Interaction between large wind farms and the atmospheric boundary layer

Interaction between large wind farms and the atmospheric boundary layer
Available online at
Procedia IUTAM 10 (2014) 307 – 318
23rd International Congress of Theoretical and Applied Mechanics
Interaction between large wind farms and the atmospheric
boundary layer
Fernando Porté-Agel*, Hao Lu, Yu-Ting Wu
Wind Engineering and Renewable Energy Laboratory (WIRE)
École Polytechnique Fédérale de Lausanne (EPFL), 1015 Lausanne, Switzerland
Accurate prediction of atmospheric boundary layer (ABL) flow and its interactions with wind turbines is of great
importance for optimizing the design and efficiency of wind farms. This study first focuses on recent efforts to
develop and validate a large-eddy simulation (LES) framework for wind-energy applications. The subgrid-scale
turbulent fluxes of momentum and heat are parameterized using tuning-free dynamic models. The turbine-induced
forces are parameterized using two types of models: an actuator disk model that allows for non-uniform force
distribution and includes rotational effects, and an actuator line model. The LES framework is validated against windtunnel measurements collected inside and above a large model wind farm. Further, this framework is used to study
wind-farm effects. Comparison of simulations of flow through both aligned and staggered wind farms shows
important effects of farm layout on the flow structure and wind-turbine performance. We also investigate the impacts
of wind farms on a stable ABL and a convective ABL.
by Elsevier
by Elsevier
Ltd. Selection and/or peer-review under responsibility of Dongxiao Zhang
Selection and/or peer-review under responsibility of the Organizing Committee of The 23rd International Congress of Theoretical
and Jens Nørkær Sørensen.
and Applied Mechanics, ICTAM2012
Keywords: atmospheric turbulence, large-eddy simulation, wind turbine
atmospheric boundary layer
Reynolds-averaged NavierStokes
large-eddy simulation
subgrid scale
Corresponding author. Tel.: +41- (0)-21-6936138; fax: +41-(0)-21-6936135.
E-mail address: [email protected]
2210-9838 © 2013 Published by Elsevier Ltd.
Selection and/or peer-review under responsibility of the Organizing Committee of The 23rd International Congress of Theoretical and Applied
Mechanics, ICTAM2012
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
actuator-disk model without rotation
blade-element momentum
actuator-disk model with rotation
actuator-line model
stable boundary layer
convective boundary layer
potential temperature
SGS stresses
SGS fluxes
rotor diameter
streamwise spacing
spanwise spacing
1. Introduction
With the fast growing number of wind farms being installed worldwide, the interaction between
atmospheric boundary layer (ABL) turbulence and wind turbines, and the interference effects among
wind turbines, have become important issues in both the wind energy and the atmospheric science
communities [13]. Accurate prediction of ABL flow and its interactions with wind turbines at a wide
range of spatial and temporal scales is of great importance to optimize the design (turbine siting) of wind
energy projects. In particular, flow prediction can be used to maximize wind-energy production and
minimize fatigue loads in wind farms. Numerical simulation can also provide valuable quantitative
insight into the potential impacts of wind farms on local meteorology. These are associated with the
significant role of wind turbines in slowing down the wind, generating turbulence, and enhancing vertical
mixing of momentum, heat, moisture and other scalars [4].
During the last decade, numerical simulation of wind-turbine wakes has become increasingly popular.
Most of the previous studies of ABL flow through isolated wind turbines or wind farms have
parameterized the turbulence using a Reynolds-averaged NavierStokes (RANS) approach [57].
However, as repeatedly reported in a variety of contexts [8], RANS is too dependent on the characteristics
of particular flows to be used as a method of general applicability. Large-eddy simulation (LES) can
potentially provide the kind of high-resolution spatial and temporal information needed to maximize wind
energy production and minimize fatigue loads in wind farms. Only recently there have been some efforts
to apply LES to simulate wind-turbine wakes [4, 912].
