Proceedings of ASME Turbo Expo 2012

GT2012
June 11-15, 2012, Copenhagen, Denmark

GT2012-69197

VORTEX GENERATORS FOR WIND TURBINE BLADES: A COMBINED WIND

TUNNEL AND WIND TURBINE PARAMETRIC STUDY

Hanns Mueller-Vahl Georgios Pechlivanoglou

Hermann Foettinger Institute Hermann Foettinger Institute
TU Berlin TU Berlin
present address: Mueller-Breslau Str. 8
Faculty of Mechanical Engineering D-10623, Berlin
Technion - Israel Institute of Technology Germany
Haifa 32000 Email: george@pechlivanoglou.com
Israel
Email: hannsmv@tx.technion.ac.il

C.N. Nayeri C.O. Paschereit

Hermann Foettinger Institute Hermann Foettinger Institute
TU Berlin TU Berlin
Mueller-Breslau Str. 8 Mueller-Breslau Str. 8
D-10623, Berlin D-10623, Berlin
Germany Germany
Email: christian.nayeri@tu-berlin.de Email: oliver.paschereit@tu-berlin.de

ABSTRACT was found to be in the range of x/c = 15% - 20%, where a

Vortex generators (VGs) are passive flow control devices comparatively low parasitic drag and a smooth post-stall lift
commonly employed to prevent flow separation on wind turbine curve were achieved. Particle Image Velocimetry measurements
blades. They mitigate the damaging fatigue loads resulting were conducted at various chordwise positions to provide insight
from stall while increasing lift and consequently lead to rotor into the interaction between adjacent streamwise vortices.
torque increase. This work summarizes a research project
aimed at optimizing the sectional as well as the full rotor-blade The experimental aerodynamic performance curves of the
aerodynamics using VGs. optimal VG configuration were used to project their effect on
wind turbine blade aerodynamics. Three different rotorblades
The effects of chordwise position, spanwise spacing and were designed and several stall and pitch regulated wind turbine
VG size were studied with force balance measurements of a models were simulated by means of a Blade Element Momentum
2D wing section. Reducing the distance between adjacent VGs (BEM) code (QBlade) developed by Smart Blade GmbH. The
produced large increases in the static stall angle and maximum performance of the rotorblades with and without VGs was
lift, but also resulted in a significant increase in drag as well simulated in order to assess their effect on the aerodynamic
as sharp lift excursions at angles exceeding the static stall performance and loads. Finally, previously measured steady
angle. The optimal chordwise position of the vortex generators state performance curves under high-roughness conditions were

1 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 used to simulate the detrimental effect of surface roughness on the turbine blades, an increase in wind speed leads to a higher
the performance of the aforementioned rotorblades. This allows angle of attack. This is particularly problematic in the region
for an estimate of the potential of the VGs to be employed as near the hub, where the Reynolds numbers are relatively low
retrofit elements for performance recovery of blades with a due to low wind speeds [1]. Since the blade structure has to
contaminated surface. absorb large bending moments there, the airfoil thickness is
comparatively high, which is not ideal for low Reynolds number
flows. Therefore, flow separation is a common problem in this
region, while the outer parts of the blades are less affected [2].
NOMENCLATURE
Vortex generators (VGs) are passive flow control devices
which are commonly employed on airfoils to prevent flow
a Wing span [m] separation. There is a rich variety of different designs, but all
AoA Angle of attack [◦ ] of them rely on generating streamwise vortices that transfer
c Wing chord [m] high momentum fluid from the freestream to the slower moving
Cd Two-dimensional drag coefficient boundary layer [3]. By energizing the retarded fluid, VGs
Cl Two-dimensional lift coefficient help overcome the adverse pressure gradients experienced at
increasing angles of attack [4, 5]. As the use of vortex generators
Cl,max Maximum two-dimensional lift coefficient
inevitably results in a drag penalty, some research was recently
Cl /Cd Two-dimensional lift - drag ratio (airfoil efficiency) directed at examining smaller, less intrusive designs, so-called
H Vortex generator height [m] sub boundary-layer VGs [3, 6–8]. However, their use is limited
L Chordwise length of vortex generator vanes [m] to applications where the flow-separation locations are relatively
Re Reynolds number fixed [8], which is not the case on wind turbine blades. Thus,
S Vortex generator span [m] it is not feasible to employ such devices on wind turbines,
u, v, w Velocity components [m/s] even though they may have some merit for improving the stall
x Chordwise position on airfoil, behavior. Nevertheless, minimizing the parasitic drag associated
with VGs is a crucial objective.
x = 0 at the leading-edge [m]
y Distance from airfoil surface [m]
The results of a research project carried out by SmartBlade
YΓ Average distance of the vortex centroids from the GmbH aimed at designing optimized VGs for wind turbine
airfoil surface [m] applications are presented herein. The project comprised several
z Spanwise positon, z = 0 at mid-span [m] wind tunnel tests as well as numerical investigations. Multiple
Z Spanwise spacing between adjacent VGS [m] geometric parameters were varied to provide an experimental
β Angle of incidence of VG vanes [◦ ] database. Force measurements were carried out with two
δ Boundary layer thickness [m] different VG models, examining the influence of the chordwise
ωx Streamwise vorticity [1/s] position and spanwise spacing between neighboring VGs.
Particle image velocity measurements were conducted to gain
ωxmax Maximum streamwise vorticity obtained for
qualitative insight into the effects of the interaction between
the respective configuration [1/s] adjacent vortices.

All VG configurations examined produced a significant in-

crease in the static stall angle and Cl,max compared to a clean
blade. The experimental data reveals a strong dependence of the
INTRODUCTION vortex generators’ performance depending on their spacing. It
Stall is a phenomenon observed on airfoils which are also supports the assumption that decreasing the spanwise spac-
pitched beyond a certain angle of attack. Adverse streamwise ing between neighboring vortex generators leads to an increased
pressure gradients cause a separation of the flow from the distance of the vortices from the airfoil surface at chordwise loca-
suction side surface, yielding reduced lift and increased drag [1]. tions further downstream. Nevertheless, VGs also cause a certain
Stall is avoided on modern multi-megawatt wind turbines since drag increase due to the conversion of the flow energy into vor-
it impairs their performance and results in fatigue loads that ticity. The shape and positioning of the VGs plays a significant
decrease the turbines’ lifetime. The temporal variations in role to the final drag penalty of each configuation. The optimal
angle of attack are mainly caused by the unsteadiness of the values for the spanwise spacing and the chordwise position are
wind approaching the turbines. For a fixed rotational speed of indicated.

