D2.04 Collation of Offshore Wind-Wave Dynamics

bbbbbb
D2.4:Collationofoff
shorewindwave
dynamics
Author(s):
H.Bredmose DTUWIND
S.E.Larsen DTUWIND
D.Matha USTUTT
A.Rettenmeier USTUTT
E.Marino UNIFI
L.Saettran NTNU

MarineRenewablesInfrastructureNetwork
Revision:02
Date:30November2012
D2.4: Collation of offshore wind-wave dynamics

Rev. 02, 30 November 2012
Page 2 of 50
ABOUT MARINET
MARINET (Marine Renewables Infrastructure Network for Emerging Energy Technologies) is an ECfunded
consortium of 29 partners bringing together a network of 42 specialist marine renewable energy testing facilities.
MARINEToffersperiodsoffreeaccesstothesefacilitiesatnocosttoresearchgroupsandcompanies.Thenetwork
also conducts coordinated research to improve testing capabilities, implements common testing standards and
provides training and networking opportunities in order to enhance expertise in the industry. The aim of the
MARINETinitiativeistoacceleratethedevelopmentofmarinerenewableenergytechnology.
Companiesandresearchgroupswhoareinterestedinavailingofaccesstotestfacilitiesfreeofchargecanavailofa
range of infrastructures to test devices at any scale in areas such as wave energy, tidal energy and offshorewind
energyortoconductspecifictestsoncrosscuttingareassuchaspowertakeoffsystems,gridintegration,moorings
and environmental data. In total, over 700 weeks of access is available to an estimated 300 projects and 800
externalusers.
MARINETisconsistsoffivemainareasoffocusorWorkPackages:Management&Administration,Standardisation
&BestPractice,TransnationalAccess&Networking,ResearchandTraining&Dissemination.Theinitiativerunsfor
fouryearsuntil2015.
Partners
Ireland
UniversityCollegeCork,HMRC(UCC_HMRC)
Coordinator
SustainableEnergyAuthorityofIreland(SEAI_OEDU)
Denmark
AalborgUniversitet(AAU)
DanmarksTekniskeUniversitet(RISOE)
France
EcoleCentraledeNantes(ECN)
InstitutFranaisdeRecherchePourl'Exploitationde
laMer(IFREMER)
UnitedKingdom
NationalRenewableEnergyCentreLtd.(NAREC)
TheUniversityofExeter(UNEXE)
EuropeanMarineEnergyCentreLtd.(EMEC)
UniversityofStrathclyde(UNI_STRATH)
TheUniversityofEdinburgh(UEDIN)
QueensUniversityBelfast(QUB)
PlymouthUniversity(PU)
Spain
EnteVascodelaEnerga(EVE)
TecnaliaResearch&InnovationFoundation
(TECNALIA)
Belgium
1Tech(1_TECH)
Netherlands
StichtingTidalTestingCentre(TTC)
StichtingEnergieonderzoekCentrumNederland
(ECNeth)
Germany
FraunhoferGesellschaftZurFoerderungDer
AngewandtenForschungE.V(Fh_IWES)
GottfriedWilhelmLeibnizUniversittHannover(LUH)
UniversitaetStuttgart(USTUTT)
Portugal
WaveEnergyCentreCentrodeEnergiadasOndas
(WavEC)
Italy
UniversitdegliStudidiFirenze(UNIFICRIACIV)
UniversitdegliStudidiFirenze(UNIFIPIN)
UniversitdegliStudidellaTuscia(UNI_TUS)
ConsiglioNazionaledelleRicerche(CNRINSEAN)
Brazil
InstitutodePesquisasTecnolgicasdoEstadodeSo
PauloS.A.(IPT)
Norway
SintefEnergiAS(SINTEF)
NorgesTekniskNaturvitenskapeligeUniversitet
(NTNU)

Page 3 of 50
Acknowledgements
The research leading to these results has received funding from the European Union Seventh Framework
Programme(FP7)undergrantagreementno.262552.
Legal Disclaimer
Theviewsexpressed,andresponsibilityforthecontentofthispublication,liesolelywiththeauthors.TheEuropean
Commissionisnotliableforanyusethatmaybemadeoftheinformationcontainedherein.
Page 4 of 50
REVISION HISTORY
Rev. Date Description Author Checkedby

01 31/102012 Draft H.BredmoseandS.E.Larsen Selfchecked
02 30/112012 Final H.BredmoseandS.E.Larsen Crosschecked


Page 5 of 50
EXECUTIVE SUMMARY
ThepresentreportconstitutestheProtocolManualforensuringharmonisationofoffshorewindandwave
simulationbeingimplementedatMaRINETfacilities.Windandwaveclimatesforfiveoffshorewindsitesinthe
NorthSeaandtheBalticSeahavebeenpresentedintermsofprobabilitydistributionsforwindspeedalongwitha
seriesoflumpedseastatesandturbulenceintensityvalues,parameterisedwithrespecttothewindspeed.Further,
extremevaluesforwindspeedandsignificantwaveheighthavebeenprovided.
Furthertothewinddistributionsandlumpedcharacteristics,theWeibullparametersforthewinddistributionand
explicitformulasfortheturbulenceintensityandsignificantwaveheightareprovided.Forthecorrelationofwave
peakperiodandsignificantwaveheight,astandardformulafromtheIEC614003codehavebeenfoundtocover
thescatterinthedata,althoughonecoefficientinthisformulamustbedecideduponbytheuser.Further,thevalue
of,theJONSWAPpeakenhancementparametermustbechosenbytheuser.Thiscanbedoneeitherfroman
explicitformulaorbythestandardchoicesof=1.0or=3.3.Herebyafulldescriptionofaunidirectionalwindwave
climatecanbeconstructed.Ifneeded,thisclimatecanbesupplementedbytheuserwiththecombineddirectional
distributionofwindandwaves,eitherbasedondataorintermsofparametricstudies.
Thescalingmethodproposedisthedynamicelasticscaling,whichmaintainstheratiosbetweenhydrodynamic,
aerodynamic,stiffnessinducedandgravitationalforces.ThisscalingpreservestheFroudenumberforthewater
phaseandthetipspeedratiofortherotor.TheReynoldsnumbersforairandwater,however,arenotconserved.A
redesignofthemodelscalebladeswillthereforebeneeded.Herethescaledthrustcurvemustbematched.Further,
ifpossible,thetorquefromtheairfoilshouldbematched.Thisrequirement,however,isdifficulttoachievedueto
thechangeinlift/dragratioatlowReynoldsnumber.Itisthereforeforeseen,thattheaerodynamictorqueandthus
producedpowerwillnotbescaledcorrectly.Asaconsequence,rollforcinginducedbythedynamicchangein
generatormomentwillnotscalecorrectly.However,thecorrectscalingofrotorthrustisfoundtohavehigher
priorityandthusjustifiesthescalingchoice.
An example of downscaling of wind and wave conditions has been supplied. The example also demonstrates how
thestructure(afloatingwindturbine)shouldbescaled.Itisdemonstratedthatthe proposedscalingyieldsmodel
scaleresultsforthrustandwaveinducedmotionthatcanbeupscaledtoprototypescalewithaperfectmatch.

Page 6 of 50
CONTENTS
1 INTRODUCTION...............................................................................................................................................7
2 CHARACTERISTICSOFOFFSHOREWINDWAVECLIMATE................................................................................8
2.1 SOURCESOFDATA......................................................................................................................................8
2.2 PARAMETERSFORWINDANDWAVESPECIFICATION............................................................................................9
2.3 ONEPARAMETERCLIMATEBASEDONWINDSPEED.........................................................................................10
2.4 STANDARDOFFSHORECONDITIONSFORTESTINGANDMODELLING......................................................................21
3 SCALINGOFWINDANDWAVECONDITIONSFORPHYSICALMODELTEST.......................................................29
3.1 REVIEWOFEXISTINGSCALINGSTUDIESANDRELEVANTNONDIMENSIONALNUMBERS...............................................29
3.2 RECOMMENDEDSCALINGMETHOD...............................................................................................................33
3.3 SCALINGOFWINDANDWAVECLIMATEPARAMETERS.......................................................................................38
3.4 EXAMPLE:SCALEDEXPERIMENTOFAFLOATINGWINDTURBINE..........................................................................38
3.5 DISCUSSION............................................................................................................................................46
4 CONCLUSIONSANDRECOMMENDATIONS.....................................................................................................47
5 REFERENCES...................................................................................................................................................48

Page 7 of 50
1 INTRODUCTION
The aim of this report is to constitute a protocol manual to ensure harmonisation of offshore wind and wave
simulationatfacilitieswithintheMaRINETproject.
Basically,thepresentationisorganisedintotwoparts:
In section 2, we review the information that is available in international standards, reports and papers describing
datasets,informationandproceduresregardingenvironmentalparametersthatmustbespecifiedforoffshorewind
turbines. The review is specifically focused on parameters that are often handled also within laboratory facilities
meaning that focus is on the wind and wave parameters. In section 2.4 the review is condensed into a
recommendationforasetofstandardwindandwaveclimates,whicharesuitableforgenericuse.Itisemphasised
thattheseclimatesarenotintendedtoformadesignbasisforanyrealstructure.
In section 3, we review the scaling necessary for physical model tests with simultaneous wind and waves. The
necessary considerations for dynamicelastic scaling is introduced and applied to the wind and wave fields for a
floatingwindturbine.Anexampleofdownscalingawindwaveclimatetomodelscaleandupscalingofthemodel
resultstoprototypescaleisgiven.Theresultsandimplicationsarediscussed.
Finallysection4and5containtheoverallconclusionandthereferencesrespectively.

Page 8 of 50
2 CHARACTERISTICS OF OFFSHORE WIND WAVE CLIMATE.
In the present section, we consider the current specification of the offshore wind and wave climate for design
purposes. We mention other parameters that would be part of an offshore observation system, but would only
rarelyenterintotestsinlaboratoryfacilitiesduetomodellingdifficulties.Finally,theclimatespecificationsfromfive
specificsitesarecomparedandagenericdescriptionisdevised.
2.1 SOURCES OF DATA
The main data sources for this report are the design basis from the EUUpWind project [1]. The report gives met
ocean data from two sites in the North Sea. One site, IJmuiden Munitiestort, is a shallow water site with 21 meter
water depth. The other site, K13 of 25 m depth is also denoted a shallow water site. After suitable supplement of
datafromanadditionalwavebuoyat50mdepth,theK13dataarefurtherarguedtoreflectanominaldepthof50
m,sincethelargerdepthwillmainlybeimportantfortheextremewaveheights[1].Inthefollowingweshallshorten
IJmuidenMunitiestorttosimplyIjmuiden.
TheUpWindsitelocationsareshowninFigure2.1.1.
Figure2.1.1.UpWindsitesforthedatasamplingfordesignbasisstudies.Left:K13,servingbothasdeepwaterandshallow
watersite.Right:IJmuidenMunitiestortshallowwatersite.From[1].
Additional to the data from the UpWind the present report draws from a number of standards developed for the
offshore wind energy industry, notably IEC614003 [2], DNV:OJJ101 [3] and ABS [4]. Further, additional data are
addedfromotherEUprojects,notablytheEUDOWNVInD(withtheBeatriceandtheSdraMidsjbanken)andthe
EUNORSEWINDprojects.ThepositionsofthefourassociatedlocationsareshowninFigure2.1.2.

