Sunteți pe pagina 1din 9

L.

Battisti
e-mail: lorenzo.battisti@ing.unitn.it

L. Zanne
S. DellAnna
DIMS, Faculty of Engineering,
Universita` degli Studi di Trento,
Via Mesiano 77, I-38050 Povo (TN), Italy

V. Dossena
e-mail: vincenzo.dossena@polimi.it

G. Persico
B. Paradiso
Laboratorio di Fluidodinamica delle Macchine,
Dipartimento di Energia, Politecnico di Milano,
Via Lambruschini 4, I-20158, Milano, Italy

Aerodynamic Measurements
on a Vertical Axis Wind Turbine
in a Large Scale Wind Tunnel
This paper presents the first results of a wide experimental investigation on the aerodynamics of a vertical axis wind turbine. Vertical axis wind turbines have recently received
particular attention, as interesting alternative for small and micro generation applications. However, the complex fluid dynamic mechanisms occurring in these machines
make the aerodynamic optimization of the rotors still an open issue and detailed experimental analyses are now highly recommended to convert improved flow field comprehensions into novel design techniques. The experiments were performed in the large-scale
wind tunnel of the Politecnico di Milano (Italy), where real-scale wind turbines for micro
generation can be tested in full similarity conditions. Open and closed wind tunnel configurations are considered in such a way to quantify the influence of model blockage for
several operational conditions. Integral torque and thrust measurements, as well as
detailed aerodynamic measurements were carried out to characterize the 3D flow field
downstream of the turbine. The local unsteady flow field and the streamwise turbulent
component, both resolved in phase with the rotor position, were derived by hot wire
measurements. The paper critically analyses the models and the correlations usually
applied to correct the wind tunnel blockage effects. Results highlight that the presently
available theoretical correction models do not provide accurate estimates of the blockage
effect in the case of vertical axis wind turbines. The tip aerodynamic phenomena, in particular, seem to play a key role for the prediction of the turbine performance; large-scale
unsteadiness is observed in that region and a simple flow model is used here to explain
the different flow features with respect to horizontal axis wind turbines.
[DOI: 10.1115/1.4004360]
Keywords: VAWT, blockage, wind Tunnel, wind turbine, aerodynamic measurements,
unsteady flows

Introduction

In recent years a renewed interest has arisen on vertical axis


concept in wind turbines (VAWT). In the urban environment,
affected by highly turbulent flows and strong vertical velocity gradients, VAWT claims for several advantages: insensitivity to yaw,
ability to withstand rapid changes of wind direction and to provide
good performances also in skewed flows [1], low noise emission
due to low tip speed ratios [2]. Finally the better integration in architectural projects also represents a key advantage.
VAWT concept, however, still presents a lot of challenges to be
solved, principally due to its intrinsic flow complexity. With its fully
three-dimensional geometry and flow structure, the blades elaborate
twice the streamtube in a revolution and interact with the wake shed
from upstream blades. Furthermore, the blade profiles work with an
oscillating angle of attack leading to blade loading unsteadiness
and possibly to dynamic stall, affecting both the wake and the tip
vortex development. The variety of geometries of rotor design, from
classical Darrieus troposkein geometry to the V and the H designs,
complicates the definition of a general model [3].
At the moment, the VAWT flow field is still far to be completely understood. In particular, there is a strong need of detailed
experimental analyses to convert an improved flow field comprehension into better aerodynamic models suitable to support novel
designs. In this way, a more refined aerodynamic optimization of
the rotor shape and performance could be achieved, and the
matching with aeroelastic codes, used to predict extreme and fatigue loads, could be improved.
Contributed by the Advanced Energy Systems Division of ASME for publication
in the JOURNAL OF ENERGY RESOURCES TECHNOLOGY. Manuscript received September 9,
2010; final manuscript received May 13, 2011; published online July 22, 2011.
Assoc. Editor: Andrea Lazzaretto.

Journal of Energy Resources Technology

At the present time, the most used aerodynamic model is the double-multiple streamtubes (DMS) [4], a blade element momentum
model (BEM) applied to two actuator disks, one representing
the upwind side of the rotor and the other one the downwind side,
each of them subdivided into multiple streamtubes. Taking
into account dynamic stall and secondary effects (e.g., parasitic arms
drag, tip losses, turbulent wake state correction, wake expansion,
and tower deficit) with dedicated submodels, the results are still
far from providing a sufficiently reliable performance estimation.
More recently, the so-called vortex methods have been developed with the aim to get a more realistic simulation of the
unsteady behavior of the wake, the tip vortex, and the blade loads.
These methods allow computing the flow field from the blade circulation, and the trailing and shed vorticity. The wake can be either prescribed or free, the latter giving much better results but
being computationally slower and affected by numerical instabilities. Normally these methods use a database for airfoils performancesas DMS doesto calculate the blade circulation. To
avoid the use of the database an unsteady panel method allows
calculating the airfoil performances [5], once coupled with the
solution of the unsteady boundary layer [6].
CFD analysis represents the most recent class of aerodynamic
performance prediction methods. In these codes, the flow field is
computed by replacing the blades elements with body forces
derived from airfoil look-up tables [7,8]. As a more complex alternative they can resolve the entire flow field including the blade
profile boundary layer.
In this uncertain scenario there is a lack of experimental data
required to validate new models of the flow behavior.
Throughout the VAWT research, Sandia Laboratories focused
the attention on free field testing [4], and only a few tests were
performed in wind tunnels (i.e., [9]). In this way, full scale

