You are on page 1of 20

RicardoSantosPereiranr1371282 November2009

ValidationofBEMcodes
Objective:
In the following pages a comparative study will be performed between two BladeElement
Momentum codes and the pressure measurements from the MEXICO obtained at yawed flow
conditions,withtheobjectiveofvalidatingthesecalculationmethodssothatposteriorlydynamicstall
modelsareincludedinthecomputation.
Motivation:
OneofthemaingoalsofthismasterthesisistocomparetheperformanceofdifferentDynamic
Stallmodelsinthepredictionoftheloadswindturbinesunderyawedflowconditionswillexperience,
andasdiscussedbefore,usuallythesesemiempiricaldynamicstallmodelsrequiretheangleofattack
historyasaninput,i.e.,Cl=f( ),with =f(t).
Howeveritisfirstlynecessarytounderstandexactlywhat'angleofattack'signifies;the'angleof
attack' isdefinedastheangleanairfoil'schordisatwiththestreamvelocity,however,inthewind
turbine environment the local velocity greatly varies when different rotor positions are considered,
since the blades indeed slow down the incoming wind, and especially in yawed conditions this
makesthe'angleofattack'conceptratherunclear.
Still,itispossibletocomputetheangleofattackatdifferentspanwisestationsandazimuthal
positions of a wind turbine using both BEM and liftingline codes. In his work, Micallef [ref to
documentwithIFW]usedaninversefreewakecodetocomputethelocalanglesofattackthroughout
therotoroftheMEXICO.Inversefreewakecodesuseloadsasinput,andconsideringtheinfluenceof
bound,shedandtrailedvorticestheycompute,locally,whatshouldbetheboundcirculationsuchthat
theKuttaconditionismetwith.Fromthiscirculation,theangleofattackisthencalculated,andbased
onitsmethodofcomputation,letuscallthisvaluethe'circulatory'angleofattack.
However,inDynamicStallmodeling,theangleofattack,whichasmentionedaboveisusually
theinputfortheunsteadyloadcomputation,istakensimplyastheangletheairfoil'schordisinstantly
atwiththestreamvelocity.Baseduponhowitisdefined,letuscallthisvaluethe'geometric'angleof
attack.Thedirectionofthestreamvelocityiswelldefinedinwindtunneltestingwheremostofthedata
usedtoderivethesemiempiricalDynamicstallmodelswastakenfrom,leavingnoroomforambiguity.
Naturally,ifthereissomeshedvorticityinwindtunneltests,thelocalvelocityattheairfoilwillbe
influenced by it, and the bound circulation necessary to meet with the Kutta condition will be
associatedwithaninstantaneous'circulatory'angleofattackwhichisdifferentfromtheusuallytaken
'geometric'angleofattack.
Thusitisunderstoodthatusingtheangleofattackhistory, =f(t),computedwithaninverse
freewakemethod,asaninputtoDynamicStallmodelsisnotexpectedtoyieldgoodpredictionsofthe
aerodynamicloads,sinceweareindeedinpresenceoftwodifferentwaystotackleandsimulatethe
problemathands.
Ontheotherhand,itisalsopossibletouseaBEMcodetocompute =f(t)andusetheangleof
attackhistoryasaninputtoDynamicStallmodelsandthusobtaintheunsteadyloads,tocomparewith
theMEXICOmeasurements.However,inaBEMcode,ateachtimeinstantthemomentumlossofthe
incomingairstreamisequatedwiththeforcesontheblades,iteratingontheinductionfactors;sincein
astandardBEMcodetheseforcesarecalculatedbystaticClvs functions,itiseasytounderstandthat
if the reduced frequency is high enough, this will lead to an inaccurate computation of the
instantaneousinductionfactorsandthustoanerroneousalphahistory.
From the stated above, one is lead to conclude that the dynamic stall models must be
implemented inside the BEM code, that is to say the instantaneous loads must be calculated
consideringtheangleofattackhistory, =f(t),sothatthemomentumbalanceyieldsamoreaccurate
estimationoftheinductionfactorsinthefollowingtimeinstant.

ItisimportanttostatethattheBEMmodelswillbevalidatedagainstMEXICOmeasurements
obtainedatmoderatewindspeeds;thisisjustifiedbythefactthatatlowwindspeedsitislikelythat
theturbineisworkingintheturbulentwakestate,whichmeansthatthetrailedvorticeswillhavea
majorinfluenceintheflowfieldandconsequentiallyintheobservedloading.Bearinginmindthatthe
goal is to use the BEM codes to simulate Dynamic Stall, which will naturally occur at high wind
speeds,itisunderstandablethatthevalidationatlowwindspeedsisnotveryrelevant.