Page 9 of 50
Figure2.1.2.Fulloverviewoflocationsreferredtointhisreport,Thesitesare:IJmuiden:1,K13:2,Beatrice:3andSdra
Midsjbanke:4.Forthetwofirstseethemoredetailedfigure2.1.1.
The wind and wave data presented in this study thus provide examples from sites in the North Sea and Baltic Sea
and further illustrate the variability between the locations. For some parameters and relations, notable similarities
are observed and might thus represent general climate characteristics with wider applicability. Here, however, it
mustbeemphasizedthattheapplicabilityofsuchgeneralrelationsareonlyvalidtotheextentthatnoexplicitand
implicit assumptions of the applied models are violated. Examples of such violating aspects could be depth
limitations,waverefraction,fetchconditionsforwindgenerationofwavesandinadequatewindsheardescription.
Withtheaboveinmind,forfulldesignbasisactivities,theimportanceofsitespecificdatacannotbeoverstated.In
thiscontext,anobviousmissinthedatasetsfromaEuropeanclimatepointofview,aredatafromtheAtlanticWest
coast, and from the Mediterranean, as well as the Baltic Sea with ice probabilities. Both because of the different
wind and wave climate ruling there and because of the practical needs for considerations about drift ice and icing
due to spray. Additional sites outside Europe would also make out a relevant extension of the study, as many
Europeanentrepreneursareactiveplayersinthenowworldwideexpansionofoffshorewindenergy.
Inspiteofthesereservations,wehavestillchosentheUpWinddesignbasis[1]asthebasicillustrativedataforthis
report,becauseitisarecent,large,comprehensiveandaccessibledatabase.
2.2 PARAMETERS FOR WIND AND WAVE SPECIFICATION
Thecorepartofthedesignbasisforanoffshorewindfarmconsistsofdataforthewindandwaveclimate.Thewind
characteristics of a given site can be determined from multiyear measurements of wind speed and wind direction,
preferablyatseveralheights.Similarandsimultaneoustimeseriesfortheheightanddirectionofthesurfacewaves
mustbeestablished.

Page 10 of 50
Thewindclimateischaracterisedby
- meanspeed
- winddirection
- windshear
- turbulencestandarddeviationandturbulenceintensity
- turbulencefrequencyspectrum
- extremewindspeedwith1yearand50yearreturnperiod.
The wind measurements must be accompanied by measuring height, z, averaging time, often 10 min averages,
samplingtime,often10min)
Thewaveclimateischaracterisedby
- significantwaveheight,H
s
(standardderivedonbasisof3hoursdata)
- peakperiod,T
p
.
- wavedirection
- frequencyspectrum
- directionaldistribution
- misalignment(relativetothewinddirection)distribution
- extremevalueofH
s
withT
p
,andthederivedheights,H
Smax
,andH
Sred
(with1and50yearreturnperiod)
The multidimensional distribution function for these parameters must be considered to evaluate the design.
Normally one can simplify the approach considerably, using physical and statistical knowledge. This is illustrated
below, where we discuss the climate conditions as function of one parameter only, the wind speed. The basis for
this approach is that the mean wind speed is the most important parameter to characterise both the wind
turbulenceandthewavefield.
Additionalinformationlikewaterdepth,tides,currentsandtemperatureinairandwatermustbemonitoredatthe
measuring site too. This information is available in the data of the UpWind design basis [1] project along with
informationonmarinegrowthandbottomsoilfeatures.Finallywenotethattheoccurrenceofwatericeandicing,
with associated loads, should be considered in relevant regions. This was not relevant, however, for the UpWind
sites.
Inthepresentreportwewillconcentrateonthewindandwaveinformationduetothefocusofmodellingandtests
attheexperimentalfacilitieswithintheMaRINETproject.Ourapproachistoseekforrobustcorrelationsbetween
the wind and wave parameters important for the evaluation of both fatigue and extreme loads on offshore wind
turbinestructures.Inordertodoso,westartbysummarisingasimplifiedoffshoreclimatedrivenbythewindspeed
inthenextsection.
2.3 ONE PARAMETER CLIMATE BASED ON WIND SPEED.
Inthissectionweseektodescribethedesignandloadparametersforoffshorewindturbines,bothwithrespectto
wind and other atmospheric parameters and with respect to the surface waves and other characteristic water
parameters. Aoneparameterapproachispresented,wherethewindspeedisconsidered afreeparameterandall
the other quantities are described conditionally to the wind speed. In the following, the parameters of this
descriptionandtheirtypicaldistributionsaresummarized.

Page 11 of 50
2.3.1 Winds and other atmospheric parameters.
ThedistributionofthemeanwindisnormallytakenasaWeibulldistribution,inthemediumtohighwindintervalas
is illustrated in the Figure 2.3.1. The distribution can either be derived for each wind direction sector, or as the
marginaldistributionforallsectors.Ithasthefollowingmathematicalform:
(2.3.1)
1
( ) exp
k k
k V V
f V
A A A
| |
| | | |
=
|
| |
|
\ . \ .
\ .
,
V being wind speed, and A and k being the scale parameter and shape parameter respectively. For the power and
load estimates, the relevant height is the hub height, presently being 80110 meter, which is the height interval
used throughout the UpWind reference 1 and in the figures and tables here, copied from that report. By
convention, however, wind climates are usually described at 10 m height. Conversion of the wind velocity
between heights can be done with equation 2.3.2, or with equation 2.3.4, which is a powerlaw wind profile.
Here we use 2.3.4 with,thepowerlawexponent,=0.14,ifnothingelseisindicated,followingtheinformationin
[1,2].

Figure2.3.1The10mwinddistributionathubheightfromtheIJmuidensite.From(1)
Whilethepowerlawprofileisacommonlyusedapproximation,thewindspeedtypicallyvarieswithheightasgiven
bythefollowingprofileexpression:
(2.3.2)
0
( ) (ln( ) ( , , ))
u z
V z z h T
z

k
-
= A ,
wherezisthemeasuringheight,u
*
,thefrictionvelocity,andz
0
theroughnesslengthofthesurface.Furtheris a
correction function that becomes important if 1) z becomes significant relative to the height of the atmospheric
boundary layer, h, or 2) the difference in temperature between the water and the air, T, becomes significant.
Togetherwiththeotherparametersintheequation,Tdescribestheatmosphericthermalstability.isnormally
neglected over the ocean, but with the wind turbines reaching hubheights larger than 100m, it may not be
defensibleanymore.Asthissubjectisstillunderactiveresearchandisnotyetincludedinthedesignstandards,
0 isassumedthroughoutthepresentreport.
Theroughnessheightz
0
canbedeterminedfromCharnocksrelation:
Page 12 of 50
(2.3.3)
2
0
/ z Cu g
-
= ,
whereCisacoefficientbetween0.01and0.015,dependentonthenearnessofacoast(andtosomeextentalsoon
windandwavehistory),andg istheacceleration duetogravity. Overwater,z
0
isasmallquantity,of theorderof
0.10.3mm.
Asanalternativetotheloglawaboveonecanusethesocalledpowerlawprofilegivenby
(2.3.4)
10 10
( ) ( )( / ) V z V z z z
o
=
,
Thepowerlawcoefficientistakenas0.14intheUpWinddesignbasis[1],whichisalsorecommendedin[2],but
bycomparisonwiththelogexpression(2.3.2)itisseenthatitwillvarywithz
0
andATandalsothezintervalused.
Thisvariationisneglectedinmostguidelineandstandardliterature.Hereitwillshowuptogetherwiththestatistical
scatter around the average behaviour. However, from both (2.3.2) and (2.3.4 with variable ), it is seen that the
Weibull distribution must change both A and k with height, not only A, as would be the case with a constant .
Indeedcloserstudiesshowthattypicallykwillincreasefromitsvalueat10mwithheightuptoaround80mfollowed
by a gradual decrease further up [16], showing a maximum variation of 0.5 across the boundary layer [, However,
thisispresentlyunderresearch.Inthepresentsimplifieddescription,theWeibullkparameteristhereforetakento
beindependentwithheight,consistentwiththechoiceofaconstantvalueof.
In the EUNORSEWIND project, the shear is measured directly from measuring stations in the North Sea and the
WesternBaltictypicallyaround100meterabovetheseasurface.ThemeasurementsarebasedonbothLIDARsand
conventional profile instrumentation. The results show a fairly large scatter between stations and for each station.
TypicaldistributionsareshowninFigure2.3.2.EstimatesfortheextremeshearvaluesaregivenintheIEC614001
designcode[6].
Figure2.3.2.Theshearpowercoefficient,,fromtwoNorthSeasites,GreaterGabbardandBeatrice.From[13].
Usingtheexpressionsof(2.3.2)or(2.3.4)onecanreferthe10meterwindspeedtohubheight,whichisthespeed
normally used in connection with design studies. As seen from the figure a good deal of variability in the
extrapolation must be expected due to the variability of the power coefficient. Equation (2.3.2) illustrates some of
thesourcestothisvariability,namelyz
0
andthefunction.
Additionally to the vertical profile of wind speed, the wind direction is logged with each recorded wind speed, and
thedirectionwilloftenbereportedintermsofwindroses,withorwithoutwindspeeddistributionsasisexemplified
Page 13 of 50
inFigure2.3.3.Thewinddirectiondistribution,however,isnotincludedinthesimpleoneparameterclimatedriven
bythewindspeed,whichisthefocusofthepresentsection.
Figure2.3.3.WindRose,winddirectiondistribution,withnumberofoccurrencesontheradialaxis,
fromtheIJmuidendata.From[1].
TheturbulenceintensityisgivenbyTI=o
1
/V(z),whereo
1
isthestandarddeviationofthewindvelocityturbulence.
Indesignstudies
1
isnotsimplyderivedasthemeanvalueofaseriesofcomputedstandarddeviations,,around
themeanwind.Ratheritisthe90%quantileoftheseriesofstandarddeviations[2].Differentfittingexpressionsare
presented in the literature [2,6], where
1
is parameterised by the wind speed at hubheight. From [1] we present
themostrecentformulationoveroceanconditionsinequation(2.3.5)aswellasinFigure2.3.4.
(2.3.5)
15
(15 )
( )
(1 )
aV
TI V I
a V
+
=
+

Herethecoefficient,a,isacoefficient around5andI
15
isareferenceturbulenceintensity at15m/s,heretakenas
0.15or0.14,see[6].NotethattheTIformulationin(2.3.5)doescontainanexplicitheightvariation.
Figure2.3.4.TurbulenceIntensityathubheightaccordingUpWind,IEC(1,2,6).From[1].

Page 14 of 50
For lower wind speeds, o and
1
(the 90% quantile
)
do not decrease as fast as V(z) for decreasing wind and TI
increasesfordecreasingwind.ForlargerwindsTIreachesalimitvalue,ordoesactuallyincreasealittle,duetoan
increasingz
0
withu
*
asgivenbytheCharnockrelation(2.3.3).