C 2011 by ASME
Copyright V

SEPTEMBER 2011, Vol. 133 / 031201-1

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

analyses (i.e., conserving Reynolds number, tip speed ratio, and


reduced frequency) can be carried out at the expense of an
increased complexity in the test management. Loading data from
open air testing are actually available [4] but a rigorous attempt
for the wake characteristics determination has not been done till
relatively recent years.
The dynamic stall phenomena was investigated by Fujisawa
and Shibuya [10] and Simao Ferreira et al. [11], that performed
PIV measurements in a wind turbine to clarify the vortex generation and shedding during the blade revolution. Hofemann et al.
[12] analyzed tip vortex evolution with a 3D stereo PIV technique.
A study of the rotor performance and wake shedding in nonskewed and skewed flow was performed with flow visualization
and hot wire measurements by Simao Ferreira et al. [13].
Unfortunately, tests in tunnels introduce further sources of
uncertainty due to the effect of blockage. Srensen et al. [14,15]
used the one-dimensional momentum approach to derive a correction for a HAWT in closed tunnels, and validated it by numerical
simulations. They also elaborated a simulation to get the correction for a special case of open wind tunnel. Nevertheless, the onedimensional approach still shows some lacks and it can be used
only till the vortex system behind the turbine does not break up in
the turbulent wake. The open question in literature is whether it is
possible to use the same models in the case of a VAWT, or better
using the Maskell theory [16] as required by the Measnet association for cup anemometer calibration.
Moreover, in the case of an open tunnel, the blockage factor
presented in Ref. [14] is considered too high and possibly other
corrections can be used, as that developed by Mercker and Wiedemann in Ref. [17].
In many cases the data are collected in free field with no controlled conditions, or in closed wind tunnel with a high blockage
(exceeding 510%) without any correction for blockage and for
the three-dimensional shape of the wake. For open-jet tunnel the
blockage effects are less evident but this normally leads to test
excessively large models characterized by high-blockage effect, often resulting in the application of even more doubtful corrections.
This lack of accuracy of design and analysis tools are the
motivation of the present work, which presents an experimental
study on the aerodynamics and performance of a VAWT. Measurements were performed in the large-scale wind tunnel of the
Politecnico di Milano, where real-scale wind turbines for micro
generation can be tested in full similarity conditions. The research

Fig. 1 Experimental set-up. The measuring section is located


1.5 D downstream of the rotor axis

covers several fields of activity. At first a critical evaluation of


correlations for wind tunnel blockage correction is carried out; to
quantify the blockage effect, both open and closed wind tunnel
configurations are considered. Moreover, a number of different
operating conditions are investigated to evaluate the effect of
loading on the turbine performance for both wind tunnel configurations. The turbine tip aerodynamic and its implications on the
overall aerodynamic performance are finally discussed with the
support of unsteady flow data and flow schematics.

The Politecnico di Milano Large Scale Wind Tunnel

The Politecnico di Milano wind tunnel is a large scale closed


loop facility, which develops along two floors of a dedicated
building. Within the facility two different test sections are available: a low speed test section (14 m  3.84 m) allowing a maximum air velocity of about 15 m=s and a high speed section
(4 m  3.84 m) allowing a maximum velocity of 55 m=s. The
wind generator is composed by two rows of 7 fans, each of them
driven by an inverter controlled electric motor. The overall installed electric power is approximately 1.4 MW.
Tests were performed in the high speed test section (Fig. 1),
located immediately upstream of the diffuser leading to the fans,
and characterized by a very low turbulence level (<1%).
Tests can be carried out in a closed configuration by positioning
the models inside of a removable test room of 4 m width  3.84 m
height, or in a free jet (open) configuration by removing the test
room and installing the models directly facing the upstream tunnel
tube (Fig. 2). The same VAWT was tested in both the closed and
the open configurations in order to analyze the influence of the
model blockage for different operational conditions.
Figure 1 reports a sketch of the experimental set-up in the freejet configuration together with the location of the aerodynamic
measurement plane and a dashed draw showing the closed test
room trace when operating in a closed environment. The model
was centered in both configurations. A blockage ratio B 0.10
results in the closed room, where the frontal section of the model
is given by AD 2H  D.