The accuracy of LES in simulations of ABL flow with wind turbines hinges on our ability to
parameterize subgrid-scale (SGS) turbulent fluxes as well as turbine-induced forces. These forces are
responsible for the development of the turbine wakes. In the next session, different wind-turbine models
are discussed. This work is dedicated mainly to the study of the characteristics of wind-turbine wakes and
their aggregated effect on wind-turbine performance as well as land-atmosphere exchanges (momentum
and heat fluxes). We describe our LES framework in Sect. 2. The LES results are presented and discussed
in Sects. 3 and 4, and a summary is provided in Sect. 5.
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
2. Large-eddy simulation framework
2.1 LES governing equations
LES solves the filtered continuity equation, the filtered momentum conservation equations (written
here in rotational form and using the Boussinesq approximation), and the filtered heat equation
§ wu wu ·
w 2u
wp * wW ij
u j ¨ i j ¸ = Q 2i G i 3 g
f cH ij 3u j i Fi ,
wxi wx j
wx j
© wx j wxi ¹
wq j
w 2 u
u j
D 2i ,
wx j
wx j
wx j
where the tilde represents a three-dimensional spatial filtering operation at scale ǻ, ui is the velocity in the
i-direction (with i = 1, 2, 3 corresponding to the streamwise (x), spanwise (y) and vertical (z) directions), ș
is the potential temperature, ș0 is the reference temperature, the angle brackets represent a horizontal
average, g is the gravitational acceleration, fc is the Coriolis parameter, p* is the modified pressure, ȡ is
the air density, Ȟ is the kinematic viscosity of air, Į is the thermal diffusivity of air, fi is an immersed force
(per unit volume) for modeling the effect of wind turbines on the flow, and Fi is a forcing term (e.g., a
mean pressure gradient). Based on the Boussinesq approximation, both ȡ and ș0 in Eq. (2) are assumed to
be constant. IJij and qj are the SGS stresses and fluxes, respectively. These SGS terms are parameterised
using Lagrangian scale-dependent dynamic models [13, 14].
2.2. Wind-turbine parameterizations
Using an actuator-disk approximation is a common approach to parameterize the turbine-induced
forces (e.g., thrust, lift, and drag) in numerical models of flow through propellers and turbines. This
approach assumes the flow surrounding a wind turbine to be inviscid and does not require resolving the
boundary-layer flow around the surface of the turbine, which decreases greatly the computational cost.
Betz [15] first applied a Rankine–Froude actuator disk method to determine the thrust force and the
power production on an ideal turbine rotor and, thus, derived the well-known Betz limit for the maximum
achievable efficiency of a wind turbine (maximum power coefficient of CP,max = 16/27). Since this
method only considers a uniform thrust load over the rotor disk and ignores the wake rotation effect, here
we refer to this method as the actuator-disk model without rotation (ADM-NR).
A major advancement in wind-turbine modeling was the introduction of the blade-element momentum
(BEM) theory. This theory considers that each blade of a wind turbine can be divided into N blade
elements (see Fig. 1), which are assumed to behave aerodynamically as two-dimensional aerofoils and to
Fig. 1. A cross-sectional aerofoil element
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
have no radial action on the flow. Based on momentum balance around the aerofoils, the aerodynamic
forces are determined using the lift and drag characteristics of the aerofoil as well as the local flow
conditions. Note that, for each blade element, the lift and drag forces are perpendicular and parallel,
respectively, to the direction of the local relative velocity. The resultant force is non-uniformly distributed
on the blade surface or over the rotor-disk area, and produces thrust as well as rotation of the flow. This
BEM approach can be applied into two types of wind-turbine models: actuator-disk model with rotation
(ADM-R) and actuator-line model (ALM). More details regarding the three wind-turbine models can be
found in [4, 12, 1618].
