsforecastsincaseofafailureinthemainsystem(powerfailure,computermalfunction,networkproblems,etc).WindForecastandWindEnergyproductionusingSKIRON

SKIRON:Real-timeReal-timedomainfromOctober2023untilNovember2023.DomainsinceNovember2023untilnow.WindForecastandWindEnergyproductionusingSKIRON

Intheelectricitygridatanymomentbalancemustbemaintainedbetweenelectricityconsumptionandgeneration-otherwisedisturbancesinpowerqualityorsupplymayoccur.

Windgenerationisadirectfunctionofwindspeedand,incontrasttoconventionalgenerationsystems,isnoteasilydispatchable.Fluctuationsofwindgenerationthusreceiveagreatamountofattention.

Managingthevariabilityofwindgenerationisthekeyaspectassociatedtotheoptimalintegrationofthatrenewableenergyintoelectricitygrids.Reasonforwindpowerforecasts

Statisticalpredictionmethodsarebasedononeorseveralmodels(linearandnon-linear)thatestablishtherelationbetweenhistoricalvaluesofpower,aswellashistoricalandforecastvaluesofmeteorologicalvariables,andwindpowermeasurements.

Modelparametersareestimatedfromasetofpastavailabledata,andtheyareregularlyupdatedduringonlineoperationbyaccountingforanynewlyavailableinformation(i.e.meteorologicalforecastsandpowermeasurements).StatisticalapproachtowindpowerFORECASTSFORTHEDAILYMARKET.LocalPredisoperationalsince2023andhasbeencontinuouslydevelopedsincethen.Reliabilityandaccuracyarethemaincharacteristicsofthesystem.

Reliabilityisbasedontheredundancyof:Hardware.Inputdata.Processes.Accuracyisobtainedthroughthecombinationofforecastswithdifferentinformation:“multi-modelensemble”.SupportVectorMachinetechnology,PCAalgorithms,dataqualitycontrol.Forecastsforoffshorewindfarms:Windfarmenergyproduction.Waves(WAM4highresolutionwaveforecasts).reducedvisibility.Operationalsince2001DESCRIPTIONOFLOCALPREDFORECASTINGSYSTEMGFSSKIRONECMWFPCAMOSENSEMBLEMOS2DELIVERYFTPCENERFTPCLIENTFTPAGENTSVMZAMUDIOCENERMultimodelensembleDESCRIPTIONOFLOCALPREDFORECASTINGSYSTEMCombinationAlgorithmDESCRIPTIONOFLOCALPREDFORECASTINGSYSTEMLocalPredincludesacombinationalgorithmdevelopedincollaborationwithDTU-IMM.

Thelevelofimprovementdependsontheerroroftheindividualforecastsandonthelevelofcorrelationbetweenthem.

Thecombinationisabletoimprovethebestindividualforecast.windfarmsclusteringanalysisEvolutionerrorindexvsthenumberofagreggatedwindfarmswindfarmsclusteringanalysisEconomicimpactobtainedbytheaggregationofwindfarmsFORECASTSFORTHEINTRADAYMARKETSIntra-dailymarketprocess.Focusonveryshort-termforecasts

TheactualSpanishelectricalmarket,allowsustocorrectthewindenergyforecastingpresentedinthedailymarketbymeansoftheintradailymarket.Thismarketisorganizedintosixsessionsandagentsthathavepreviouslyparticipatedinthedailymarketcanpresentnewprogramofproduction.

Thenewpredictionsmustbepresentedbetweentheopeningandclosinghoursofthesession.