Turbine Model and Instrumentation

The physical model of the turbine is a small VAWT characterized by three straight blades as represented in the picture of Fig. 2.
The machine was designed and built for research purposes by the
company Tozzi Nord Wind Turbines sited in Trento (Italy) and
instrumented by the turbomachinery laboratory of the university
of Trento. The main turbine geometrical data are reported in Table 1. The wind turbine was equipped with strain gauge bridges
installed on the supporting mast 1.2 m below the rotor mid
section, to measure the streamwise and lateral aerodynamic

Fig. 2 Picture of the wind turbine and traversing system in the


open tunnel configuration

031201-2 / Vol. 133, SEPTEMBER 2011

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Transactions of the ASME

Table 1

Main characteristics of the three-blade VAWT

Blade height (2H)


Rotor diameter (D)
Blade profile
Solidity (Nc=D)

1.457 m
1.030 m
NACA0021
0.250

forces; a torque meter was installed to measure the shaft


mechanical torque. The rotational speed and the rotor phase were
provided by a 13 bit encoder.

Description of the Instrumentation

Aerodynamic measurements were performed on a plane normal


to the upstream wind direction located 1.5 rotor diameters downstream of the VAWT shaft, as previously shown in Fig. 1.
Two aerodynamic probes were used to define the time and
phase averaged flow field on the whole measuring plane:

Directional pneumatic five hole probe: The probe was calibrated on a low speed jet over an angular range of 6 24 deg
both in pitch and yaw direction. This probe was used in order
to obtain the local total pressure, static pressure and the 3D
velocity vector. Uncertainty in total and static pressure measurements can be assumed to be lower than 5 Pa, while for
angular measurement an uncertainty of 0.2 deg has been
evaluated.
Single sensor hot wire anemometer: The probe head, characterized by a sensor wire diameter of 5 lm normal to the probe
stem, was operated in constant temperature mode. The probe
was mounted with the wire in vertical direction, to minimize
the effects due to vertical velocity components. So operating,
no specific directional calibration of the hot wire probe is
required. Hot wire data allowed to define, for each point of the
measurement grid, the time averaged component of the flow
velocity and the periodic velocity component, as a function of
the rotor angular position. An estimation of the turbulence intensity was also provided by the hot wire measurements.

Because of the higher accuracy of the hot wire technique in


evaluating the velocity magnitude, the time averaged value of the
local kinetic head derived from hot wire measurements was used
to improve the accuracy in evaluating the velocity vector and the
static pressure field from the 5 hole probe measurements.
Probes were traversed on a measurement grid of 3700
mm  900 mm defined by 21 points along horizontal transverse
direction (Y) and 9 points along vertical spanwise direction (Z),
extending from rotor midspan (Z=H 0.0) to 170 mm above the
rotor tip (Z=H 1.24). Test carried out on the whole plane are
denoted as 3D, while the ones performed only at the mid-span
section are denoted as 2D.
The operating conditions considered for the current tests reproduced tip speed ratios k reported in Table 2, obtained by varying
the rotational speed in the range from 400 rpm to 500 rpm and
operating the tunnel at wind speeds ranging from 10 to 20 m=s.
The k 1.6 condition is in the ascending branch of the CP-k curve
Table 2

Summary of the test conditions

Open tunnel tests

k [-]

CT [-]

V0 [m=s]

2D
3D
3D

1.62
1.56
2.50

0.46
0.46
0.68

16.14
13.14
10.50

Closed tunnel tests


2D
3D
3D

1.64
1.63
2.54

0.50
0.49
0.78

16.05
13.30
10.55

Journal of Energy Resources Technology

where the blades stall for a nonnegligible part of the rotor revolution, while for k 2.5 this rotor works near the maximum CP condition. The maximum rotational speed was limited to avoid
structural collapse of the wind turbine rotor, while the minimum
wind speed was limited by the accuracy in pressure measurements. Table 2 reports the main specifications of the tests considered in the following.

3D Wake Profile

Figures 3 and 4 show the general shape of the velocity field on


the measurement plane for k 1.6 and k 2.5, in both open and
closed wind tunnel tests. A frame representing the model location
is also superimposed.
The wake appears to be not symmetric and deformed turnwise,
according to the rotational direction of the wind turbine, i.e., counter
clockwise when viewed from above. A similar characteristic was
also found by Simao Ferreira et al. in Ref. [13]. These phenomena
are due to the blade tip vortex interactions, that are stronger in the
upcoming blade region (on the left of the figures), and weaker in the
retreating blade one (on the right in the front view of the rotor). This
asymmetry reduces as the tip speed ratio increases, as the variation
of fluid dynamic conditions experienced by the blades (in terms of
relative velocity, blade angle of attack, and vortex convection velocity) also reduces during a full revolution, consequently reducing the
fluctuation of blade bound vorticity. At low tip speed ratios the
blade normally stalls for a nonnegligible portion of the revolution,
producing a more intense bound circulation in the upcoming region
and a stronger tip vortex [4].
This conclusion was also drawn from the simulations presented
by Dixon et al. [5] that predicted such a behavior with a 3D
unsteady panel method. The present experimental results confirm
the capability of vortex methods to capture the general 3D flow
field. At low tip speed ratios the core flow is characterized by
weak velocity gradients for the open and the closed tunnel configurations. At high tip speed ratio the more intense work extraction
(and CT) determines a higher velocity drop in the core flow. Furthermore stronger gradients are observed in the closed chamber
tests due to blockage effects.