3. Simulation of turbulent flow inside and above wind farms
To validate this recently-developed LES framework, we choose a simulation case study,
corresponding to a wind-tunnel wind-farm experiment performed by Chamorro and Porté-Agel [19] in an
ABL wind tunnel under neutral stratification. In that experiment, a turbulent boundary-layer flow was
developed in the 16 m × 1.7 m × 1.7 m test section of the tunnel. At the downwind part of the test section,
the flow had a free-stream velocity of Uf § 3.0 m˜sí1 and a boundary-layer depth of į § 0.68 m; the
friction velocity and surface roughness length are u‫ = כ‬0.12 m˜sí1 and zo = 0.03 mm, respectively. The
wind farm had an aligned configuration and consisted of 30 miniature, horizontal-axis, three-bladed wind
turbines arranged in 10 rows and three columns that were spaced Sx = 5 apart in the streamwise direction
and Sy = 4 apart in the spanwise direction, where d = 0.150 m is the rotor diameter (see Fig. 2). Each
turbine consists of a three-bladed GWS/EP-6030×3 rotor attached to a small DC generator motor at a hub
height (Hhub) of 0.125 m. The normalized angular velocity distribution measured in the aligned wind farm
is shown in Fig. 2. Moreover, the effect of the wind-farm configuration on the flow is investigated using
LES. To achieve that, a staggered wind farm, where the even turbine rows are shifted laterally by 2d with
respect to the aligned layout, is also considered. The angular velocity of the turbines in the staggered farm
was measured (see Fig. 2) under the same inflow condition.
To simulate the entire wind-farm wake, the horizontal computational domain spans a distance Lx =
72d = 10.8 m in the streamwise direction and Ly = 12d = 1.8 m in the spanwise direction. In the
experiment, the boundary-layer depth į = 0.68 m grew slightly along the streamwise direction due to the
increased effective surface roughness induced by the wind farm. To allow for this effect in the
simulations, the computational domain has a vertical height Lz = 0.89 m, which is slightly higher than the
depth of the incoming boundary-layer flow. A constant streamwise pressure gradient is used to drive the
flow within the boundary layer. The domain is uniformly divided into (Nx × Ny × Nz) = (648 × 108 × 108)
grid points in the streamwise, spanwise and wall-normal directions, respectively. More detail regarding
the numerical set-up can be found in [16].
Fig. 2. (a) Schematic of the aligned wind-farm configuration; (b) Normalized measured angular velocity distribution of the wind
turbines at different downwind positions
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
3.1. Model validation
Figure 3 shows contours of the normalized time-averaged streamwise velocity and the streamwise
turbulence intensity obtained from the wind-tunnel experiment and the simulations with the ADM-R and
the ADM-NR. As expected, the turbine wakes (regions of reduced mean velocity) are clearly visible
behind each turbine. Also evident is the cumulative effect of multiple wakes, leading to the formation of a
“wind-farm wake” with two distinct regions: in the first region, below the top-tip level, the mean velocity
deficit adjusts relatively rapidly and reaches an equilibrium after only two–three rows of wind turbines.
This is consistent with field observations showing that the power output from operational offshore aligned
wind farms decreases significantly (with respect to the first row) for the second and, to a lesser extent, the
third row of turbines, while it remains relatively unchanged after that [e.g., [20]]. In the second region,
above the turbines, the flow experiences a larger downwind variation as the cumulative farm wake
expands. The edge of the farm wake, defined here as the height where the time-averaged wake velocity is
99% of the mean inflow velocity at that height, is shown in Fig. 3. The location of the simulated wake
edge is very similar for both turbine models and it is in good agreement with the measurements. The wake
edge grows with downwind distance and reaches a height of about 0.4 m (twice the turbine height) behind
the tenth row of turbines (x /d • 45).