AlthoughwehaveherespecifiedboththewindshearandtheturbulenceintensityasanalyticalfunctionsofV,they
will appear with statistical scatter around these expected values, when they are based on direct measurements.
Furthertothenormalturbulencedefinedby(2.3.5)onecanalsospecifyanextremeturbulenceintensityforcertain
loadstudies,seeFigure2.3.4.
Turbulencecanbecharacterisedbythefrequencyspectrum,S(f),ofthehorizontalwindspeed,scalingwiththe
turbulencestandarddeviation,, anditsspatialandtemporalscales.Theliteratureshowsseveralformsthathave
agreatdealofoverlapastheyallrepresentthesameatmosphericphysics,withsimilarcombinationsofdatafitting
andtheory.HerewecitetheKaimalformaspresentedin[3]:
(2.3.6)
2 10
5/3
10
4
( )
(1 6 )
Lf
V
fS f
Lf
V
o =
+
,
wherefisfrequency(Hz),oisthestandarddeviationoftheturbulenceandListhecharacteristicturbulencelength
scale,takenas:
(2.3.7)
5.67 60
340.2 60
z for z m
L
for z m
<
=

>
The turbulence spectrum above describes the frequency distribution as well the wave number spectrum, because
theatmosphericturbulencetoagoodapproximationobeyTaylorshypothesisoffrozenturbulence,meaningthatf=
Vk
V
/2t,wherek
V
isthewavenumberalongthemeanwinddirection.
Figure 2.3.1 illustrates how the overall wind speed distributions are normally well approximated by Weibull
distributions. However, these distributions are less satisfactory for extreme events. Here, it can be shown
mathematically that the most common realistic extreme value distribution functions have high value tails that
convergetowardsaGumbeldistribution,undertheassumptionofstationarity.Fromtheobservedwindrecords,one
cannowgenerateadistributionofmaximumwindspeedsoveragivenbasictimethatmustbelargeenoughforthe
maximum values to be independent of each other. Using the characteristics of the Gumbel distribution, one can
next estimate extreme wind events that will happen in averageover a certain period, denoted return period, e.g.
onceeveryyear,50yearor100year,eventhoughthetimeseriesavailablearenotablyshorterthanthelargerof
thesereturnperiods.AccordingtothecharacteristicsoftheGumbeldistribution,therelationbetweentheexpected
extremewindanditsreturnperiodcanbefoundfrom:

(2.3.8)
0 0
ln( ln(1 / )) ln( / ) ln
T
V T T T T A b T o | o | = ~ + + ,
where and are the most probable value and the standard deviation of the series of maximum values,
respectively.Further,Aandbareconstants,derivedfromafittothemaximumvaluesplottedversustheirestimated
probability . NotethatthemiddleapproximationrequiresthatT>>T
0
.Finally,Aandbcanbeestimatedfromactual
dataseries,asisillustratedinFigure2.3.5.
Page 15 of 50
Figure2.3.5.ExtremewindspeedversusreturnperiodfromUpWindIJmuidendataathubheight.From[1].
From Figure 2.3.5 one can from the red line data fit determine the extreme wind with associated probability or
returnperiod,aspresentedinTable2.3.1[1].Note,thatthereissomefreedomforthelinefitassociatedwitheq.
(2.3.8).Whiletheredlineappeartobebasedonthefulldatarange,acloserfittotheextremevalueobservations
could beobtainedifthefitwasperformed closertothetailof thedistribution.Thiswouldyieldsomewhatsmaller
extremewinds.
Table2.3.1.Summaryofextremewindversusreturnperiodathubheight.IJmuidendata.FromUpWind[1]
Obviously these kinds of extreme value estimations can be used on other measured parameters than the wind
speeds, e.g. the shear or the turbulence intensity, but here one often prefer to work with distributions and
conditional distributions, e.g. given certain winds speeds, as have been done for the extreme turbulence model in
Figure2.3.4.Specificcombinationsofwindandwaveparametersforextremeloadcasesareprovidedinthedesign
code[6].
2.3.2 Waves and other oceanic parameters

Whenthewindblowsoverthewater,surfacewaveswillbegenerated.Forreasonablyhomogeneousandstationary
windfields,thewavescangenerallybedescribedbyafairlynarrowbandfrequencyspectrumthatintegratestothe
variance of the wave field. Like the atmospheric turbulence the wave spectra show characteristic behaviour that
have resulted in several analytical forms based on mixtures of theory and data fitting. The form presented here is
extractedfrom[2].TheformulationstartswiththePiersonMoskowitzspectrum.
Page 16 of 50
(2.3.9)
2 4 5 4
( ) 0.3125 exp( 1.25( ) )
p
PM s p
f
S f H f f
f
= ,
whereH
s
isthesignificantwaveheight,fthefrequencyinHzandf
p
thepeakfrequency(=1/T
p
).Thetwoparameters
mustbedeterminedfromotherequationsorfromfittingtoactualdata.Theydependonthedurationandstrength
oftheactingwind.GenerallyH
s
increasesandf
p
decreaseswithfetchordurationofthewind.
TheJONSWAPspectrumisformulatedasamodificationtothePiersonMoskowitzspectrumandismoreapplicable
to fetch limited situations and growing waves, as is found for most offshore wind turbine sites in the North Sea. It
reads:
(2.3.10) ( ) ( ) ( )
JS PM
S f C S f
o
= ,
where is denoted the peak enhancement parameter, is a function of frequency and C is a normalisation
factor. The forms of the two spectra are illustrated in Figure 2.3.6.
Figure2.3.6.SketchofthewavespectrumaccordingtothePiersonMoskowitzspectrumandaccordingtotheJONSWAP
spectrum.From[9].

For 1 the PiersonMoskowitz spectrum is recovered. Formulas for and C are provided in 2. The value of
is often taken to be 3.3 for storm waves and 1.0 for fatigue calculations. It can also be estimated with
basis in the spectral parameters 6 where H
s
must be inserted in metres and T
p
in seconds:

2.3.11

Even for a constant wind speed of wind direction
0
, all waves do not propagate along the wind direction, but in
an angle interval around
0.
This is normally expressed by a directional wave spectrum [2,15] , for example the
cosine2sspectrum
Page 17 of 50
(2.3.12)
0
( , ) ( ) ( , ), ( , ) cos ( )
s
S f S f D f with D f u u u u u = ~ .
Here D(f,) normalises to 1 by integration over , and s is described by a complex function of f and the
characteristicsofthespectralfunctionin(2.3.11),see[15].
Another reason for waves propagating in different directions from the local wind is that the wind is neither
homogenous nor stationary, and waves generated at other times and places may propagate across the measuring
sitefollowingthedirectionofthewind,whenandwheretheywereformed.Theywilltypicallyappearatthelower
frequencyendofthelocallygeneratedS
JS
(f)spectrum,becauselongwavesdissipateslowerthanshortwaves.They
aredenotedswells,andwillatsomesitesbesignificant.
Asafinalreasonforthewavepropagatingdirectiontobeoffthewinddirection,onemustmentionwavediffraction
thatcanhappenforthelongerwavespropagatingontolowerwaterdepthorduetointeractionwithacurrent.Just
as for the wind one can establish a wave rose, summarising the climatic direction of the wave fields. Such one is
illustratedinFigure2.3.7.
Figure2.3.7.Waverose.From[1]
Asforwinds,one couldalsofromwavedata deriveclimaticdistributionsof thecharacteristicwaveparameters,H

S
and f
p
, but since wind speed is the dominating driver of the wave field, it is more useful to consider the wave
parameters conditional to the wind parameters. Rather than a wave rose, as above, one focus on the directional
difference between the wind and the wave direction, and rather than the climatic distribution of f
p
and H
S
one
focusesontherelationoftheseparameterstothewindspeed.Thelatteristypicallydonethroughascattermatrix
forthedistributionof(H
s
,T
p
)conditionalonthewindspeed.Suchmatricescanbefoundin[1],whileinthepresent
report, we report only averaged values. Additionally, since the directional wave distribution is site specific in the
present report, we shall use only unidirectional waves along the wind direction. However, for a real design one
would need to consider the full scatter diagrams of wind and wave direction. In Table 2.3.2 we illustrate the
combinedmarginaldistributionofwindandwavedirectionforallwindspeeds,takenfromoneoftheUpWindsites
[1].Sincethewavedirectiondistributionasstatedissitespecific,wehavenottriedtomergethedifferentTablesof
[1]fromthedifferentUpWindsites.

Page 18 of 50
Table2.3.2.Distributionofsimultaneouspropagationdirectionsforwindandwaves,takenfromtheK13siteinUpWind.
Takenfrom[1].
2.3.3 Lumped windwave climate for fatigue calculation

Whilethefullwindwaveclimateatasiteisamultidimensionalparameterspacewithamultidimensionalstatistical
distribution function, it can be practically expressed in terms of a oneparameter climate conditional to the wind
speed. In this approach, central values of all other parameters are determined by suitable averaging through the
parameterspace.Theoverallpurposeofsuchasimplerclimatedescriptionistosimplifythefatiguecalculationsfor
the structure. Therefore, the averaging of the climate parameters is done with respect to fatigue contribution and
not simply with respect to probability of occurrence. An example of such a lumped windwave climate is given in
Table2.3.3whichshowsthewaveandwindparametersasfunctionofmeanwindspeedathubheighttogetherwith
thefrequencyofoccurrenceforeachwindspeedbin.Thestatisticsinthetablearelumpedaccordingtothemethod
ofKhn[1,43].Two turbulenceintensityvaluesareshown,namelythoseassociatedwith the normalandextreme
turbulencemodels,see[1].Itshouldbeemphasizedthattheestimatesofturbulenceintensityarederivedfromthe
normalturbulencemodelusedalsoinFigure2.3.4andin(2.3.4),basedonassumptionsaboutthedistributionofTI
foragivenwindspeed.
Thetwovaluesof1.0and3.3arequiteusual,andweshallreturntothisinsection2.4.Finally,wepointoutthatH
S
and T
p
values are weighted values for a given wind speed. Individual values will scatter statistical around these
averages.
It should be mentioned that the fatigueweighting of the wave data cited here was done for bottom fixed wind
turbines. Hence, for a floating wind turbine, the weighting is likely to be different and might thus lead to other
weighted values of H
s
and T
p
. Currently an extensive study is conducted by NREL, University of Massachusetts, and
University of Stuttgart to address this question. An large set of aeroelastic simulations (12 million of 10min
duration)ismadeandwillbeusedtoestablisharecommendationforsimplifyingloadcasesforextremeandfatigue
calculations for floating wind turbines. First results from the study will be presented at OMAE 2013: L. Haid, G.
Stewart, Jason Jonkman, Matthew Lackner, Denis Matha, Amy Robertson SimulationLength Requirements in the
LoadsAnalysisofOffshoreFloatingWindTurbines.

Page 19 of 50
Table2.3.3.Lumpedwindwaveclimateconditionaltowindspeedathubheight(here85meterabovemeansealevel).The
table provides turbulence intensities, significant wave height, period of the peak frequency, and peakedness for the wave
spectrum, probability ofoccurrence ,or correspondingly duration ofoccurrenceper year. Data are from the K13site from
theUpWindproject.Takenfrom[1].

2.3.4 Extreme values for wave climate
Thelumpedstatisticsofdatainthetableaboveisverysuitableforfatigueloadstudies.For theextremeloadsone
must use the extreme values of both winds and wave heights as estimated by the Gumbel statistics in the text
above,correspondingtoa1yearand50yearreturnperiod.Thisisillustratedinfigure2.3.8,alsotakenfromtheK13
siteintheUpWindreport[1].
Figure2.3.8.ExtremeH
S
versusreturnperiodbytheGumbelmethodfortheK13shallowdepthsiteofUpWind.From[1].
Page 20 of 50
FromFigure(2.3.8),wecansummarisetheextremewavesasfunctionofthereturnperiod:
Table2.3.4.ExtremesignificantwaveheightsasderivedfromtheGumbelcorrespondingtotheUpWindK13shallowdepth
site.From[1].
TheH
S
valueistakendirectlyfromtheGumbelfitmethodoftheFigure2.3.8.SimilarremarksasforFigure2.3.5on
therangeof linefittingandtheimplicationfortheextremevaluesofTable2.3.4apply.H
max
isbasedonthatH
S
is
derived from an average over 3 hours and that the waves are Rayleigh distributed, giving rise to a maximum value
being ~ 1.86 times H
S
[2]. The peak period corresponding to the extreme wind speeds must be bounded by the
followingformulacriterion,[2]:
(2.3.12) 11.1 ( ) / 14.3 ( ) /

S S
H U g T H U g s s ,
whereoneshouldselectthepeakperiodthatresultsinthestrongestloads.

Page 21 of 50
2.4 STANDARD OFFSHORE CONDITIONS FOR TESTING AND MODELLING
Inthissectionweextractasetofoffshorestandardconditionsfortestandmodelling.Theextractionismadefrom
dataoftheUpWinddesignbasisandfromtheBeatriceandSdraMidsjbanksites.
2.4.1 Wind climate parameters in the extended North Sea and Baltic Sea
ThewindconditionsovertheextendedNorthSeaandBalticSeaisillustratedinFigure2.4.1and2.4.2takenfromthe
EUNORTHWINDreporting[13,14]onthemarinewindsofNWEurope.Figure2.4.1showsthemesoscalemodelling
result of the average 100 meter wind speed 200611, while the four plots of Figure 2.4.2 depicts the marine wind
climate at 10 meter height measured by radar from 10 year satellite measurements [14]. The figures illustrate the
variabilityofwindparametersintheextendedNorthSeaandBalticSea.
Figure2.4.1.Thevariationofthe100mmeanwindacrosstheNorthWesternEuropeanwaters,asmodelledbymesoscale
modelling,Norsewind.From[13].