2 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TABLE 1. DIMENSIONS OF THE VORTEX GENERATORS

VG configuration VG1 VG2

height (H) 10 mm 20 mm
length (L) 20 mm 40 mm
span (S) 30 mm 60 mm
β 18◦ 18◦
fin thickness 0.8 mm 0.8 mm

EXPERIMENTAL SETUP FIGURE 1. SCHEMATIC OF VORTEX GENERATOR GEOME-

Vortex Generators TRY
A schematic of the counter-rotating common flow down
vortex generators investigated here is presented in Fig. 1. The
geometry was based on a design used at the Technical University

Boundary Layer Displacement Thickness [mm]

of Denmark [2]. The streamwise vortices produced by each VG
induce a downward velocity component on each other, while
the effect of neighboring VGs is to move the vortices away
from the airfoil surface. The main advantage of counter-rotating
configurations lies in a higher efficiency in momentum transfer.
However, a motion of the vortices from the wall has been
reported [7, 9], which is commonly not the case in a co-rotating
system [10]. Two different VG sizes were investigated; VG2
(H = 3.3%c) is similar to VG1 (H = 1.7%c) in geometry but
twice as large in all dimensions. The geometric parameters of
the respective VG configurations are listed in Tab. 1.

No experimental data is available for the boundary layer

thickness distribution in this particular configuration. For this
reason it was decided to perform 2D airfoil simulations with the
2D panel method code XFOIL [11]. XFOIL is a viscous-inviscid FIGURE 2. BOUNDARY LAYER DISPLACEMENT THICKNESS
2D potential flow simulation code coupled with boundary layer (SUCTION SIDE) ESTIMATES DERIVED WITH XFOIL
equations. It is able to predict the boundary layer behavior at
pre-stall conditions with very high accuracy for low and medium
Reynolds Number flows. The XFOIL simulations produced leads to the onset of flow separation. Therefore, choosing a
values of δ ≈ 0.3 mm at x/c = 0.05 and δ ≈ 0.6 mm at x/c = 0.2 sufficiently small spacing is essential to attain a large static stall
(see Fig. 2). At the trailing edge, the boundary layer thickness angle. On the other hand, positioning the VGs too close together
reaches values of approximately δ ≈ 15 mm. The height of also produces adverse effects, as the mutual interaction between
the vortex generators was selected in such a way that they neighboring vortices due to their induced velocity causes them
protrude from the boundary layer, which ensures reliable stall to move further away from the airfoil surface, reducing their
suppression even for sudden changes in AoA with a varying effectiveness. Preventing the vortices from rising too rapidly is
location of separation. The selection of the height of the VGs is desirable in order to transport high-energy air to the regions deep
always a trade-off between lift increase and drag penalty. within the boundary layer [6]. Spacing the vortices too closely
has also been shown to be responsible for a quick decay of
In this study, the spanwise spacing Z between adjacent vorticity [3,6,7]. Furthermore, the increase in drag caused by the
vortex generators was given particular attention, as it is a key presence of the VGs can be expected to be roughly proportional
parameter influencing both the static stall angle and the drag to their number per unit span. Hence, an optimal value of Z
coefficient. When the vanes are too far apart, the boundary layer corresponding to a compromise between the aforementioned
in the intermediate region is not energized effectively, which effects was expected to exist. Determining this optimal value

3 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 based on the static stall angle and the coefficients of lift and drag
was one of the main goals of this study.

The angle of incidence β also has a significant impact on

the vortex generators’ performance. If it is too small, the VGs
do not generate sufficiently strong vortices, and are therefore
ineffective at stall control. Excessive vane deflections, on the
other hand, lead to vortex breakdown, which also decreases
the control authority and produces excessive drag [12]. An
angle of 18◦ was selected here, which corresponds to that used
by Velte et al. [2] and also represents the optimum value for
FIGURE 3. VORTEX GENERATOR CONFIGURATION 1 AT-
sub boundary-layer VGs (h/δ = 0.2) observed by Godard and
TACHED TO THE AIRFOIL AT x=20%c
Stanislas [10].

The vortex generators were attached to the blade with adhe-

sive tape to allow for fast and simple changes of the configura-
tion. For the force measurements, they were positioned across
the entire span of the airfoil as shown in Fig. 3. Templates were
used to ensure equal spacing between adjacent VGs.

Wind Tunnel
All measurements were carried out in the wind tunnel facil-
ity of ISTA, TU Berlin. The closed loop wind tunnel is powered
by a 500 kW speed regulated DC motor. The test section has a
length of 10m, a cross-sectional area of 2 × 1.4 m2 and a contrac-
tion ratio of 6.25 : 1 (see Fig. 4). The turbulence level of the free FIGURE 4. SCHEMATIC OF THE AIRFOIL TEST SECTION AT
stream is below 0.5% in the test section. The investigated airfoil TU BERLIN
was a NACA 63(3) -618 with a chord length of 600 mm. It was
connected to an external 6-component balance, which is discon-
Particle Image Velocimetry
nected from the test section. Splitter walls were installed to min-
Particle image velocimetry measurements were carried out
imize the influence of boundary layer effects with the wing ex-
at three chordwise positions (x/c=40%, 60% and 80%) in planes
tending over the entire span (a=1.554m) between the walls. The
oriented perpendicular to the airfoil surface. A Quantel ‘Big Sky
gaps (≈ 0.5 mm) at the sides were covered with adhesive tape to
Laser’ operating at a wavelength of 532 nm was employed, gen-
reduce the pressure losses at the tips of the airfoil. Since the tape
erating 170 mJ pulses with a duration of 8 ns. A PCO ‘Sensicam’
was not in contact with the splitter walls, the force measurements
CCD camera with a resolution of 12 Bit was mounted on the
were not impaired. All tests were carried out at a wind speed of
wind tunnel floor 1500 mm downstream of the trailing-edge at a
29 ms . The resulting Reynolds number based on the chord length
height of ≈ 250 mm. Its spanwise position corresponded to that
was Re ≈ 1.1 · 106 . of the vortex generator at the center of the test section, which was
The use of a single airfoil naturally somewhat limits the vol- also the spanwise center of the PIV interrogation region. The
ume of results since modern wind turbine blades use a much camera was equipped with a Nikon Nikkor HC lens ( f = 50 mm).
larger number of different airfoil sections along their span. The
overall performance however of the VGs as well as their behav- The streamwise vortices generated by the VGs are located
ior in terms of lift and drag was at the focus of this research effort close to the boundary layer. Therefore, reflections of the laser
and therefore parametric investigations with other airfoil sections light from the airfoil surface would have significantly impaired
are out of the scope of the current paper. Finally the currently se- the quality of the data in the region under investigation. In order
lected airfoil is a quite common mid-span and out-board airfoil to minimize the resulting error, the position of the camera was
section used on several wind turbine designs since the late 90’s adjusted in such a way that the reflections were not visible. As
and thus the performance results of this airfoil with several VG a consequence of this, no velocity data is available for the lower
configurations are relevant for the wind turbine industry. boundary layer region, since this part of the laser light sheet was

4 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 covered by the trailing edge of the airfoil. employed to eliminate outliers.

