Aerodynamic design methodology for wind tunnel tests of wind [617461]

Aerodynamic design methodology for wind tunnel tests of wind
turbine rotors
Ilmas Bayati, Marco Belloli∗, Luca Bernini, Alberto Zasso
Politecnico di Milano, Department of Mechanical Engineering, Italy
article info
Article history:
Received 15 February 2017Received in revised form9 May 2017Accepted 10 May 2017
Available online 24 May 2017
Keywords:
Wind turbines
Aero-elastic scaling
Carbon- fibre
Wind tunnelabstract
This paper illustrates the methodology and the experimental veri fication of the design of a 1/75 aero-
elastic scaled rotor of the DTU 10 MW reference wind turbine for wind tunnel tests. The aerodynamic
design was focused on the minimization of the difference, in terms of thrust coef ficient, with respect to the
full scale reference. From the Selig low-Reynolds airfoils database, the SD7032 one was chosen for thispurpose and a corresponding constant section wing was tested at DTU red wind tunnel, providing force
and distributed pressure coef ficients for the design, in the Reynolds range
−×30 250 103and for different
angles of attack. The aero-elastic design algorithm was set to de fine the optimal spanwise thickness over
chord ratio (t/c), the chord length and the twist, in order to match at least the firstflapwise scaled natural
frequency. An aluminium mould for the carbon fibre autoclave process was CNC manufactured based on
B-Splines CAD de finition of the external geometry given as an output of the design procedure. Wind tunnel
tests at were carried out Politecnico di Milano on the whole 1/75 wind turbine scale model, con firming the
successful aerodynamic design and manufacturing approaches. The experimental modal analysis carried
out to verify the structural consistency of the scaled blade is also reported.
&2017 Elsevier Ltd. All rights reserved.
1. Introduction
Wind tunnel tests of wind turbine scale models represent an af-
fordable and effective way for assessing the aerodynamics of windturbines saving time, costs and uncertainties related to full scale ex-
perimentation. However, the main limitation in rotor scaling proce-
dure for wind tunnel tests is the impossibility of matching Reynolds
number with respect to full scale. Thi s paper illustrates the non-trivial
aero-elastic optimal design, the realization and the experimental ver-ification of the wind tunnel 1/75 scale rotor of the DTU 10 MW wind
turbine. More speci fically, this work was developed for floating off-
shore wind turbine (FOWT) applications ( Lifes50 ț,B a y a t ie ta l . ,2 0 1 3 ,
2014 ); nevertheless, the methodology reported and the conclusions
drawn are of general validity in scaling rotors of wind turbines.
Similar efforts in scaling wind turbines have been recently made
(Bredmose, 2014 ). Furthermore, a deep analysis of the scaling ef-
fects can be found in ( Bottasso et al., 2014 ) regarding previous ac-
tivities at Politecnico di Milano wind tunnel: this work deals withthe de finition of a procedure for aero-elastic model design, and
good results, in term of thrust and torque value matching, were
obtained as well as a correctly scaled blade structural behaviour
also considering bend-twist scaling ( Campagnolo et al., 2014 ).A further study on the scaling effect of the turbine rotor aero-
dynamics was carried out in ( Make, 2014 ), where it was found,
both numerically and experimentally, that the Reynolds dis-
crepancy caused a different behaviour of the model scale rotor,and by adjusting the chord length by an increment of 25% wasobtained so that the model rotor matched target scaled thrust.
Similar results were obtained by DTU in ( Bredmose et al., 2015 ),
also in this case the rotor blades were geometrically adjusted inorder to overcome the Reynolds scaling limit which, together withthe use of low Reynolds airfoil and turbulence generators, allowedto obtain good results for the rotor aerodynamic performance.
The Reynolds scaling problem is even more important when
dealing with offshore related testing, in this case Froude scaling ismandatory ( Bredmose, 2014 ) worsening the Reynolds mismatch.
The DTU 10 MW wind turbine, which is the reference of this
work, was firstly designed in the framework of the Light Rotor
project in 2012 ( Bak et al., 2012 ), starting from the upscaling of the
reference 5 MW turbine from NREL ( Jonkman et al., 2009 ). Later
the Light Rotor project design evolved in the nowadays publiclyavailable reference design, released by DTU ( Baket al, 2013 ). The
DTU 10 MW is being used as reference design in numerous currentresearch activities related to wind energy development, rangingfrom wind farm optimization to offshore wind turbine simulationor also for numerical tools benchmark and validation. Table 1Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/jweiaJournal of Wind Engineering
and Industrial Aerodynamics
http://dx.doi.org/10.1016/j.jweia.2017.05.004
0167-6105/ &2017 Elsevier Ltd. All rights reserved.
∗Corresponding author.J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227

reports the main DTU 10 MW speci fications in term of dimensions,
masses and operating wind speed.
