See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/225089248

Near-wake flow structure downwind of a wind turbine in a turbulent boundary

layer

Article in Experiments in Fluids · May 2012

DOI: 10.1007/s00348-011-1250-8

CITATIONS READS

193 550

3 authors:

Wei Zhang Corey Markfort

Cleveland State University University of Iowa
41 PUBLICATIONS 1,094 CITATIONS 45 PUBLICATIONS 1,139 CITATIONS

SEE PROFILE SEE PROFILE

Fernando Porté-Agel
École Polytechnique Fédérale de Lausanne
289 PUBLICATIONS 14,437 CITATIONS

SEE PROFILE

All content following this page was uploaded by Wei Zhang on 21 March 2023.

The user has requested enhancement of the downloaded file.

Exp Fluids (2012) 52:1219–1235
DOI 10.1007/s00348-011-1250-8

RESEARCH ARTICLE

Near-wake flow structure downwind of a wind turbine

in a turbulent boundary layer
Wei Zhang • Corey D. Markfort • Fernando Porté-Agel

Received: 10 May 2011 / Revised: 29 November 2011 / Accepted: 7 December 2011 / Published online: 21 December 2011
Ó Springer-Verlag 2011

Abstract Wind turbines operate in the surface layer of was used to measure velocities in multiple vertical stream-
the atmospheric boundary layer, where they are subjected wise planes (x–z) and vertical span-wise planes (y–z). In
to strong wind shear and relatively high turbulence levels. particular, we identified localized regions of strong vor-
These incoming boundary layer flow characteristics are ticity and swirling strength, which are the signature of
expected to affect the structure of wind turbine wakes. The helicoidal tip vortices. These vortices are most pronounced
near-wake region is characterized by a complex coupled at the top-tip level and persist up to a distance of two to
vortex system (including helicoidal tip vortices), unstead- three rotor diameters downwind. The measurements also
iness and strong turbulence heterogeneity. Limited infor- reveal strong flow rotation and a highly non-axisymmetric
mation about the spatial distribution of turbulence in the distribution of the mean flow and turbulence structure in
near wake, the vortex behavior and their influence on the the near wake. The results provide new insight into the
downwind development of the far wake hinders our capa- physical mechanisms that govern the development of the
bility to predict wind turbine power production and fatigue near wake of a wind turbine immersed in a neutral
loads in wind farms. This calls for a better understanding of boundary layer. They also serve as important data for the
the spatial distribution of the 3D flow and coherent tur- development and validation of numerical models.
bulence structures in the near wake. Systematic wind-tun-
nel experiments were designed and carried out to
characterize the structure of the near-wake flow downwind 1 Introduction
of a model wind turbine placed in a neutral boundary layer
flow. A horizontal-axis, three-blade wind turbine model, Turbulent wakes downwind of horizontal-axis wind tur-
with a rotor diameter of 13 cm and the hub height at bines have been studied extensively through field obser-
10.5 cm, occupied the lowest one-third of the boundary vation, laboratory experiments and numerical simulations
layer. High-resolution particle image velocimetry (PIV) (Crespo et al. 1999). Knowledge of wake development
behind the turbine rotors is important for the minimization
of wake interference effects and the optimization of wind
W. Zhang  C. D. Markfort turbine performance in wind farms. As Whale et al. (1997)
Saint Anthony Falls Laboratory, Department of Civil pointed out, the mean wake characteristics, their relation to
Engineering, University of Minnesota,
the incident wind field and the local topography, influence
Minneapolis, MN 55414, USA
e-mail: wzhang@umn.edu the total energy resource at a potential wind farm site.
Furthermore, the turbulence structure, and particularly the
C. D. Markfort
e-mail: mark0340@umn.edu turbulence intensity distribution in the wake, affects the
fatigue loading of downwind turbines. These factors should
F. Porté-Agel (&) be considered for optimizing the layout of wind farms. The
Wind Engineering and Renewable Energy Laboratory (WIRE),
design of wind farms, therefore, can benefit significantly
École Polytechnique Fédérale de Lausanne (EPFL),
ENAC-IIE-WIRE, 1015 Lausanne, Switzerland from detailed information about these fundamental wake
e-mail: fernando.porte-agel@epfl.ch characteristics.

123
1220 Exp Fluids (2012) 52:1219–1235

Aerodynamic research of wind turbines has contributed and all components of the turbulent stress tensor in the near
significantly to the success of modern wind energy. Inter- wake.
ested readers are referred to a comprehensive review of Turbulence intensity is used as a measure of the fatigue
wind-turbine-wake research conducted by Vermeer et al. loads on various components of the turbine. Also, the power
(2003). Wind-tunnel experiments have been widely used to fluctuations of a wind turbine depend primarily on the tur-
study wind turbine wakes (e.g., Whale et al. 2000; Ver- bulence intensity of the wind. Models for turbulence levels
meer 2001; Medici and Alfredsson 2006; Dobrev et al. in turbine wakes do not consider cross-sectional variability.
2008). These measurements have provided valuable In many applications, it is of importance to estimate reliable
insights into the aerodynamics of wind turbines, such as turbulence stresses in the wake. For example, to calculate
studies by NREL at NASA Ames (Simms et al. 2001) and the unsteady loads induced by turbulence, it is critical to
the recent European-funded MEXICO wind-tunnel exper- understand the spatial correlation structure of the turbulence
iments which included detailed flow-field measurements in in all three directions. To date, no accurate model exists for
the wake of a 4.5 m diameter rotor (Snel et al. 2007; describing a turbulent wind field within the turbine wake,
Sørensen 2011). Data collected under well-controlled especially in the near-wake region.
conditions are particularly useful to validate the perfor- The major source of three-dimensionality of the near
mance of numerical wake models. However, most previous wake is the complicated vortical system that forms
laboratory studies were performed under uniform, free- immediately behind the rotor. Particle image velocimetry
stream flow conditions (see review by Vermeer et al. (PIV) measurements are able to yield the turbulence sta-
2003), which are not fully representative of the non-uni- tistics of the wake flow and also reveal the detailed vortex
form characteristics of the atmospheric boundary layer structure of the wake by mapping the vorticity contours.
where wind turbines operate. Only recently, wind-tunnel One of the earliest applications of PIV to study wind tur-
experiments have been designed and carried out with the bine wakes is by Whale et al. (1997). They measured mean
purpose of understanding the effects of the incoming flow and turbulent wake characteristics at distances of l.l and 1.5
(free-stream vs. boundary layer over rough and smooth rotor diameters away from the model turbine in the near
surfaces) and thermal stability (stable vs. neutral) on the wake and compared their results with full-scale data col-
wake structure by Chamorro and Porté-Agel (2009) and lected in the field. Additionally, Whale et al. (2000) carried
Chamorro and Porté-Agel (2010), respectively. Also, out a series of experiments to characterize tip vortex
Hancock and Pascheke (2010) investigated the behavior of properties of a two-blade turbine in the free-stream flow in
a stand-alone and multiple wind turbine wakes in a stable a water tank. Grant and Parkin (2000) also made PIV
boundary layer wind tunnel. Cal et al. (2010) studied the measurements of two- and three-blade wind turbines in
vertical transport of kinetic energy from a neutral boundary order to track the tip vortices. Medici and Alfredsson
layer flow to an array of model wind turbines using wind- (2006) studied the wind turbine wake under yaw condi-
tunnel experiments. These works highlight the importance tions. Dobrev et al. (2008) presented quantitative infor-
of considering the effects of boundary layer flows on wind mation of the wake downstream of a model wind turbine.
turbine wake structures. Most of the experiments mentioned previously were done
Previous measurements behind turbine models have with a free-stream inflow condition, without simulating an
mainly been limited to measurements of the stream-wise or incoming boundary layer flow. As stated previously, it is
axial velocity component in order to determine the wake essential to correctly simulate the boundary layer with high
development and the velocity deficit. The velocity deficit in shear and high turbulence level, which significantly affects
the far wake is usually assumed as axisymmetric and self- the wake structure (e.g., Chamorro and Porté-Agel 2009).
similar in analytical models (Burton et al. 2001). However, In this study, we report new wind-tunnel data collected
the near wake is characterized by a complex coupled vortex in the near wake of a model wind turbine immersed in a
system (including helicoidal tip vortices), unsteadiness and neutrally stratified boundary layer. Previous work by
strong turbulence heterogeneity. Limited information about Chamorro and Porté-Agel (2009) discussed the velocity
the spatial distribution of turbulence in the near wake, the deficit and enhanced turbulence in the far wake of the wind
vortex behavior and their influence on the downwind turbine in great detail. Our particular focus is on under-
development of the far wake hinders our capability to pre- standing and quantifying the development of the near wake
dict wind turbine power production and fatigue loads in by revealing the tip vortex properties and characterizing the
wind farms. For instance, Hansen and Butterfield (1993) spatial distribution of turbulence statistics, including mean
pointed out that knowledge of all three velocity components velocities, turbulence intensities and Reynolds shear
in the flow field is essential in order to predict rotor loads in a stresses. This study seeks to improve our understanding of
wind farm. Therefore, it is of great interest to understand the the turbulent wake structure, especially in the complex
full velocity field, including all three velocity components near-wake region.