Blockage Modeling

Since the wind tunnel test section has a confined volume, the
aerodynamic measurements obtained from the wind tunnel tests
do not resemble those obtained in infinitely spaced boundaries,
such as the case of in open field.
The test of wind turbines in wind tunnels is based on fluiddynamic similarity rules, which scale the actual sizes to the model
ones according to the Reynolds and Strouhal numbers (the Mach
number is not of concern because the process can be considered
as incompressible). Tests are usually devoted to assess rotor
global performances (power curve) and to evaluate improved
models for the turbulent wake. Nevertheless both tasks have to
deal with the accurate correction of the tunnel blockage effect.
When rotating devices as wind turbine rotors are actually tested,
additional complexities arise from the influence of test conditions
on the wake shape. Highly loaded rotors have a more expanded
wake compared to lightly loaded rotors. Moreover, blade passing
events cause turbulent structures to be shed downstream and lead
to periodic fluctuations in the velocity and pressure fields.
A number of approaches are available to study the actual airflow
around the wind turbine rotor and to estimate the influence of blockage [1822] for an unshrouded propeller or a wind turbine. Glauert
[18], developed the basic theory for thrusting propellers, which, by
limited extension, can be adapted to HAWT wind turbines. Glauert
used a momentum-balance=actuator-disk model to provide a procedure for setting the wind tunnel inlet flow speed so as to generate
the same thrust as the one experienced in unbounded conditions.
The resulting simple relationship has been used throughout the
propulsion industry but encounters difficulty for wind turbines, as
SEPTEMBER 2011, Vol. 133 / 031201-3

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Fig. 3

Contours of non dimensional velocity V=V0 on the measurement plane at k 5 1.6 for open and closed configurations

a singularity occurs when the thrust coefficient equals unity. Mikkelsen and Srensen [15] revisited the issue and, still using a
momentum-balance=actuator-disk model, provided a new velocity
adjustment relation that avoids Glauerts model singularity. Comparison to Navier Stokes CFD studies confirmed the predictions of
the simple actuator-disc model for blockage levels of practical interest. To date though, no method has been put forward for estimating the blockage correction factors for unshrouded VAWT.
Regardless of the tunnel configuration, a unique model can be
adopted to compute the one-dimensional streamwise thrust stressing the turbine. With reference to Fig. 5, the flow field within the
test chamber is shown. Five main sections are identified: section
0, upstream; sections 1 and 2, wind turbine inlet and outlet respectively; section D, wind turbine (disk section); section 3, far downstream (measuring section). The stream is entering the test section
with a uniform velocity V0 over the whole wind tunnel section
_ core ) and an
and divides itself into a core fractional mass flow (m
_ outer ). The former defines a stream
outer fractional mass flow (m
tube which passes through the upstream area A0, the turbine area
ADdelivering the work and providing the thrustand the downstream area A3 (wake area). The outer air mass flow entering the
volume outside of the central streamtube mixes with a fractional
_ R ), possibly exchanged with the external environment.
mass (m
This fraction vanishes when a closed wind tunnel is of concern.
Considering the core streamtube, the thrust operated by the
flow stream on the disk can be easily evaluated by the static pressure drop across the disk itself as
p1  p2 AD T

(1)

The total pressure can be assumed constant from section 0 to


section 1 and from section 2 to section 3; considering a uniform

velocity distribution on section 3, and being sections 1 and 2 infinitely close compared to the wake streamwise extension,
V1 V2 VD and Eq. (1) can be rewritten as

T AD

1
1
p0 qV02  qVD2
2
2



1
1
 p3 qV32  qVD2
(2)
2
2

leading to
1
T p0  p3 AD qV02  V32 AD
2

(3)

Since pressures and velocities are locally measured in sections 0


and 3, thrust T can be directly deduced by the Eq. (3) and the
result can be compared with the streamwise thrust measured by
the strain gauges installed on the supporting mast of the model.
Since on the downstream measuring plane the actual measured
velocity distribution is not uniform in both directions, a methodology for the evaluation of an equivalent velocity V3 and of a corresponding section A3 is described in the following.
As previously discussed, the blockage model requires the
downstream flow field to be divided in a wake region (core
flow) and in a free-stream region (outer flow). To identify the
edges of the wake region, an analysis of the nondimensional total
pressure drop distribution (Cpt) at measurement traverses has been
performed for an each spanwise position. Thanks to the sharp
edges of the wake, its boundaries can be easily identified for all
turbine operational conditions. An example of typical wake profiles in terms of Cpt distributions at several spanwise coordinates
is reported in Fig. 6 (closed tunnel, k 2.5) which shows how the
wake boundaries are sharply defined. The total pressure drop