2. Scaling the reference design
Thefirst step of model design was the comparison between the
turbine speci fications and the Polimi Wind Tunnel (GVPM) ( Zasso et al.,
2005 )t e s ts e c t i o nd i m e n s i o n sa n d flow performance. The GVPM is a
closed circuit facility with two test rooms: a 4 /C24m high speed low
turbulence and a 14 /C24m low speed boundary layer test section. The
high speed section is characterized by very low turbulence, <Iu 0.15% ,
and high speed, maximum velocity of 55 m/s, in the low speed section
t h et u r b u l e n c ei n d e xi sh i g h e r , <Iu 2%, with a reduced maximum
velocity 15 m/s. The low speed section is 36 m long, 14 m wide and 4 m
high, allowing very large scale wind engineering tests, useful for civil
engineering application or low bloc kage aerodynamic related tests.
Trying to avoid an excessive miniaturization of the turbine model com-
ponents, the wind tunnel tests are performed in the low speed section.
In Eq. (1)the scale factor is de fined as the ratio between a
general DTU 10 MW turbine parameter and the corresponding
wind tunnel model parameter.
λ=
()p
p 1reference
model
The dimensional analysis technique is fundamental in model
design for wind tunnel. A series of non-dimensional groups areusually taken into account, the most used are the Reynolds num-ber, Froude Number, Strouhal Number, Cauchy number, etc.
Usually the length scale,
λL,i sd e fined from simple considerations
about the wind tunnel dimension, then one of the non-dimen-
sional group is selected to be kept constant from full scale to
model scale. The choice is made considering which are the most
important parameters that in fluence tests results. For example
Froude scaling is typically used for the presence of non-negligible
gravity dependant loads (e.g. long-span bridges, hydrodynamic
forces). In floating offshore wind turbine scale tests in oceanbasins, Froude scaling is mandatory due to the presence of physical
waves. Froude number is de fined as in Eq. (2)
=
()FrV
gL 2
where Vis the velocity, gis the gravitational acceleration and Lis
the length. Fixing the length scale factor λLdue to the dimension of
the model, the velocity scale factor λVis consequently de fined as
λL, resulting in very low speeds for the tests.
F o rt h i sp a r t i c u l a rp r o j e c tt h e λLhas to be selected in the range:
−709 0, the lower limit comes from the maximum wind tunnel
model diameter of 2.5 m, this ensures that the blade tip is far enoughfrom the tunnel ceiling and floor during the rotor revolution, thus
avoiding the wall boundary layer. The higher limit avoids to have anexcessive miniaturization of the model components.
The
λVhas a fixed range of possible values: −1.5 3 , due to a
comparison between the cut out speed, 25 m/s, of the DTU 10 MW
and the maximum wind tunnel speed, 15 m/s.
A discrete number of possible combinations for the scales were
evaluated, a good compromise was found in λ=75L and λ=2V .O n c e
defined the length and velocity scales, the scales of the principal
physical quantities were derived from dimensional analysis. Table 2
reports the most important scaled turbine characteristics.
The blade design aims at matching as close as possible the
scaled values of the turbine aerodynamic thrust and torque. It isworth mentioning that, since this scaled design is related to thestudy of a floating system, the thrust matching is of higher im-
portance since the floating system dynamics is more in fluenced by
thrust than torque ( Bredmose et al., 2015 ).
3. Wind tunnel model blade design input
In this scenario, the main goals in the blade design can be
summarized as follows:
/C15matching the reference thrust coef ficient
/C15matching the scaled first blade flapwise natural frequency
/C15matching the scaled blade weight
It is pretty clear that the blade design is challenging both from
an aerodynamic and structural point of view. In Fig. 1 the blade
design procedure is reported.
3.1. Reference design input
The DTU 10 MW reference and wind tunnel model turbine
operational parameters are reported in Table 3 , as combinations of
wind speed, V, rotor rotational speed, Ωand Tip Speed Ratio, TSR
( =Ω ·TSR R V /). The model wind speed operational value are re-
duced by λVand the model rotational speed is reduced by λλ/VL,
this ensures that the TSR does not change when scaling to wind
tunnel dimensions. Keeping TSR similitude ensures to have thesame aerodynamic kinematics, as it is discussed in the following.
3.2. Model airfoil
One of the most critical aspect in the model blade design is the
airfoil selection, as a matter of fact, the main limitation in reproducingthe reference aerodynamic performance is related to the Reynoldsnumber reduction when working at wind tunnel scales.
Referring to Eq. (3), Renyolds number depends on the air
density
ρ, wind speed U, the blade chord length cand air dynamic
viscosity μ. The scale factor for Reynolds number is therefore de-
fined as λ λλ=Re L V equal to 150 (i.e. the wind tunnel Reynolds
number is 150 times smaller the full scale one). This could result inTable 1
DTU 10 MW turbine speci fications.
Parameter value units
Cut in wind speed 4 m/s
Cut out wind speed 25 m/sRated wind speed 11.4 m/sRotor Diameter 178.3 mHub Diameter 5.6 mHub Height 119.0 mMinimum Rotor Speed 6.0 rpm
Maximum Rotor Speed 9.6 rpm
Blade Prebend 3.332 mRotor Mass 228,0 tonnNacelle Mass 446,0 tonnTower Mass 628,4 tonn
Table 2Wind Tunnel Model turbine speci fications.