Whenhighwindspeeds areconsidered,itisexpectablethattheairfoilstalls,atleastinsome
portionoftheblades,andasdiscussedbeforethisstallwillhaveadynamicnature.SincetheBEM
codes under consideration usestaticClvs functions,itisnotexpectablethatgoodagreementis
foundbetweencalculationsandmeasurementsatthesehighwindspeeds,andsoitmakeslittlesenseto
comparesuchresults.
Finally,oneshouldalsokeepinmindthatathighwindspeeds,i.e.atlowtipspeedratios,the
inductionfactorswillhaveasmallermagnitude[includereferencethatexplainsthis],whichmeansthat
theadvanceretreatingbladeeffectwillbedominant.Asmentionedbefore,BEMcodesiterateonthe
inductionfactors,butsincethesefactorswillhaveasmallermagnitudetheircontributiontotheangleof
attack'seen'bythebladewillalsobesmallcomparedtothegeometriceffect,andconsequentiallythe
accurate computation of these induction factors becomes less important in the prediction of the
unsteadyloadsthewindturbinebladeswillexperience.
ConsideredModels:
TwodistinctBladeElementMomentummodelswillbeconsidered,whichfromnowonwillbe
referredtoas'Classical'BEMand'Empirical'BEM:
ClassicalBEM
ThisdesignationreferstothealgorithmforBEMcodeinyawedflowsuggestedin[referenceto
the'WindEnergyHandbook'],andwasmainlyadaptedfromthepreviousworkbyMicallef[reference
toDaniel'sBEMcode].
Thismodelassumestheinducedvelocitytobegivenby:
u=u
avg
(1+K(X) F(j)sin())
,whereKdependsonthewakeshape,Fisaflowexpansionfunctionwhere isthefractionradius
andistheazimuthangle.Thiscodesiteratestheinductionfactorsateachradialpositionbyaveraging
over the azimuth, using vortex cylinder theoryforthethrustforces,andcomputesbothazimuthally
averagedaxialandtangentialinductionfactors.
Inthismodeltheflowexpansionisalsoincluded,aswellasbothtipandrootlossfactors,and
alsothewakerotation,computedassumingasinglevortexbeingtrailedfromthecenterofrotation.
Oncetheazimuthallyaveragedinductionsarecomputedthismodelcalculatestheloadvariations,and
the azimuthal variations can be found by using the thrust and torque gradients. A more detailed
descriptionisgivenin[referencetoDaniel'sBEMcode].
Thenormalandtangentialforcecoefficientsillustratedbelowarederivedfromthecomputedlift
anddragcoefficients,usingalsothecalculatedanglesofattack,accordingto
C
n
=C
l
. cos(o)+C
d
.sin(o)
C
t
=C
l
.sin(o)C
d
. sin(o)
EmpiricalBEM
This method of computation uses the empirical model from Schepers [reference to Gerard's
model] of the axially induced velocities in a yawed flow wind turbine; it was derived from an
experimentperformedatTUDelft,wherethevelocitydistributionwasmeasuredatseveralspanwise
andazimuthalpositionsandfitwithasecondharmonicFourierseries,takingtheform:
u=u
avg
(1A
1
cos(P
1
)A
2
cos(2 P
2
))
where A
1
, A
2
, P
1
and P
2
arefunctionsoftherelativeradius,oftheazimuthandoftheempirically
derivedconstants.
Itshouldbenotedthatthisaxialinductionvelocitiesempiricalmodelwasderivedonlyforthe
designcondition ,andthatmeasurementsonlycoveredspanwise stationsupto80%. Forpositions
closertothetiptheresultsweresimplyextrapolatedfromtheazimuthallyaveragedinduction,which
meansthatthetiplosswasnottakenintoaccount.
As explained in [reference to Gerard's model], this empirical model has been previously
extensively validated, from the output forces,withmeasurements,performingveryoftenbetterthan
othermodels,andneverworse.
NowregardingtheBEMmodelimplemented,theinplanevelocitiessimplyconsidertheskewed
inflow,whichcanbederivedfromgeometricconsiderations;inthisapproachtheinducedtangential
velocitiesareneglected,howeveritisnotexpectedthatthiswillgreatlyaffecttheforce'scalculation
since their magnitude is very small when compared with the inplane velocities due to the blade's
rotation.
From both the inplane and axial velocities, the angle of attack at each spanwise position is
calculated, and using static Cl vs and Cd vs functions, the forces on each blade element are
calculated.Finallyintegratingoverthespanandsummingthecontributionofallbladeswegetthetotal
Thrust,anditerateontheaverageaxialinductioncoefficientaccordingto
C
T
=
T
0.5j.(U
Wind
. cos)
2
. A
and C
T
=4a(1a)
Oncetheprocessconverges,thenormalandtangentialforcecoefficientsarecomputedinan
identicalwayhaswasdoneforthe'Classical'BEM.