The sets of velocity fields acquired for each configuration

EXPERIMENTAL PROCEDURE were averaged to determine the mean velocity field. The mean
Force Measurements vorticity distribution was then derived by numerically calculating
The force measurements constitute the main part of this
experimental study. Several parameters were varied to identify ∂ ∂
ωx (~x) = w(~x) − v(~x) (1)
the optimum configuration. Spanwise spacings of Z = 3H, 5H ∂y ∂z
and 7H, were investigated with the VGs positioned at x/c = 20%.
The spacing Z is measured at the downstream side of the VGs
from the mean velocity profiles. The average distance of the vor-
where the vortices are formed. Furthermore, various chordwise
ticity centroids from the airfoil surface YΓ can be obtained as
positions in the range of x/c = 5% - 30% were examined for the
follows [13]:
first, smaller VG design.
R
The airfoil was pitched from AoA = −10◦ to 30◦ with an in- |ωx (~x)| · y dA
YΓ = R . (2)
crement of 0.5◦ . At each angle of attack, a set of 20,000 samples |ωx (~x)| dA
was taken at a sample rate of 10 kHz and subsequently averaged.
The airfoil was then pitched back to smaller angles of attack This term was numerically calculated in the region of the flow
until the measured aerodynamic coefficients corresponded pre- field fully comprising the vortices. The integration limits were
cisely to the ones previously recorded at the same angle, which chosen in such a way that the regions with |ωx | > |ωxmax | · 5%
was always the case below the static stall angle, where the flow were included.
is fully attached on the suction side. This procedure was fol-
lowed throughout all measured configurations to detect possible
hysteresis effects caused by the presence of the vortex genera- BEM SIMULATION
tors. For the sake of clarity, most of the plots presented here only The BEM Approach
show the data obtained while pitching up the airfoil. The transition from section-wise steady state performance
characteristics to wind turbine performance characteristics
Particle Image Velocimetry is achieved with the use of the Blade Element Momentum
Method [14]. This is currently the standard wind turbine
For the measurements of the flow field, the VGs were
simulation methodology for industrial and academic purposes.
attached at x/c = 20% with spacings of Z=3H, 5H and 7H. The
The BEM simulation theory utilizes the steady state lift and drag
PIV measurements were carried out at an angle of attack of
performance curves of individual airfoils and combines them
7◦ , where the blades would typically be operated in the case
in order to compute the overall aerodynamic performance of
of no control. However, when vortex generators are used, the
a rotorblade consisting of the airfoils. Engineering correction
stall angle is increased, and the optimum AoA would be larger.
algorithms such as the tip and root corrections proposed by
This difference in optimum angle of attack is illustrated with the
Prandtl and Shen [14, 15], the highly loaded rotor corrections
force measurement results.
proposed by Glauert and modified by Buhl [14, 16] and the
”Himmelskamp effect” [17] correction proposed by Snel [18]
For each vortex generator configuration, 600 image pairs
are applied in order to correct simulation inaccuracies.
were recorded at a sample rate of 5 Hz with a pulse separation
of 5 µ s. Assuming that the measured velocity fields can be
For the current research project the simulation code
regarded as statistically independent at this sample rate [2], this
QBlade [19] is used for the investigation of the VG performance
number of samples is more than sufficient to extract the time
on wind turbine rotorblades. It has been successfully validated
mean velocity field. The program ‘VidPiv’ by ILA was used for
against other well established BEM codes [19].
the PIV data processing, and a multipass scheme was adopted
to maximize the quality of the data. At first, a grid of 24 by
24 pixels was analyzed using standard correlation. This was
followed by an adaptive correlation algorithm that uses an offset Simulation Methodology
between the base and the cross interrogation areas, once again In order to identify the turbine performance variations
examining a grid of 24 by 24 pixels with an overlap of 50 % brought about by the implementation of VGs, a set of rotor-
between grid cells. This is equivalent to a separation of roughly blades was designed and simulated. The goal was to examine the
2.5 mm in both y and z directions. Adaptive local filters were effect of the optimized VG configuration on small stall blades

5 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TABLE 2. DESIGN CHARACTERISTICS OF TEST ROTOR- extrapolation schemes (Fig. 5).
BLADE DESIGNS
The Reynolds number values during the wind tunnel tests
Blade 1 2 3 were much lower than the ones achieved by the same airfoil sec-
tion when used on large multi MW wind turbine blades, but still
Design λ 6.0 6.5 7.0
high enough in order to assume a post-critical Reynolds Number
Max. Twist [o ] 26.5 26.5 24.0 performance. Often Reynolds number corrections are applied
on wind tunnel measurements in order to bring the experimental
Length [m] 4.0 20.0 40.0
values closer to the operation regime of actual rotorblade airfoil
Nom. Power [kW] 10 500 2000 sections. The effect of VGs and roughness however with respect
to Reynolds number variation is generally beyond the scope of
Max. Cp 0.42 0.41 0.45
such correction models. For this reason and since the current
blade designs are meant to be comparative theoretical cases, all
the airfoil sections of all the blades used the same set of perfor-
mance polars. In this way the airfoil section performance is a
”constant” among blade designs.




Parametric investigations were performed with the afore-
 mentioned blade designs which include several different types of
wind turbine control methods (e.g. Stall regulation, fixed speed
 & collective blade pitch, variable speed & collective blade pitch
e.t.c). For space economy, only the most significant results are
 presented in this document.


RESULTS
 Force Measurements
Different Vortex Generator Geometries. The per-
 formance of the different vortex generator sizes is presented in
Fig. 6. In both cases, the VGs were positioned at x = 20% c with
 a spacing of Z = 5H. The data acquired during the pitch-down
      
A
A A  motion is represented by plus signs for AoA ≤ 15◦ for the con-
figurations with VGs and for AoA ≤ -10◦ for the baseline. Some
FIGURE 5. EXTRAPOLATED LIFT CURVE FOR THE COM- variation of lift and drag was observed at angles of attack beyond
PLETE RANGE OF ANGLES OF ATTACK WITH AND WITHOUT the stall angle, but these variations appeared random, and no
VORTEX GENERATORS considerable hysteresis behavior was observed. The deviations
were thus attributed to stochastic unsteady phenomena linked to
the flow separation on the suction side.
as well as large pitch rotorblades. Three different rotorblade
designs were generated and their performance with and without Both vortex generator designs significantly increase the
VGs was simulated. Table 2 presents the main design charac- static stall angle, as can be seen in Fig. 6(a). The best perfor-
teristics of the respective blade designs. All the blades were mance is achieved with VG1, which yields a maximum value
designed with the NACA 63(3) -618 as the airfoil section. This of the lift coefficient Cl,max = 1.91 at an angle of attack of 20.5◦ .
was done to allow for a direct comparison of the performance This represents an increase by 28% in the lift coefficient and an
variations due to the existense of VGs on various blades. increase of 6.5◦ in the static stall angle compared to the baseline
Naturally, actual rotorblade designs are based on a large variety case, where Cl,max = 1.50 is reached at AoA = 14◦ . VG2 produced
of airfoil sections, but the focus of this study is to demonstrate a maximum lift of Cl,max = 1.79 at AoA = 18.5◦ . Predictably,
the general effects of VGs on different rotorblades rather than VG2 caused a much larger increase in drag (see Fig. 6(b)).
to investigate their effect on an actual blade design. The blade
chord distributions follow the ”Schmitz” principle [20, 21]. The finding that the static stall angle achieved with VG1 is
The measured wind tunnel polar curves (clean, rough and significantly larger than that obtained for VG2 is quite stunning.
VG configurations) were extrapolated to 360 degree angle of The main reason why the larger geometry was expected to per-
attack curves based on the Viterna [14] and Montgomerie [22] form less well from an overall perspective was the large degree