Fig. 3. Contours of the normalized time-averaged streamwise velocity (left panel) and the streamwise turbulence intensity (right
panel) in the vertical plane at zero span ( y = 0): wind-tunnel measurements (top), ADM-R (middle), ADM-NR (bottom). Circles
denote the edge of the measured farm wake. White dashed lines denote the edge of the simulated farm wake
The simulation results in Fig. 3 show clear differences between the predictions of the wake velocity by
the two wind-turbine models. In particular, the LES with the ADM-R produces velocity profiles that are
in good agreement with the measurements throughout the wind farm. In contrast, the ADM-NR clearly
overpredicts the mean velocity (i.e., it underpredicts the velocity deficit) in the wake behind each turbine.
This is consistent with the previous simulations of the wake of a stand-alone turbine presented by [12, 17].
It should be noted that the thrust coefficient CT used for each turbine in the ADM-NR is obtained based
on the overall thrust force computed using the BEM theory in the ADM-R. Consequently, the failure of
the ADM-NR model to reproduce the velocity magnitude in the wake regions is attributed to the
limitations of two major assumptions made in the ADM-NR (but not in the ADM-R): (a) the turbine-
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
induced rotation effect is ignored, and (b) the axial thrust force is uniformly distributed over the rotor disk
area, thus ignoring the radial variation of the force. As pointed out by [12], the latter assumption has the
stronger effect of the two.
Due to the cumulative effect of the multiple wakes from upstream wind turbines, the maximum level
of the turbulence intensity found behind the wind turbines increases substantially in the wakes behind the
first four rows of turbines, and reaches a plateau after the fifth row. That maximum turbulence intensity is
found behind each turbine at the top-tip level. This is due to the high mean shear (see Fig. 3) and
associated mechanical (shear) turbulence production in that region. In particular, a peak of turbulence is
found at approximately three rotor diameters behind each of the turbines, except for the first one. This is
due to the fact that the inflow to the first row of wind turbines is much less turbulent, leading to a slower
recovery of the wake (due to less efficient mixing with the surrounding flow), compared with the wakes
of the other turbines. This also explains why the turbulence intensity peak is found further downwind
from the turbine in the case of a stand-alone wind turbine [12, 17].
The distribution of the simulated turbulence intensity obtained with the two turbine models (ADM-NR
and ADM-R) shows a similar qualitative behaviour as the one reported in the experiment. However,
significant differences are found in the ability of the two models to match quantitatively the measured
turbulence intensity levels. The magnitude of the turbulence intensity obtained with the ADM-R, and
particularly its maximum value at the top-tip level, is found to be in acceptable agreement with the windtunnel measurements. The ADM-NR tends to systematically underestimate the peak of turbulence
intensity behind most of the turbines. Below the top-tip height, and further than two rotor diameters
downwind of the turbines, both models underestimate the turbulence intensity, with a slightly worse
prediction from the ADM-NR model.
3.2. Layout effects
In this subsection, a comparison of the LES results for the aligned and staggered wind farms is
presented. The numerical set-up in the previous model validation case is adopted in the staggered windfarm wake simulation, except for the horizontal position and the angular velocity of the turbines.
Figure 4 shows contour plots of the normalized time-averaged streamwise velocity and the streamwise
turbulence intensity on a horizontal x-y plane at the hub level and in a vertical y-z plane at 3d downstream
behind the third and ninth turbines for the two farms. In this figure, it is clear that the farm layout has a
strong effect on the structure of the cumulative wakes and, consequently, on the distribution of the
different turbulence statistics. In the aligned case, the wake regions are centred around the turbine rows
and grow radially with distance downwind, only interacting laterally after approximately the eighth row.
In the staggered farm, lateral wake interactions are obvious even after the third row of turbines due to the
fact that the wind farm offers a larger “frontal area” to the incoming flow. Moreover, since the effective
distance between downwind turbines is now 10d, the wakes have a longer distance to recover before the
next turbine, which results in a higher efficiency of the turbines (i.e., faster rotating speed) in extracting
momentum from the flow, compared with the aligned counterpart. This explains the higher angular
velocity of the turbines and the more uniform distribution of the velocity within the wind farm.