123
Exp Fluids (2012) 52:1219–1235 1221

2 Experimental setup PIV or S-PIV system from TSI, Inc. was used to measure
three velocity components in 2D planes from two orthog-
2.1 Boundary layer wind tunnel onal views: vertical stream-wise (x–z) planes and vertical
span-wise/cross-sectional (y–z) planes. A dual-head Nd:Yag
Experiments were carried out in the recirculating boundary laser (Quantel-USA) was used as the light source which can
layer wind tunnel at the Saint Anthony Falls Laboratory of produce a laser pulse of 190 mJ. The laser beam was
the University of Minnesota. The main test section, transferred by an optical arm to the wind-tunnel test section
designed for atmospheric boundary layer studies, has a and transformed into a light sheet using cylindrical and
development length of 16 m and a cross-section of spherical lenses. Two PowerView Plus 4MP 12-bit CCD
1.7 9 1.7 m2. A turbulent boundary layer was developed cameras (2,048 9 2,048 pixels) fitted with 105 mm lenses
with the help of a tripping mechanism (80 mm picket were setup with Scheimpflug mountings to capture particle
fence) at the entrance of the wind-tunnel test section. The images in x–z planes. In addition to S-PIV measurements,
test section floor is made up of a series of smooth alumi- conventional 2-Dimensional, 2-Component (2D2C) PIV
num plates for which temperature can be controlled by an was also performed in multiple y–z planes, at x/d = 1, 2, 3
automated heating and cooling system, to achieve the and 5, with the help of a reflection mirror located at eight
desired thermal stratification condition (neutral, stable or rotor diameters downwind from the measurement plane.
unstable). The range of controlled temperature of both the Figure 1 shows a schematic diagram of the measurement
air and the floor is within ±0.25°C between 5 and 80°C. In setup.
this experiment, the temperature of the air (Tf) and the floor Olive oil particles, tracers for PIV measurements, were
(Ts) were kept at a constant temperature of 30°C to simu- generated using a homemade atomizer. Six cylindrical
late a neutral boundary layer in the test section. The free- reservoirs were filled half-full with olive oil, and com-
stream turbulence intensity is approximately 2% at pressed air was forced through Laskin nozzles to generate
Uf = 2.5 m s-1. More details about the wind tunnel can be 1 lm oil droplets which were released into the upper part
found in Carper and Porté-Agel (2008). of the chamber. The atomized olive oil droplets went into
the entrance of the test section, where they were allowed to
2.2 Measurement methodology mix with the airflow sufficiently and fill homogeneously
prior to measurements.
Two measurement techniques were employed to acquire Calibration of the 2D2C PIV was accomplished by
detailed flow information. A high-resolution stereoscopic imaging a grid target placed in the laser light sheet. This

Fig. 1 Schematic of the

experimental setup, coordinate
system and measurement
locations (top). Geometric
parameters of the model wind
turbine (bottom)

123
1222 Exp Fluids (2012) 52:1219–1235

way a scale factor of the current PIV setup was obtained to probe, at seven inclination angles and multiple wind
convert the length scale from the measurement plane to the velocities at 30 ± 0.25°C. A cubic-spline lookup table
image plane. S-PIV calibrations followed the multi-plane calibration method was then used to determine instanta-
procedure described by Raffel et al. (2007). Particle ima- neous velocities from the two instantaneous voltages.
ges were evaluated to determine instantaneous velocity Calibration of the anemometer was performed before and
fields using a multigrid method having second-order after the experiments to verify the calibration throughout
accuracy (Insight 3G program from TSI, Inc.). The final the entire experiment. During the calibration and mea-
interrogation window of 32 9 32 pixels is analyzed with surements, the temperature fluctuations were held with-
50% interrogation window overlapping. Erroneous vectors, in ±0.25°C to avoid bias errors due to thermal drift of the
less than 1% of the total calculated vectors, were replaced voltage signal. More details on the calibration procedure
by vectors interpolated through a Gaussian scheme from can be found in Bruun (1995). Velocities measured with
valid neighboring vectors. The field of view (FOV) and the hot-wire can be considered accurate to at best 1%. The
spatial resolution of representative PIV measurements are accuracy of the turbulence intensity in a wall-bounded flow
summarized in Table 1. The mean velocity field was depends on a complex combination of probe length,
obtained by ensemble averaging 1,500 instantaneous wall-normal positioning and sensor Reynolds number (see
velocity fields. All fluctuating velocity fields, derived by Jorgensen 1996; Hutchins et al. 2009).
subtracting the mean velocity field from the instantaneous
velocity fields, were used to compute turbulence intensities 2.3 Experimental conditions
and Reynolds shear stresses. Because the flow is time
dependent, these results represent a statistical average of The experiments were conducted with a free-stream
the flow field for a large number of rotor revolutions. The velocity of approximately 2.5 m s-1. A boundary layer was
measurement uncertainty level of the velocity vectors is allowed to grow to d & 0.6 m at the measurement location.
estimated to be within 2% and approximately 6% for the The friction Reynolds number of this neutrally stratified
vorticity using a central difference scheme. boundary layer (based on the friction velocity of
In addition, a hot-wire anemometer (HWA) was used to u ¼ 0:12 m s1 ) is Rem = u*d/m & 4.6 9 103. The aero-
measure instantaneous stream-wise and wall-normal dynamic surface roughness length z0 was estimated to be
velocity components with high temporal resolution at 0.021 mm from the log-law profile using averaged stream-
selected downwind locations. The probe is made of two wise velocity in the surface layer (approximately lowest
5 lm tungsten wires, which are connected to an A. A. Lab 10–15% of the boundary layer). The normalized standard
Systems AN-1003 10-channel constant temperature ane- deviation of the stream-wise and vertical velocity compo-
pffiffiffiffiffiffi
mometry (CTA) system. The sensor was mounted on a nents near the surface is u0 2 =u = 2.47–2.25 and
traversing system (Velmex, Inc) to measure vertical pro- pffiffiffiffiffiffiffi
files at multiple locations. As shown in Fig. 1, the cross- w0 2 =u = 1.08–1.14, falling into the typical value range
wire anemometer was positioned at selected downwind of the surface layer values in a neutrally stratified atmo-
stream-wise locations in the central plane (y = 0) of the spheric boundary layer simulated in wind tunnels:
pffiffiffiffiffiffi pffiffiffiffiffiffiffi
turbine wake (x/d = 2, 3, 5, 10, and 20). Data were col- u0 2 =u = 2.1–2.5 and w0 2 =u = 1.0–1.2 (e.g., Grant
lected in vertical profiles starting at z = 1 cm up to z = 40 1992; Gong and Ibbetson 2009; Athanassiadou and Castro
cm with intervals of 1 cm between measurement points. 2001; Khurshudyan et al. 1981). Figure 2 shows the ver-
Data were sampled for 60–120 s at a frequency of 1,000 or tical profiles of the mean stream-wise and vertical veloci-
2,000 Hz depending on the specific stream-wise location. ties and turbulence intensities measured in the boundary
The hot-wire anemometer was calibrated in a well- layer normalized by the rotor diameter d and the incoming
controlled small calibration unit against a Pitot-static mean velocity measured at the rotor hub height (Uhub).
The miniature wind turbine model is designed in a way
that the key geometric parameters are comparable to those
Table 1 Field of view and spatial resolution of PIV measurement
cases of large-scale wind turbines in the field. There is only a
limited number of wind-tunnel experiments designed to
Measurement station Field of Spatial Vectors
view resolution
study the wake of rotating wind turbine models. Most
(mm2) (mm) experiments involve the use of stationary models (e.g.,
porous disks), as summarized in Vermeer et al. (2003) and
x–z planes (x/d = 0–3) 146.9 9 135.4 2.3 128 9 118 Medici and Alfredsson (2006). The model wind turbine
y–z planes (x/d = 1, 2, 149.8 9 188.7 3.3 92 9 116 employed here is a functional turbine that consisted of a
3, 5)
three-blade GWS/EP-5030 9 3 rotor driving a small DC