6 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 of parasitic drag, whereas it was not expected to be considerably by the position of the VGs. The chordwise position does,
less effective at stall suppression. The PIV results show that VG2 however, play a crucial role for the post-stall behavior of the
generates far stronger vortices because of its larger dimensions, airfoil. Positioning the vortex generators too far downstream
but this does not pay off in terms of separation control perfor- (x = 30% c) results in a very sudden decrease of Cl . As described
mance. The distance of the vortices from the airfoil surface ap- above, this behavior has to be avoided to ensure a long lifetime
parently plays an important role in this context, as it impacts the of the turbine blades. For the other configurations, the post-stall
ability of the vortices to energize the lower regions of the bound- characteristics of Cl are quite similar. Positioning the VGs close
ary layer. This aspect will be discussed further on the basis of to the leading-edge tends to attenuate the drop in lift experienced
the PIV results. beyond the stall angle.

Another difference lies in the aerodynamic performance

Spanwise Spacing. The effect of varying the spanwise
at angles of attack in the range of AoA = 5◦ - 15◦ . Positioning
spacing is presented in Figures 7 for the first vortex generator
the VGs further downstream leads to a larger lift coefficient as
geometry. A similar comparison was carried out with the second
well as less drag, see Figures 9(a) and 9(b). The drag reduction
VG design (Figures 8). In both cases, the vortex generators were
can be interpreted as a consequence of a decrease in surface
placed at x = 20%c. The results for the different VG sizes are
friction. These effects cause a considerable difference in airfoil
very similar: The smallest spacing examined generated by far
performance (Fig. 9(c)) at angles of attack where a NACA
the largest increase in the static stall angle, but also caused the
63 3 618 turbine blade is commonly operated. In this respect,
largest increase in Cd . In the case of VG1, stall occurs at 26.5◦
it is preferable to position the VGs further downstream to
with Z = 3H, which is the largest stall angle achieved with any of
maximize the power output of the turbine.
the configurations examined. This corresponds to an increase by
12.5◦ compared to the baseline value. With Cl,max = 1.96, which
Judging from these results, the optimal position of the vortex
is equivalent to a relative increase of 31%, this configuration also
generators appears to be in the range of x = 15 - 20%c, as the best
yields a particularly large value of the maximum lift coefficient.
compromise between performance gain and post-stall behavior
However, an especially sharp drop in lift is observed at angles
is reached here.
exceeding 26.5◦ . This more abrupt onset of stall caused by the
presence of vortex generators is a known phenomenon [23]. On
a wind turbine, it could result in severe, abrupt load excursions,
which would harm the blades and components of the drive train
and therefore have to be avoided. With larger values of Z, the
drop is far less severe.

The data attained for VG2 show a similar trend, see Fig.
8(a). The static stall angles measured with these larger vortex
generators were 23.5◦ (Z = 3H) and 18.5◦ (Z = 5H) respectively.
In the case of Z = 7H, a rather smooth transition is observed,
and the stall angle, which is slightly lower than in the case
of Z = 5H, can not be defined precisely. With the smallest
spacing, large fluctuations in lift are observed at angles exceed-
ing 23◦ . This once again indicates that such small spanwise
spacings bear the risk of abrupt load excursions at high angles of
attack as a result of abrupt flow separation on the suction surface.

The increase in drag observed at small spacings and the re-

sulting decrease in performance over the linear region are mainly
a result of the larger number of VGs attached to the airfoil surface
in these configurations.

Chordwise Vortex Generator Position. The force

measurement results attained for the first vortex generator
configuration with (Z = 5H) at various chordwise positions are
presented in Figures 9. The static stall angle is hardly affected

7 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 2 2

1.5 1.5

1 1
Cl

Cl
0.5 0.5

0 0 Baseline
Baseline VG1, Z=3H, 20%c
VG1, Z=5H, 20%c VG1, Z=5H, 20%c
VG2, Z=5H, 20%c VG1, Z=7H, 20%c
−0.5 −0.5
−10 −5 0 5 10 15 20 25 30 −10 −5 0 5 10 15 20 25 30
AoA [°] AoA [°]
(a) (a)

2 2

1.5 1.5

1 1
Cl

Cl

0.5 0.5

0 0 Baseline
Baseline VG1, Z=3H, 20%c
VG1, Z=5H, 20%c VG1, Z=5H, 20%c
VG2, Z=5H, 20%c VG1, Z=7H, 20%c
−0.5 −0.5
0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.05 0.1 0.15 0.2 0.25 0.3
Cd Cd

(b) (b)

50 50

40 40

30 30

20 20
Cl / Cd

d
C /C

10 10
l

0 0

−10 −10
Baseline
Baseline VG1, Z=3H, 20%c
−20 −20
VG1, Z=5H, 20%c VG1, Z=5H, 20%c
VG2, Z=5H, 20%c VG1, Z=7H, 20%c
−30 −30
−10 −5 0 5 10 15 20 25 30 −10 −5 0 5 10 15 20 25 30
AoA [°] AoA [°]
(c) (c)

FIGURE 6. FORCE MEASUREMENT RESULTS FOR VARIOUS FIGURE 7. IMPACT OF THE SPANWISE SPACING BETWEEN
VORTEX GENERATOR GEOMETRIES, ATTACHED AT x = 20%c ADJACENT VORTEX GENERATORS, VG DESIGN 1, LOCATED
WITH A SPACING OF Z = 5H AT x = 20%c

8 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 2
2

1.5
1.5

1
1
Cl

Cl
0.5
0.5
Baseline
0 Baseline VG1, Z=5H, 5%c
VG2, Z=3H, 20%c 0 VG1, Z=5H, 10%c
VG2, Z=5H, 20%c VG1, Z=5H, 15%c
VG2, Z=7H, 20%c VG1, Z=5H, 20%c
−0.5 VG1, Z=5H, 30%c
−10 −5 0 5 10 15 20 25 30 −0.5
AoA [°] −10 −5 0 5 10 15 20 25 30
AoA [°]
(a)
(a)
2
2