Thus,ineachintradailysession,wecorrectamaximumoffivepredictions,andtakingintoaccounttheclosinghourofthesession,wecanusefromfourthtoeighthstepahead.Thereforetheimportanceoftheshorttimeforecastandsotheneedofashorttimemodelfallshere.Intra-dailymarketprocess.Focusonveryshort-termforecasts

Anewmodelforshort-termpredictionhasbeendevelopedtakingintoaccounttheSpanishmarketrules.Thismodelisfocusedinshortforecastinghorizons.First,itusesonlinepowerproductiondataofthewindfarmstobuilddifferenttimeseriesmodels(Box-JenkinsmethodologyandaversionofHoltWintersAlgorithm).Ontheotherhand,itutilizesexistingforecastsforthedailymarketproducedbyCENER’sLocalPredmodelbasedonmesoscaleNWPandMOScorrections.Finallyitimplementsacombinationalgorithmthatofferstheoptimalforecastforeachhorizon.MethodologyoftheshorttimeforecastingmodelofCENERResultsobtainedWepresenttheresultsobtainedfromtheCENERshorttimemodelappliedonamediumwindfarmfromSpainbetweenFebruaryandDecember.Wepresenttheimprovementofthenewmodelagainstthepersistenceasshorttimepredictionandagainstthedailymarketforecasting.EUROPEANPROJECTSR+DEuropeanprojects(VIandVIIFrameworkProgram):UPWIND

“Findingdesignsolutionsforverylargewindturbines”POW’WOW

“PredictionOfWaves,WakesandOffshoreWind”ANEMOS

"DevelopmentofaNextGenerationWindResourceForecastingSystemfortheLarge-ScaleIntegrationofOnshoreandOffshoreWindFarms"ANEMOS.PLUS

“AdvancedToolsfortheManagementofElectricityGridswithLarge-ScaleWindGeneration”SAFEWIND

“Multi-scaledataassimilation,advancedwindmodellingandforecastingwithemphasistoextremeweathersituationsforasafelarge-scalewindpowerintegration”

PUBLICATIONS

[1]Martí,I.,Nielsen,T.S.,Madsen,H.,etal.Predictionmodelsincomplexterrain.ProceedingsoftheEuropeanWindEnergyConference.Copenhagen,July2023.[2]Martí,I.,Usaola,J.etal.Windpowerpredictionincomplexterrain.LocalPredandSipreólico.ProceedingsoftheEuropeanWindEnergyconference,June2023.[3]Martí,I.etal.Windpowerpredictionincomplexterrain:fromthesynopticscaletothelocalscale.“Thescienceofmakingtorquefromwind”.Delft.TheNetherlands,2023.[4]M.Gastón,L.Frías,M.J.SanIsidro,I.Martí.Exploringthelimitsofwindfarmgroupingforpredictionerrorcompensation.EWEC2023.[5]

L.Frías,M.Gastón,I.Martí.Anewmodelforwindenergyforecastingfocusedintheintra-dailymarkets.EWEC2023.[6]L.Frías,E.Pascal,U.Irigoyen,E.Cantero,Y.Loureiro,S.Lozano,P.M.Fernandes,I.Martí.SupportVectorMachinesinthewindenergyframework.Anewmodelforwindenergyforecasting.EWEC2023.ThedailySKIRONforecastscanbevisitedat

.CloudCover,Snow,Wind,Temperature,Precipitation,Localforecaststomaincities,etc...Skironreal-timemeteorologicalproducts:

Anomalies:Windand/orEnergyDensity:Anomaliesmapswithan1kmx1kmresolution.ItisnecessarytopossessasimulateddatabasewithSKIRONtobeabletocalculatethesemaps.It'spossibletocalculatetheanomaliesforeverydesiredperiod(daily,weekly,monthly,seasonally,yearly,….etc)Thereferenceperiodusedtoobtaintheseproductcanalsobechangedaccordinglytotheclient’sneeds.There'salsothepossibilitytogeneratevariabilitymapstohelpidentifyunstable(regardingwindresource)regions.TheyaredeliveredinGISformatwithseverallayersattached(Anomaly;topography,windfarmlocations,etc)Skironreal-timemeteorologicalproducts:FigureInterpretation:AttheEbroValley,theFebruarywindvelocitywas20%lessthanthelast6yearsmean.Typesofanomalies:WindandEnergyDensityMonthlyYearlySeasonly...Advantage!!:Allowstoindentifyinaveryintuitiveway,ifagivenregionhadregistedabove/underaveragewinds.→Easytoidentifyapossiblecauseforover/underproductionofawindfarm.Skironreal-timemeteorologicalproducts-AnomaliesTheuseofmesoescalemodelsforwindresourceassessmentisarecentactivity.Thestandardmethodologymayvary,accordinglywiththemodeluser,butthegoalisthesame:TotakeadvantageoftheNWPmodelscapacitytopredictwind.Methodology??-Themaingoalistodeterminethewindclimatologyinsteadofthereal-timeprediction.Insteadofpredictinthefuture,longperiods(years)oftimearesimulatedusingastoredinputdataarchive(GFS,Reanalysis).SKIRON:WindresourceAssessmentThemethodologiesusedintheseclimaticsimulationscanbeverydifferent,butsomeaspectsarecommon:1. InitialDataThemodelusedtocalculatethewindmapneedsinitialinputdataandinitialboundaryconditions.Therearefewavailablesourcesofthatinformationtosuchlongperiods:Reanalysis(NCAR/NCEP,ECMWF,JRA)orthestoredoutputsfromtheGFSandECMWFmodels.2. ClimaticSimulation.Itisnecessarytoobtainalongtermrepresentativesimulationoftheareainquestion.Theonlyavailableoptionsaretryingtosimulatethelargestavailableperiod:5,10,15ormoreyears,orsimulateaclimatologicalrepresentativeyeartotheregionofinterest.SKIRON:WindresourceAssessmentExamplesofmethodologiesusedtoobtainarepresentativeWindAtlas.SKIRON:WindresourceAssessmentCENERtestcase:SKIRON:WindresourceAssessmentInCenercase,24weatherstationsinNavarra,thatfullfillalltherequirements,havebeencarefullyselectedandtheresultshavebeenanalyzedtakingintoconsiderationthecomplexityoftheterrain,meanwindvelocityandstation.MAE,RMSEandBiaswerecalculated:Ifwelookonlytothesimulationswithreanalysisdata,itcanbestatedthatthelowestMAEisachievedwiththe0,03ox0,03oresolution,butthelowestBiasisachievedinthe0,1ox0,1oresolution.Ifweanalyzeallthesimulationsmade,it'seasytoseethattheoptimalconfigurationtotheSKIRONmodelisGFSasinputdataandanhorizontalresolutionof0,05ox0,05o.SKIRON:WindresourceAssessment“Climatic”Simulations:HorizontalResolution:0.05ox0.05o(~5kmx5km)?VerticalResolutions:50Etaverticallevels.NonestingTemporalresolution:outputfrequency=1h(48hhorizon)Inputs:GFS12UTC,SST,SnowcoverandSnowdepthAvailableperiodatCENER(fromJune2023untilnow–8years)CENERcomputacionalResources:618processors640GBRAMUnixOperatingSystemSKIRON:WindresourceAssessmentMEASUREDSKIRON0.05ox0.05oGFS1ox1oHOURLYWINDVALUESVALIDATIONFILTERINGWINDMAPHOURLYTIMESERIESGISWEBSERVICESKIRON:WindresourceAssessment01/01/2023:…………..48hourlymaps31/12/2023:…………..48hourlymaps......(Firstrun)?(Lastrun)?(1274runs)?Toruneachyear,182simulations,eachonewithan48hourspredictionhorizonarelaunched.Sothenumberofsimulationsneededtoruneachcaseare:

NoYearsx182=NoSimulationsSKIRON:WindresourceAssessment++++++++7(y)x365(d)x24(h)=61320hourlywindmapsTheresultfromallthesimulationsisonewindmapforeveryhourinthechosenperiod.ADVANTAGE!!:Thepossibilitytovalidatethewindmapwithrealmeasures.SKIRON:WindresourceAssessmentInTunisia,CENERusedthemeasurementsfrom17weatherstations,withanemometersinstalledat20mand40m.ThatallowedustoperformanexhaustivevalidationofCENERmethodologytoobtainwindmapsandvirtualseries.TUNISIAWINDMAPSKIRON:WindresourceAssessment–TunisiaIntheGreatLakesWindMap,CENERusedwindmeasuresfrom50weatherstations,mostofthemhadanemometersatan10mheight.Thiswasagoodtestcasetovalidatethesimulatedwind,bothonshoreandoffshore.SKIRON:WindresourceAssessment–GreatLakesWiththegoalofvalidatetheoffshorevirtualseriesgeneratedwithSKIRON,awindmapfortheNorthSeawasgenerated:

FINOValidation:6months:January-June2023Datacoverage:90%SKIRON:Resolution:0.05o,50verticallevels,1hrForecastmaximumhorizon:48hrMeasures:Windat:33,40,50,70,80,90mTemperatureat:30,40,50,70,100mRHat:33,50,90mTowereffectcorrectedinthewindvalues.SKIRON:WindresourceAssessmentOffshore–FinoExcellentresultsbothinVelocityanddirectionSKIRON:WindresourceAssessmentOffshore–FinoSonicsCupsFino154.014oN6.5905oESKIRON:WindresourceAssessmentOffshore–FinoSKIRON:WindresourceAssessmentOffshore–FinoMeanWindMapfromthedesiredregionwithafinalhorizontalresolutionof1kmx1km(Inversedistanceweightedinterpolation);Themeanwindcouldbecalculatedathubheight;Windrosesandprobabilitydistributioninrepresentativepointsinjpgformat.DeliveredinGISformatwithseverallayersofinformation(MeanWind,Topography,Protectedareas;Electricgrid,etc)POLANDWINDMAPTUNISIAWINDMAPNORTHSEAWINDMAPEASTEUROPEWINDMAPSKIRON:WindresourceAssessmentOffshore–DerivedproductsGISFormatcompatible KMZFormat–GoogleEarth SKIRON:WindresourceAssessmentOffshore–DerivedproductsEnergyDensityMapsMeanEnergyMapsfromthedesiredregionatan1kmx1kmhorizontalresolution.Theenergyiscalculated,pointbypoint,toeveryhourusingthepressureandtemperaturesimulatedwithSKIRON.Thismeansthattheairdensityusedit'salsocalculatedandnotaveragedtotheentiredomain.Theenergydensityobtainedisrepresentativeofthesimulatedperiod.It'scalculatedatthehubheight.It'sdeliveredinGISformat(Energydensity+standardlayers)Weibullparametersmaps,Individualmapsfortheparameters(A,k)totheentiresimulatedregionatafinalhorizontalresolutionof1kmx1km.AlsodeliveredinGISformat,withthestandardinformation.SKIRON:WindresourceAssessmentOffshore–DerivedproductsCENERsimulateddomains:Wind:IberianPeninsula;Poland;Romania;GreatLakes;Mexico;CentralAmerica,Chile;Peru;Fino,Brazil(NERegion)SolarRadiation:Australia;NorthAfrica;UnitedEmiratesCENERSimulatedDomains(Wind/SolarRadiationMaps)Withthegoaltosuppressthelackofmeasurementsinsomelocations,CENERhasdevelopedandvalidatedamethodologycapableofgeneratingvirtualwindseries(velocityanddirection),energydensity,temperature,pressure,etc,usingthemesoescalemodelSKIRON.Inordertoobtainthewindseriesatadesiredlocation,hourlywindoutputsfromSKIRONareusedasinputtothemicroscalemodelWasPandthencorrectedbytheAdfactorgivenbytheWasP/CFDsimulation.WindresourceAssessment–DownscalingandVirtualWindSeriesAdvantage!!:AllowstodetectthelocaleffectscausedbythelocaltopographythatSKIRONcan'tresolve.SKIRON0.05ox0.05oHOURLYWINDVALUESWAsPSRTMAdFACTORHOURLYTIMESERIESHIGHRESOLUTIONWUNDMAPSWindresourceAssessment–DownscalingandVirtualWindSeriesHourlywindoutputsfromSKIRONareusedasinputtothem