Page 22 of 50
Figure2.4.2.The10metermarinewindclimateacrosstheNorthWesternEuropeanseas.Thewhiteareascorrespondtolack
ofdata,mainlyduetoprecipitation.From[14].
The figures indicate that one should expect a fairly gradual change in the wind characteristic of the North Western
Europeanwaters,withexceptionofthequiteclosetocoastalareas.At10mabovethesurface,atypicalWeibullA
parameter is seen to be about 9 m/s, with a kvalue about 2. The A and kvalue seems of the order of the ones
derivedfromthe10mdistributionsinFigure2.4.3andpresentedinTable2.4.1.
2.4.2 Wind and wave climate for five specific sites

Wenowcomparethewindandwaveclimateparametersforfivesites,namelythethreesitesoftheUpWinddesign
basis, the Beatrice site and Sdra Midsjbanken. The UpWind data are extracted from [1], while the Beatrice and
SdraMidsjbankenresultsaretakenfrom[10,11].ItshouldbenotedthatthewavedatafromSdraMidsjbanken
arenotbasedonmeasurementsbuthavebeencalculatedwithclosedformwavegrowthmodels.
AfirstsummaryisprovidedinTable2.4.3whichliststheWeibullwinddistributionparameters,theextreme(1,50)
year wind speed and significant wave heights, the water depth and the lumped wave climate parameters for V
10
=
20m/s.Giventhedifferenceingeographicallocation,thesimilaritybetweenthewindandwaveparametersisquite
remarkable. Further, the 10 meter A and k values of the table compare quite well with the satellite data of Figure
2.4.2.
Page 23 of 50
Site A
10
K H
S
V
10
=20m/s
T
p
,
V
10
=20m/s
V
50
m/s
V
1
m/s
H
S,50
m
H
S,1
M
Depth
m
IJmuiden 7.9 2.1 4.2 8.7 31.5 23.8 7.6 5.7 21
K13shallow 9.3 2.0 3.5 8.0 34.1 25.9 8.2 6.1 25
K13Deep 9,3 2.1 3.5 8.0 34.1 25.9 9.4 7.1 50
Beatrice 8,7 1,9 3.8 6.5 38.5 44
Sdra
Midsjbanken
8,2 2.1 3.3 7.5 35.2 15
Table 2.4.1 Summary of parameters from the 5 sites considered. Note that the wave parameters from Sdra Midsjbanken
aremodelledandnotbasedonmeasurements.OntheK13Deepsite,thedepthisindicatedby50,becauseofthewaythe
K13Deepdatasetismade,seesection2.1.TheH
S
andT
p
datafromSdraMidsjbankenaremarkedwithalso,because
theyarebasedonwavegrowthmodelsnotdataAtBeatriceandSdraMidsjbankennootherextremevaluesthanthe50
yearwindspeedweredetermined.
Next, two tables of lumped windwave climates are provided in Table 2.4.2 and 2.4.3, pertaining to the IJmuiden
shallowwatersiteandtheK13shallowwatersite.ThesimilartablefortheK13deepwatersite(notshownhere)is
identical to the one of the K13 shallow water side except for 1) the probabilities, which are associated with the
differenceintheWeibullparametersforthewindspeeddistributionatthetwosites;and2)theextremewavedata,
whicharenotlistedinthesetables.
Table2.4.2LumpedstatisticsfromtheIJjmuidensite.Thewindspeedreferstothehubheight(83.9m).From[1].

Page 24 of 50
Table2.4.3LumpedstatisticsfromtheK13shallowsite.Thewindspeedreferstothehubheight(85.2m).From[1].
Thesimilarityanddifferencesbetweenthedifferentsitesarenowanalysedintermsofgraphicalcomparison.Figure
2.4.3 shows the wind distributions together with raw data and the fitted Weibull distributions. The Weibull
parameters are those of Table 2.4.1. In general a reasonable match to the Weibull curve is observed. Further, as is
also reflected by the similarity of the Weibull parameters in Table 2.4.1, the wind distributions from the different
sitesarefairlysimilar.
Figure2.4.3The10mwindspeeddistributionatthe5sitesconsidered,comparedtoaeachotherandananalyticalWeibull
function.
Page 25 of 50
The turbulence intensity at hub height is compared in Figure 2.4.4. The turbulence intensity for the three UpWind
siteswerebasedontheformulain(2.3.5)ratherthanonmeasurementsandarethereforeingoodagreement.The
Beatricevaluesaresmaller.Thevalueof(2.3.5)fora=5m/sandI
15
=0.14areshownonthefigureaswell.
Figure2.4.4.Normalturbulenceintensityversuswindspeedatthehubheightfrom4ofthesitesconsidered.Theblackcurve
isthenormalturbulenceformulafrom[1]givenin(2.3.5),whichisthebasisfortheTIestimatesfromtheUpWindsites.The
BeatriceTIisdeterminedfromTlvaluesovertheNorthSeaextrapolatedbyuseof(2.3.3).NoindependentTIwasestimated
atthefifthsite.
ThedependenceofsignificantwaveheightH
s
towindspeediscomparedinFigure2.4.5.Thecurvesforthefoursites
are seen to be very similar. This suggests that a generic relation might exist, although it should be noted that the
valuesarebasedonfatiguelumpedH
s
valueswhichmightthusinducesomedependencetostructuralparameters.
Nevertheless,ananalyticalfunctionwasfittedtothedata.Itwasfoundthatthecurve
(2.4.1)H
S
/H
0
=1+2.6(V/V
0
)
3
/(1+(V/V
0
)
2
)H
0
=1m,V
0
=13m/s
providesareasonablefittothedata.
Page 26 of 50
Figure2.4.5.H
S
versusthe10meterwindspeedforthe4sitesconsideredwithwavemeasurements.TheH
S
valuesforthe
fifthsitewereestimatedfromwavemodelsonlyandarenotshownhere.ThefitisdefinedasH
S
/H
0
=1+2.6
(V/V
0
)
3
/(1+(V/V
0
)
2
),withH
0
=1mandV
0
=13m/s.
ThecorrelationofpeakperiodandsignificantwaveheightisshowninFigure2.4.6.FortheUpWinddata,thetwo
curvesofK13shallowandK13deepareidentical.Further,thecorrelationforIJmuidenisseentobequitesimilarto
thatfromK13.ThecorrelationfromBeatrice,however,isseentohavesubstantiallysmallervaluesofT
p
thanthe
UpWindsites.Thismaybeduetoadifferenceintheappliedmethodforfatiguebasedlumping,asthelumpedT
p
valuesofBeatricewereweightedwith1/T
p
toreflectthenumberofstresscyclesassociatedwithagivenperiod.On
thefigure,thecurves
(2.4.2)T
p
=asqrt(H
s
/g)
witha=11.1anda=14.3havealsobeenincluded.Thisreflectstherequirementof(2.3.12).Itisseenthatthedatais
withinthesebounds,exceptfortheBeatricedatawhichisverycloseandsometimessmaller(T
p
wise)thanthelimit
associatedwitha=11.1.
AllthreecurvesshowapeculiarupwardturnforsmallH
s
values.Thisbehaviourisnotreflectedinthegenericcurves
associatedwith(2.4.2)andmaythuseitherbysitespecificorsimplyaconsequenceofthefatiguelumping.
Page 27 of 50
Figure2.4.6.T
p
versusH
S
,estimatedfromdataatthefoursitesconsidered.FortheUpWindsites(2),thedataaretabulatedin
section2.3.TheBeatricecurveisfrom[10]andisonthelowerbound,whichmayreflectthenearnessofthecoastatthissite.
ThetwoanalyticalcurvesreflectboundsforT
p
in(2.3.12).
TheJONSWAPpeakenhancement parameter,, isdeterminedfrom(2.3.11) anddepictedinfigure2.4.7.Itisseen

that the UpWind sea states show a general increase in for increasing wind speed, towards a maximum value of
approximately2.5.TheBeatricevaluesgrowreadilytoavalueof=5,whichisverylarge.Thismaybeexplainedby
the relatively small T
p
values, which again might be a consequence of the fatiguelumping method rather than the
actualsiteswaveclimate.
Figure2.4.7.Thepeakfactor versus the 10 meter wind speedforthe4sitesconsidered.ThehighvaluesfromBeatrice

reflecttherelativelylowT
p
valuesforthissite,seeFigure2.4.4.
Page 28 of 50
2.4.3 Recommendation
On the basis of the analysis and comparisons of the windwave climates at the five selected sites, the following
approach for the generation of generic windwave climates for model test purposes and generic numerical
computationisrecommended:
1. ThewinddistributionischosenasaWeibulldistributionwithparameterssimilartotheonesinFigure2.4.3/
Table 2.4.1. The wind speeds are extrapolated to hub height by use of the power law profile given in
equation (2.3.4). A power law coefficient of = 0.14 is applied, consistent with IEC 16043 [2]. When
choosing the Weibull parameters, one should consider both the geographic variation, illustrated in this
section,andtheuncertaintyassociatedwiththeirheightextrapolation,discussedinsection2.3.
2. With basis in the chosen wind distribution, a number of wind speed bins with associated probability are
determined.
3. Foreachwindspeed,theassociatednormalturbulenceintensityisdeterminedfromFigure2.4.4.
4. Foreachwindspeed,aH
s
valueisdeterminedfromFigure2.4.5orthefittedformula(2.4.1).
5. The associated T
p
value is estimated from Figure 2.4.6. The data shows quite some scatter but falls within
theboundsdeterminedfrom(2.3.12).
6. The JONSWAP peak enhancement factor is chosen as =1.0 for fatigue studies or =3.3 for ultimate load
cases.Alternatively,onecandeterminefromequation(2.3.11).
7. Extreme values of wind speed and wave height are chosen from Table 2.4.1. A choice consistent with the
chosenWeibulldistributionisrecommended.
Herebyalumpedwindwaveclimateisestablishedalongwithextremevaluesforwindspeedandwaveheight.The
climates are not intended as a replacement of any real data for a specific site. But with no data available the
procedurecanservetoexemplifytherelevantparametersandtheircorrelationwithwindspeed.