1.5
1.5

1
1
Cl

Cl

0.5
0.5
Baseline
0 Baseline VG1, Z=5H, 5%c
VG2, Z=3H, 20%c 0 VG1, Z=5H, 10%c
VG2, Z=5H, 20%c VG1, Z=5H, 15%c
VG2, Z=7H, 20%c VG1, Z=5H, 20%c
−0.5 VG1, Z=5H, 30%c
0 0.05 0.1 0.15 0.2 0.25 0.3 −0.5
Cd 0 0.05 0.1 0.15 0.2 0.25 0.3
Cd
(b)
(b)
50
50
40
40
30
30
20
20
Cl / Cd

d
C /C

10
10
l

0
0
Baseline
−10
Baseline −10 VG1, Z=5H, 5%c
VG2, Z=3H, 20%c VG1, Z=5H, 10%c
−20 VG1, Z=5H, 15%c
VG2, Z=5H, 20%c −20
VG2, Z=7H, 20%c VG1, Z=5H, 20%c
−30 VG1, Z=5H, 30%c
−10 −5 0 5 10 15 20 25 30 −30
AoA [°] −10 −5 0 5 10 15 20 25 30
AoA [°]
(c)
(c)
FIGURE 8. IMPACT OF THE SPANWISE SPACING BETWEEN
FIGURE 9. VG DESIGN 1 WITH A SPANWISE SPACING OF
ADJADCENT VORTEX GENERATORS, VG DESIGN 2, LOCATED
Z = 5H, ATTACHED AT VARIOUS CHORDWISE POSITIONS
AT x = 20%c

9 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 ωx
Particle Image Velocimetry results 0.15
50
Examples of mean vorticity profiles measured at x = 80%c 0.12

y [mm]
0.09

are presented in Fig. 10 (VG1) and Fig. 11 (VG2) for various 25

0.06
0.03

spanwise spacings. The pairs of counter-rotating streamwise -0.03

-0.06

vortices are reproduced well with this experimental technique. 0 -100 -50 0 50 100
-0.09
-0.12

The fact that the vortices are located at rather distinct positions z [mm] -0.15

in the mean velocity field and do not cover large areas even at (a) VG 1 with a spanwise spacing of Z=3H, located at 20%c
x=80% c shows that they are not subjected to a large degree of
ωx
jitter. 0.15
50 0.12

y [mm]
0.09
0.06
At this chordwise position, no data is available for the 25 0.03
-0.03
region below y ≈ 8 mm, as indicated by dashed lines. As a -0.06
-0.09
consequence of this, some vortices located close to the airfoil 0 -100 -50 0 50 100 -0.12
z [mm] -0.15
surface are ‘cut off’ at the bottom. For the identification of
the mean distance based on Eqn. 2, this has to be accounted (b) VG 1 with a spanwise spacing of Z=5H, located at 20%c
for. To avoid a biased data evaluation, the region including the
ωx
streamwise vortices was cut symmetrically on both sides in the 50
0.15
0.12
y-direction where necessary.

y [mm]
0.09
0.06
25 0.03
-0.03
The vorticity plots presented in Figures 10 and 11 contain -0.06
-0.09
some interesting qualitative information on the position and 0 -100 -50 0 50 100 -0.12
z [mm] -0.15
strength of the vortices for different configurations. For obvious
reasons, the smaller VGs generate smaller vortices with a lower (c) VG 1 with a spanwise spacing of Z=7H, located at 20%c
maximum strength: Those produced with VG1 are roughly half
FIGURE 10. MEAN VORTICITY FIELDS GENERATED BY VG
as large in both spanwise and vertical directions compared to
GEOMETRY 1, MEASURED AT 80%c AT AN ANGLE OF ATTACK
those obtained with VG2. On the average, the absolute value of
OF AoA = 7◦
peak vorticity is approximately 70% of that generated with the
larger VGs.

The pairs of vortices located at z ≈ -60 mm and z ≈ -10 mm governing the degree of their destructive interference [6, 7, 24].
in Fig. 10 affiliate with the same physical vortex generator, However, no such effect is apparent from the vorticity profiles.
which was not moved between the measurements. The same is Since no data is available close to the airfoil surface, integrating
true for the vortices located at z ≈ -120 mm and z ≈ -40 mm in the vorticity to yield the average circulation as a function of the
Fig. 11. One main effect of spacing the VGs further apart is spacing is not feasible. In any case, as no considerable effect is
that the interjacent gaps are extended. Since the VGs produce an observable, the variation in vorticity decay does not appear to be
adverse upwelling of the boundary layer in these regions, this is pronounced enough to have a significant impact on separation
clearly detrimental for the suppression of stall. At the same time, control performance.
the distance between pairs of vortices originating from the same
VG increases as they are convected downstream (remember that The curves showing the average distance YΓ of the vortices
the spacing of the trailing-edges of the vanes is S = 30 mm and from the airfoil surface presented in Fig. 12 indicate the same
S = 60 mm respectively, see Tab. 1). This spanwise deflection qualitative trends for both VG dimensions. However, the rela-
is probably caused by the presence of the airfoil surface. As a tionship of the VG height and YΓ is not linear. Close to the VGs,
result of this, the regions where high momentum fluid is supplied at x/c=40%, YΓ is approximately 11.5 mm for VG1 and 14.5 mm
to the boundary layer are stretched out further downstream, for VG2 regardless of the spanwise spacing. This corresponds
whereas the spanwise extent of the regions experiencing an to YΓ = 1.15 H (VG1) and YΓ = 0.73 H (VG2) respectively. This
upwelling of the boundary layer is slightly reduced. When the disparity is quite surprising, also because the boundary layer
spacing between adjacent VGs is decreased, the deflection is thickness was predicted to be far smaller than the device height
less pronounced. in this region. This observation may be explained in part with
the absence of reliable data close to the airfoil surface.
The impact of the vortex spacing on the decay of vorticity
is an other aspect worth mentioning. It is well known that With increasing streamwise distance, the vortices lift off
the distance between parallel vortices is an important factor from the surface, especially when the spacing between adjacent

10 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 vortex generators is comparatively small. In this respect, the ωx
0.15
50
results for different VG dimensions are in very good agreement. 0.12

y [mm]
0.09
The outward deflection can be interpreted as a result of the 25
0.06
0.03
induced velocity caused by neighboring vortices. With the -0.03
-0.06
first VG design and Z = 3H, the average distance from the 0 -100 -50 0 50 100
-0.09
-0.12
surface increases from YΓ ≈ 1.15 h at x = 40% c to YΓ ≈ 2.2 H z [mm] -0.15

at x = 80% c, which is an increase by 91%. The percentage (a) VG 2 with a spanwise spacing of Z=3H, located at 20%c
increase in YΓ is very similar for the second vortex generator
ωx
design with the same spacing: It grows from YΓ ≈ 0.74 H 0.15
50
(x = 40% c) to YΓ ≈ 1.44 H (x = 80% c), which corresponds 0.12

y [mm]
0.09

to a 95% increase. With a spacing of Z = 7H, the relative 25

0.06
0.03

increase in YΓ between the same chordwise positions is smaller: -0.03

-0.06

43% (VG1) and 64% (VG2) respectively. Somewhat similar 0 -100 -50 0 50 100
-0.09
-0.12

observations were made by Ashill et al. [6] and by Betterton z [mm] -0.15

et al. [7], who compared counterrotating sub boundary-layer (b) VG 2 with a spanwise spacing of Z=5H, located at 20%c
VGs with joint vanes to those with vanes that were spaced by 1H.
ωx
0.15
50 0.12
The distance of the streamwise vortices from the airfoil