Important differences between the two layouts are also found in the turbulence intensity distribution.
The enhancement of turbulence intensity and, consequently, the potential negative impacts of the
associated fatigue loads are much higher in the aligned farm. In the staggered farm, the distance between
“immediately downwind” turbines is longer, which allows for the turbulence to dissipate to lower levels
before reaching the next downwind turbine, thus reducing the cumulative turbulence enhancement effect.
It should be noted that the maximum turbulence intensity region corresponds to a U-shaped area at the
rotor edge and above hub height, where the local shear and associated production of kinetic energy are
In order to further illustrate the growth of the cumulated farm wakes and their dependence on farm
layout, Fig. 5 depicts a three-dimensional representation of the distribution of the simulated wake edge
over the aligned and staggered wind farms. The wake edge is defined as the location where the time-
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
averaged wake velocity is 99% of the mean flow velocity at height. From this figure, it is obvious that in
the aligned wind farm there is no lateral interaction between the turbines until the eighth row of turbines.
In contrast, the wakes merge relatively soon, leading to a more uniform spanwise distribution of the wake
edge. As a result, the growth of the cumulated farm wake resembles more a classical “internal boundary
layer” in the case of the staggered farm than in the case of the aligned farm.
Fig. 4. Contours of the normalized time-averaged streamwise velocity (left panel) and the streamwise turbulence intensity (right
panel) on the horizontal x-y plane at the turbine hub height (top) and in a vertical y-z plane at 3d downstream behind the third and
ninth turbines (bottom) for the both aligned and staggered wind farms
Fig. 5. Isosurface of internal wake layer distribution: (a) Aligned and (b) staggered wind-farms
4. Wind farms in stable and unstable boundary layers
4.1. Wind farms in a stable boundary layer
Of special interest for wind-energy applications is the study of thermally stratified stable boundary
layers (SBLs). SBLs are relatively shallow and are characterized by strong shear and a relatively high
wind near the top. At that height, the wind can become super-geostrophic and form the so-called lowlevel jet. As a result, compared to neutral and convective ABLs, SBLs provide larger energy potential, but
also larger structural fatigue loads associated with the strong shear. SBLs are particularly challenging to
simulate accurately due to the large shear and anisotropy of the flow. An LES inter-comparison study was
carried out in the context of the global energy and water cycle experiment atmospheric boundary layer
study (GABLS) initiative [21] to simulate a moderately stable boundary layer. This case is used here as a
baseline case.
In order to study the effect of a wind farm on the GABLS case, we have “immersed” (using the ALM) a
V112-3.0MW wind turbine in the GABLS domain. This wind turbine has a rotor diameter of d = 112 m. Two
x-direction dimensions, corresponding to two typical wind-turbine spacings, are studied: (1) Lx = 8d = 896 m
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
(the corresponding LES is abbreviated as the 8d case); (2) Lx = 5d = 560 m (the corresponding LES is
abbreviated as the 5dd case). Periodic boundary conditions are applied horizontally to simulate an
infinitely large wind farm. Readers can find more numerical details from previous articles [4, 21]. It
should be noted that the baseline case attains a quasi-steady state in 89 h [21]. Therefore, in order to
examine the wind-turbine effects relative to the baseline case, we introduce the wind turbines only in the
last hour of simulation.
Figure 6a shows the formation (initial stages) of blade-induced three-dimensional helicoidal tip
vortices, detected using vorticity iso-surface. Due to the strong shear and non-uniformity of the incoming
boundary-layer flow, helicoidal vortices are stretched as they travel faster at the top tip level compared
with the bottom tip level. Figure 6b shows the mean vertical profiles of wind speeds. In agreement with
other studies [22], when wind turbines are installed, there is an increase off the boundary height. The
baseline case clearly shows a super-geostrophic nocturnal jet peaking nearr the top of the boundary layer.