ProcessingoftheresultsfromtheMEXICO
Sincetheyareplottedforcomparisonandtookasareferencetocomputetheassociatederror,it
isnecessarytoexplainhowtheresultsfromtheMEXICOexperimentwereprocessed.
Forthenormalandtangentialforces,acubicinterpolationofthepressuremeasuredbyeach
sensorwastaken,andintegratedoveracubicinterpolationoftheairfoilsurface.Toobtaintheforce
coefficients,thesecomputedforcesweredividedbythemaximumpressureoccurringovertheairfoil,
takenasthemaximumofthecubicinterpolationofthepressuresensors.
C
n
MEXICO
=
F
n
Interp
P
max
Interp
Fromthefollowedapproachthecomputedforcesandforcecoefficientscanbeexpectedtoyield
resultswhichareaccurateenoughtocomparemodelsagainst.
ViscousContribution
Itshouldalsobestatedthatbysimplyintegratingthepressuresovertheairfoilonedoesnot
actually compute the Drag force, since obviouslythere isalsoa viscouscontribution. However, the
magnitudeofthiscontributionisquitesmall,andtheassociatederrorcanbecomputed[referenceto
viscous/pressuredragcomputation]resultingin:
e
Cn
0.0005 and e
Ct
0.08

Beingsothisdiscrepancywill,forthemoment,notbetakenintoaccount.
Results:
Thenormalandtangentialforcecoefficientswillnowbeplotted,versustheazimuthangle,for
eachofthespanwisestationsandconsideringdifferentyawanglesandincomingwindspeedsof15and
18m/s.Weshouldalsobareinmindthatthefirstwindspeedcorrespondstothedesigncondition.
Intheplots,weshouldobservethefollowinglegend:
Blue,forthetangentialforcecoefficientfromtheMEXICOmeasurements
Turquoise,fortangentialforcecoefficientfromtheEmpiricalBEM
Green,forthetangentialforcecoefficientfromtheClassicalBEM
Red,forthenormalforcecoefficientfromtheMEXICOmeasurements
Pink,forthenormalforcecoefficientfromtheEmpiricalBEM
Black,forthenormalforcecoefficientfromtheClassicalBEM
Moreover,foreachspanwisestationtheaverageandmaximumrelativeerrorisshowninatable,
includingcomparisonsfornormalandtangentialforceandnormalandtangentialforcecoefficient,for
bothBEMmodels.
25%Span
Illustration1:U=15m/s,Beta=15deg
Illustration2:U=15m/s,beta=30deg
Illustration3:U=15,beta=45
Illustration4:U=18m/s,Beta=15deg
Illustration5:U=18m/s,Beta=30deg
Illustration6:U=18m/s,Beta=45
From the results shown above there are several things to observe; firstly, it is clear that the
empiricalBEMiscapableofpredictingthetrendswhichtheforcecoefficientsdisplay,throughouta
wholerevolution;onecanalsoseethatthemagnitudeofthecalculatedvalueswiththeempiricalBEM
isinagreementwiththemeasuredvalues.AsfortheclassicalBEMonecansaythatitsloadprediction
issomewhatpoorer.
One must bare in mind that theaxialvelocity,and thusthe angleof attack, willbedirectly
related with the product U
Wind
. cos() .It is thus understandable that both models cannot really
predicttheloadsatU=18m/sand=15deg,sincethentheairfoilwillbe(dynamically)stalledovera
largeportionoftherevolution,andasmentionedbefore,nounsteadyeffectsaretakenintoaccountin
themodelsunderconsideration.
Also,since25%isthespanclosesttotheroot,therotationaleffectswillbeverystronganditis
naturalthatstalldelayandthusincreasedliftispresent,ascanbeseeninfig6;sincetheimplemented
models are based on 2D static data, one can understand the poor agreement often found at 0 deg
azimuth,i.e.wheretheanglesofattackwillbelarger.Betterresultsarelikelytobemetwithifsome
rotationalcorrectionisincluded,orifaspecificsoftwareisusedtoobtaintheforcecoefficients,suchas
Xfoil.
35%Span
Illustration7:U=15m/s,Beta=15deg
Illustration8:U=15,Beta=30deg
Illustration10:U=18m/s,Beta=45deg
Illustration9:U=15,Beta=45deg
Intheresultsabove,thetrialscorrespondingtoU=18m/s,=15degandU=18m/sand=30
degwerenotdisplayedsincethepressuredistributionovertheairfoilwasextremelyincoherent,andso
itisassumedthatapressuresensormalfunctionhasoccurredmeaningtheresultsshouldbediscarded.
Itispossibletoobservesimilartrendsarefoundatthisspanwisesectionfromwhatwasfoundat
the 25% position, which is understandable since these are indeed stations under a considerable
influenceoftheroot,andthusthemodelsshouldshowasimilarlevelofagreement.