y [mm]
0.09
0.06
surface was anticipated to have a noticeable effect on the overall 25 0.03
-0.03
separation control performance. It was surmised that config- -0.06
-0.09
urations with Z = 3H might suffer from reduced effectiveness 0 -100 -50 0 50 100 -0.12
z [mm] -0.15
resulting from the increased interaction between vortices. Their
larger distance from the surface was expected to impair their (c) VG 2 with a spanwise spacing of Z=7H, located at 20%c
ability to transport high momentum air to the retarded boundary
layer flow. However, the force measurements do not show FIGURE 11. MEAN VORTICITY FIELDS GENERATED BY VG
this effect. The data presented in Figures. 7 and 8 clearly GEOMETRY 2, MEASURED AT 80%c AT AN ANGLE OF ATTACK
demonstrates that the smallest spacing investigated produced by OF AoA = 7◦
far the largest increase in static stall angle and Cl,max . It appears
30
that the phenomenon of the vortices lifting off the surface is not VG 1, Z=3H
VG 1, Z=5H
pronounced enough to produce a noticeable negative effect. VG 1, Z=7H
VG 2, Z=3H
VG 2, Z=5H
These results indicate that in the range of values of Z 25 VG 2, Z=7H
considered here, a different mechanism is predominant: When
the spacing is too large, the regions between neighboring VGs
YΓ [mm]

are not energized sufficiently, leading to the onset of stall there.

This appears to be the reason why Z = 3H showed the best 20
performance at postponing stall. The spacing would presumably
have to be decreased below Z = 3H to demonstrate the adverse
effect caused by the induced velocity from neighboring vortices.
15

Nevertheless, the values of Z examined here cover the range

that is of interest for wind turbine applications. In order to pre-
vent large fatigue loads, a compromise has to be found between a 10
high static stall angle and a large value of Cl,max on one hand and 40 50 60 70 80
x/c [%]
a smooth lift curve on the other. Since very sharp drops in lift
were observed with Z = 3H, a smaller spacing is not expedient
for wind turbine blades. FIGURE 12. AVERAGE DISTANCE OF STREAMWISE VOR-
TICES FROM THE AIRFOIL SURFACE AT VARIOUS CHORD-
WISE POSITIONS AS A FUNCTION OF SPANWISE VG SPACING,
AoA = 7◦

11 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use




*
)
+
+
.
 
     ++,--
  ,












, -+/01-,0


23




 


 
FIGURE 13. TWIST DISTRIBUTIONS OF THE 3 TESTED

 
BLADES  





  


FIGURE 15. POWER COEFFICIENT OF 4 m RADIUS ROTOR
WITH AND WITHOUT VORTEX GENERATORS


!"#$!%&
'"(
production and the overall performance of the wind turbine.
In fact, the addition of VGs on small wind turbine rotorblades
has such a profound effect that they can be utilized for the
fine-tuning of the blade performance within a very wide range.
With a suitable placement of VGs, it is possible to precisely
adjust the blade power and load performance in order to match
the requirements of a wide variety of wind turbine systems.

FIGURE 14. POWER CURVE OF 10 kW TURBINE WITH 4 m The analysis of the aerodynamic efficiency of the rotor
BLADES WITH AND WITHOUT VORTEX GENERATORS equipped with the 4 m long blade reveals the effect of the VGs
on the overall performance (Fig. 15). The rotor efficiency is
significantly improved at low tip speed ratios (high wind regime)
BEM Simulation results
and reduced at high tip speed ratios (low speed regime). The
Small Size Stall Regulated HAWTs
reduction in the high tip speed ratio regime is mostly due to the
The 4 m long blade was designed for a 10 kW (8.6 m rotor additional drag generated by the VGs.
radius) stall regulated wind turbine. The blade has a maximum
twist of 26 degrees (Fig. 13) and operates at a range between 20
and 120 rpm. Its effective cut-in wind speed is 2.5 m/s and its It is worth noting once again that vortex generators as a
nominal wind speed is 10 m/s. The maximum power of 12 kW retrofit solution are considered as performance tuning elements
is achieved at a wind speed of 12 m/s. The 10 kW stall turbine and their number, size and position should always be assessed
was modeled and simulated with the QBlade BEM simulation according to the individual requirements. The current perfor-
code with and without VGs on its blades. Figure 14 shows mance plots represent the performance of blades fully equipped
the computed power curve of the turbine. The performance with VGs, which is often not the optimal configuration. In some
difference is quite dramatic, which was anticipated since the cases, for instance, VGs should be used only in the inner blade
presence of VGs significantly delays stall and increases lift. region in order to reduce/eliminate flow separation effects due to
insufficient blade twist without adding drag at the outer region.
As the power regulation of stall turbines is achieved by However, these individual cases are beyond the scope of the
means of flow separation, the delay of stall brought about by the current paper and are therefore not presented.
presence of the VGs is expected to significantly affect the power

12 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 46 57
9
6 58
556
Cp
0.5



55

0


 -0.5

565 5 6565:
;<8=>5
-1


-1.5
20m Blade "Clean"
 20m Blade "VGs"
-2 20m Blade "rough surface"

  -2.5
      0 5 10 15 20 25
Wind Speed [m/ s]

FIGURE 16. LIFT AND DRAG POLARS OF THE NACA 63(3) - FIGURE 17. CP CURVES FOR A 20 m ROTORBLADE IN
618 TEST WING WITH VORTEX GENERATORS AND LEADING ”CLEAN” AND ”ROUGH” SURFACE CONDITION AS WELL AS
EDGE ROUGHNESS WITH FULL SPAN VORTEX GENERATORS