However, the extraction of energy by the turbines, leads to a distortion of the velocity field (compared
with the baseline case) and an elimination of the low-level jet in the wind farm simulations. Also, as
expected, the closer the distance between the wind turbines, the larger extraction of kinetic energy from
the flow.
Fig. 6. (a) Vorticity isosurface of the 5dd case at t = 30 s; (b) Vertical distributions of mean x-direction velocity U
and y-direction velocity V
Besides extracting kinetic energy and generating turbulence, wind-turbine blade motions also mix
fluid parcels. Figure 7a compares the potential temperature profiles obtained from the baseline case and
two wind-turbine cases. Blade motions enhance the vertical mixing and transfer more thermal energy
from higher levels to lower levels. This leads to an increase of temperature (about 0.5 K warming) below
the top tip level and a decrease between the tip-height and the SBL height. The investigation of fluxes is
of interest because local meteorology is considerably affected by the overall exchanges of momentum,
heat, moisture, etc. The magnitude of the surface buoyancy flux decreases with time as shown in Fig. 7b.
Specifically, over the last 15 min, the 8dd case yields a buoyancy-flux magnitude of approximately
í3.8×10í4 m2˜sí3 (reduced §15%), corresponding to a heat flux of í13.5 W˜m-2; the 5dd case yields a
buoyancy-flux magnitude of approximately í3.2×10í4 m2˜s-3 (reduced § 28%), corresponding to a heat
flux of í11.4 W˜m-2. Regarding the overall thermal-energy budget, this reduced heat flux is consistent
with the increase of air temperature in the boundary layer as shown in Fig. 7a.
4.2. Wind farms in a convective boundary layer
Though a very-coarse-resolution simulation [2] reported cooling effects by wind farms in a daytime
ABL, a high-fidelity study of wind-farm effects on day-time convective atmospheric boundary layers
(CBLs) has yet to be performed. In CBL flows, the surface of atmospheric boundary layer is warmer than
the surrounding air, in response to solar heating. The warmer surface leads to a positive (upward)
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
buoyancy flux, which creates a thermal instability and generates turbulence. In order to reduce
uncertainties when studying the interactions between a wind farm and a CBL, we start with a CBL
baseline case (without wind turbines) that has been well tested with LES. We conduct three-dimensional
LESs with the actuator disk model with rotation [12] to investigate the impact of wind farms on this CBL
case. Siemens SWT-2.3-93 wind turbines, with a rotor diameter of 93 m and a hub height of 80 m, are
‘‘immersed’’ in the flow, as shown in Fig. 8a. The horizontal domain is larger than four boundary layer
depths, which is enough to resolve large waves. In order to understand farm layout effects, the framework
is applied to study several cases of aligned wind farms with different streamwise and spanwise spacings.
Fig. 7. (a) Vertical profiles of potential temperature; (b) Surface buoyancy flux evolutions
As an example, Fig. 8b shows vertical profiles of averaged (horizontally and over a certain time period)
wind speed obtained from the aligned Sx ×Sy = 5×5 case and the baseline case. In line with the situation in
the stable condition, results reveal the extraction of momentum by the turbines. However, the boundary
layer height in the CBL flow is relatively larger, thus wind turbines cause a relatively smaller increase of
2.5% in comparison with 10% in the stable condition.