Parameter value units scale
Cut in wind speed 2 m/s
λ=2V
Cut out wind speed 12.5 m/s λ=2V
Rated wind speed 5.7 m/s λ=2V
Rotor Diameter 2.37 m λ=75L
Maximum Rotor Speed 360 rpmλ λλ==−37.5fL V1
Rotor Mass 0.54 kgλ λ== × 4.22 10ML3 5I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 218

a completely different aerodynamic behaviour of the blade pro file
at model scale.
ρ
μ=··
()ReVc
3
The Reynolds discrepancy forces to use different airfoil shape
than the one used by DTU at full scale. In particular, choice wenton the SD70xx airfoil series from the Selig-Donovan low Reynolds
database ( Lyon et al., 1998 ). From similar previous experience
(Bredmose et al., 2015 ), the selected airfoil was the SD7032: Fig. 2
reports the model airfoil shape and the DTU 10 MW one at tip. TheSD7032 has a thickness over chord length of roughly 10%, whereasthe FFA-W3-240 thickness is 24%. The lower thickness leads to
limited structural performance but it also makes the airfoil less
sensible to flow separation at low Reynolds value conditions, like
the ones encountered in the wind tunnel testing.
The aerodynamc coef ficients for the SD7032 for Reyndolds
numbers equal to 100
×103and 300 ×103are available in ( Lyon
et al., 1998 ). However, for more suitable Reynolds number data, a
new series of wind tunnel tests were performed on a 2D section
model of the airfoil. The 2D sectional model is 497 mm span-wise
long and 130 mm chord-wise long; it was manufactured with thesame carbon fiber technology used for the turbine blade final
model, this ensured the correct reproduction of surface finisheffect on the airfoil performance, as well as in terms of trailing
edge effective thickness. Fig. 3 shows the two main parts made of
carbon fiber (4 layers of 0.2 mm thickness), the wing ends and the
external attachment rods were machined from steel. The central
aluminium part includes a total of 32 pressure taps drilled directly
on the surface.
Tests were carried out at the red wind tunnel facility located at
the Lyngby DTU campus (Denmark), Fig. 4 . Two aerodynamic
coefficients were measured:
/C15blade section lift coef ficient, Cl: data from pressure taps
/C15blade section drag coef ficient, Cd: data from wake rake pressure
(placed downstream the model at around 5 chord length).
A total of eight Reynolds number were tested,
= [ × × ××× ×Re 50 10 ; 60 10 ; 75 10 ;100 10 ;125 10 ; 150 10 ;3 3 333 3
×× ] 200 10 ; 250 10 .33Figs. 5 and 6 show the obtained aero-
dynamic coef ficients for SD7032 used in the aerodynamic design
of the model blade for different Reynolds numbers; more detailsabout the sectional tests can be found in ( Lifes50 ț).
4. Design process
Standard turbine rotor design procedures are based on the
blade element approach ( Hansen, 2008 ;Manwell et al., 2009 ),
starting from the hypothesis of no radial dependency of the results
the design is carried out for each blade section independently. The
developed procedure for the wind tunnel model design hereinreported starts from the same blade element approach but instead
Table 3
Wind turbine operational parameters, V (m/s), Ω(RPM) and TSR ( /C0).
V(10 MW) Ω(10 MW) TSR(10 MW) V(model) Ω(model) TSR(model)
4.0 6.0 14.0 2.0 225.0 14.0
8.0 6.4 7.5 4.0 240.9 7.511.0 8.8 7.5 5.5 331.4 7.516.0 9.6 5.6 8.0 360.0 5.6
20.0 9.6 4.5 10.0 360.0 4.5
24.0 9.6 3.7 12.0 360.0 3.7Fig. 1. Blade design I/O.
Fig. 3. 2D section model exploded-view drawing.Fig. 2. Model airfoil (SD7032, solid line) and reference airfoil (FFA-W3-240, dashed
line) shapes.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 219

of merely maximising the rotor aerodynamic performance, i.e.
power ef ficiency, the design objective is the matching of a few
selected parameters of the reference full scale turbine.
The aerodynamic and the structural optimization were carried
out in combination, in an iterative loop, until the design reaches anoptimal solution.
4.1. Aerodynamic design
The aerodynamic design deals with the de finition of the chord
and twist value distribution of the model blade. In order to matchthe DTU 10 MW scaled thrust value, it is necessary to match thescaled lift value along the blade, since in common working con-
dition, the section normal load,
Pn, is generated almost entirely by
the section lift, as in Fig. 12 .
The wind tunnel model is successfully designed if the model
section lift Lwtmequals the scaled reference one, LMW 10 , along the
entire blade span, Eq (4).
ΛΛ=
()LL
4MW
VLwtm10
2
So that it is necessary to consider section lift matching and
same working condition for the reference and the model turbine(i.e. same TSR).