low wind speed performance of the blade with full span VGs is
Medium Size Stall Regulated HAWTs
inevitably worse that the case of the ”clean” blade. However,
Larger rotorblades for medium sized stall regulated turbines the addition of VGs significantly modifies the performance
can also be significantly improved with the use of VGs. One of the turbines at higher wind speeds and allows for the fine
of the most significant issues of stall regulated wind turbines is tunning of blade designs as well as the ”partial correction” of
their sensitivity to blade contamination. Leading edge erosion, the surface roughness issues of the blades, at least to some extent.
sand accumulation or salt crust formation increase the blade
surface roughness with significant adverse effects on the power
performance and energy yield [25]. In order to simulate this
effect, the NACA 63(3) -618 wing was tested in the wind tunnel Large Pitch Regulated HAWTs
with a rough leading edge surface (Fig. 16). The rough surface Modern large utility scale wind turbines are equipped
was generated with a thin tape applied in the leading edge region with electromechanical pitch systems in order to adjust the
(10% c suction side to 10% c pressure side). The surface of this aerodynamic performance of their rotors. Even though these
tape was covered by fine sand grain particles with an average turbines theoretically do not experience flow separation, in
diameter of 0.15mm. From previous experiments of the authors, reality this is often the case. Due to manufacturing limitations,
this type of leading edge roughness leads to similar performance the structural twist is often small, thus forcing the inner blade
degradation with the NACA Standard Roughness and also with region to operate at high angle of attack. Furthermore, rotor-
the roughness level found on heavily contaminated wind turbine blade contamination is a significant power deficit factor for
rotorblades [25]. these turbines as well [25]. Therefore, the use of VGs as retrofit
solutions or as design-integrated elements is often necessary.
The 20 m long blade was designed for a 500 kW stall For this simulation, a fictitious 40 m long blade was designed
regulated wind turbine and it is based on a theoretical design for a 2 MW pitch regulated wind turbine. Once again, the blade
incorporating a single airfoil. This was done in order to in- was designed with a single airfoil (NACA 63(3) -618) in order to
vestigate the overall VG performance and is not intended to allow direct performance comparison based on the wind tunnel
replicate actual blade designs. The blade has a maximum twist measured polars. The blade has a maximum twist of 24 degrees
of 26◦ (Fig. 13) and operates in a range of 12 to 15 rpm. Its (Fig. 13) and operates in a range of 5 to 15 rpm with an effective
effective cut-in wind speed is 2.5 m/s and its nominal wind speed cut-in wind speed of 2.5 m/s and a nominal wind speed of 11 m/s
is 12 m/s. Figure 17 shows the adverse effect of blade surface (Fig. 18).
roughness on the efficiency of the said rotorblade. The use of
VGs on stall regulated wind turbine blade designs helps reduce Under normal operation at a moderate wind site (Weibull
the performance decrease over the lifetime of the turbine. The Parameters k = 2, A = 6.2), such a wind turbine produces ap-

13 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Power [W]

2.0x106

1.5x106

1.0x106

5.0x105
40m Blade "rough surface"
40m Blade "VGs"
40m Blade "Clean"

0
0 5 10 15 20 25
Wind Speed [m/ s]

PQN S Q [
FIGURE 18. POWER CURVES FOR A 40 m ROTORBLADE IN

]
FIGURE 19. Cp CURVES FOR A 40 m ROTORBLADE IN

L
”CLEAN” AND ”ROUGH” SURFACE CONDITIONS AS WELL AS

PL QN S Q \
”CLEAN” AND ”ROUGH” SURFACE CONDITIONS AS WELL AS

MH
AIK [
WITH VORTEX GENERATORS LOCATED AT THE INNER 1/3 OR

G
WITH VORTEX GENERATORS LOCATED AT THE INNER 1/3 OR

F
THE ROTORBLADE SPAN

@
THE ROTORBLADE SPAN

J
G PQN S Q [N
T
U
V
W
X
Y
proximately 3.31 MWh/year according to the power curve of

EF QN S N []
a ”clean” blade shown in Fig. 18. With a sell-back price of

AC?
 

@ \
0.092 Euro/kWh (sell-back price in Germany), this turbine

DB
A@OPQNNRNOQNQON^_`aOàbc
NZN[[
 

generates a revenue of roughly 305000 Euro per year. According


to simulations based on the measured polars of the NACA
63(3) -618 with leading edge roughness (Fig. 16 ) and previous 
field experiences and measurements [25], the same turbine with



contaminated rotorblades generates around 3.25 MWh/year, 
which corresponds to approximately 6000 Euro less revenue.

B
This does not take into account that the regions already operating
near stall (inner blade region) are in reality far more severely
affected by contamination. The implementation of VGs on such 
wind turbine blades is able to bring the energy production closer
to the ”as-designed” levels mostly by improving the performance 
of the inner blade region operating at high angles of attack and   
  ]
by preventing early stall at the outboard parts of the blade.
FIGURE 20. TOTAL ROOT BENDING MOMENT OF THE 40 m
The operational drawback in terms of aerodynamic loads ROTOR BLADE WITH AND WITHOUT VORTEX GENERATORS
when implementing VGs on a rotorblade is small since the load
increase on the rotorblade is minimal (Fig. 20). Furthermore, the
root bending moment values presented in Fig. 20 correspond to a The performance and efficiency of the blades is significantly
rotorblade equipped with VGs across 1/3 of the span. Naturally, reduced by surface roughness throughout the operational spec-
the installation of VGs only at the inner region of a rotorblade trum of the turbine. The installation of VGs at the blade root re-
has a smaller influence on the aerodynamic loads (i.e. root gion and transition area reduces such power losses. However, de-
bending moment) compared to a full span installation. Finally, spite that, the use of VGs reduced the overall efficiency compared
it is worth noting that when the VG installation is limited to to the baseline simulation. This is a result of the drag increase
the root region, the adverse effects due to higher skin friction caused by the addition of VGs. Unfortunately, no measurements
drag are reduced. This is especially obvious from a comparison were performed with VGs in combination with surface rough-
between the effect of full span VGs on the 20m blade (Fig.17) ness, therefore the actual performance of the VGs cannot be pre-
and the effect of root region VGs on the 40m blade (Fig.18 and cisely estimated. Previous literature and especially the work of
19). Reuss et al. [23] shows that the addition of VGs on an airfoil sec-
tion with roughness is able to restore the lost lift as well as the
stall angle of the airfoil.