123
Exp Fluids (2012) 52:1219–1235 1223

Table 2 Operational parameters of three-blade wind turbine models

used in wind-tunnel studies
Studies Tip speed Power Thrust
ratio, k coefficient, Cp coefficient, CT

The present study 3.7 0.37 0.42

Medici and 0.42–4.87 0–0.32 0.13–0.35
Alfredsson (2008)
Cal et al. (2010) 4.9 0.29 0.32
Hancock and 6.05 0.293 0.497
Pascheke (2010)

number (Medici and Alfredsson 2006). This highlights the

value of wind turbine wake data acquired in wind tunnels
for the purpose of understanding the fluid dynamics. Fur-
thermore, as long as the aerodynamic characteristics of the
Fig. 2 Characteristics of the NBL simulated over a smooth surface:
a Normalized mean stream-wise (circle) and vertical (diamond) rotor blades are known in the modeled Reynolds number
velocities and b Stream-wise (circle) and vertical (diamond) turbu- range, wind-tunnel data of turbine wakes provide valuable
lence intensities. Height above the surface z is normalized with the benchmarks for comparison with and validation of numer-
rotor diameter d
ical models (Vermeer et al. 2003).
Throughout this paper, we use the following definitions:
generator to extract energy from the wind. The blades of the wind velocities U, V, W are mean wind speeds in the
this rotor are nearly flat plates, similar to those of the small stream-wise (x), lateral/span-wise (y) and vertical
homemade wind turbine models used in Lebrón et al. (z) directions, and u0 , v0 and w0 are deviations of the
(2010) and Cal et al. (2010). The tower is 10.5 cm and the instantaneous wind speed components (u, v, and w) from
rotor diameter d is 13 cm, with the bottom tip at 4 cm the mean wind speed.
above the ground and the top tip at 17 cm. As shown in
Fig. 1, the bottom tip of the turbine is set at the height
about 0.6 times the turbine radius, which is similar to that 3 Results and discussion
of large-scale turbines; and the turbine rotor plane is within
the lowest one-third of the boundary layer. 3.1 Signature of tip vortices
The rotor operates at 1,140–1,160 rpm with a free-
stream velocity of 2.5 m s-1, giving a tip speed ratio Tip vortices are distinct features of horizontal-axis wind
kð¼ Xð2p=60Þðd=2Þ=Uhub , that is, ratio of the tangential tip turbine wakes; they are shed from the tip of each blade and
velocity to the undisturbed wind speed measured at the hub transported downwind and eventually dissipated. Observed
height) of approximately 3.7. This tip speed ratio falls into both in free-stream flow and boundary layer flow, tip
the range of k for typical three-blade wind turbine models vortices follow a helical path and dissipate faster in the
used in wind-tunnel tests, from 3 to 6.7 (Vermeer et al. latter than in the former conditions (Pedersen and Antoniou
2003). Note that full-scale wind turbines operate in a wider 1989; Porté-Agel et al. 2011). In order to identify coherent
range of tip speed ratios, with the optimum values around vortex structures, we first used PIV data to map the mean
6–8. For the conditions of this experiment (k = 3.7), the and instantaneous fields of vorticity and swirling strength.
thrust coefficient CT and the power coefficient Cp are Contours of the ensemble-averaged stream-wise vortic-
estimated to be 0.42 and 0.37. The coefficients are close to ity (xx = 0.5 (qv/qz - qw/qy)) calculated using PIV data
those of wind turbine models used in previous wind-tunnel in the vertical span-wise (y–z) planes are in Fig. 3. The
experiments, as listed in Table 2. stream-wise vorticity contours show a ‘‘double ring’’ pat-
In general, the Reynolds number of the atmospheric flow tern at x/d = 1. The intensive vorticity can be interpreted
cannot be matched by wind-tunnel experiments. The as the signature of the helicoidal tip vortices advected
Reynolds number in this study is lower than that of a full- through the measurement plane and the shear layers formed
scale wind turbine flow by a factor of about 100. Even on the edge of the wake. In particular, the magnitude of
though the influence of Reynolds number on turbine wakes vorticity is noticeably higher at the edge of the upper half
is not fully understood, previous studies have shown that of the rotor-swept area, owing to the non-uniformity of the
primary wake characteristics such as wake rotation and tip incoming boundary layer flow. In addition, at the center of
vortices can be reproduced even at relatively low Reynolds the rotor-swept area, strong negative vorticity is generated

123
1224 Exp Fluids (2012) 52:1219–1235

Fig. 3 Contours of the

1.5 1.5
ensemble-averaged stream-wise
vorticity (xx) in the span-wise 0.0013 0.00
13

planes of the near wake. 0.0

056
0.0
027

a x/d = 1; b x/d = 2; c x/d = 3; Vorticity ωx (1/s) Vorticity ωx (1/s)

3
0.00 1

01
13 00 -0.0030
0.0070 -0.0 -0 0.0070

0.0
and d x/d = 5. The dashed line 1 -0.0059
0.0041 1
.00
30
0.0041

0 13
3
01
indicates the turbine rotor plane 0.0013 0.0013

0
0.0
0.0

-0.003
059
-0.0
-0.0016 -0.0016
-0.0044 -0.0044

z/d

z/d
13
-0.0073 0.0013 -0.0073

0.0013
0.00
-0.0016
-0.0101 -0.0101
-0
-0.0059 -0.0130 .0
0 16
-0.0130

13
01
0.5 0.0041 0.0013 0.5 0

0.00
.0
-0
0.0
01 0.0013
3
3 0.0
01 01
3
0.0

44
05
- 01 3

0
7
02

0.0

.0
-0
0.0

0.0
0 (a) =1 0 (b) =2
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

1.5 1.5

0.0027 0.0013

13
00 -0.0001
0.