Theflow angle at full scale and model scale must be exactly the
same. It is well accepted ( Hansen, 2008 ) to consider the induction
factor of the wake, a and a ’inFig. 12 , only in fluenced by the lift
force and TSR. Therefore, it is possible to de fine an unique flow
angle
ϕfor both the reference and the model turbine, Eq. (5).F o r
the model design it was considered the same pitch angle, θ, of the
reference turbine and two different twist angles, βMW10and βwtm,
the angle of attack, αis defined consequently in Eq. (5).
αϕ θ β
αϕ θ β=− ( + )
=− ( + ) ()5MW MW
wtm wtm10 10
Eq.(6)comes from the substitution of the lift value in Eq. (4)
with the lift coef ficient, Cl, times the air dynamic pressure,
ρV 0.5MW 102at full scale and ρV 0.5wtm2in wind tunnel:
⎛
⎝⎜⎞
⎠⎟ ρλαλρα () = ( )
()VClcVwtm Cl c 0.5 0.5
6MW
VMW MWMW
Lwtm wtm wtm102
10 1010 2
The Eq. (6)is simpli fied in Eq. (7).
ϕθ βλϕθ β (− ( + ) ) = (− ( + ) )
()ClcCl c
7MW MWMW
Lwtm wtm wtm 10 1010
For pitch regulated turbines in standard working condition it is
reasonable to consider that the blade is working away from stall atleast in the region far from the hub. Therefore the blade is in the linear
aerodynamic region and the lift coef ficient can be well approximated
with a linear curve with a
Klslope and Cl0zero value, Eq. (8).
αα
αα() = · +
() = · + ()Cl Kl Cl
Cl Kl Cl 8MW MW MW MW MW
wtm wtm wtm wtm wtm10 10 10 10 100
0
where Klis the lift coef ficient first derivative with respect to the angle Fig. 5. SD7032 drag coef ficients from DTU red wind tunnel test.
Fig. 4. DTU red wind tunnel test section.Fig. 6. SD7032 lift coef ficients from DTU red wind tunnel test.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 220

of attack αin the airfoil linear region and Cl0is the lift coef ficient value
at null angle of attack.
Substituting this in Eq. (6)thefirst aerodynamic constraint
equation is found, Eq. (9).
ϕθ βλ
ϕθ β(· ( − ( + ) ) +) ·
=( ·( −( + ) )+ ) · ()Kl Clc
Kl Cl c 9MW MW MWMW
L
wtm wtm wtm wtm10 10 100 10
0
Also the lift derivative with respect to the flow angle, ϕ,i s
imposed to be matched by the model design, in Eq. (10).
λ·= ·
()KlcKl c
10MWMW
Lwtm wtm 1010
This is done in order to ensure that the unsteady behaviour of
the turbine is well reproduced by the model since the turbine isalso going to be tested for unsteady condition ( Bayati et al., 2016a ).
Moreover, it is reasonable to consider that, by first approximation,
the unsteady behaviour of an aerodynamic body function of thefirst derivatives of the drag and lift curves calculated around the
steady angle of attack ( Cheli and Diana, 2015 ).
The Eqs. (9) and (10) can be rearranged in the final governing
system Eq. (11) from which the model chord and twist are com-
puted along the entire blade span.
⎧
⎨⎪⎪
⎩⎪
⎪λ
ββ=·
=− +
()ccK l
Kl
Cl
KlCl
Kl 11wtmMW
LMW
wtm
wtm MWMW
MWwtm
wtm10 10
10100
100
InFig. 7 the model airfoil, SD7032, is compared with the DTU
10 MW tip airfoil, FFA-W3-240 in terms of lift coef ficient Cl versus
angle of attack, for the range /C02ț9. The lift slope for the DTU
airfoil is higher than the model one, thus the ratio Kl Kl /MW wtm10 is
greater than one which implies a model chord bigger than the
geometric scaled reference one.
Fig. 8 reports the ratio between the model chord and the scaled
reference value, Fig. 9 reports the difference between model twist
and the reference one. The dashed line are the raw calculationoutput, the calculation results near the root region were discardedsince calculation output seemed inconsistent by visual inspection.Also the tip region design has been simpli fied to a constant chord
and pitch variation also due to manufacturing issues. The chordoutput was also interpolated with a cubic smoothing spline.4.2. Structural design
The aerodynamic design de fined the chord and twist distribu-
tion along the blade span. The remaining degree of freedom is theblade cross section relative thickness distribution (t/c). There aretwo constraints in the blade relative thickness de finition:
/C15t/c at the blade tip is equal to the nominal t/c of the SD7032
airfoil (t/c ¼10%)
/C15t/c must converge to 100% at the blade root to match the cir-
cular root section
The blade t/c is de fined in three different regions reported in
Fig. 10 . Region I is the root region where the blade has a circular cross-
section, region III is the tip region, where the cross-section is theSD7032 pro file and lastly in the transition region II the cross-section
must converge from the circular shape to the SD7032 shape.