Fig. 8. (a) Instantaneous streamwise velocity contours and turbine-induced vortex structures; (b) Wind speed profiles obtained from
the aligned Sx × Sy = 5 × 5 case and the baseline case
In contrast to the warming effects under stable conditions, the wind farm leads to a slight decrease of
temperature in the CBL as shown in Fig. 9a. The magnitude of cooling is about 0.040 K near the surface
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
and about 0.025 K throughout the entire boundary layer, which is even less than a tenth of magnitude of
warming in stable condition. The warming effect in stable condition is most likely caused by the
enhanced mixing, which pulls down warmer air from higher altitudes. Under unstable conditions,
turbulent wakes mix cool air down and warm air up, producing a cooling near the surface. In comparison
with the ABL flows under stable conditions, the existing mixing under unstable conditions is already very
large. Hence, the turbine-enhanced turbulent mixing may play a smaller role. However, as shown in
Fig.9b, the vertical profiles of the heat flux reveal largely enhanced entrainment fluxes, indicating that the
cooling below the entrainment zone has created substantial negative heat flux (downward warming). The
entrainment/surface flux ratio has been increased approximately from 0.29 to 0.48. Regarding the overall
thermal-energy budget, the cooling throughout the entire boundary layer is consistent with the reduced
surface heat flux as shown in Fig. 10. The discrepancy between the surface heat fluxes obtained from
baseline case and each wind farm case is increasing with time, and all wind farm cases yield
approximately 5% reductions in the surface-heat-flux magnitude.
Fig. 9. Vertical profiles of (a) potential temperature and (b) heat flux obtained from the aligned Sx ×Sy = 5 × 5 case
and the baseline case
Fig. 10. Time evolution of the area-averaged surface buoyancy flux
5. Summary
This paper presents recent efforts to develop and validate a large-eddy simulation framework for wind
energy applications. The tuning-free dynamic models are used to parameterize the SGS stress tensor and
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
the SGS heat flux. Two types of models are used to parameterize the turbine-induced forces: actuator disk
models and actuator line models.
The proposed LES framework is validated against high-resolution velocity measurements inside and
above the aligned wind farm in an ABL wind tunnel. In general, the characteristics of the simulated
turbine wakes (average velocity and turbulence intensity distributions) are in good agreement with the
measurements. In the case of the ADM-R, accounting for rotation and non-uniform distribution of the
forces yields improved predictions.
Comparison of the simulation results for the aligned and staggered wind-farm cases shows a strong
effect of wind-farm layout on the turbulent flow structure inside and above wind farms. In particular, the
cumulative wakes are found to have little lateral interaction (with no interaction before the eighth row of
wind turbines) in the case of the aligned wind farm. In contrast, the lateral interaction between the wakes
is much stronger and happens throughout most of the wind farm in the case of the staggered wind farm.
As a result, the growth of the cumulative wake from the staggered farm resembles more a classical
internal boundary layer compared with that from the aligned farm.
Further, we investigated the impacts of wind farms on a stable ABL and a convective ABL. Our
results clearly show that the largest discrepancies appear in the distribution of momentum. The
differences in the distribution of potential temperature are relatively smaller. Previous studies at relatively
low resolutions have shown that wind farms could have noticeable effects on the global climate [3], and
on local meteorology [2]. In agreement with these studies, our results show that the wind-turbine motions
enhance the vertical mixing of heat, resulting in near-surface warming under stable conditions and
cooling under convective conditions. They also lead to lowered surface heat flux magnitudes.
The wind in the lowest part of the atmosphere is the most important atmospheric variable for windpower meteorology. The results presented in this paper show that LES has the potential provide reliable
detailed information of wind-turbine wakes, which is needed to optimize wind-farm design (maximize
energy output and minimize fatigue loads) and also to develop more accurate parameterizations of
turbulent fluxes in weather/climate models.
This research was supported by the Swiss National Science Foundation (200021 132122), and the
National Science Foundation (EAR-0537856 and ATM-0854766). Computing resources were provided
by the Minnesota Supercomputing Institute, and the project of Swiss National Supercomputing Centre
(ID s306).
Vermeer LJ, Sørensen JN, Crespo A. Wind turbine wake aerodynamics. Progress Aero Sci 2003; 39: 467–510.
Baidya Roy S, Pacala SW, Walko RL. Can large wind farms affect local meteorology? J Geophys Res 2004; 109: 1–6.