Vorticity ωx (1/s) -0 Vorticity ωx (1/s)

-0.0

.0
-0.0001 -0.0001 00
01

0.0070 1 0.0070
-0.0

1
6

00
1 0.0041 1 0.0041
0 16

.0
-0
-0.0001 -0.0001
-0.0016
0.0013 -0.0001
0.0013
-0.0016 -0.0016
-0.0044 -0.0044
z/d

z/d
0.0

-0.0073 -0.0073
-0.0016

-0.0001
01

-0.0001
3

-0.0101 -0.0101
-0.0130 -0.0130
1
00

-0.0001
-0.0

13
0.5 0.0013
0.
00 0.5 0.0013 -0.0001

0.0013 -0.0001

0 01
.0 -0.0001
-0

-0
.00
-0.0001
-0.0001 -0.0001 01

0 (c) =3 0 (d) =5
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

around the hub, which is the signature of the hub/root

vortices. As expected, the intensive vorticity expands and
diffuses with downwind distance, indicative of wake
expansion. At x/d = 5, the vorticity distribution becomes
nearly homogeneous over the entire rotor-swept area, and
the magnitude of xx is higher than that of the incoming
flow.
Since concentrated vorticity can be generated by either
coherent vortices or shear layers, swirling strength (kci)
(Adrian et al. 2000; Zhou et al. 1999) was used to improve
the identification of vortex cores. Defined as the imaginary
Fig. 4 Contours of the ensemble-averaged span-wise vorticity (xy)
part of the complex eigenvalues of the complex velocity overlaid with swirling strength (kci) distribution in the center plane of
gradient tensor, the swirling strength is used to differentiate the wake (y = 0, x/d = 0–3). The contours of swirling strength are
spiraling motion around an axis normal to the 2D measured presented using red lines. The horizontal dashed line indicates the
plane from rotation caused by a shear layer. Figure 4 shows turbine hub height
the color-coded span-wise vorticity (xy = 0.5(qw/qx -
qu/qz)) distribution overlaid with swirling strength (red vorticity as well as swirling strength contours is found at
lines) in the central wake plane at x/d = 0–3. The distri- the top-tip height, but not at the bottom-tip level. This is
bution of swirling strength is well matched with vorticity. due to the fact that tip vortices dissipate slower and thus
Strong positive and negative vorticities in the upper and persist over a longer distance at the top-tip height, where
lower part of the wake immediately behind the nacelle the incoming flow has a relatively low turbulence level
indicate alternating vortex shedding. A band of intensive compared to a higher turbulence level at the bottom-tip

123
Exp Fluids (2012) 52:1219–1235 1225

height. The hub/root vortices are significant prior to The period of tip vortex shedding can be estimated with
x/d = 1.5 and then rapidly dissipate beyond this distance. the distance between two tip vortices (&0.25d) and the
The top-tip vortices are still visible as far as x/d = 2. Lu stream-wise advection velocity (1.8 m s-1) at the top-tip
and Porté-Agel (2011) also found top-tip vortices persist height in instantaneous velocity fields (see Fig. 5). Using
until about two rotor diameters downwind of the turbine, in this approach, the shedding frequency of top-tip vortices is
their high-resolution large-eddy simulations (LES, with an estimated to be approximately 55 Hz. Additionally, spec-
actuator line model) of the wake behind a wind turbine tral analysis was performed to examine the distribution of
immersed in a stable boundary layer. energy across a range of frequencies in the wake. The
Top-tip vortices can also be identified in the instanta- spectra were calculated by taking the fast Fourier transform
neous velocity fields immediately downwind of the turbine (FFT) of the velocity fluctuations (u0 and w0 , 216 samples)
(x/d = 0–1), as shown in Fig. 5. Four or five tip vortices are utilizing a square window of 212 points. These velocity
visible at the top-tip height, but not at the bottom-tip fluctuations were measured by HWA at 1,000 Hz for 120 s.
height. Reviewing the series of instantaneous velocity The analysis shows the classical production subrange and
fields, we notice that tip vortices occur consistently and not inertial subrange for the surface layer turbulence, identified
randomly. Qualitatively the tip vortices observed here are as regions that follow power law scaling with -1 and -5/3
very similar to those reported in Dobrev et al. (2008). slope, respectively (Katul et al. 1995). There are also
The present PIV data reveals a complicated vortex sys- localized high-energy signatures at a frequency corre-
tem, composed of tip vortices and hub/root vortices in the sponding to periodic coherent structures, that is, the tip
near-wake downwind of a model wind turbine. The hub/root vortices and large-scale wake meandering motions.
vortices are significant prior to x/d = 1.5 and dissipate fast. A concentration of turbulent kinetic energy is observed
The top-tip vortices remain until x/d = 2. Previous work to at frequencies of 55 Hz (3ft), 110 Hz (6ft) and 165 Hz (9ft),
visualize tip vortices shed from the wind turbine blades which are multiples of the frequency of consecutive
using PIV was reported in Dobrev et al. (2008), Grant et al. rotating blades 3ft (ft is the rotation frequency of the tur-
(1997), Grant and Parkin (2000) and Sherry et al. (2010). bine). Specifically, at x/d = 0.5, 1.0 and 1.5, a distinct
Among them, Whale et al. (2000) and Sherry et al. (2010) primary peak associated with the presence of tip vortices is
also revealed the hub/root vortices immediately behind the observed at the top-tip level (Fig. 6). The primary peak is
hub. In a low turbulence flow or in the free-stream, previous at a frequency of 3ft and the first and second harmonic
measurements show that the tip vortices are present beyond frequencies are seen at x/d = 0.5. The number of observed
a distance of three rotor diameters, while the vortices behind harmonic frequencies decreases with increasing distance
the hub dissipate faster. One should note that the relative from the turbine. At x/d = 1, the primary and the first
strength of hub/root vortices versus tip vortices observed in harmonic frequency can be seen while only the primary
a wind tunnel may not be characteristic of field scale flows frequency remains at x/d = 1.5 and 2. Note that the fre-
due to the ratio of hub to rotor blade size of a wind turbine. quency peak is not observed at the bottom-tip level
This ratio for wind turbines in the field is much smaller than (Fig. 6f), where the turbulence level of the incoming flow
that of the model wind turbine used in the present and most is relatively higher than that at the top-tip height and the
wind-tunnel studies, which suggests that relatively weaker wake interacts with the ground. This observation agrees
hub/root vortices may occur in the field. with the PIV results presented previously.

Fig. 5 Tip vortex signature in

instantaneous velocity fields at
x/d = 0–1; a Snapshot 1;
b Snapshot 2. The contours of
swirling strength are presented
using red lines

123
1226 Exp Fluids (2012) 52:1219–1235

−1 −2
10 10
(a) top-tip ( = 0.5) (b) top-tip ( = 0.5)
−2
10 −3
β = −1 10

−3
10
−4
s )

s )
10
−1

−1
−4
10
β = −5/3
2

2
Su(f) (m

Sw(f) (m
−5
10
−5
10
−6
10
−6 3f = 55
10 t
3ft = 55 6ft = 110
6f = 110 −7 9f = 165
−7 t 10 t
10

−8 −8
10 −1 0 1 2 3
10 −1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
f (Hz) f (Hz)

−2 −2
10 10
(c) top-tip ( = 1) (d) top-tip ( = 1.5)
−3 −3
10 10

−4 −4
s )
s )

10 10
−1
−1

2
2

S (f) (m
Sw(f) (m

−5 −5
10 10
w

−6 −6
10 10
3ft = 55 3ft = 55
6f = 110
t
−7 −7
10 10

−8 −8
10 −1 0 1 2 3
10 −1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
f (Hz) f (Hz)

−2 −2
10 10
(e) top-tip ( = 2) (f) bottom-tip ( = 2)
−3 −3
10 10

−4 −4
Sw(f) (m2 s )

s )

10 10
−1

−1
2
S (f) (m

−5 −5
10 10
w

−6 −6
10 10
3f = 55
t

−7 −7
10 10

−8 −8
10 −1 0 1 2 3
10 −1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
f (Hz) f (Hz)

Fig. 6 Power spectrum of the velocity fluctuations (u0 and w0 ) at the x/d = 0.5, c Sw(f) at x/d = 1, dSw(f) at x/d = 1.5, e Sw(f) at x/d = 2
top-tip and bottom-tip heights in the center plane (y = 0) of the wake (top-tip) and f Sw(f) at x/d = 2 (bottom-tip)
at selected downwind locations. a Su(f) at x/d = 0.5, b Sw(f) at

We also investigated the strength of the tip vortices in Fig. 7. The spectra show stronger tip vortex signature at
the side-tip planes. The power spectra of u0 at the hub the near-tip side (y/d = -0.5) compared to the far-tip side
height and 0.5d, downwind of the turbine are examined in (y/d = 0.5). This is due to the fact that the wake rotates in a