The extension of region I was de fined by technological con-
straints, in particular, a machined aluminium component has to be
glued to blade root in order to allow a proper blade assembly. The
remaining parameter is the transition region extension,
lII. This
dimension has a great impact in the structural performance of the
Fig. 7. Reference and model airfoil polar lift coef ficient comparison, at rated con-
dition for velocity scale factor λ=2V .Fig. 8. RAW and corrected output of Eq. (11) as function of the non-dimensional
blade station: model and scaled 10 MW chord ratio.
Fig. 9. RAW and corrected output of Eq. (11) as function of the non-dimensional
blade station: model and 10 MW twist difference.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 221

blade, it is obvious that higher lIImeans higher model blade
stiffness since the second moment of area of the circular section is
greater than the SD7032 one.
The lIIlength has been numerically optimized in order to ensure
the matching of the first scaled flapwise blade frequency ( Table 4 ). A
Matla bsimplemented beam FEM model code was iteratively ran until
the correct firstflapwise blade modal frequency. The FEM model is
discretized by 193 beam elements, for each one the correspondent
properties of blade section is assigned, the model is constrained as a
cantilever beam in order to compute the modal analysis.
As reported in Fig. 11 , from a structural point of view, the blade
section is characterized by:
/C15 (· ) EJ N m12bending stiffness about the first principal axis, flap-
wise direction
/C15 (· ) EJ N m22bending stiffness about the second principal axis,
edgewise direction
/C15 () XmE elastic section center.
/C15ν()rad principal axes orientation with respect to the chord line
/C15 ( ) mk gm /mass per unit length
At this design phase the blade was considered made only by
unidirectional high modulus carbon fiber layer aligned with the
blade radial direction. Discarding the anisotropy in the material
the elastic modulus and the density of the carbon fiber was takenfrom standard commercially available data ( =EM P a 135 ,
ρ= kg m 1560 /3). One layer of carbon fiber was considered for a
total thickness of 0.26 mm. Knowing the material properties andthickness and considering the blade chord and twist output from
the aerodynamic design the blade section mechanical properties
were easily computed, see ( Hansen, 2008 ) for theoretical details.
The center of mass for each section is assumed to be coincident
with the geometric center (isotropy).
Once the
lIIlength is optimized, the t/c blade pro file is com-
pletely de fined and the last phase of the structural design is the
recalculation of the blade section shape and aerodynamic coef fi-
cient. In region I and region III the section shape and aerodynamic
are perfectly known, being respectively equal to the circular sec-tion and SD7032 section. In region II, the section shape and
aerodynamic coef ficients are calculated as function of the radial
position from simple linear interpolation based on the t/c localvalue. In Figs. 13 and 14 the section shape and lift coef ficient are
shown for four different radial blade positions.
4.3. Design loop
The aerodynamic and structural design are part of an iterative
loop whose steps are summarized as follows:
1. Computing model airfoil
Klwtmand Clwtm0from model section lift
coefficient
2. Calculating model chord and twist from Eq. (11)
3. Estimating section mechanical propertiesFig. 10. Model blade t/c de finition.
Table 4
Reference structural permanences.
Mode reference frequency (Hz) scaled frequency (Hz)
1stflap mode 0.61 22.87
1st edge mode 0.93 34.872ndflap mode 1.74 65.25
2nd edge mode 2.76 103.50
Fig. 11. Definition of the blade structural principal axes (not in scale).Fig. 12. Blade section velocities and loads.
Fig. 13. Section shape at different radial position and t/c: section shape.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 222

4. Optimizing lIIfor blade flapwise frequency
5. Updating blade section shape and lift coef ficient and loop from
(1)
The design is considered done when the lIIvariation from one
iteration to the following is below 5%. Fig. 15 shows the output of
four subsequent iterations, the design starts from the DTU 10 MWreference value for chord, twist and t/c, then the optimized solu-
tion is computed. It is visible how the chord grows higher than the
fullscale reference, in particular near the maximum chord position,the twist is almost equally reduced along the blade by a couple ofdegrees and the t/c curve goes more rapidly to the tip airfoilthickness than the 10 MW one.
4.4. Design numerical validation
The aerodynamic response of the designed scaled blade was
checked using the AeroDyn module of NREL FAST software(Moriarty and Hansen, 2005 ). AeroDyn is one of the most used,
freely available, rotor aerodynamic solver both in industry andacademic research projects. For this application the main settingsthat has been adopted, are reported:
/C15no stall model;
/C15Equil in flow model (standard BEM theory ( Hansen, 2008 ));
/C15Prandtl hub and tip loss factors;
/C15ρ=( ) kg m 1.225 /3;ν=( )−em s 1.45 5 /2.
A total of 23 equally spaced wind speeds, ranging from cut in to
cut out, were simulated for both DTU 10 MW and wind tunnelmodel. Fig. 16 shows, as an example, the output of the simulations
in terms of thrust coef ficient (the axes are scaled dimensions). The
agreement is fairly good in particular up to the rated wind speed(condition of maximum thrust) thus the scaled blade design wasconsidered a success, even more considering the strong Reynoldsdiscrepancy and the completely different airfoils (e.g. ef ficiency)
used in the scaled model design.