031201-4 / Vol. 133, SEPTEMBER 2011

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Transactions of the ASME

Fig. 4

Contours of non dimensional velocity V=V0 on the measurement plane at k 5 2.5 for open and closed configurations

profile is constant until almost the tip of the blade (Z=H 0.8),
and starts deviating consistently only in the tip region. The drop is
reabsorbed almost completely just above the turbine maximum
height (Z=H 1.2), except for the nonsymmetrical local peak
induced by the blade load variation throughout the full revolution.
A combined analysis of Figs. 3, 4, and 6, indicates that the 3D
effects vanish quite rapidly as the tip of the blade is reached.
Figure 7 shows a comparison of the nondimensional velocity
distribution in the wake for the closed and open tunnel tests. It is
to be noted that the highest nondimensional velocity rate measured in the closed wind tunnel contributes to the increase of thrust

Fig. 5

coefficient (Eq. (3)). The fairly complete superposition of results


for the open or for the closed configuration at constant k proves
the measurement reliability.
Once the wake edges were defined, a mass averaging procedure
was performed on the velocity and total pressure distributions at
fixed spanwise coordinate, to finally determine the equivalent
wake velocity and the outer equivalent velocity (see Fig. 5). The
corresponding extension of the wake region, for a fixed spanwise
coordinate, was further evaluated on the basis of the mass flow
_ core ). The same prorate measured within the wake streamtube (m
cedure was also extended to the whole measuring plane, leading

Simplified flow field model upstream and downstream of a wind turbine

Journal of Energy Resources Technology

SEPTEMBER 2011, Vol. 133 / 031201-5

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Fig. 6 Wake shape in terms of Cpt distribution at different


spanwise locations (test closed, 3D, k 5 2.5)

to the definition of an over-all wake velocity (V3) and of a


wake streamtube section.
According to Eq. (3), the actual streamwise thrust acting on
each spanwise section of the machine can be easily computed on
the basis of the above defined aerodynamic quantities. Finally, the
integration of the spanwise thrust distribution derived from aerodynamic measurements allows the evaluation of the bending load
acting on the model mast. This is then compared with the thrust
measurement performed by the strain gauges. In Fig. 8, strain
gauge measured thrust coefficients Ct,m are compared to the thrust
coefficient obtained by the data reduction procedure, Ct,aer.
Results show a very good agreement witnessing the good accuracy of both measurement techniques, and, in particular, of the
applied data reduction methodology.
To generalize wind tunnel results to unbounded conditions, the
measurements need to be corrected for wind tunnel wall effects.
Due to the wind tunnel walls constraining the flow velocity outside of the wake is increased compared to the unbounded condition (V30 > V0 ), as also shown by the nondimensional velocity
distributions reported in Fig. 7. Consequently, the static pressure
downstream of the model, outside of the wake, is reduced and in

Fig. 8 Comparison of thrust coefficients CT obtained by strain


gauge measurement and by aerodynamic measurement data
reduction for the open and closed configurations

general p < p0 . Far downstream of the model, the static pressure


becomes constant across the whole tunnel section, and no further
expansion occurs along the wake streamtube.
The static pressure in the wake of closed configuration tests is
hence reduced compared to the unbounded conditions, namely
p3 < p0 . Thus the static pressure difference p3  p0 generates a
thrust T higher than that observed in unbounded conditions (see
Eq. (3)).
The absence of physical walls in the test section of an open jet
wind tunnel (such as the case of the open configuration) strongly
reduces the blockage effects, leading to an almost complete recovery
of the undisturbed pressure in the downstream far field (see Fig. 5).
Measurements at station 3 reveal that the static pressure difference between inside and outside the core streamtube is small,
leading to a maximum 12% contribution in the thrust coefficient
for both open and closed test chambers. This evidence allows stating that, at station 3, the wake can be considered completely
developed.
Glauert [18] proposed to determine an equivalent undisturbed
wind speed V0 that in the unbounded flow condition gives the
same thrust of the undisturbed wind speed V0 in the wind tunnel
application. Thus, by denoting with the apex prime the equivalent
unbounded conditions, we have
VD VD0

(4)

CT V02 C0T V002

(5)

The one-dimensional momentum theory for the equivalent


unbounded conditions yields

1
T 0 qAD C0T V002 2VD0 qAD V00  VD0
2

(6)

By combining Eqs. (4)(6), it is found


V00 VD CT

V0 V0 4 VVD

(7)

Since the velocity and the area of the wake in section 3 are computed by the aerodynamic measurements, VD can be easily estimated as
Fig. 7 Comparison of the wake shape at rotor midspan in
closed and open tunnel tests at k 5 1.6 (2D and 3D tests)