123
Exp Fluids (2012) 52:1219–1235 1227

−1 −1
10 10

−2 −2
10 10 β = −1
β = −1
−3 −3
10 10
s )

s )
−1

−1
−4 β = −5/3 −4 β = −5/3
10 10
2

2
Su(f) (m

Su(f) (m
−5 −5
10 10

−6 −6
10 10 3ft = 55
3ft = 55
6ft = 110
6ft = 110
−7 −7
10 10

−8 −8
10 −1 0 1 2 3
10 −1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
f (Hz) f (Hz)

Fig. 7 Power spectrum of the stream-wise velocity fluctuation (u0 ) at the hub height and 0.5d downwind of the turbine. The near-tip side
y/d = -0.5 (left) and the far-tip side y/d = 0.5 (right)

counterclockwise direction (viewed from the downwind of is between 0.12 and 0.2 and in our case about 0.35. We,
the turbine), thus bringing helicoidal vortices from below therefore, conclude that the low frequency motion detected
(where the background turbulence level is higher and the here may be the overall effect of large-scale meandering of
wake interacts with the ground) at y/d = 0.5 and from the wake structure. It should be noted that this self-induced
above (where the background turbulence is weaker) at instability is different from the wake meandering reported
y/d = -0.5. in field studies. In the field, large-scale atmospheric
The power spectra of w0 at the hub height in the center structures, which are comparable in size to the wake, are
plane (y = 0) of the wake shows a strong low frequency the main contributors to wake meandering (Bingöl et al.
peak at about 6 Hz at both x/d = 1 and 2 (Fig. 8). This is 2007; Larson et al. 2008). This wake meandering is cor-
consistent with previous work by Medici and Alfredsson related with the variability in the wind direction, resulting
(2006), who found a spectral peak at a lower frequency in ‘‘smoothed’’ or more homogeneous velocity profiles
(&8 Hz) compared to the tip vortex shedding frequency, (Ainslie 1988; Whale et al. 1997). The type of bluff-body-
and attributed it to ‘‘meandering’’ of the whole wake. The induced wake meandering observed here has not been
resulting Strouhal number (calculated with the turbine rotor reported in field experiments, where near-wake data are
diameter d and the local free-stream velocity) in their cases scarce. It is also worthwhile mentioning that the ratio of the

−2 −2
10 10

−3 −3
10 10

−4 −4
Sw(f) (m2 s−1)
S (f) (m2 s )

10 10
−1

−5 −5
10 10
w

−6 −6
10 10

f = 6 f = 6
−7 −7
10 10

−8 −8
10 −1 0 1 2 3
10 −1 0 1 2 3
10 10 10 10 10 10 10 10 10 10
f (Hz) f (Hz)

Fig. 8 Power spectrum of the vertical velocity fluctuation (w0 ) at the hub height in the center plane (y = 0) of the wake. x/d = 1 (left) and
x/d = 2 (right)

123
1228 Exp Fluids (2012) 52:1219–1235

nacelle size to the rotor diameter of wind turbines is much

smaller in the field, which might make the self-induced
instability effect not as prominent as in the wind tunnel.

3.2 Mean wake characteristics

The near-wake region is characterized by three-dimen-

sional heterogeneously distributed turbulent flow that
results from the complex interaction between the incoming
boundary layer flow and the turbine-induced forces. These
forces are responsible for the reduction in the axial velocity
as well as the resulting rotation in the near-wake region.
They also produce coherent vortex structures behind the
turbine as described earlier. Highly three-dimensional flow
structure is typically observed in the near wake of either
stand-alone wind turbines (e.g., Wu and Porté-Agel 2011)
or turbines in wind farms (see Porté-Agel et al. 2011; Lu
and Porté-Agel 2011). In the following, we present the
spatial distribution of the three velocity components mea-
sured in multiple vertical stream-wise (x–z) and vertical
span-wise (y–z) planes in the near wake.
As mean kinetic energy is extracted from the wind by
the wind turbine, there is a reduction in wind speed behind
the rotor. Figure 9 shows an overall view of the mean wake
structure in terms of three velocity component contours
covering x/d = 0–3 and z/d = 0.3–1.3 (rotor-swept area) in
the central wake plane (y = 0). A low momentum zone
appears with nearly zero stream-wise velocity immediately
downwind of the rotor (x/d B 0.5). Flow reversal behind
the nacelle and blade roots is observed in the region of
Fig. 9 Contours of mean velocities at x/d = 0–3. The stream-wise
x/d B 0.4. The flow reversal zone behind the nacelle is velocity U (top) is normalized by Uhub, while the vertical velocity
fairly large compared to the nacelle size, which indicates W (bottom) and the lateral velocity V (mid) are not normalized
that it may be caused by both the nacelle and the root
vortices. A large pressure gradient induced by strong near- wake as discussed by Chamorro and Porté-Agel (2009).
wake rotation may also contribute to the amplification of This facilitates modeling the velocity deficit in the far wake
the flow reversal zone. This complicated flow region, as axisymmetric and self-similar.
immediately downwind of the rotor, is strongly related to To capture the span-wise variation of the mean stream-
the rotor characteristics, such as the nacelle size, rotor wise velocity (U), contours of the stream-wise velocity are
blade geometry and hub-blade configuration. Expansion of plotted in the center and two side-tip vertical planes in
the flow reversal zone was also reported by Massouh and Fig. 10. Compared with the near-zero velocity zone in the
Dobrev (2007), who explored detailed 3D flow properties central wake plane (y = 0), the stream-wise velocity in the
behind a model wind turbine. Note that there is a lack of two side-tip planes is much higher. However, the velocity
near-wake data (x/d B 1) in the field measurements, hence or momentum deficit is still visible at the hub height and
it is not clear if this flow reversal exists behind the full- further downwind. The stream-wise velocity has higher
scale nacelle or not. Beyond this zone, the stream-wise values in the far-tip plane (y/d = 0.5) than the near-tip
velocity gradually increases and the wake continues to plane (y/d = -0.5), which is attributed to the effect of wake
recover downwind. With respect to the rotor axis, the rotation. This non-symmetric span-wise distribution of
vertical distribution of stream-wise velocity is non-sym- U was also observed from previous wind-tunnel data in a
metric in this neutral boundary layer flow. The non-sym- vertical cross-sectional plane at x/d = 5, presented by Wu
metric distribution of the stream-wise velocity extends and Porté-Agel (2011). It should be noted that the span-
throughout the wake region. Subtraction of the incoming wise variation of the stream-wise velocity distribution still
flow results in the stream-wise velocity deficit, which is exists in the far wake (x/d = 5), where the flow rotation is
nearly symmetric with respect to the rotor axis in the far weak.

123
Exp Fluids (2012) 52:1219–1235 1229

The wake of a wind turbine is complicated by the fact

that the rotor is composed of a finite number of blades that
can independently affect the flow mainly through shedding
vorticity (Vermeer et al. 2003). Moreover, due to conser-
vation of angular momentum, the torque exerted by the
wind on the rotor blades, which produces their rotation
(clockwise, viewed from downwind of the turbine), leads to
flow rotation in the opposite (counterclockwise) direction.
The lateral velocity component (V) shows rotation of
the wake, with the positive into and the negative out of the
x–z plane in Fig. 9b. It is noticeable that the negative
velocity contours persist longer compared to the positive
contours in the lower part of the wake. This trend is again
confirmed by the velocity fields captured in vertical span-
wise planes (Fig. 11a, b). The maximum lateral velocity is
about 0.4 m s-1 (20% of the stream-wise velocity magni-
tude) and occurs close to the wind turbine. With increasing
Fig. 10 Contours of the stream-wise velocity distribution in the downwind distance, the lateral velocity decreases due to
central and two side-tip planes immediately downwind of the wind the turbulent mixing as the wake recovers. At x/d = 5,
turbine. (x/d = 0.2–1.2, y/d = 0 and ±0.5) wake rotation is not noticeable in the mean velocity field.