60%Span
Illustration11:U=15ms,Beta=15deg
Illustration12:U=15m/s,Beta=30
Illustration13:U=15m/s,Beta=45deg
Illustration14:U=18m/s,Beta=15deg
Illustration15:U=18m/s,Beta=30deg
Illustration16:U=18,Beta=45deg
AtthisspanwisestationitisagainpossibletoseethattheempiricalBEMmodelcloselyfollows
thetrendsobservedintheempiricalresults,whiletheclassicalBEMgenerallyperformsnotsowell.
Onealsogetstheimpressionthatthemagnitudeofthepredictedloadsaresomewhatcloserthanatthe
moreinboardsections,whichcanbeconfirmedbyanalyzingtheerrortables;thisbetteragreementcan
beexplainedbythefactthat,atthismidspanposition,theairfoilislesslikelytostall,andthusthe
(unaccountedfor)unsteadyeffectswillhaveasmallerinfluence,andalsobecausetherotationaleffects
toowillhavealessnoticeableimpactontheloading.
82%Span
Illustration17:U=15m/s,Beta=15deg
Illustration18:U=15m/s,Beta=30deg
Illustration19:U=15m/s,Beta=45deg
Illustration20:U=18m/s,Beta=15deg
Illustration21:U=18m/s,Beta=30deg
MindingtheresultsaboveonecanseethatforthisoutboardsectiontheempiricalBEMagainis
abletofollowthetrendstheempiricalresultsshow,usuallyinabetterfashionthantheclassicalBEM
model.Regardingthemagnitudeoftheloadingpredictions,itispossibletoobservethat,atsmallyaw
angles, some over estimation of the aerodynamicforcesseemsto takeplace, which can perhaps be
causedbythefactthattiplossesarenottakenintoaccountintheempiricalBEMmodel.Whenlarge
yawanglesareconsidered,however,onecanseethereisverygoodagreementbetweentheempirical
BEMmodelandthemeasuredloading.
Illustration22:U=18m/s,Beta=45deg
92%Span
Illustration23:U=15m/s,Beta=15deg
Illustration24:U=15m/s,Beta=30deg
Illustration25:U=15m/s,Beta=45deg
Illustration26:U=18m/s,Beta=15deg
Illustration27:U=18m/s,Beta=30deg
For this outboard section the obtained results are similar to the 82% span case, which is
understandablesincethesetwostationsarelocatednearthetipofthebladeandsowillbeunderits
influence.Again,wecanseeaslightloadoverestimation,possiblyattributedtothenoninclusionof
thetiplosseffects,whichislesspronouncedatlargeryawangles.
Illustration28:U=18m/s,Beta=45m/s
DiscussionandConclusion:
Becauseitisthequantitywiththegreatestcontributiontothetotalforcecoefficient,andalso
sincesomeofthepreviouslydiscussedDynamicStallmodelsonlyincludethepredictionofCl,itwas
thoughtoberelevanttocomparetheaveragenormalforcecoefficienterrorofthetwoBEMmodels,
for each of the considered wind speeds and spanwise positions, averaged over all considered yaw
angles.
For15m/s:
Andfor18m/s:
Minding the tables above, it is clear that the Empirical BEM model performs better than
Classicalone,atallspanwisestationsandconsideredwindspeeds,andsooneisledtotheconclusion
thattheempiricalshouldbepreferredandusedtoimplementtheDynamicStallmodels.
SpecificallyregardingtheperformanceoftheEmpiricalBEMmodelwithdifferentspanwise
sections, the results seem to indicate thatthismodel will yieldbestpredictionswhen themidspan
sectionisconsidered.Asexplainedbefore,atthestationsneartheroottherotationaleffectswillhavea
largeinfluenceindelayingstall,whichisnottakenintoaccountinthemodel,andatthemoreoutboard
sectionsthelargerdiscrepancycanperhapsbeattributedtothefactthatthemodeldoesnotincludetip
losseffects.
Finally, one should bear in mind that these BEM models were not tuned to perfection,
specificallyregardingtheprocedureusedtogettheforcecoefficientsfromtheangleofattack,sincein
themodelspresentedabovethiswassimplydonebyinterpolatingfromatablewithdataobtainedina
2D,staticenvironment.However,DynamicStallmodelsuseadifferent,moresophisticatedapproach,
usuallytakingtheslopeoftheClvsalphacurveattheoriginandapplyingsomedeficiencyfunctions,
andthiswillnaturallybeimplementedintheBEMcodeposteriorly.Still,theobjectiveofthissection
wastovalidateandselecttheBEMmodeltousefromnowon,andforthatitwasthoughtthatthis
simplerapproachwouldsuffice.

You might also like