VD V3

031201-6 / Vol. 133, SEPTEMBER 2011

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

A3
AD

(8)

Transactions of the ASME

Fig. 9 Wind speed correction factors as function of the thrust


coefficient. Experiments and theoretical predictions with the
Glauert, Maskell, and M-W (Merker-Wiedemann) methods

Since the CT coefficient


 determined in the wind tunnel tests is also
known, the ratio V00 V0 can
 be easily computed by Eq. (7).
In Fig. 9, the ratio V00 V0 is shown as a function of the thrust
coefficient and compared to the original Glauert correlation [18],
to the correlations by Mikkelsen and Srensen [15] and Maskell
[16] for closed tunnel and Merker-Wiedemann [17] for open test
section. The former has been plotted for two limiting cases: the
first case, for B 0.1, is based on the actual geometrical ratio; the
second case, B 0.25, corresponds to a purely 2D blockage i.e.. it
considers the tunnel height not influencing the wake evolution.
Figure 9 shows that the actual aerodynamic behavior of the
VAWT is different from that predicted by simple actuator disk, or
simple blockage models, also in the limiting case of B 0.25. At
first the experimental corrections for the closed tunnel are by far
more remarkable compared to the ones predicted by usual corrections. Moreover the measurements show that the magnitude of the
correction reduces as CT increases. This result is quite surprising
because the trend is opposite to that given by the aforementioned
models, but indicates that the complex fluid-dynamics of a
VAWT cannot be modeled by a simple 1D momentum approach,
but rather by an actuator cylinder approach.
A possible explanation of this trend is that, as CT increases,
both the k coefficient and the blade rotor loading enhance; so, a
higher bound shed vorticity is generated by the blades, but with
weaker blade bound vorticity and trailing tip vortex. It can be figured out that the vortices shed by the tip of the upstream running

Fig. 10 (a) Phase-resolved velocity magnitude at k 5 2.5, midspan, (b) Phase-resolved turbulence intensity at k 5 2.5, midspan, (c) Phase-resolved velocity magnitude at k 5 2.5, Z=H 5 1.1, and (d) Phase-resolved turbulence intensity at k 5 2.5,
Z=H 5 1.1

Journal of Energy Resources Technology

SEPTEMBER 2011, Vol. 133 / 031201-7

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Fig. 11 Schematic of the vortex structures released by the


blades at mispan and tip sections of the turbine

blades entrain some mass flow into the core flow from the virtual
bases of the rotor defined by the blade tips. The effect is to
increase the mass flow through the wind turbine and, as a consequence, in the wake, thus modifying the magnitude of the velocity
correction. This hypothesis is confirmed by the simulation of
Dixon [5] showing a roll-up phenomenon of the upwind tip vortices toward the inner part of the rotor. These phenomena will be
further discussed and modeled throughout the unsteady flow
analysis.
The results of these experiments also raise a further question
concerning the rotor wind tunnel blockage study, i.e., the applicability of the actuator disk model for stalled conditions. Maskell
[16] showed that the blockage corrections for stalled wing and
bluff bodies are about five times the corrections for attached flow.

Unsteady Flow Field

The analysis reported in the previous sections has shown that


the aerodynamic of VAWT is complex, with significant deviations
with respect to the most known horizontal axis machines.
In this section, the unsteady-periodic component of the flow
field measured by the hot wire probe is presented, with the aim to
improve the level of comprehension of the flow field. The periodic
component is meant to be the unsteady fluctuation of flow velocity
resolved in phase with the rotor position.
The passing of the rotor blades induces a periodic unsteadiness
in the flow field; its amplitude can be used as a proper marker of
large-scale viscous structures released by the blades, commonly
neglected by classical theories.