14 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 CONCLUSIONS with Z = 7H.
An experimental and numerical study has been carried out
The aerodynamic performance simulations of various rotor-
to determine the optimal design parameters of vortex generators
blade designs and sizes produced the following results:
intended for the use on wind turbines and to simulate their effect
on the aerodynamics of various turbine blade geometries. The • Vortex Generators are able to substantially vary the
force measurements yielded the following results: performance of stall regulated turbine rotorblades and can
be subsequently used as performance tuning elements.
• The smaller vortex generator with H = 1.7% c shows
a better overall performance, producing both a higher • The adverse effects of blade contamination and erosion on
increase in the static stall angle and a larger maximum the lift coefficient can be corrected to a large extent with the
lift coefficient. This difference is attributed to the closer proper use of vortex generators. Nevertheless, the additional
proximity of the streamwise vortices to the airfoil surface. drag of the VGs further adds to the efficiency degradation
It also suffers from a lesser drag penalty than the larger of the contaminated blades.
version (H = 3.3% c).
• Both stall and pitch rotorblades clearly benefit from the
• The best performance in terms of stall suppression was use of vortex generators as a practical and low cost retrofit
reached with a spacing of Z = 3H: An increase of the solution.
static stall angle by 12.5◦ with respect to the baseline value
was produced with the smaller VG geometry, which is • Wind turbine rotorblade designs which include vortex gen-
accompanied by an increase in Cl,max by 31%. However, erators from the time of installation can possibly reduce the
this configuration does not only result in a high increase performance deficit problems associated with roughness ef-
in drag, but also produces sharp load excursions at angles fects on the blade.
exceeding the static stall angle. In order to avoid these
potentially harmful variations in lift, a spacing of Z ≈ 5H
appears preferable. ACKNOWLEDGMENT
The authors would like to acknowledge the kind support of
• The optimal chordwise position of the vortex generators is Smart Blade GmbH and its CEO Dipl.-Ing. F.U. Lang for the
in the range of x/c = 15% - 20%, as the best compromise be- current research project.
tween airfoil performance and a smooth post-stall lift curve
is reached here.
The impact of the spanwise spacing between adjacent vor- REFERENCES
tex generators on the distance of the streamwise vortices from [1] Nickerson, J. D., 1986. “A study of vortex generators at
the airfoil surface was determined with particle image velocime- low reynolds numbers”. In 24th AIAA Aerospace Sciences
try measurements. This method turned out to be well suited for Meeting, Reno, NV; United States; 6-9 Jan. 1986, AIAA
visualizing and quantifying the position of the vortices. The fol- Paper 1986-0155.
lowing observations were made: [2] Velte, C. M., Hansen, M. O. L., Meyer, K. E., and Fuglsang,
P., 2008. “Evaluation of the performance of vortex genera-
• Decreasing the spacing Z results in an increased interac- tors on the du 91-w2-250 profile using stereoscopic piv”. In
tion of neighboring vortices, causing them to move further International Symposium on Energy, Informatics and Cy-
away from the surface in the downstream region of the bernetics: Focus Symposium in the 12th World Multicon-
wing. Between x = 40% c and x = 80% c, the distance of the ference on Systemics, Cybernetics and Informatics (WM-
vortex centroids from the wall increased by 43% (VG1) and SCI 2008) 4.
64% (VG2) respectively with Z = 7H. When the spacing [3] Wendt, B. J., and Hingst, W. R., 1994. “Measurements
was reduced to Z = 3H, the relative increase was 91% (VG1) and modeling of flow structure in the wake of a low profile
and 95% (VG2). Even though the difference in deflection is ’wishbone’ vortex generator”. In 32nd AIAA Aerospace
quite considerable, the force measurements do not indicate Sciences Meeting and Exhibit, Reno, NV; United States;
a corresponding detrimental effect on stall suppression 10-13 Jan. 1994, AIAA Paper 1994-0620.
effectiveness. [4] Storms, B. L., and Jang, C. S., 1993. “Lift enhancement of
an airfoil using a gurney flap and vortex generators”. Jour-
• A spanwise deflection of the vortices, most likely caused nal of Aircraft(3), pp. 542–547, AIAA Paper 93–0647.
by the proximity of the airfoil surface, was observed. This [5] Griffin, D. A., and Russell, D. A., 1994. “Experimen-
effect is especially pronounced for the small VG geometry tal study of a method for predicting the effect of vortex-

15 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 generators”. In 25th AIAA Fluid Dynamics Conference, C., 2010. “Integration of a wind turbine blade design tool
Colorado Springs, CO; United States; 20-23 June 1994, in xfoil/xflr5”. In DEWEK 2011, DEWI.
AIAA Paper 1994-2378. [20] Gasch R., Twele J., B. P. C. W., 2010. Windkraftanla-
[6] Ashill, P., Fulker, J., and Hackett, K., 2001. “Research gen: Grundlagen, Entwurf, Planung und Betrieb. Vieweg
at dera on sub boundary layer vortex generators (sbvgs)”. + Teubner.
In 39th AIAA Aerospace Sciences Meeting and Exhibit, [21] G., S., 1955. “Theorie und entwurf von windraedern opti-
Reno, NV, United States, 8-11 January 2001, AIAA-Paper maler leistung”. Wissenschaftliche Zeitschrift der Univer-
2001-0887. sitaet Rostock.
[7] Betterton, J. G., Hackett, K. C., Ashill, P. R., Wilson, M. J., [22] Montgomerie, B., 2004. “Methods for root effects, tip ef-
Woodcock, I. J., Tilman, C. P., and Langan, K. J., 2000. fects and extending the angle of attack range to + − 100o
“Laser doppler anemometry investigation on sub boundary with application to aerodynamics for blades on wind tur-
layer vortex generators for flow control”. In 10th sympo- bines and propellers”. FOI Swedish Defence Research
sium on application of laser techniques to fluid mechanics, Agency, Scientific Report FOI-R-1035-SE.
Lisbon, p. 1012. [23] Reuss, R. L., Hoffman, M. J., and Gregorek, G. M., 1995.
[8] Lin, J. C., 2002. “Review of research on low-profile vortex Effects of surface roughness and vortex generators on the
generators to control boundary-layer separation”. Progress naca 4415 airfoil. Tech. rep., National Renewable En-
in Aerospace Sciences. Vol. 38, no. 4-5, pp. 389–420. ergy Laboratory, Golden, CO.; Ohio State Univ., Colum-
[9] Pauley, W. R., and Eaton, J. K., 1988. “Experimental study bus, OH, United States.
of the development of longitudinal vortex pairs embedded [24] Bridges, J., W. M. B. C., 2003. “Control of jet noise through
in a turbulent boundary layer”. AIAA Journal, 26(7), mixing enhancement”. NASA/TM-2003-212335.
pp. 816–823. [25] Pechlivanoglou, G., Fuehr, S., Nayeri, C., and Paschereit,
[10] Stanislas, M., and Godard, G., 2006. “Control of a deceler- C., 2010. “The effect of distributed roughness on the power
ating boundary layer. part 1: Optimization of passive vortex performance of wind turbines”. In Proceedings of ASME
generators”. Aerospace Science and Technology, 10(3), 04, IGTI Turbo Expo 2010 ASME/IGTI June 14 - 18, 2010,
pp. 181–91. Glasgow, Scotland, UK.
[11] Drela, M., and Giles, M., 1986. “Viscous-inviscid analy-
sis of transonic and low reynolds number airfoils”. AIAA
Journal, p. 10.
[12] Broadley, I., and Garry, K. P., 1997. “Effectiveness of
vortex generator position and orientation on highly swept
wings”. In 15th AIAA Applied Aerodynamics Conference,
Atlanta, GA; United States; 23-25 June 1997, AIAA Paper
1997-2319.
[13] Ortega, J. M., Bristol, R. L., and Savas, O., 2003. “Experi-
mental study of the instability of unequal-strength counter-
rotating vortex pairs”. Journal of Fluid Mechanics, 474,
pp. 35–84.
[14] Hansen, M. O., 2008. Aerodynamics of Wind Turbines (2nd
Edition). Earthscan.
[15] Shen, W., Mikkelsen, R., Sorensen, J., and Bak, C., 2005.
“Tip loss corrections for wind turbine computations”.
[16] Buhl, M., 2005. “A new empirical relationship between
thrust coefficient and induction factor for the turbulent
windmill state”. Technical Report, NREL/TR-500-36834.
[17] Himmelskamp, H., 1950. “Profiluntersuchungen an einem
umlaufenden propeller”. Mitteilungen aus dem Max-Plank-
Institut fuer Stroemungsaufgaben.
[18] Snel, H., Houwking, R., van Bussel, G., and Bruining, A.,
1993. “Sectional prediction of 3d effects for stalled flow
on rotating blades and comparison with measurements”. In
Proceedings of EWEC, H.S. Stevens & Associates.
[19] Pechlivanoglou, G., Marten, D., Nayeri, C., and Paschereit,

16 Copyright
c 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use