4.5. Blade shape export to CAD file
From the chord, twist and t/c distributions the blade shape is
entirely known, at first step the blade is generated as a points
cloud generated by 721 point at 193 different blade radial position(Fig. 17 ). The blade shape was converted to igesfile format, from a
point cloud, using a 3D B-Spline space interpolation based on a in-house
Matla bsblades converter, for a direct manufacturing
implementation.
4.6. Blade production
From the blade CAD file a CNC machined mould was realized,
Fig. 18 . The blades, Fig. 19 , were realized with prepreg using a
vacuum bag oven process, the mould is divided into two parts that
are the pressure and suction side of the blade, the blade layers areplaced in the mould and pushed against the mould surface with aninflated plastic balloon. The final carbon fiber layup differs from
the design one for purely technological reasons: a 90 deg glassfiber layer is added to the single unidirectional carbon fiber one, inFig. 14. Section lift coef ficient at different radial position and t/c: section lift
coefficient.
Fig. 15. Output of the design algorithm at different iteration loop as chord, twist and t/c value vs the non dimensional blade radial position.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 223

order to add some torsional stiffness to produced blade. The
manufactured blade has a weight of 230 g, thus more then doubleof the target mass. It could have been possible to reduce further
the weight of the blade using thinner carbon fiber layer but the
main limitation was related to possible issues in the extractionphase of the model from the mould.
5. Design veri fication
5.1. Modal analysis
The model blade has been structurally examined using a modalFig. 16. Comparison of the thrust coef ficient computed using FAST, for a velocity
scale factor λ=2V , nominal (scaled) rotational speed and nominal pitch angle.
Fig. 17. Blade points cloud.
Fig. 18. Aluminium mould.
Fig. 19. Carbon fiber blades.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 224

analysis approach. The authors performed an impulse response
test using a series of piezoelectric accelerometer and an in-strumented hammer. The blade was constrained to the groundthrough a rigid beam in a cantilever con figuration, as visible in
Fig. 20 .
A limited number of accelerometers were employed not to have
an excessive mass compared to the blade, and placed at differentdistance from the blade constrained section. Table 5 reports the
precise indication of the accelerometers displacement in the span-wise direction, the disposition is nearly equally spaced, while thesensors distances from the leading edge were varied from onesensor to the other. This ”zig-zag ”disposition was employed in
order to have a sensing pattern able to discern torsional frombending modes.
A series of impact tests responses were acquired, the impact
position was at 0.1 m from the blade root, exciting the blade in apoint close to the grounded section ensure that the firstflap-wise
modes will be excited by the impact. From the impact tests themode shapes of the first two blade modes were reconstructed,
looking at Figs. 21 and 22 the displacement is compatible with
pure flapwise modes, the two mode are respectively the first and
second flapwise mode.
The experimental estimated modes frequencies are slightly
lower than the scaled design ones, this could be a consequence ofthe different layer layup and more importantly of the overweightof the produced blade. Table 6 reports the relative error between
the design and measured frequencies for the first two flapwise
modes.
In order to deal with the lower natural frequency of the man-
ufactured blade than expected ( Table 6 ), the chance of modifying
the scaling factor of the model was taken into consideration. Asexplained in Sec.2, for this particular application the velocity scale,
λV, could be varied in a limited range and looking at Table 2 the
frequency scale, λf, is inversely proportional to the velocity scale so
an higher λVwill turn out in lower scaled model frequency.
Table 7 reports the design versus measured frequency con-
sidering a λ=3V and thus a λ λλ==−25fL V1. For this velocity scales
the blade is more rigid then expected but the error on the first
flpawise mode, which is the most important issue for the aero-
elastic response of the turbine, is now close to 10% that is con-
sidered by the authors a good result, also con firmed by the per-
formance of the rotor in wind tunnel tests at greater velocity scale
factor with respect the design one, as explained in the following.
More details can also be found in ( Bayati et al., 2017a ).
Fig. 20. Experimental setup.
Fig. 21. First mode shape.Table 5
Accelerometers position details.
accelerometer number position span-wise distance
from root (m)distance from lead-
ing edge (m)
Acc1 PosA 0.23 0
Acc2 PosB 0.38 0.066
Acc3 PosA 0.52 0Acc4 PosA 0.77 0.036Acc5 PosA 1.08 0Fig. 22. Second mode shape.
Table 6
Comparison between design and measured blade frequencies.
Mode Design frequency
(Hz)Experimental frequency
(Hz)Relative error
1stflap mode 22.87 17.1 /C025.5%
2ndflap
mode65.25 56.4 /C013.6%I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 225

5.2. Wind tunnel tests
The wind tunnel tests were performed in the Polimi wind
tunnel using a model of the entire turbine described in details in
(Bayati et al., 2016b ), focusing, as first step, on the steady aero-
dynamic characterization.
Firstly, test with λ=2V was performed, Figs. 23 and 24 show the
results of the wind tunnel model compared to the scaled DTU 10 MW
performance in terms of thrust and torque. The agreement on the
thrust force is excellent for all the tested conditions, the maximumthrust force,
∼N70, is well reproduced by the turbine model. Also the
torque matching, that was not the main model target, is very good up
to maximum tested wind speed at the rated condition.