Fig. 11 Lateral velocity

1.5 1.5
contours in span-wise planes at
selected downwind locations. -0
.1
6 -0.16
a x/d = 1; b x/d = 2; c x/d = 3;
-0 .

and d x/d = 5
0.00

-0.2 -0.20
0.00 -0.24
24

4 V (m/s) V (m/s)
6

-0.32 -0.16 -0.24

.1

0.32 0.32
-0

1 0.24 1 0.24
0.16 0.16
0.08 0.08
0.00 0.00
z/d

z/d

-0.08 -0.08
-0.08

0.04 -0.16 -0.04 -0.16

-0.08
-0.24 -0.24
-0.32 -0.32
0.00

0.16 0.00 0.04

-0.24

0.5 0.5
.16

2
.1
-0

-0
-0.0 -0.08
8

0 (a) =1 0 (b) =2
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

1.5 1.5

-0.16 -0.20

-0.24 -0.24
0.00 -0.20 V (m/s) 0.00 -0.16 V (m/s)
-0.08

-0.12 -0.16 0.32 -0.12 0.32

1 0.24 1 0.24
0.16 0.16
0.08 0.08
0.00 0.00
z/d

z/d

-0.08 -0.08
-0.12

-0.08
-0.12 -0.16 -0.16 -0.16
-0.24 -0.24
-0.32 -0.32
0.00 -0.04 0.00 -0.12
0.5 0.5
-0.08 -0.12
-0

-0.08
.0
8

0 (c) =3 0 (d) =5
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

123
1230 Exp Fluids (2012) 52:1219–1235

Similarly, the vertical velocity distribution also indicates of the turbine-induced forces and including rotation effects
wake rotation in both x–z (in Fig. 9c) and y–z planes. (using blade-element momentum theory) yields improved
Contours of the average vertical velocity at the vertical predictions of wind turbine wakes with respect to standard
cross-section planes and downwind location x/d = 5 are actuator disk models, especially in the near-wake region (Wu
shown in Fig. 12. The results show the non-symmetry and Porté-Agel 2011; Porté-Agel et al. 2011).
(with respect to the hub height) of the velocity distribution
owing to the incoming boundary layer flow and the pres- 3.3 Turbulent wake characteristics
ence of the ground surface. The maximum vertical velocity
is approximately 0.24 m s-1, which is 75% of the maxi- The ensemble-averaged PIV data clearly indicates that the
mum of the lateral velocity magnitude at x/d = 1. maximum of the vertical and lateral velocity components,
In summary, due to the effect of the wind turbine, the approximately 10–20% of the magnitude of the stream-
stream-wise velocity is reduced and the lateral and vertical wise velocity, may not be neglected in the near wake. It is
velocities increase close to the rotor. With increasing dis- also important to understand the spatial distribution of
tance to the rotor, the stream-wise velocity increases and the other turbulence statistics, such as turbulence intensities
other velocities reduce. From the velocity field measured in and Reynolds shear stresses, in the near wake. Previous
the multiple vertical stream-wise and vertical span-wise studies of wind turbine wakes found that the turbulence
planes, one can see that the near-wake region is characterized recovered more slowly than the mean velocity in the wind
by high three-dimensionality and strong flow rotation. turbine wakes (Vermeer et al. 2003).
Recent LES studies capture the same flow characteristics and Figure 13 shows the high magnitude of the stream-wise
have shown that accounting for the non-uniform distribution turbulence intensity (Iu = ru/Uhub) within the rotor-swept

Fig. 12 Vertical velocity

contours in span-wise planes at 1.5 -0.1 1.5
6 -0.12
selected downwind locations. -0.16 -0.12
a x/d = 1; b x/d = 2; c x/d = 3;
and d x/d = 5 -0 . W (m/s) 0.0 W (m/s)
4

04 0
-0.0

-0.04
0.32 0.32
1 0.24 1 0.24
-0.

0.16 -0.12 0.16

1

0.08 0.08
6

0.20

0.04
-0.0

0.00 -0.04 0.00

z/d

z/d
0.08
4

-0.08 -0.08
-0.1

0.12
-0.24

-0.16 -0.08 -0.16

-0.12
6

0.08

-0.24 0.00 -0.24

-0 -0.32 -0.32
.0 -0.04
0.5 4 0.5
0.08

0.08
0.08

0.08

0 (a) =1 0 (b) =2
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

1.5 1.5
-0.12 -0.12
-0 .
12 -0.12

0.00 W (m/s) -0.08 W (m/s)

-0.08 -0.08
0.32 0.32
1 0.24 1 0.24
08 0.16 -0.04 0.16
-0 . 0.08 0.08
0.0

0.00 -0.04 0.00

-0.04 -0.04
0
z/d

z/d

0.04 -0.08 0.00 -0.08

-0.04 -0.16 0.00 -0.16
-0.08 -0.04
0.00 -0.24 0.00 -0.24
-0.32 -0.32
0.00 0.00
0.5 0.5

0.08 0.08
0.08 0.08

0 (c) =3 0 (d) =5
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

123
Exp Fluids (2012) 52:1219–1235 1231

the maximum values of Iv and Iw are about 0.18 in the

middle of the rotor-swept plane. This trend agrees well
with that observed in the vertical stream-wise plane (in
Fig. 13b, c). However, the contours of Iv and Iw are not
completely axisymmetric in these cross-sectional mea-
surement planes (x/d = 1, 2 and 3). Instead, the distri-
bution of the turbulence intensities is tilted, owing to
pronounced rotation in the near-wake region. With
increasing downwind distance from the rotor, the magni-
tude of turbulence intensity decreases and the area of
influence spreads out as the wake expands. At x/d = 3,
the high turbulence intensities are distributed close to the
edge of the rotor-swept area. From the results presented
here, it is evident that the areas of high Iw and Iv deviated
slightly from the central vertical plane (y = 0) in the
wake. It should be noted that we are not able to measure
the spatial distribution of Iu in the span-wise planes with
the 2D2C PIV. However, Wu and Porté-Agel (2011)
reported previous wind-tunnel results of Iu measured with
a cross-wire and found a similar tilting distribution, as
shown here for Iw and Iv, in the span-wise plane at
x/d = 5. Moreover, we obtained vertical profiles of Iu and
Iw at selected downwind locations with the HWA (in
Fig. 16). The maximum value of Iw is about 67% of the
maximum Iu, which occurs at x/d = 2 and 3. This agrees
with the PIV data. Also, note that the maximum turbu-
lence intensities occur around the upper edge of the rotor-
swept plane. These results demonstrate the importance of
all three components of turbulence intensity, Iu, Iv and Iw,
Fig. 13 Turbulence intensity contours in the near-wake region (x/d = since they are of comparable magnitude in the near-wake
0–3). The stream-wise turbulence intensity Iu = ru/Uhub (top), region.
vertical turbulence intensity Iw = rw/Uhub (mid) and lateral turbu- Furthermore, the Reynolds shear stress distribution in
lence intensity Iv = rv/Uhub (bottom) the vertical stream-wise plane are presented in Fig. 17.
High levels of the stream-wise, wall-normal Reynolds
area, which is about 0.15–0.18 at x/d B 1.5. The maximum
stream-wise turbulence intensity occurs at the upper and shear stress (u0 w0 ) occur in the shear layers immediately
lower parts of the shear layers immediately behind the behind the nacelle and again at the top-tip level at
nacelle, where shear-produced turbulence is substantial. In x/d = 2.5–3. The distribution is very similar to that of Iu,
the region of x/d = 0.5–1.2, the maximum lateral and resulting from the strong wind shear and associated
vertical turbulence intensities (Iw and Iv) are found at the mechanical production of turbulent kinetic energy. The
hub height. Further downwind of the turbine, Iu decays in maximum value of the Reynolds stress component v0 w0 is
the shear layers and it reaches a maximum at the top-tip about half of u0 w0 and is also found behind the nacelle. It
level at about x/d = 3. This finding is consistent with the reduces with increasing downwind distance, except at the
previous measurements as well as the LES results reported top-tip level (about x/d = 3) where a relatively high
by Porté-Agel et al. (2011), which found the maximum magnitude is also found. The distribution of the Reynolds
stream-wise turbulence intensity occurs at the top-tip level, stress component v0 w0 in the vertical span-wise planes
about two to five rotor diameters downwind of the turbine. (not shown here) is roughly symmetric with respect to the
The significant turbulence enhancement is associated with rotor axis in the near wake (x/d = 1 and 2). It should be
strong wind shear and high mechanical production of tur- noted that PIV results from two orthogonal views are
bulent kinetic energy around the top-tip level. consistent with each other. As the downwind distance from
The variation of the turbulence intensities in the span- the turbine increases, the wake expands and the Reynolds
wise direction can be seen in Figs. 14 and 15. At x/d = 1, shear stresses decrease.