Page 29 of 50
3 SCALING OF WIND AND WAVE CONDITIONS FOR PHYSICAL
MODEL TEST
In this section, we review the relevant scaling considerations for offshore wind turbines that are subject to
simulteneous loads from waves and wind. A recent study for a floating wind turbine is reviewed too. Next, the
suggested scaling methodology is derived and a scaling method for structural parameters and wind/wave climate
parameters are devised. The scaling method is applied to a simple numerical model for a floating TLP (Tension Leg
Platform) wind turbine, subjected to an external climate of section 2.4. It is demonstrated that a perfect
reproduction of structural response at model scale can be achieved, provided that the structural loads are
reproduced correctly. This requirement implies a redesign of the blades due to the smaller Reynolds number at
modelscaleandassumesinsensitivityofthehydrodynamicloadcoefficientstothehydrodynamicReynoldsnumber.
Thelimitationsofthescalingprocedurearediscussed.
3.1 REVIEW OF EXISTING SCALING STUDIES AND RELEVANT NONDIMENSIONAL

NUMBERS
Althoughoffshorewindturbinesarepredominantlybottomfixed,mostoftheexistingstudiesofsimultaneouswind
and wave loads in offshore wind energy are concerned with floating wind turbines. This is attributed to the larger
responseofthesupportstructurecomparedtobottomfixedstructures.
Until now, only little material has been published regarding FOWT (Floating Offshore Wind Turbine) scale model
testing.ThescaletestingeffortsrelatedtoprototypesofFOWT(e.g.StatoilsHywindprototypeorEDPsWindFloat)
havenotbeenpublished.ThemostcomprehensivepublishedFOWTscaletestwasperformedintheUnitedStates
by collaboration between the National Renewable Energy Laboratory, NREL, and the University of Maine, UMaine.
Experiences from this test are reviewed in the following. Further, a review of relevant scaling laws and a list of
relevantpublicationsaddressingthisissueareprovided.
3.1.1 List of Symbols for the review

c
s
=speedofsound
C =wavecelerity
Fr =Froudenumber
g =accelerationofgravity
l =characteristiclength
KC=KeuleganCarpenternumber
Ma =Machnumber
=rotorangularvelocity[radians/sec]
=fluiddensity
R =bladeradius
Re =Reynoldsnumber
St =Strouhalnumber
TSR =tipspeedratio
=fluiddynamicviscosity
U =freestreamfluidvelocity

Page 30 of 50
3.1.2 Important NonDimensional Numbers
Nondimensional numbers are dimensionless fractions or products resulting from relations between parameters
defined by the fundamental quantities L, M, and T (length, mass, and time, respectively). The two most important
nondimensionalnumberstoconservewhenmodelingascaledrotorsystemaretheLockNumberandtheReynolds
Number [27,28]. These reflect the most important parameters involved with aerodynamic and elastic forces. The
Froude Number is the most important non dimensional number to preserve when modeling floating structures for
hydrodynamicsimilarity[29,30].

Geometric Scaling Factor ()
Thegeometricscalingfactordescribestheratioofthephysicallengthbetweenthefullscaleturbineandthescaled
test model. The value of is constrained by the size of the testing facility, feasibility of model construction, scaling
lawsimilituderequirements,andtheprojectsbudget.

Froude Number (Fr)
The Froude number is a dimensionless number that defines the ratio of inertial forces to gravitational forces in a
fluid. When a free surface wave propagates within a fluid, the Froude number can be defined in the following way
[29]:
3.1.1 Fr
wave
C/gl
In windwave basin model tests, the Froude number is the most frequently conserved dimensionless number to
ensurehydrodynamicsimilitude.WhileviscouseffectsdoaffectFOWTplatformhydrodynamics,theyareconsidered
small in comparison to inertial effects at both test and model scales. Viscous effects are mostly manifested in the
thinboundarylayeraroundfloatingbodies[31].

Reynolds Number (Re)
TheReynoldsnumberisdefinedastheratioofinertialforcestoviscousforcesinaflow.Foracylinderexposedtoa
fluidwithvelocityu,theReynoldsnumberis
3.1.2 ReDU/
TheReynoldsnumberisimportantinfluidsystemsbecausequalitatively,flowoverabodyactssimilarlyforidentical
Reynoldsnumbers.Herebye.g.liftanddragcoefficientscanbeparamterizedintermsoftheReynoldsnumber[33].
Because of the small length scales at which models are tested, there is almost always Reynolds number mismatch
between model and full scales that cannot be resolved unless there is a change in the fluid of the system (i.e.
viscosityordensity)[27,32].Althoughthishasbeendonesuccessfullyinthehelicopterindustry,webelieveitisout
of the scope of existing windwave basin test facilities. This mismatch is often combated by using airfoils that have
coefficients of lift and drag with minimal dependence on Reynolds number. This is especially important in the low
Reynoldsnumberregimesencounteredinmodeltesting[27].
KeuliganCarpenter number (KC)

TheKeuliganCarpenternumberisacharacteristicnumberforplanaroscillatoryflowsdefinedas
(3.1.3)KC=U
m
T
w
/D,
where U
m
is the maximum velocity in the outer flow, T
w
is the period of the flow and D is the cylinder diameter. It
providesameasureforthelengthofahorizontalparticlepathintheouterflowrelativetothecylinderdiameter.For
Page 31 of 50
small KC numbers, there will be no flow separation around the cylinder, where for large KC numbers, vortex
sheddingwilloccurduringtheflowcycle.Asaruleofthumb,thebalancebetweeninertiaandviscousforcesisequal
to20/KC[15].SmallKCnumbersthusimpliesdominantinertialloading,whilelargeKCnumbersareassociatedwith
dominantviscousloading.
Tip Speed Ratio (TSR)

Thetipspeedratioistheratiooftherotorsvelocitytothewindsfreestreamvelocity:
(3.1.4) TSRR/U
MaintainingaconstantTSRisoneofthemostbasicproceduresusedtopreservesimilarityinscalingawindturbine
system[34].IftheTSRandthegeometryofthebladesarescaledsimilarly,theflowgeometryoverthebladeswillbe
preserved.ThisstatementisonlyanapproximationbecausetheReynoldsnumberwillvaryasthescaleoftheblade
changes,thereforecausingflowdissimilarity.However,TSRconsistencyshouldstillberealizedinmodeltesting,and
iscompatiblewithFroudesimilitudewhenusedtoconstrainrotationalfrequenciesandwindspeed[29].
Mach Number (Ma)

The Mach number is the ratio of the flow velocity to the speed of sound, and is an influential parameter in gas
dynamics[35]
3.1.5 MaU/c
s

The conservation of the Mach number is often employed in aerodynamic model scaling because it characterizes
compressibility effects [30]. At Mach numbers below 0.3, flow can be considered incompressible [35]. Figure 3.1.1
depicts the Mach number for a full size and model size (1:45) NREL 5 MW reference rotor. It is seen that
compressibilityeffectscanbeignoredatbothscalesbecausetheMachnumberremainsbelow0.3.Thismeansthat
conservingtheMachnumbersexactvaluebetweenscalesisnotnecessaryforaerodynamicsimilitude[29].

Figure3.1.1.MachnumbercalculationforafullscaleNREL5MWreferenceturbineandits1/45
th
scalemodel(calculated
usingBEMtheory)[29].TheMachnumbersinbothsystemsarebelow0.3,whichvalidatestheassumptionofincompressible
flowatfullandmodelscales.Thisdatawasgeneratedinasimulationwherethewindspeedwas11.4m/sandtherotorTSR
was6.958.
Page 32 of 50

Strouhal Number (St)
TheStrouhalnumberisthedimensionlessvortexsheddingfrequencyforabodyemergedinaviscousfluid
3.1.6 Stf D / V
IfFroudescalingisemployed,theStrouhalnumbercannotalwaysbemaintainedbecauseofitsdependenceonthe
Reynolds number. However, the Strouhal number stays at a value of 0.2 for a range of Reynolds number and its
effects have been neglected in past experiments [29]. The Strouhal number is typically conserved to ensure
similitude in vortex induced vibration (VIV) driven effects, which may not be important for FOWT (specifically TLP)
modeltesting.
3.1.3 Dominant scaling methodologies for FOWT

Geometric Scaling
The simplest method of creating a scale model is to scale the full turbines geometry by a factor of , thereby
creatingaminiatureturbinethatpreservesallrelativelengthdimensions.However,theperformanceofturbinesat
different length scales depends on more than their geometry and crucial dissimilarities will exist if this is the only
scalingmethodchosen.Forinstance,everynondimensionalnumber(except)intheprevioussectiondependson
environmental parameters independent of the models geometry, meaning that the dynamic and kinematic effects
describedbysuchnondimensionalnumberswilldifferbetweenthemodelandfullscaleturbinesunlesseffortsare
madetoequatethevaluesofthesenumbersatallscales.

Dynamic and Elastic Scaling
In order for model size turbines to act dynamically similar to full size turbines, one must ensure that the models
aerodynamicsandhydrodynamicsmatchthoseofthefullscalesystem[31].Additionally,dynamicsimilarityrequires
thatnaturalfrequenciesandgyroscopicmomentsareproperlyscaledfromthefulltomodelscale[36].Ithasbeen
well established by the oil and gas industry that Froude number equality between the model scale and full scale is
most effective in achieving hydrodynamic similarity. Froude scaling has been succesfully applied in several model
studiesoffloatingwindturbinesincludingthoseofNRELandUMAINE.
FroudescalinghasfurtherbeenvalidatedbyNRELFASTsimulations[36].However,thesetestsassumethatReynolds
number dependent aerodynamic parameters (airfoil Cl, Cd, and viscous damping) were kept constant between
modelandfullscales,anassumptionthatisdifficulttorealizeinphysicalmodeling.InadditiontoFroudescaling,the
TSR should be held constant at different scales in order to yield consistency in FOWT kinematics and flow
geometries.Thiswillbefurtherdetailedinthisreport.
The scaling methodology proposed in this report is the dynamic and elastic scaling, based on preservation of the
Froudenumberandtipspeedratio.Thiswillbedetailedinsection3.2.
3.1.4 UMAINE scaled FOWT testing review

TheUniversityofMainedesignedtheirtowertomimicthedynamicbehavioroftheOC3Hywindtowerbyscalingthe
lowest frequency modes and center of gravity according to Froude similarity [29]. The mass of the tower failed to
directly follow Froude scaling, because a massscaled tower construction was impractical and difficult in initial
Page 33 of 50
designs.UMainejustifiedandcompensatedforthisdissimilaritybyscalingthecombinedmassofthetower,nacelle,
androtor.Thenacellewasoverweight,andthelightweighttowerwouldhelpcompensate.Thetowerhadaforeaft
fundamentalbendingfrequencythatwasonly5.4%lowerthandesiredandthecenterofgravityforthemodelwas
3.3%higherthandesired[29].Themodeltowerdiameterwas3.7timessmallerthanaFroudescaleddiameter,and
UMaine considered thistobeadvantageousbecauseitdecreasedaerodynamicinteraction betweenthetowerand
theturbine,somethingtheavailableNRELsoftwarecouldnotsimulate[37,29].
UniversityofMainesgoalwhendesigningbladeswastokeepthemlightweightandtohavetheairfoilcrosssections
toremainas consistentaspossiblewiththefullscalemodel.Prioritywasgiventogatheringinformationabout the
global responses of the FOWT system, not the blade deformation, rotor dynamics, and higher order aeroelastic
effects[29].
Geometric redesign was also avoided. Blades were constructed out of carbon fiber to eliminate as much blade
flexibility as possible and reduce the number of variables in testing. This made blade tip deflections negligible and
limitedthechancesofbladefailureduringexperiments.Thelowweightofcarbonfiberwasadvantageousbecause
ofthesmallblademassrequiredforFroudeconsistencyinrotormassandgyroscopicmoments.
A mold and bladder process was used to fabricate 15 blades with a mass of 0.130 kg. This was .010 kg lower than
Froudescaledmass,whichbenefitedUMainebymakingupfortheoverweightnacelleandtestingwires[29].
The blade deflections were only 2% of theoretical maximum blade deflections, which satisfied UMaine because of
the blades nearrigid nature. The rotor was unable to produce Froude scaled thrust when the testing wind speed
wasconstrainedbythewindspeedtowavecelerityratio(),sotestswereruninfasterwinds.Doingsoeliminated
the possibility of testing at TSRs higher than five, because the rotor had not been designed for such high angular
velocities[36].
Besidesthescalingissuesdescribedabove,UMAINEusedacabletocollectthedatafromtheFOWT,thatmighthave
influenced the floater motions during the test. Therefore it is recommended, that a lightweight cable or a wireless
dataacquisitionsystemisusedinfuturetests.
3.2 RECOMMENDED SCALING METHOD
The proposed scaling method is derived in this section. It is consistent with the one desribed by Martin [29] apart
from the inclusion of a different water density at prototype scale and model scale in the present scaling. Such a
differencewilloftenoccurduetotheuseoffreshwaterinmodeltests,opposedtotheseawateratprototypescale.
Formodeltestswithcombinedwindandwaveforcing,themaininterestistheglobalmotionofthestructuresubject
to the aero and hydrodynamic loads plus the loads from mooring and gravity. The ratio of these loads must
therefore be preserved between prototype scale and model scale. Further, as the loads are dynamic in time, the
frequencies of the loads, the structural frequencies and the rotor frequency must scale consistently. These
requirementsdefinethescalingofallrelevantquantitiesandarederivedinthefollowing.Thescalingcanbeapplied
toe.g.modeltestsoffloatingwindturbines,asillustratedinFigure3.2.1.TheproposedscalingpreservestheFroude
numberandKCnumberforthehydrodynamicmotionandthetipspeedratiofortherotor.TheReynoldsnumberfor
waterandair,however,arenotpreserved.Thisleadstotherequirementofaredesignofthebladestoachievethe
correctthrustforce.Further,theinvarianceofthehydrodynamicforcecoefficients(intertiaanddragcoefficients)to
the hydrodynamic Reynolds number should be checked and significant differences be compensated, see e.g. [44].
Also, but of less practical impact, the Weber number for water (surface tension) and Mach number for air
(compressibility)arenotpreserved.