Figs. 25 and 26 report the same results at λ=3V , so at lower
wind tunnel speed. In this case an higher discrepancy was ob-
served between the measured performance and the reference
ones, possibly due to the Reynolds number decrease. In order to
correct the discrepancy in the results, different collective pitch
angle values were tested, between /C01 deg and /C05 deg from the
nominal pitch angle. It was possible to get the experimental curve
closer to the reference one, in fact the model thrust curve almost
matches the target values but this was done at expense of the
torque matching (lowering the pitch angle the thrust increases
while the torque decreases). However, this could be accepted since
the thrust matching is the primary design goal.
Beside the steady aerodynamic curves, wind tunnel tests al-
lowed for a first analysis of the unsteady scaled model response
(Bayati et al., 2016a ). As stated in Sec.4.1, the aerodynamic design
of the scaled blade was done comparing the lift coef ficient deri-
vatives of the full scale and the model scale airfoil, this shouldensure that the unsteady response of the scaled turbine due to
dynamic variation of the operational parameters is similar to the
DTU 10 MW one. In order to check the unsteady response, aTable 7
Comparison between design and measured blade frequencies (using λ=3V ,
λ=25f ).
Mode Design frequency
(Hz)Experimental frequency
(Hz)Relative error
1stflap mode 15.25 17.1 12.1%
2ndflap
mode65.25 56.4 29.6%
Fig. 23. Wind tunnel test results compared with the reference scaled DTU 10 MW
performance: thrust.Fig. 24. Wind tunnel test results compared with the reference scaled DTU 10 MW
performance: torque.
Fig. 25. Wind tunnel test results, for λ=3V : thrust.
Fig. 26. Wind tunnel test results, for λ=3V : torque.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 226

linearization approach was used both numerically and experi-
mentally: a small wind speed variation was imposed at differentmean wind speed without changing the rotor speed or the bladepitch. The variation in the thrust force was computed numerically,
using the FAST model of the DTU 10 MW and of the scaled model
one, as well as sampled during the wind tunnel tests. Fig. 27 shows
the computed or acquired variation of thrust force at four meanwind speed condition:
1. Below Rated (
=V 2.33 m/s at wind tunnel scale);
2. Rated ( =V 3.67 m/s at wind tunnel scale);
3. Above Rated ( =V 5.33 m/s at wind tunnel scale);
4. Above Rated 2 ( =V 6.67 m/s at wind tunnel scale
Looking at the results it is possible to notice how the deriva-
tives of the thrust force are very similar between numerical modelof the full scale and model scale turbine as well as for the measuredwind tunnel response of the model turbine. This con firms the ef-
fective design approach for wind turbine rotors to be tested in wind
tunnel with the aim of investigating also the aerodynamics of
floating offshore wind turbines, where the unsteady phenomenon
become relevant, see ( Bayati et al., 2017b ;Bachynski et al., 2015 ).
6. Conclusions
The paper reported the design methodology for an aero-elastic
scale rotor of a 10 MW wind turbine. Theoretical and technicalaspects were discussed and the satisfactory experimental ver-ification of such a design was commented, revealing an effective
approach for this kind of application. The wind tunnel tests have
shown the very good aerodynamics performance, in particularconsidering that the main target of the design was the matching ofthe DTU 10 MW thrust curve. The velocity scale factor
λ=2V was
taken as reference for the aero-elastic design, and the related wind
tunnel results have shown excellent agreement with the target.Furthermore, also an other velocity scale factor,
λ=3V , was as-
sessed during the test, showing very good agreement as well, al-
though inevitably a bit worse than the initial target. However,
tuning the collective pitch of the blades, the objective of matching
the target thrust is reached, also for λ=3V . Furthermore, assuming
λ=3V test con figuration, also the firstflapwise natural frequency
of the blade, assessed experimentally through the reported modalanalysis, is consistent beside the non-quanti fiable variables in the
actual manufacturing process that can be hardly accounted in thedesign phase. Therefore, in these conditions, the scaled rotor canbe considered consistent in an aero-elastic sense. Furthermore, ithas been shown that good agreement between the target and theexperimental aerodynamic derivatives of the thrust force, makesthe model reliable in the dynamic response about a given dynamicstate (i.e. linearization) with important outcome in the investiga-tion of floating offshore wind turbine dynamics.
Acknowledgements
This project has received funding from the European Union's
Horizon 2020 research and innovation programme under grant
agreement No 640741.
References
Bachynski, E., Chabaud, V., Sauder, T., 2015. Real-time hybrid model testing of
floating wind turbines: sensitivity to limited actuation. Energy Procedia 80,
2–12.http://dx.doi.org/10.1016/j.egypro.2015.11.400 .
Bak, C., Bitsche, R., Yde, A., Kim, T., Hansen, M.H., Zahle, F., …, Behrens, T., 2012.