123
1232 Exp Fluids (2012) 52:1219–1235

Fig. 14 Lateral turbulence

0.
1.5 1.5 0.07

06
intensity Iv (rv/Uhub)
distribution in span-wise planes
at selected downwind locations. 0.0

08
0.08 0.1
8 0

0.
a x/d = 1; b x/d = 2; c x/d = 3;

0.11
Iv Iv

0.
0.12 0.13

10
and d x/d = 5

08
0.18 0.09 0.18

0.
1 0.16 1 0.16
10

0
0.14 0. 0.14

0.1

0.12
4
0.17 0.12 0.12

0.1
0.14

0.10
0.10 0.12 0.10

z/d

z/d
0.08 0.08

3
16

0.0
0.13

0.1
0.06 0.08 0.06
0.

0.1
0.12

8
0.14 0.04 0.12 0.04

2
0.02 0.02
0.00 0.00
0.5 0.5

08
0.09

8
0.07

0.0
0.
0.
0.11

10
08

0.08

07
0. 0.07

0.
0 (a) =1 0 (b) =2
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

1.5 0.07 1.5 0.08

0
0.1
08

0.11 0.08
0.

0.12

0.11 Iv 0.11 0.10 Iv

0.18 0.18
1 0.16 1 0.16
0.1

0 0.14 0.10 0.14

1

0.1 0.09
0.12 0.12
0.11 0.10 0.10 0.10
z/d

z/d
0.08 0.08
0.08 0.11 0.06 0.08 0.06
0.11 0.09
0.04 0.04
0.02 0.02
0.00 0.00
0.5 0.09 0.08
0.5 0.08
0.11 0.09

0.08 0.08

0 (c) =3 0 (d) =5
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

4 Summary and conclusions Due to the effect of the wind turbine, which extracts
momentum from the flow while inducing rotation, the
Wind-tunnel measurements of the turbulent wake down- stream-wise velocity decreases strongly and the lateral
wind of a model wind turbine were conducted in a neutrally and vertical velocities increase immediately behind the
stratified boundary layer flow. We carried out high-reso- rotor. As the wake grows with increasing downwind dis-
lution PIV measurements in two orthogonal views and tance from the rotor, the stream-wise velocity increases
HWA measurements at multiple downwind locations to and the other velocity components decrease. From the
capture the signature of coherent tip vortices and map the velocity fields measured in the multiple vertical stream-
spatial variation of the mean flow and turbulence statistics wise wake planes and vertical span-wise planes, it is
in the near wake. shown that the near-wake region is characterized by high
The spatial distribution of vorticity and swirling strength three-dimensionality, turbulence heterogeneity and strong
measured with PIV reveals the presence of top-tip vortices, flow rotation. Accounting for the non-uniformity and
which persist up to about 2–3 rotor diameters downwind of rotation of the flow induced by the turbine is important to
the turbine, longer than the hub/root vortices in the near obtain accurate simulation of wind turbine wakes, as
wake. The power spectrum of velocity fluctuations con- recently shown by Wu and Porté-Agel (2011) and Porté-
firms the vortex shedding with strong spectral peaks at Agel et al. (2011).
multiples of the frequency of consecutive rotating blades at Spatial distributions of turbulence intensities and Rey-
the top-tip and side-tip levels. Using neither PIV nor HWA nolds shear stress display strong variability, with the three
data, can we find a similar tip vortex signature at the bot- turbulence intensities having comparable values in the near
tom-tip height, which is attributed to the higher background wake. The significant turbulence enhancement at a distance
turbulence level compared to the top-tip height. of three rotor diameters is associated with strong wind

123
Exp Fluids (2012) 52:1219–1235 1233

Fig. 15 Vertical turbulence

1.5 1.5
intensity Iw (rw/Uhub) 0.06
distribution in span-wise planes 0.0
8
0.0
6 0.0
8
at selected downwind locations. 6 0.10
0.0 0.1
0

0.08
a x/d = 1; b x/d = 2; c x/d = 3; 0.08 6
Iw 0.0 Iw

0.
8
0.12
and d x/d = 5

10
0.0
0.18 0.18

8
0.08

0.0
14

0.1
1 0. 0.16 1 0.16

4
0.10
0.14 0.1 0.14
2

0.10
0.12 0.12

0.
0.16

12
0.10 0.10

z/d

z/d

0.10
0.12 0.08 0.08

0.1
0.06
0.06 0.06

0.08
0.04 0.04

0.12
0.13

0.0
0.1 0.02 0.02
4 0.0
0.12

8
0.00 0.00
0.5 0.10 0.5 8
0.1
0
0.08

0.06
0.06 0.0

06
6 0.0
6

0.

0.0
8
0.0

6
6
0.0
0 (a) =1 0 (b) =2
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

1.5 1.5
0.0 0.08
7

0.08 0.08
0.10 0.09

0.07
7
0.0

0. Iw Iw
0.09
10
0.18 0.18
9

0.08
0.0

1 0.11
0.16 1 0.10
0.16
0.14 0.14
0.12 0.12
11 0.10 0.10
0. 0.
z/d

z/d
0.10 0.08 09 0.08
1 0.09
0.1 0.06 0.06
0.04 0.04
08
0.02 0. 0.02
0.00 0.00
0.09

0.5 0.08 0.5

0.
08
0.09 0.07
0.07 0.07
0.06
0.07 0.06
0.06 0.06
0.06 0.06

0 (c) =3 0 (d) =5
-0.5 0 0.5 -0.5 0 0.5
y/d y/d

Fig. 16 Vertical profiles of the 3 3

x/d= 2 x/d= 2
stream-wise (left) and vertical x/d= 3 x/d= 3
(right) turbulence intensity x/d= 5 x/d= 5
2.5 x/d=10 2.5 x/d=10
at selected downwind locations x/d=20 x/d=20
(x/d = 2, 3, 5, 10, and 20) NBL NBL
2 2
z/d

z/d

1.5 1.5

1 1

0.5 0.5

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
σu/U hub σw/U hub

shear and high mechanical production of turbulent kinetic community. Currently, the effects of a convective bound-
energy at the top-tip level. ary layer on the structure of wind turbine wakes are being
The effect of atmospheric stability (stable and convec- analyzed. Additionally, future research will include the
tive boundary layers) on the turbulent structure of wind study of other factors such as the coupled effects of com-
turbine wakes is of great interest to the wind energy plex topography and boundary layer flow, as well as the