Page 34 of 50
Itshouldalsonotedthatwhiletheproposedscalingleadstoaconsistentthrustforcefromtherotor,theredesigned
blades at the model scale Reynolds number might not reproduce a consistent rotor torque. This will lead to an
inconsistent power production and an inconsistent dynamic generator moment. The latter moment contributes to
therollforcingoftheplatform.Hence,toavoidthisimperfection,areproductionoftheaerodynamictorqueofthe
redesigned blades should be pursued. Most important, as already mentioned, however, is a correct thrust
reproduction.

3.2.1 Scaling of geometric length and mass properties
Thegeometriclengthscaleratioisdenedby
(3.2.1)
Foracorrectdynamicscalingtheratioofstructuralmassanddisplacedwatermassmustbepreserved:
(3.2.2)
Asthevolumescaleslike
3
thisyields
(3.2.3)
Figure3.2.1:AfloatingwindturbineinawaveflumewithopenjetwindtunnelatDTU(Hansen&Laugesen[38]).Thesketch
totherightshowstheforcesinvolved.Thecircledquantitiesaredimensionlessnumbersthatcharacterisestheenvironment
andthemotion.

Page 35 of 50
ThisdenesthescalingofthemassmomentofinertiaJ
(3.2.4)
andthescalingoftheareamomentofinertiaforastructuralcrosssectionI
(3.2.5)
3.2.2 Froude scaling of hydrodynamic forces

Acorrectdynamicscalingrequiresthattheratioofinertialforcetogravitationalforceispreserved
(3.2.6)
Further,thetravelleddistanceofawaterparticleoveragiventimemustscalewiththegeometriclength:
(3.2.7)
which denes the global time scale. This constitutes the classical Froude scaling of the hydrodynamic forces and
motion.
3.2.3 Aerodynamic scaling

For dynamic similarity, the ratio of the aerodynamic thrust force to the gravity force on the structure must be
preserved
(3.2.8)
Itisdesirablethattherelativevelocitybetweenthestructureandairispreserved.Thereforetheairvelocitiesmust
scalelike
1/2
.ItisthereforenecessarytoscaletheC
T
valueswiththedensityratiotoobtain.
(3.2.8)
Hereby,however,theproposedscalingwillnotpreservetheReynoldsnumberintheairorwater.Forthisreasona
redesignofthebladeswillbeneededtoobtainthedesiredvalueofC
Tm
.
Thechosenscaling,however,preservesthetipspeedratio(TSR)

Page 36 of 50
(3.2.9)
astherotorfrequencyscalesinverselywithtime
p
/
m
=
1/2
.
3.2.4 Scaling of structural stiffness
The structural stiffness must be scaled such that the natural frequencies scale consistently with time. Further the
structural deection must scale directly with the length scale. The appropriate scaling can be derived from the
dynamicbeamequation
(3.2.10)
Here
s
A is the structural mass per length, x transverse deection, z the spatial coordinate along the beam axis, E
Youngsmodulus,Iareamomentofinertiaandpthetransverseloadperunitlength.Itisobservedthattherstterm
on the left hand side and the transverse load term scale like (
wp
/
wm
)
2
while the middle term scales like E. A
consistentscalingoftheequationofmotionisthereforeobtainedfor
(3.2.11)
3.2.5 Check of natural frequency and gyroscopic force

The above derivation defines the necessary scaling. A check on the natural frequency and gyroscopic force is
provided.ThenaturalfrequencyofacantileverbeamoflengthLisgivenby
(3.2.12)
Applicationoftheabovescalingyields
(3.2.12)
whichisconsistentwiththetimescalingt~
1/2
.
Thegyroscopicmomentfromthechangeoftheorientationoftherotoris
whereisistherotationalvectorandJisthemassmomentofinertia.Thegyroscopicmomentsscalesaccordingto
(3.2.13)
whichisconsistentwiththescalingofforcetimeslength.
Page 37 of 50
3.2.6 Properties that may not scale consistently

Ithasalreadybeenmentionedthatcertainpropertieswillnotscaleconsistentlywiththeproposedscaling.Theseare
1. theaeroandhydrodynamicReynoldsnumbers.
2. thehydrodynamicWebernumber
3. theaerodynamicStrouhalnumber
4. theaerodynamicMachnumber
5. theaerodynamictorque
6. theaerodynamicpower
7. thegeneratortorqueanditscontributiontorollforcing
While(1)iscompensatedbyaredesignofthebladesandacheckofthehydrodynamicdragandinertiacoefficients,
(26)areconsideredofsmallsignificancefortheglobalplanarmotionofthewindturbine.Thescalingthusensuresa
consistentinplanemotionofthefloatingwindturbine.Fortheoutofplanemotion(sway,roll,yaw)thegyroscopic
moments will scale correctly, while the rollforcing from a dynamic generator torque (7) will not necessarily scale
correctly.Acorrectionofthiseffectisopenforfutureresearch.
3.2.7 Summary of the scaling

TheproposedscalingissummarisedinTable3.2.1
Property Scalingfactor
Length
Mass (
wp
/
wm
)
3
massmomentofinertia(J) (
wp
/
wm
)
5
areamomentofinertia(I)
4
Velocity
1/2
acceleration 1
Time
1/2
frequency
1/2
Angle 1
Force (
wp
/
wm
)
3
Moment (
wp
/
wm
)
4
stiffness(E) (
wp
/
wm
)
Stress (
wp
/
wm
)
Power (
wp
/
wm
)
7/2
Thrustcoefficient(C
T
) (
wp
/
wm
)
Table3.2.1Summaryofphysicalquantitiesandscalingfactors.

Page 38 of 50
3.3 SCALING OF WIND AND WAVE CLIMATE PARAMETERS
Thedynamicandelasticscalingcanbedirectlyappliedtothewindandwaveclimatestoscalefromprototypescale
tomodelscale,asthescalingsforlengthandvelocityhasbeendefined.Thisyieldsthefollowingscaling:
Property Scalingfactor
geometricheight(z)
windspeed(V)
1/2
turbulentwindspectrumS
w
3/2
turbulentwindfrequency(f)
1/2
turbulenceintensity 1
windprofilepowercoefficient()
1
waterdepth
velocity
1/2
significantwaveheight
peakperiod
1/2
windwavemisalignment 1
Table3.3.1.Scalerelationsforwindandwaveclimate.

3.4 EXAMPLE: SCALED EXPERIMENT OF A FLOATING WIND TURBINE
Inthefollowingthescalingprocedureisillustratedwithanumericalexample.AsimplemodelforaTLPfloatingwind
turbine is derived. The wind turbine is subjected to stochastic wind and wave loads at prototype scale. Next, the
wind climate, wave climate and wind turbine are scaled to scale 1:50 and the smallscale response is computed in
thenumericalmodel.Theprototypescaleandsmallscaleresponsesmomentarecompared.Itisdemonstratedthat
a perfect scaling can be achieved if the aerodynamic thrust and hydrodynamic loads are reproduced correctly at
modelscale.Imperfectionsduetoincompleteaerodynamicscalingarediscussed.
3.4.1 The TLP wind turbine

The wind turbine is the NREL 5 MW reference turbine (Jonkman et al [39]) placed on the NRELMIT TLP platform
(Matha [40]). A sketch of the structure is shown in Figure 3.4.1. The purpose of the example is to illustrate the
scaling in a simplified setting that includes hydrodynamic forces from a wave climate, aerodynamic forces from a
wind climate, gravitational forces and structural elasticity. Only two degrees of freedom are retained and several
simplifyingassumptionsaremade.Thesearelistedbelow
- onlytwodegreesoffreedomareconsidered:platformsurgeandtowerflexibility
- thetowerisconsideredmasslessandwithuniformstiffness.Theoriginaltowermassisdistrubutedontothe
topmassandthefloatermass.
- smalldeflectionsfortheplatformisassumed
- thetethersareassumedmasslessandstiff
- therotormass,nacellemassareconsideredasapointmass,placedontopofthetower.Halfoftheoriginal
towermassisincluded.
- thefloaterisconsideredslender(D/L<0.2)
- thewavesareconsideredsmalltoallowapplicationoflinearwavetheory
- spatiallycoherentturbulenceisassumed
Page 39 of 50
Figure3.4.1.ModelsketchforsimplemodelofaTLPwindturbine.
ThestructuralparametersarelistedinTable3.4.1.
Property Symbol Valueprototypescale Scaling Value, model scale

(1:50)
Floatermass m
1
8.774x10
6
kg (
wp
/
wm
)
3
68.48kg
Topmass m
2
518.5x10
3
kg (
wp
/
wm
)
3
4.047kg
Towerstiffness EI 300x10
9
Pa (
wp
/
wm
)
5
936.6Pa
Hubheight H
H
90m 1.80m
Floaterdiameter D
f
18m 0.36m
Floaterdraft d
f
47.89m 0.958m
Waterdepth h 200m 4.00m
Tetherlength L
T
152.11m 3.04m
Rotordiameter D
R
126m
2
2.52m
Densityofair
a
1.29kg/m
3
1 1.29kg/m
3
Densityofwater
w
1025kg/m
3
wp
/
wm
1000kg/m
3
Table3.4.1:Parametersforfloatingwindturbineatfullscaleandmodelscale.
Page 40 of 50
3.4.2 Equations of motion
TheequationsofmotionconsistofNewtonssecondlawforthefloaterandthetopmass.Thedisplacementofthe
masses relatively to an earthfixed reference point is denoted x
1
and x
2
, respectively, as shown in Figure 3.4.1. The
restoringforcefromthemooringsystemcanbederivedbycalculationoftheexcessbuyancyforceF
BE
(3.4.1)
which constitutes the vertical component of the total tether force T. Next, the force triangle for the tether force,
excessbuyancyforceandhorizontaltetherforceF
x
gives
(3.4.2)
where is the tether angle with vertical, measured positive in the counterclockwise direction. Under the
assumptionofsmalldeflectionangles,thisisapproximatedbyx
1
/L
T
whereby
(3.4.3)
The elastic bending of the tower constitutes the connection force between the tower and floater, tower and top
mass, respectively. The tower is considered massless and no dynamic loads are thus associated with the tower.
From standard EulerBernoulli beam theory, the force needed to displace the end point of a cantilever beam by a
distance x is F=3 EI x/L
3
, where EI is the stiffness and L is the length. The equations of motion can thereby be
expressedas
(3.4.4a)
(3.4.4b)
whereF
H
isthehydrodynamicexcitationforce,A
11
isthehydrodynamicaddedmassforthefloater(derivedinnext
paragraph)andF
W
istheaerodynamicthrustforcefromthewind.
3.4.3 Hydrodynamic force