To perform such a data-reduction, the velocity measurements


were first triggered on the rotor instantaneous position (phaselocking); then, they were phase-averaged over the blade passing
periods available in each acquisition time-span (about 25 periods
were available for each acquisition) to obtain the periodic flow
component over the blade passing period. Correspondingly, the
RMS of the instantaneous-to-periodic flow velocity was also evaluated for each phase of the period, from which a phase-resolved
turbulence intensity was computed.
In Fig. 10 the phase-resolved velocity magnitude and turbulence intensity are reported for k 2.5 (open configuration) at
two different spans, i.e., the midspan section and a section
placed just above the blade tip. These plots are arranged as twodimensional fields in which the abscissa is the nondimensional
Y=D position along the measurement traverse, and the ordinate
represents the phase s, connected to the rotor angular position (a
whole revolution period is reported, i.e., three blade passing
periods).
In these maps the mean spatial gradients (the turbine wake profile, for example) appear as vertical bands, while unsteady effects
can be recognized as perturbation of the main horizontal gradients.
In Fig. 11 a sketch of the vortex structures released by the
blades at the two sections are depicted. At midspan the vortex pattern is dominated by the shed vorticity, made unsteady by the
blade circulation variation throughout a revolution. Since the axis
of the unsteady shed vorticity is parallel to the rotor shaft, velocities in a plane perpendicular to the rotor axis are induced, in accordance to the application of the BiotSavart law.
A different vorticity configuration is sketched at the tip section, where the tip vortices show an axis perpendicular to the
rotor shaft. The tip vortices can induce velocities both in the
horizontal and the vertical directions, so they are responsible
both for a part of the horizontal wake velocity induction and for
the entrainment of air in the rotor and in the wake from their top
and bottom sides.
Figure 10(a) clearly shows that, in the midspan section, the role
of the periodic unsteadiness is negligible; this means that the viscous wakes (shed vorticity) of the single profiles are almost completely mixed out in the measurement plane. The distribution of
the turbulence intensity (Fig. 10(b)) at midspan is slightly more
sensitive to the rotor blade unsteadiness, but the structure of the
turbulent field is space-dominated, with high peaks in the margins
of the wake of the machine (especially on the right side). This
high turbulence region is connected to the shed vorticity, which is
due especially to dynamic stall as it was already shown by the experimental results of Fujisawa and Shibuya [14]. In particular, as
the blade starts the downwind passage, it interacts with this shed
vorticity which is higher at the rotor margin.
The situation at the tip section of the wake (Figs. 10(c) and
10(d)) is very different: the velocity defect in the wake is much
lower than that at midspan, and a distinct unsteady modulation of
the velocity field on the right side of the wake is now clearly visible. The higher relevance of unsteadiness at this section is further
demonstrated by the turbulence field, which is now characterized
by very high values of turbulence intensity in the whole wake
region (and not only on the wake margins). The structure of the
turbulence field inside the wake is now dominated by an unsteady
perturbation, where the maximum peaks of turbulence are found,
linked to the blade passing events. The maps in Fig. 10(c) and
especially Fig. 10(d) indicate that a large-scale viscous structure
is generated by the tip of the rotor bladesthe tip vortex of the
profiles; the relevance of this structure, measurable in a section
placed relatively far from the blades, is much higher than the one
of the single wakes of the profiles.
The tip region is, therefore, effectively affected by fully 3D and
unsteady flow phenomena that play a relevant role on the definition of the streamtube passing through the machine and measured
downstream. Hence, it strongly influences all the integral quantities that determine the performance of the wind turbine. The modeling of such a tip effect is therefore crucial for the setting up of

031201-8 / Vol. 133, SEPTEMBER 2011

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

Transactions of the ASME

accurate models for the prediction of vertical axis wind turbine


performance.

Conclusions

In this paper the first results of a wide experimental investigation on the aerodynamics and performance of a VAWT have been
presented.
The measurement technique and the data reduction methodology employed for the present investigation were shown to be
accurate and reliable, on the basis of thrust measurement comparison and data repeatability.
The wake analysis shows congruent behavior at different tip
speed ratios and thrust coefficients, and indicates a wake dimension comparable to the wind turbine swept area. The flow is
mostly two-dimensional almost up to the tip, where large scale
turbulent structures appear in a very confined region. This suggests that modeling the tip effects is crucial for the prediction of
VAWT performances.
Experimental results evidence that the upstream velocity corrections, commonly applied to account for blockage effects in
wind tunnels, are of doubtful validity in the case of VAWT,
leading probably to underestimated correction rates for all the
considered cases. Furthermore upstream velocity corrections
based on experimental results decrease as the thrust coefficient
increases.
The key-problem of the existence of velocity components parallel to the VAWT shaft, responsible for the entrainment of an unexpected flow rate, is discussed with the support of tip unsteady data
and of a tip vortex evolution model.

Acknowledgment
The authors thank the Company Tozzi-Nord Wind Turbines
(Italy) for partial funding the research. The authors also acknowledge Mr. Claudio De Ponti for his valuable contribution to the experimental set-up.

Nomenclature
A
B
c
Cpt
CT
D
H
_
m
N
p
Tu
T
V
V0
X, Y, Z
X
q
k
s

area (m2)
blockage ratio (B AD =AT )
blade chord
pt;0  pt;3 =0; 5  q  V02 non dimensional total pressure
drop
thrust coefficient
turbine diameter (m)
half blade height (m)
mass flow (kg=s)
number of blades
static pressure (Pa)
turbulence intensity
thrust (N)
velocity (m=s)
corrected velocity (m=s)
streamwise, transversal, spanwise coordinates
streamwise non-dimensional distance from rotor axis
air density (kg=m3)
tip speed ratio (x D=2=V0)
rotor phase (time divided by the blade passing period)

Journal of Energy Resources Technology

x rotational speed (rad=s)


C vorticity

Subscripts
0
1
2
3
D
aer
m
t
T

wind tunnel upstream


model upstream
model downstream
measurement section
disk section
from aerodynamic measurements
measured with strain gauges
total
tunnel