123
1234 Exp Fluids (2012) 52:1219–1235

Ainslie JF (1988) Calculating the flow field in the wake of wind

turbines. J Wind Eng Ind Aerodyn 27:213–224
Athanassiadou M, Castro IP (2001) Neutral flow over a series of
rough hills: a laboratory experiment. Boundary Layer Meteorol
101:1–30
Bingöl F, Mann J, Larsen GC (2007) Laser measurements of wake
dynamics. The proceedings at EWEC, Milan
Bruun HH (1995) Hotwire anemometry, principles and signal
analysis. Oxford University Press, UK
Burton T, Sharpe D, Jenkins N, Bossanyi E (2001) Wind energy
handbook. 1st edn. Wiley, New York
Cal RB, Lebrón J, Castillo L, Kang HS, Meneveau C (2010)
Experimental study of the horizontally averaged flow structure in
a model wind-turbine array boundary layer. J Renew Sustain
Energy 2:013–106
Carper MA, Porté-Agel F (2008) Subfilter-scale fluxes over a surface
roughness transition. Part I: measured fluxes and energy transfer
rate. Boundary Layer Meteorol 126:157–179
Chamorro LP, Porté-Agel F (2009) A wind-tunnel investigation of
wind-turbine wakes: Boundary-layer turbulence effects. Bound-
ary Layer Meteorol 132(1):129–149
Chamorro LP, Porté-Agel F (2010) Thermal stability and boundary-
layer effects on wind-turbine wakes: a wind-tunnel study.
Boundary-Layer Meteorol 136:515–533
Crespo A, Hernandez J, Frandsen S (1999) Survey of modelling
methods for wind turbine wakes and wind farms. Wind Energy
2:1–24
Dobrev I, Maalouf B, Troldborg N, Massouh F (2008) Investigation
of the wind turbine vortex structure. In: 14th international
symposium on applications of laser techniques to fluid mechan-
ics, Lisbon, Portugal
Gong W, Ibbetson A (2009) A wind-tunnel study of turbulent flows
over model hills. Boundary-Layer Meteorol 49:113–148
Grant ALM (1992) The structure of turbulence in the near-neutral
atmospheric boundary-layer. J Atmos Sci 49:226–239
Grant I, Parkin P (2000) A DPIV study of the trailing vortex elements
from the blades of a horizontal axis wind turbine in yaw. Exp
Fig. 17 Normalized Reynolds shear stresses, u0 w0 =Uhub 2
(top), Fluids 28:368–376
2
v0 w0 =Uhub 2
(mid), u0 v0 =Uhub (bottom) in the central wake plane at Grant I, Parkin P, Wang X (1997) Optical vortex tracking studies of a
horizontal axis wind turbine in yaw using laser-sheet, flow
x/d = 0–3
visualisation. Exp Fluids 23:513–519
Hancock PE, Pascheke F (2010) Wind tunnel simulations of wind
interaction of multiple wakes in wind farms with various turbine wake interactions in neutral and stratified wind flow. In:
layouts/configurations. 10th EMS annual meeting, 10th European Conference on
Applications of Meteorology (ECAM) Abstracts, September
Acknowledgments This research was supported by the Swiss 13–17, 2010 in Zurich, Switzerland
National Science Foundation (grant 200021-132122), the National Hansen AC, Butterfield CP (1993) Aerodynamics of horizontal-axis
Science Foundation (grant ATM-0854766), NASA (grant wind turbines. Annu Rev Fluid Mech 25:115–149
NNG06GE256), customers of Xcel Energy through a grant (RD3-42) Hutchins N, Nickels TB, Marusic I, Chong MS (2009) Hot-wire
from the Renewable Development Fund, and the University of Min- spatial resolution issues in wall-bounded turbulence. J Fluid
nesota Institute for Renewable Energy and the Environment. C. M. Mech 635:103–136
would like to acknowledge funding from NSF IGERT (Grant DGE- Jorgensen FE (1996) The computer-controlled constant temperature
0504195) and NASA Earth and Space Science Fellowship (Grant ameometer: aspectes of the set-up, probe calibration, data
NNX10AN52H). Thanks also go to the research engineer James acquisition, and data collection turbulence. Meas Sci Technol
Tucker for his efforts in preparation of the experimental facility and 12:1378–1387
instruments. Computing resources were provided by the University of Katul GG, Chu CR, Parlange MB, Albertson JD, Ortenburger TA
Minnesota Supercomputing Institute and the Swiss National Super- (1995) Low-wave-number spectral characteristics of velocity
computing Centre. and temperature in the atmospheric surface-layer. J Geophys Res
Atmos 100:14243–14255
Khurshudyan LH, Snyder WH, Nekrasov IV (1981) Flow and
dispersion of pollutants over two-dimensional hills. Env Prot
References Agency Rpt No EPA-600/4-81-067 Research Triangle Park, NC
Larson GC, Madsen HA, Thomsen K, Larson TJ (2008) Wake
Adrian RJ, Christensen KT, Liu ZC (2000) Analysis and interpreta- meandering: a pragmatic approach. Wind Energ 11:377–395
tion of instantaneous turbulent velocity fields. Exp Fluids Lebrón J, Castillo L, Cal RB, Kang HS, Meneveau C (2010)
29:275–290 Interaction between a wind turbine array and a turbulent

123
Exp Fluids (2012) 52:1219–1235 1235

boundary layer. In: 48th AIAA aerospace sciences meeting a comparison of predictions to measurements. NREL/TP-500-
including the new horizons forum and aerospace exposition, 4–7 29494
January 2010, Orlando, Florida Snel H, Schepers JG, Montgomerie B (2007) The MEXICO project
Lu H, Porté-Agel F (2011) Large-eddy simulation of a very large (model experiments in controlled conditions): the database and
wind farm in a stable atmospheric boundary layer. Phys Fluids first results of data processing and interpretation. J Phys Conf Ser
23, 065101:1–065101:19. doi:10.1063/1.3589857 75:012–014
Massouh F, Dobrev I (2007) Exploration of the vortex wake behind of Sørensen JN (2011) Aerodynamic aspects of wind energy conversion.
wind turbine rotor. J Phys Conf Series 75, 012036:1–012036:9. Annu Rev Fluid Mech 43:427–448
doi:10.1088/1742-6596/75/1/012036 Vermeer LJ (2001) A review of wind turbine wake research at tudelft.
Medici D, Alfredsson PH (2006) Measurements on a wind turbine In: Proceedings of ASME wind energy symposium, ASME, New
wake: 3D effects and bluff body vortex shedding. Wind Energy York, AIAA-2001-0030, vol 39, pp 103–113
9:219–236 Vermeer LJ, Sørensen JN, Crespo A (2003) Wind turbine wake
Medici D, Alfredsson PH (2008) Measurements behind model wind aerodynamics. Progr Aero Sci 39:467–510
turbines: further evidence of wake meandering. Wind Energ Whale J, Papadopoulos KH, Anderson CG, Helmis CG, Skyner DJ
11:211–217 (1997) A study of the near wake structure of a wind turbine
Pedersen TF, Antoniou I (1989) Visualisation of flow through a stall- comparing measurements from laboratory and full-scale exper-
regulated wind turbine rotor. Wind Eng 13:239–245 iments. Sol Energy 56:621–633
Porté-Agel F, Wu YT, Lu H, Conzemius RJ (2011) Large-eddy Whale J, Anderson CG, Bareiss R, Wagner S (2000) An experimental
simulation of atmospheric boundary layer flow through wind and numerical study of the vortex structure in the wake of a wind
turbines and wind farms. J Wind Eng Ind Aerodyn 99:154–168 turbine. J Wind Eng Ind Aerodyn 84:1–21
Raffel M, Willert CE, Wereley ST, Kompenhans J (2007) Particle Wu YT, Porté-Agel F (2011) Large-eddy simulation of wind-turbine
image velocimetry: a practical guide. 2nd edn. Springer, Berlin wakes: evaluation of turbine parametrizations. Boundary-Layer
Sherry M, Sheridan J, Jacono DL (2010) Horizontal axis wind turbine Meteorol 132:129–149
tip and root vortex measurements. In: 15th international Zhou J, Adrian RJ, Balachandar S, Kendall TM (1999) Mechanisms
symposium on applications of laser techniques to fluid mechan- for generating coherent packets of hairpin vortices in channel
ics, Lisbon, Portugal flow. J Fluid Mech 387:353–359
Simms D, Schreck S, Hand M, Fingersh LJ (2001) NREL unsteady
aerodynamics experiment in the NASA-AMES wind tunnel:

123

View publication stats