Thehydrodynamicloadingisderivedfromsimplelinearwavetheoryundertheassumptionofsmallwavesteepness.
ThefreesurfaceelevationisexpressedasaFourierseries
(3.4.5)
wheretheamplitudesa
p
arerelatedtothewavespectrumby
(3.4.6)
andtheradianfrequency
p
andwavenumberk
p
arerelatedthroughthedispersionrelation
Page 41 of 50
(3.4.7)
andfurther
p
isasetofrandomphases,uniformlydistributedontheinterval[0;2].
Thehorizontalfluidvelocitiesaregivenby
(3.4.8)
and allows computation of the hydrodynamic force on a horizontal crosssection of height dz through the Morison
equation[45]undertheassumptionofsmallfloaterdiametertowavelength:
(3.4.9)
HereC
m
isthehydrodynamicaddedmasscoefficient,C
D
isthedragcoeffientandA
f
isthecrosssectionalareaofthe
floater.Stillundertheapproximationoflinearwavetheory,thehydrodynamicforcingisintegratedfromthebottom
ofthefloatertothestillwaterlevel.Further,thecontributionfromx
1,tt
ismovedtothelefthandsidewhereitforms
theA
11
termofthemassmatrix:
(3.4.10)
(3.4.11)
ThehydrodynamiccoefficientsarechosenbasedontheReynoldsnumberandKCnumbertobeC
D
=0.7andC
m
=0.8.
3.4.4 Aerodynamic force
Thewindforceisbasedonatimeseriesofhorizontalwindvelocityathubheight
(3.4.12)
whereV
hub
isthemeanwindspeedatthehubandtheFourieramplitudesarebasedonthewindspectrumS
f
(3.4.13)
andfurther
m
isasetofrandomphases,uniformlydistributedontheinterval[0;2].Theinstantaneousthrustcan
thenbeexpressedby:
(3.4.14)
whereC
T
isthethrustcoefficient.Undertheapproximationofsteadywindspeed,C
T
canbeobtainedfromasteady
BEM model as function of wind speed, see e.g. Hansen [41]. The curve applied in the present example is shown in
Page 42 of 50
Figure3.4.2andresemblestheonefortheNREL5MWreferencerotor.Inthesamefigure,alsothethrustforceF
wind
is plotted. It is seen how the thrust force increases up to the rated speed of 11.4 m/s, where the rated power of
5MW is obtained. Beyond the rated wind speed, the thrust force decreases due to the active blade pitch system,
whichpitchesthebladestoreducetheaerodynamictorqueinordertomaintaintheratedpower.
Figure3.4.2.ThrustcoefficientandthrustasfunctionofinflowspeedV.
The pitching of the blades is controlled by an active control system. It has been shown by Larsen and Hanson [42]
that the standard onshore control algorithm can lead to an instability for the platform pitch motion. The control
system therefore needs to be modified. In the present simplistic model, a simple Pcontrol is used to model the
reaction time of the bladepitching, based on the instantenous difference between the current C
T
value and the
targetC
T
valuefortheinstantenousrelativewindspeedVx
2,t
.
3.4.5 Wind and wave climate
Basedonthegenericwaveandwindclimatesofsection2.4,asetofprototypewindand waveconditionsarenow
chosen.Next,theclimateparametersarescaledtomodelscale.
TheWeibulldistributionsforwindheightatz=10mofFigure2.4.1canbetransformedtothehubheightbyascaling
of the A parameter by the assumed powerlaw profile. It is further assumed that the shape parameter, k, does not
changewithheight.Theapproximationassociatedwiththisisdiscussedinsection2.
(3.4.15)
TheadjustedWeibullparametersprovideacompleteprobabilitydistributionforthemeanwindspeedathubheight.
Forthepresentexample,wechooseasingleclimateforasinglerealizationwiththenumericalmodel.Wechoosea
hub wind speed of V
hub
=18m/s. Next, the turbulence intensity is found by the formula (2.3.5) which with a=5 and
I
15
=0.14yields

Page 43 of 50
(3.4.16)
The standard deviation to be used in combination with the Kaimal spectrum is then =TI V = 2.45 m/s. Next the
significantwaveheightischosenwithbasisinFigure2.4.3.WechoosethefittedcurvefortherelationbetweenV
hub
andH
s
(2.4.1)andobtain
(3.4.18)
ThepeakperiodischosenbasedonFigure2.4.4whichshowsquitesomescatter.Ratherarbitrarilyavalueofa=12
forthecoefficientinformula(2.4.2)ischosentoyield
(3.4.19)
Finally,fortheJonswapspectrum, needstobedetermined.Hereformula (2.3.11)suggests=3.83whichseems

unrealisticasitexceedsthestandardizedvalueof3.3.Forthisreason=3.3ischosen.Itshouldbenotedherethat
forafloatingwindturbine,thedepthwillusuallybelargesuch thatthePiersonMoscowitzspectrumislikelytobe
morerealistic.Thisisrecoveredwithreason=1.0.However,forthepresentexample=3.3isapplied.

Oncethewindandwaveclimateparametersaredetermined,theycanbescaledtomodelscaleaccordingtoTable
3.4.1.TheresultislistedinTable3.4.2alongwiththeprototypevaluesandthescalerelation.
Property Symbol Value prototype

scale
Scaling Value, model
scale(1:50)
hubheight z
hub
90m 1.8m
windspeed10mheight V
10
13.2m/s
1/2
1.87m/s
hubwindspeed V 18m/s
1/2
2.55m/s
powerlawcoefficient

0.14 1 0.14
turbulenceintensity TI 0.136 1 0.136
standarddeviationforturbulence

2.45m/s
1/2
0.347m/s
lengthscaleforKaimalspectrum L 340.2m 6.80m
WindvelocityFourieramplitude b Fromeq(3.4.13)
1/2
Fromeq(3.4.13)
significantwaveheight H
s
3.37m 0.0674m
peakperiod T
p
7.03s
1/2
0.994s
Jonswappeakenhancementparameter 3.3 1 3.3
Table3.4.2.Windandwaveclimate.Prototypescaleandmodelscale.

Page 44 of 50
3.4.6 Results for perfect aerodynamic scaling
Results for the realization at prototype scale are shown in Figure 3.4.3. The left column shows time series of wind
speedathubheight,freesurfaceelevationforthewaves,floaterdisplacementandnacelledisplacement.Theright
column shows the corresponding power spectra. It can be seen that the wind signal has spectral energy in the full
frequencyintervalshown,whilethewavesignaldoesnothaveanyenergyforfrequenciesbelow0.1Hz.Further,the
floatersurgesignalisdominatedbytheresponsetothewindatthefirstnaturalfrequencyof0.017Hz(60s)whichis
associatedwiththerestoringforceofthetethers.Responsetothewaveforcingisevidentbothfromthetimeseries
and the power spectrum. For the nacelle displacement, a similar response can be seen, although with a larger
contentoftowervibration.Thisisduetothewindsforcingofthetowernaturalfrequencyat0.25Hz(4s).
Figure3.4.3.Responseatprototypescaleforsimultaneouswindandwaveloading.
Next, the numerical model was rerun at model scale. All the model parameters and external climate parameters
werescaledaccordingtoTables3.2.1and3.3.1.ThescaledparametervaluesarelistedinTables3.4.1and3.4.2.The
numericaltimeseriesareshowninFigure3.4.4,plottedontopofthefullscalevalues.Priortothecomparison,the
Page 45 of 50
modelscaleresultswerescaledupagaintoprototypescale.Themodelscaleresultsareshownasblackdotsontop
oftheprototypescaleresults(redline).Aperfectmatchisseen.Inhindsight,thisisnotasurprise,sincethescaling
isconsistentwiththegoverningequations.Theexamplethusillustratesthatitispossibletoscaledownthewindand
waveclimatesandscalebackthemodelresponsetofullscale.Thegoodmatch,however,reliesonthemodelsability
to reproduce the aerodynamic (and hydrodynamic) loads at model scale. For both of them, the Reynolds number
dependence is likely to induce scale effects. As already stated, a redesign of the blades will be thus necessary to
maintainthethrustcurveatthereducedReynoldsnumber.
Figure3.4.4.Comparisonofprototypescaleresults(redline)andmodelscaleresults(blackdots).Themodelscaleresults
havebeenscaleduptoprototypescalebeforeplotting.
3.4.7 Consequences of imperfect aerodynamic scaling
To illustrate the consequences of an imperfect reproduction of the aerodynamic loads at model scale, a
complementarycomputationwascarriedout,withathrustcurvethatdeferredfromthetargetcurve.Themodified
thrustcurveisshowninFigure3.4.5inprototypescale.Further,theresponseatprototypescale(withcorrectthrust
curve) and model scale (with imperfect thrustcurve) are compared in the figure. Again, for this comparison, the
modelscaleresultswerescaledbacktofullscale.Itisseenhowthereductionintheaerodynamicthrustaffectsthe
floaterandnacelleresponses.
Page 46 of 50

Figure3.4.5.Consequencesofimperfectaerodynamicscaling:Left:Imperfectaerodynamicthrustcurveandtargetthrust
curve.Right:Comparisonofprototypescaleresponseandmodelscaleresponse.Themodelscaleresponsewasproduced
withtheimperfectthrustcurveandscaledbacktoprototypescale.
3.5 DISCUSSION
Throughtheexampleithasbeendemonstratedthatthedevisedscalingmethodallowsforaperfectreproductionof
model response at model scale, provided that the structural loads can be reproduced correctly. This implies a re
design of the blades to the lowered Reynolds number at model scale. The redesigned blades must be able to
reproduce the thrust curve of the full scale blades. While the above example only covers the plane motion of the
wind turbine, a real wind turbine will also have response in the outofplane direction. These can be induced by
transverse waves, gyroscopic effects, the dynamic change of rotor moment and through interaction with the
mooring system. A true reproduction of these features requires a correct reproduction of the rotor torque as well.
Thismaybedifficulttoachievesimultaneouslywiththefullfilmentofcorrecttipspeedratioandandthrustcurve.In
suchcases,whereaperfectreproductionofthefullscalebehaviouratlabscaleisnotpossible,thephysicalmodelis
ofcoursestillvalidandtrueinitsownright.Itcanthusbeusedtovalidateanumericalresponsemodel,whichcan
laterbeappliedatfullscale.

Page 47 of 50
4 CONCLUSIONS AND RECOMMENDATIONS
ThepresentreportconstitutestheProtocolManualforensuringharmonisationofoffshorewindandwave
simulationbeingimplementedatMaRINETfacilities.Windandwaveclimatesforfiveoffshorewindsitesinthe
NorthSeaandtheBalticSeahavebeenpresentedintermsofprobabilitydistributionsforwindspeedalongwitha
seriesoflumpedseastatesandturbulenceintensityvalues,parameterisedwithrespecttothewindspeed.Further,
extremevaluesforwindspeedandsignificantwaveheighthavebeenprovided.
Furthertothewinddistributionsandlumpedcharacteristics,theWeibullparametersforthewinddistributionand
explicitformulasfortheturbulenceintensityandsignificantwaveheightareprovided.Forthecorrelationofwave
peakperiodandsignificantwaveheight,astandardformulafromtheIEC614003codehavebeenfoundtocover
thescatterinthedata,althoughonecoefficientinthisformulamustbedecideduponbytheuser.Further,thevalue
of,theJONSWAPpeakenhancementparametermustbechosenbytheuser.Thiscanbedoneeitherfroman
explicitformulaorbythestandardchoicesof=1.0or=3.3.Herebyafulldescriptionofaunidirectionalwindwave
climatecanbeconstructed.Ifneeded,thisclimatecanbesupplementedbytheuserwiththecombineddirectional
distributionofwindandwaves,eitherbasedondataorintermsofparametricstudies.
Thescalingmethodproposedisthedynamicelasticscaling,whichmaintainstheratiosbetweenhydrodynamic,
aerodynamic,stiffnessinducedandgravitationalforces.ThisscalingpreservestheFroudenumberforthewater
phaseandthetipspeedratiofortherotor.TheReynoldsnumbersforairandwater,however,arenotconserved.A
redesignofthemodelscalebladeswillthereforebeneeded.Herethescaledthrustcurvemustbematched.Further,
ifpossible,thetorquefromtheairfoilshouldbematched.Thisrequirement,however,isdifficulttoachievedueto
thechangeinlift/dragratioatlowReynoldsnumber.Itisthereforeforeseen,thattheaerodynamictorqueandthus
producedpowerwillnotbescaledcorrectly.Asaconsequence,rollforcinginducedbythedynamicchangein
generatormomentwillnotscalecorrectly.However,thecorrectscalingofrotorthrustisfoundtohavehigher
priorityandthusjustifiesthescalingchoice.
An example of downscaling of wind and wave conditions has been supplied. The example also demonstrates how
thestructure(afloatingwindturbine)shouldbescaled.Itisdemonstratedthatthe proposedscalingyieldsmodel
scaleresultsforthrustandwaveinducedmotionthatcanbeupscaledtoprototypescalewithaperfectmatch.