Kirk-Davidoơ DB, Keith DW. On the climate impact of surface roughness anomalies. J Atmos Sci 2008; 65: 2215–2234.
Lu H, Porté-Agel F. Large-eddy simulation of a very large wind farm in a stable atmospheric boundary layer. Phys Fluids
Alinot C, Masson C. Aerodynamic simulations of wind turbines operating in atmospheric boundary layer with various
2011; 23: 065101.
thermal stratifications. ASME Wind Energy Symposium AIAA. 2002; 2002–42.
Gómez-Elvira R, Crespo A, Migoya E, Manuel J, Hernández F. Anisotropy of turbulence in wind turbine wakes. J Wind Eng
Ind Aerodyn 2005; 93: 797–814.
Kasmi AE, Masson C. An extended țíİ model for turbulent flow through horizontal-axis wind turbines. J Wind Eng Ind
Aerodyn 2008; 96: 103–122.
Sagaut P. Large eddy simulation for incompressible Àows. 3rd ed. Springer-Verlag: Berlin Heidelberg; 2006.
Fernando Porté-Agel et al. / Procedia IUTAM 10 (2014) 307 – 318
Jimenez A, Crespo A, Migoya E, Garcia J. Advances in large-eddy simulation of a wind turbine wake. J Phys Conf Ser 2007;
75: 012041.
[10] Jimenez A, Crespo A, Migoya E, Garcia J. Large-eddy simulation of spectral coherence in a wind turbine wake. Environ Res
Lett 2008; 3: 015004.
[11] Calaf M, Meneveau C, Meyers J. Large eddy simulation study of fully developed wind-turbine array boundary layers. Phys
Fluids 2010; 22: 015110.
[12] Wu YT, Porté-Agel F. Large-eddy simulation of wind-turbine wakes: Evaluation of turbine parametrisations. BoundaryLayer Meteorol 2011; 138: 345–366.
[13] Stoll R, Porté-Agel F. Dynamic subgrid-scale models for momentum and calar fluxes in large-eddy simulations of neutrally
stratified atmospheric boundary layers over heterogeneous terrain. Water Resour Res 2006; 42: W01409
[14] Stoll R, Porté-Agel F. Large-eddy simulation of the stable atmospheric boundary layer using dynamic models with different
averaging schemes. Boundary-Layer Meteorol 2008; 126: 1–28.
[15] Betz A. Windenergie und Ihre Ausnutzung durch Windmüllen. 1926.
[16] Wu YT, Porté-Agel F. Simulation of turbulent flow inside and above wind farms: model validation and layout effects.
Boundary-Layer Meteorol 2012; 146: 181–205.
[17] Porté-Agel F, Wu YT, Lu H, Conzemius RJ. Large-eddy simulation of atmospheric boundary layer flow through wind
turbines and wind farms. J Wind Eng Ind Aerodyn 2011; 99: 154–168.
[18] Sørensen JN, Shen WZ. Numerical modeling of wind turbine wakes. J Fluids Eng 2002; 124: 393–399.
[19] Chamorro LP, Porté-Agel F. Turbulent flow inside and above a wind farm: a wind-tunnel study. Energies 2011; 4: 1916–
[20] Barthelmie RJ, Hansen K, Frandsen ST, Rathmann O, Schepers JG, Schlez W, et al. Modelling and measuring flow and wind
turbine wakes in large wind farms offshore. Wind Energ 2009; 12: 431–444
[21] Beare RJ, MacVean MK, Holtslag AAM, Cuxart J, Esau I, Golaz JC, et al. An intercomparison of large-eddy simulations of
the stable boundary layer. Boundary Layer Meteorol 2006; 118: 247–272.
[22] Frandsen S, Barthelmie R, Pryor S, Rathmann O, Larsen S, Hojstrup J, et al. Analytical modelling of wind speed deficit in
large off shore wind farms. Wind Energ 2006; 9: 39–53.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Related manuals

Download PDF