References
[1] Mertens, S., van Kuik, G., and van Bussel, G., 2003, Performance of a H-Darrieus
in the Skewed Flow on a Roof, ASME J. Sol. Energy Eng., 125, pp. 433440.
[2] van Bussel, G. J. W., Mertens, S., Polinder, H., and Sidler, H. F. A., 2004,
R : Concept and Realisation of a Small VAWT for the Built EnvironTURBYV
ment, The Science of making Torque from Wind, 1921 April, Delft.
[3] Battisti, L., Brighenti, A., and Zanne, L., 2009, Analisi delleffetto della scelta
dellarchitettura palare sulle prestazioni di turbine eoliche ad asse verticale,
Atti del 64 Congresso Nazionale ATI.
[4] Paraschivoiu, I., 2002, Wind Turbine DesignWith Emphasis on Darrieus
Concept (Polytechnic International Press, Montreal, 2002).
[5] Dixon, K., Simao Ferreira, C. J., Hofemann, C., Van Bussel, G. J. W., and Van
Kuik, G. A. M., 2008, A 3D Unsteady Panel Method for Vertical Axis Wind
Turbines, Proceedings of EWEC Brussels.
[6] Oler, J. W., Strickland, J. H., Im, B. J., and Graham, G. H., 1983, Dynamic
Stall Regulation of the Darrieus Turbine, Sandia National Laboratories, Albuquerque, New Mexico, SAND837029.
[7] Allet, A., and Paraschivoiu, I., 1995, Viscous Flow and Dynamic Stall Effects
on Vertical-Axis Wind Turbines, Int. J. Rotating Mach., 2(1), pp. 114.
[8] Shen, W. Z., Zhang, J. H., and Srensen, J. N., 2009, The Actuator Surface
Model: A New NavierStokes Based Model for Rotor Computations, ASME J.
Sol. Energy Eng., 131, p. 011002.
[9] Blackwell, B. F., Sheldal, R. E., and Feltz, L. V., 1976, Wind Tunnel Performance Data for the Darrieus Wind Turbine With NACA0012 Blades, Sandia
National Laboratories, Albuquerque, New Mexico, SAND760130.
[10] Fujisawa, N., and Shibuya, S., 2001, Observations of Dynamic Stall on Darrieus Wind Turbine Blades, J. Wind. Eng. Ind. Aerodyn., 89(2), pp. 201214.
[11] Simao Ferreira, C. J., van Bussel, G. J. W., Scarano, F., and van Kuik, G., 2007,
2D PIV Visualization of Dynamic Stall on a Vertical Axis Wind Turbine Proceedings of the AIAA=ASME Wind Energy Symposium.
[12] Hofemann, C., Simao Ferreira, C. J., Van Bussel, G. J. W., van Kuik, G. A. M.,
Scarano, F., and Dixon, K. R., 2008, 3D Stereo PIV Study of Tip Vortex Evolution on a VAWT, Proceedings of EWEC, Brussels.
[13] Simao Ferreira, C. J., van Kuik, G., and van Bussel, G., 2006, Wind tunnel
hotwire measurements, flow visualization and thrust measurement of a VAWT
in skew, Proceedings of the AIAA=ASME Wind Energy Symposium.
[14] Srensen, J. N, Shen, W. Z., and Mikkelsen, R., 2006, Wall Correction Model
for Wind Tunnels With Open Test Section, AIAA J., 44(8), pp. 18901894.
[15] Mikkelsen, R., and Srensen, J. N., 2002, Modelling of Wind Tunnel Blockage, Proceedings of the Global Windpower Conference and Exhibition.
[16] Maskell, E. C., 1963, A Theory of the Blockage Effects on Bluff Bodies and
Stalled Wings in an Enclosed Wind Tunnel, ARC=R & M3400.
[17] Mercker, E., and Wiedemann, J., 1996, On the Correction of the Interference
Effects in Open Jet Wind Tunnels, SAE Paper No. 960671.
[18] Glauert, H., 1947, The Elements of Aerofoil and Airscrew Theory, 2nd ed.,
Cambridge University, Cambridge, England.
[19] Loeffler, A. L., Jr., and Steinhoff, J. S., 1985, Computation of Wind Tunnel
Wall Effects in Ducted Rotor Experiments, J. Aircr., 22(3), pp 188192.
[20] Barlow, J. B., Rae, W. H., and Pope, A., 1999, Low Speed Wind Tunnel Testing,
3rd ed., John Wiley and Sons Inc., New York.
[21] Fitzgerald, R. E., 2007, Wind Tunnel Blockage Corrections for Propellers,
MS thesis, University of Maryland, College Park, MD.
[22] Hansen, M. O. L., 2000, Aerodynamics of Wind Turbines, James and James
Publishing, London.

SEPTEMBER 2011, Vol. 133 / 031201-9

Downloaded From: http://energyresources.asmedigitalcollection.asme.org/ on 10/28/2013 Terms of Use: http://asme.org/terms

S-ar putea să vă placă și