Page 48 of 50
5 REFERENCES
1. UpWind:Fisher,T.,W.deVriesandB.Schmidt:UpWinddeliverable,DesignBasisWP4:OffshoreFoundations
andSupportStructuresUpWindProject.UniversittStuttgart,2010.
2. IEC614003:WindturbinesPart3:Designofoffshorewindturbines,2009
3. DNVOJJ101.OFFSHORESTANDARD:Designofoffshorewindturbinestructures.DNV,2007.
4. ABS(AmericanBureauofShipping),OffshoreWindTurbineInstallation,AmericanBureauofShipping,2010
5. ABS(AmericanBureauofShipping),DesignStandardsforoffshorewindfarms,AmericanBureauofShipping,
2011
6. IEC614001,Designrequirements,2008.
7. IEC614001,Designrequirementsamendment1,2010
8. DS/EN614003,WindturbinesPart3:Designrequirementsforoffshorewindturbines,2009.
9. J.Mann,Windfieldsimulation,Prob.Engng.Mech.Volume14,no4,269282,1998
10. E.R.Jrgensen,S.E.Larsen,S.Frandsen,N.J.TarpJohansenandJennyTrumars,BeatriceEnvironmentaldata,
RisI2354FP6IntegratedprojectDOWNVIndcontract503202.
11. S.E.Larsen,N.E.TarpJohansen,S.FrandsenandE.R.Jrgensen,SdraMidsjbankenEnvironmentalData
Initialanalysis,RisI2505.FP6IntegratedprojectDOWNVIndcontract503202
12. Pena, T. Mikkelsen, S.E. Gryning, C.B. Hasager, A.N. Hahmann, M. Badger, I. Karagali and M. Coutney,
Offshore vertical wind shear. Final report NORSEWIND work task 3,1, DTU Wind Energy reort 0005 (EN),
2012
13. N.Hahmann,JuliaLange,AlfredoPeaandCharlotteB.Hasager:TheNORSEWInDnumericalwindatlasfor
theSouthBaltic,DTUWindEnergyEReport0011(EN).
14. Karagali,A.Pena,M.BadgerandC.B.Hasager,WindcharacteristicsoftheNorthSeaandtheBalticSeasfrom
QuickScattsatellite.WindEnergy.2012
15. BMSumerandJFredse"Hydrodynamicsaroundcylindricalstructures".WorldScientific2006.
16. Wierings,J.Shapesofannualfrequencydistributionsofwindspeedobservedonhighmeteorologicalmasts.
BoundLayerMeteorol,1988,47,85
17. M.K.Ochi.Oceanwaves:thestochasticapproach.CambridgeOceanTechnologySeries,2005.
18. Y. Goda.RandomSeasandDesign Of MaritimeStructures.Advanced Series OnOceanEngineering,Volume

15WorldScientificPublishingCompany,2000.
19. O. Ditlevsen. Stochastic model for joint wave and wind loads on offshore structures. Structural Safety,
24:139163,2002.

Page 49 of 50
20. P.W. Cheng, G.J.W. van Bussel, G.A.M. van Kuik, and J.H. Vugts. Reliabilitybased design methods to
determinetheextremeresponsedistributionofoffshorewindturbines.WindEnergy,6:122,2003.
21. K. Johannessen. Joint distribution for wind and waves in the northern north sea. In Proceedings of the
EleventhInternationalOffshoreandPolarEngineeringConference,2001.
22. P.Agarwal,L.Manuel.Simulationofoffshorewindturbineresponseforlongtermextremeloadprediction.
EngineeringStructures,Volume31,Issue10,October2009,Pages22362246
23. Hildebrant and T. Schlurmann: Breaking wave kinematics, local pressure, and forces on a tripod support
structure.CoastalEngineering,114,2012.
24. O. J. Andersen and J. Lvseth: The Frya database and maritime boundary layer wind description. Science
DirectMarineStructures,173192.2006.
25. O.J.AndersenandJ.Lvseth:Stability modificationsoftheFryawindspectrum.Journ.WindEngeneering

IndustrialAerodynamics,236242,2010.
26. O. J. Andersen and J. Lvseth: Gale force maritime wind. The Frya data base. Part1: Sites and
instrumentation.Reviewofthedatabase.Jour.WindEngeneeringIndustrialAerodynamics,97109,1995
27. Bottasso,Carlo."DevelopmentofaWindTunnelModelforSupportingResearchonAeroServoElasticityand
ControlofWindTurbines."Diss.PolitecnicodiMilano,2011.
28. Wolowicz , Chester, James Bowman, and William Gilbert. United States of America. NASA. Similitude
RequirementsandScalingRelationshipsasAppliedtoModelTesting(NASATechnicalPaper1435).Hampton,
Virginia:LangleyResearchCenter,1979.
29. Martin, Heather. METHODOLOGY FOR WINDin/WAVE BASIN TESTING OF FLOATING OFFSHORE WIND
TURBINES.UniversityofMaine,June10152012.
30. Singleton, Jeffrey. "IMPORTANT SCALING PARAMETERS FOR TESTING MODELSCALE HELICOPTER ROTORS."
Diss.U.S.ArmyResearchLaboratories,VehicleTechnologyCenterNASALangleyResearchCenter.
31. Newman,JohnNicholas.MarineHydrodynamics.1.Cambridge:MITPress,1977.
32. Dagher, Habib. "Floating Offshore Wind". University of Maine, DeepCwind Consortium. 27 Apr. 2011.
AdvancedStructures&CompositesCenter.25June2012.
33. Hall,Stephen."DynamicModeling."SportAviatioin.July1987:288298.
34. Jameson,Peter.InnovationonWindTurbineDesign.Sussex:JohnWileyandSonsLtd.,2011.78.
35. Anderson,John.FundamentalsofAerodynamics.4th.Singapore:McGrawHill,2007.41,499.
36. Jain, Anant, Amy Robertson, Jason Jonkman, Andrew Goupee, Richard Kimball, and Andrew Swift. United
StatesofAmerica.U.SDepartmentofEnergy.FASTCodeVerificationofScalingLawsforDeepCwindFloating
WindSystemTests.OakRidge,TN:OfficeofScientificandTechnicalInformation,2012.
37. Jonkman,J.M."DynamicsModelingandLoadsAnalysisofanOffshoreFloatingWindTurbine."Diss.National
RenewableEnergyLaboratory,2007.

Page 50 of 50
38. Hansen,AMandLaugesenRExperimentalstudyofthedynamicresposeofaTLPwindturbine.BScthesis,
DTUMechanicalEngng.Lyngby,Denmark,June2012.
39. Jonkman, J, Butterfield, S, Musial, W, Scott, G. Definition of a 5 MW reference wind turbine for offshore
systemdevelopment.NREL,US,2009.
40. Matha, D. Model development and loads analysis of an offshore wind turbine on a tension leg platform
withacomparisontootherfloatingwindturbineconcepts.MScthesis.Univ.Colorado,US.April2009.
41. Hansen,MOLAerodynamicsofwindturbines.Earthscan,London,UK,2008.
42. Larsen, TJ and Hanson, TD A method to avoid negative damped low frequency tower vibrations for a
floatingpitchcontrolledwindturbine,Proc.ScienceTorqueWind.2007.
43. MKhn,"Dynamicsanddesignoptimisationofoffshorewindenergyconversionsystems,"PhDThesis,Delft
UniversityofTechnology,Delft,TheNetherlands.2001.
44. J.N.Newman,MarineHydrodynamics,1977,MITPress
45. MorisonJR,OBrienMP,JohnsonJWandSchaafSATheforcesexertedbysurfacewavesonpiles.J.Petrol.
Techn.,PetroleumTransactions,AIME189pp149154.1950.

D2.04 Collation of Offshore Wind-Wave Dynamics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

D2.04 Collation of Offshore Wind-Wave Dynamics

Uploaded by

Copyright:

Available Formats

bbbbbb

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

Rev. Date Description Author Checkedby

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

2.2 PARAMETERS FOR WIND AND WAVE SPECIFICATION

D2.4: Collation of offshore wind-wave dynamics

2.3 ONE PARAMETER CLIMATE BASED ON WIND SPEED.

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

2.3.2 Waves and other oceanic parameters

Asforwinds,one couldalsofromwavedata deriveclimaticdistributionsof thecharacteristicwaveparameters,H

D2.4: Collation of offshore wind-wave dynamics

2.3.3 Lumped windwave climate for fatigue calculation

D2.4: Collation of offshore wind-wave dynamics

(2.3.12) 11.1 ( ) / 14.3 ( ) /

D2.4: Collation of offshore wind-wave dynamics

2.4 STANDARD OFFSHORE CONDITIONS FOR TESTING AND MODELLING

D2.4: Collation of offshore wind-wave dynamics

2.4.2 Wind and wave climate for five specific sites

D2.4: Collation of offshore wind-wave dynamics

TheJONSWAPpeakenhancement parameter,, isdeterminedfrom(2.3.11) anddepictedinfigure2.4.7.Itisseen

Figure2.4.7.Thepeakfactor versus the 10 meter wind speedforthe4sitesconsidered.ThehighvaluesfromBeatrice

D2.4: Collation of offshore wind-wave dynamics

3.1 REVIEW OF EXISTING SCALING STUDIES AND RELEVANT NONDIMENSIONAL

3.1.1 List of Symbols for the review

D2.4: Collation of offshore wind-wave dynamics

KeuliganCarpenter number (KC)

Tip Speed Ratio (TSR)

Mach Number (Ma)

3.1.3 Dominant scaling methodologies for FOWT

3.1.4 UMAINE scaled FOWT testing review

3.2 RECOMMENDED SCALING METHOD

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

3.2.2 Froude scaling of hydrodynamic forces

3.2.3 Aerodynamic scaling

D2.4: Collation of offshore wind-wave dynamics

3.2.5 Check of natural frequency and gyroscopic force

3.2.6 Properties that may not scale consistently

3.2.7 Summary of the scaling

D2.4: Collation of offshore wind-wave dynamics

3.4.1 The TLP wind turbine

Property Symbol Valueprototypescale Scaling Value, model scale

3.4.3 Hydrodynamic force

3.4.5 Wind and wave climate

D2.4: Collation of offshore wind-wave dynamics

Finally,fortheJonswapspectrum, needstobedetermined.Hereformula (2.3.11)suggests=3.83whichseems

Property Symbol Value prototype

D2.4: Collation of offshore wind-wave dynamics

3.4.7 Consequences of imperfect aerodynamic scaling

D2.4: Collation of offshore wind-wave dynamics

D2.4: Collation of offshore wind-wave dynamics

18. Y. Goda.RandomSeasandDesign Of MaritimeStructures.Advanced Series OnOceanEngineering,Volume

D2.4: Collation of offshore wind-wave dynamics

25. O.J.AndersenandJ.Lvseth:Stability modificationsoftheFryawindspectrum.Journ.WindEngeneering

D2.4: Collation of offshore wind-wave dynamics

You might also like