Light Rotor: the 10-MW reference wind turbine, Proceedings of EWEA 2012-European Wind Energy Conference. European Wind Energy Association(EWEA) .
Bak, C., et al., 2013. The DTU 10-MW Reference Wind Turbine. Technical University
of Denmark, DTU Wind Energy, Denmark .
Bayati, I., Belloli, M., Facchinetti, A., Giappino, S., 2013. Wind tunnel tests on floating
offshore wind turbines: a proposal for hardware-in-the-loop approach to va-
lidate numerical codes. Wind Eng. 37 (6), 557 –568. http://dx.doi.org/10.1260/
0309-524X.37.6.557 .
Bayati, I., Belloli, M., Ferrari, D., Fossati, F., Giberti, H., 2014. Design of a 6-DoF ro-
botic platform for wind tunnel tests of floating wind turbines. Energy Procedia .
http://dx.doi.org/10.1016/j.egypro.2014.07.240 .
Bayati, I., Belloli, M., Bernini, L., Zasso, A., October 2016a. ”Wind tunnel validation of
AeroDyn, within LIFES50 țproject: imposed Surge and Pitch tests. J. Phys. Conf.
Ser. 753 (5) . http://dx.doi.org/10.1088/1742-6596/753/9/092001 .
Bayati, I., Belloli, M., Bernini, L., Fiore, E., Giberti, H., Zasso, A., October 2016b. On the
functional design of the DTU10 MW wind turbine scale model of LIFES50 ț
project. J. Phys. Conf. Ser. 753 (5) . http://dx.doi.org/10.1088/1742-6596/753/5/
052018 .
Bayati, I., Belloli, M., Bernini, L., Giberti, H., Zasso, A., 2017a. On the scale model
technology for floating offshore wind turbines. IET Renew. Power Gener. . http:
//dx.doi.org/10.1049/iet-rpg.2016.0956 , IET Digital Library
Bayati, I., Belloli, M., Bernini, L., Zasso, A., 2017b. A formulation for the unsteady
aerodynamics of floating wind turbines, with focus on the global system dy-
namics, 36th International Conference on Ocean, Offshore and Arctic En-gineering, Trondheim (Norway), OMAE2017 –61925 .
Bottasso, C.L., Campagnolo, F., Petrovi
ć, V., 2014. Wind tunnel testing of scaled wind
turbine models: beyond aerodynamics. J. Wind Eng. Ind. Aerodyn. 127, 11 –28.
http://dx.doi.org/10.1016/j.jweia.2014.01.009 , ISSN 0167-6105.
Bredmose, H., Mikkelsen, R., Hansen, A.M., Laugesen, R., Heilskov, N., Jensen, B.,
Kirkegaard, J., 2015. Experimental study of the DTU 10 MW wind turbine on aTLPfloater in waves and wind, EWEA Offshore 2015 Conference, Copenhagen .
Bredmose, H., 2014. Contribution to InnWind Deliverable 4.22, Scaling Laws for
Floating Wind Turbine Testing. DTU Wind Energy .
Campagnolo, F., Bottasso, C., Bettini, P., 2014. Design, manufacturing and char-
acterization of aero-elastically scaled wind turbine blades for testing active andpassive load alleviation techniques within a ABL wind tunnel, The Science ofMaking Torque from Wind .
Cheli, F., Diana, G., 2015. Advanced Dynamics of Mechanical Systems. Springer .
H2020 Lifes50 țProject of ficial website: http://lifes50plus.eu .
Hansen, M.O.L., 2008. Aerodynamics of Wind Turbines, second ed. Earthscan Pub-
lications Ltd .
Jonkman, J., Butter field, S., Musial, W., Scott, G., 2009. De finition of a 5-MW Re-
ference Wind Turbine for Offshore System Development Technical ReportNREL/TP-500 –38060. NREL National Renewable Energy Laboratory .
Lyon, C.A., Broeren, A.P., Gigure, P., Gopalarathnam, A., Selig, M.S., 1998. Summary of
Low-speed Airfoil Data vol. 3. SoarTech Publications, Virginia Beach, VA .
Make, M., 2014. Predicting Scale Effects on Floating Offshore Wind Turbines MsC
thesis. TUDelft .
Manwell, J., McGowan, J., Rogers, A., 2009. Wind Energy Explained: Theory, Design
and Application, second ed. John Wiley & Sons, Ltd .
Moriarty, P.J., Hansen, C., 2005. AeroDyn Theory Manual vol. 15. National Renewable
Energy Laboratory NREL, p. 50036313, no. January .
Zasso, A., Giappino, S., Muggiasca, S., Rosa, L., 2005. Optimization of the boundary
layer characteristics simulated at Politecnico di Milano Boundary Layer WindTunnel in a wide scale ratio ranges. IRIS Politecnico di Milano .Fig. 27. Thrust variation due to wind speed variation. ∘- wind tunnel; *- DTU
10 MW FAST model; □- scaled rotor FAST model; dashed line – reference DTU
10 MW steady thrust.I. Bayati et al. / J. Wind Eng. Ind. Aerodyn. 167 (2017) 217 –227 227