Content uploaded by Brad Cochran
Author content
All content in this area was uploaded by Brad Cochran on Jan 05, 2016
Content may be subject to copyright.
Integrating Wind Energy into the Design of Tall Buildings –
A Case Study of the Houston Discovery Tower
WINDPOWER 2008
Brad C. Cochran, MS
Rick R. Damiani, PhD, P.E. (Europe)
CPP, Inc.
1415 Blue Spruce Drive
Fort Collins, CO 80524
Corresponding Author: Brad Cochran
970-221-3371
bcochran@cppwind.com
WindPower 2008, Houston, Texas June 2-4, 2008 1
Harvesting Wind Power from Tall Buildings
Brad C. Cochran
1
, Rick R. Damiani
1
1. CPP, Wind Engineering and Air Quality Consultants, 1415 Blue Spruce Drive, Fort Collins, CO,
80524, USA
Abstract
Integrating wind energy systems into building design is a small but growing trend, and high rises with
their elevated wind speeds seem particularly suited to the technology. Designs that incorporate wind
turbines are increasingly being seen on the drawing boards for skyscrapers across the globe. Leading
the development of this fledgling field, wind engineers at CPP have evaluated the potential for wind
turbine integrated buildings in both the U.S. and China. This paper will focus on a project conducted for
the Discovery Tower in downtown Houston, TX, which is slated to break ground in early 2008.
Keywords: Small wind turbines, tall buildings, wind resource assessment
Introduction
Tall building designers are showing an
increasing interest in reducing the environmental
impact of the construction and operation of their
buildings. One of the approaches currently used, is
the incorporation of on-site power generation.
This is primarily achieved by integrating solar and
wind devices into the design of the building. One
such example is the Discovery Tower, which is
currently under construction in Houston, TX.
The requirements for optimizing the
performance on wind generators in an urban
environment are quite different from those
pertaining to wind farms in open sites. This entails
the use of different design approaches to assess
the optimum placement of wind turbines within
the building envelope, the most suitable generator
types for the building environment, and to
estimate annual energy production for the wind
turbines.
This paper will examine aspects involved in
wind power generation on tall buildings, and how
those were specifically addressed for the
Discovery Tower.
Wind climate
When contemplating the incorporation of
wind power generation into a tall building design,
the first consideration must be the local wind
climate of the area. Bluntly, if there is not
sufficient wind resource in the area, then the
potential for successful use of turbines will be
very limited.
Wind conditions in urban environments tend
to be very different. The effect of urban
environments on a boundary-layer is shown in
Figure 1. This shows how buildings slow the wind
near the ground, and increases turbulence.
Turbines work most efficiently in low-turbulence
environments; therefore care needs to be taken in
specifying turbine types that will cope with both
existing turbulence levels and potential future
changes as a result of urban development.
Urban development is likely to pose one of
the greatest challenges to increasing use of
turbines on tall buildings. In city center locations,
height restrictions often mean that many tall
buildings are of similar heights. Even if a building
is very tall, if all the surrounding buildings are of
similar height then the potential for efficient
turbine installation is significantly reduced.
Unlike rural wind farms, where the nearest
anemometer may be located many miles away,
most cities have reasonable lengths of records
from nearby airports. This is not, however, to say
these are necessarily good records. It is not
uncommon to see anomalous directionality
characteristics due to poor anemometer siting
close to buildings. Wherever possible, records
from multiple stations should be used. A rule of
thumb is to use a minimum of 10 years of records
to ensure statistical robustness. Trends are also
sometimes apparent that don’t reflect climate
changes, but are more often indicative of changing
urban development close to the anemometer site.
WindPower 2008, Houston, Texas June 2-4, 2008 2
In all cases, the first stage is to correct the data
back to the equivalent of open-country exposure
to simulate the readings that would be experienced
in the absence of any development. There are a
number of methods for doing this, and the
approach codified by ESDU (1993a and 1993b)
and based on the work of Deaves and Harris is
among the most common.
In areas where available anemometer records
are suitable to describe the local wind
environment, a simple wind speed transfer
approach can be utilized, as illustrated in Figure 1.
In this approach, the wind speeds at the
anemometer are extended to a gradient height
(200 m to 600 m above grade), where the local
terrain has little or no impact on the wind speeds,
using a power law relationship. That same wind
speed is then assumed to exist at the gradient
height (which may not be at the same height
above grade) above the site. The wind speeds are
then transferred down to the site using either the
power law with a site specific exponent, or by
measuring the vertical velocity profile in an
atmospheric boundary layer wind tunnel.
When there are no reliable anemometer
records within a reasonable distance of the site,
meso-scale modeling can be used to determine the
wind climate of the area. This uses input from
historical meteorological records from, maybe,
hundreds of kilometers away to regenerate the
weather systems affecting the site. This is an
approach that is commonly used for rural turbine
locations.
The directionality of the wind is also
important. The incorporation of turbines into tall
buildings tends to favor limited wind directions,
perhaps within a 45 degree sector, depending
upon the building configuration and the location
of the wind turbines on the building.
Figure 2 shows the wind frequency
distribution at the Hobby Airport in Houston, TX,
while Figure 3 shows the alignment of the
Discovery Tower. The wind turbines are located
along the southwest side of the roof. The wind
rose indicates that the predominant winds in
Houston are from the SSE through S. The
discovery tower is aligned such that the broad
sides of the structure are along a SE to NW axis.
As a result, the roof-top turbines are likely to be
somewhat sheltered from the predominant winds,
whereas the most favorable wind conditions for
the turbines (winds from the SW) are fairly
infrequent. As shown later on, this mismatch will
significantly impact the wind energy potential for
the roof top wind turbines.
Basic tall building
aerodynamics
As discussed in the previous sections, it is
desirable to locate turbines in regions of high
wind speed and low turbulence. Describing the
wind flow around a tall buildings can be quite
complex and has been studied in depth for many
years (Cermak, 1975 and 1976). A simplified
sketch of the mean flow is shown in Figure 4.
There will be positive pressure on the windward
face and negative pressure on the side and leeward
faces. As air naturally flows from areas of high
pressure to areas of low pressure, the most
effective locations for wind turbines will be either
in the accelerated shear layers around the edge
and top of the building, or in specially developed
passages linking the areas of positive and negative
pressure. Note that wind speeds close to the center
of a flat roof may be low as this area is often in a
region of separated flow. Whereas with a pitched
or tiered roof, the center may be the location of
the greatest wind resource.
Figure 1: Transfer of Wind Speed from Hobb
y
Airport to
the Discovery Tower site
WindPower 2008, Houston, Texas June 2-4, 2008 3
Shaping of tall buildings to increase
efficiency of wind turbines
Shaping of tall buildings can be used
effectively to enhance the performance of wind
turbines. Two examples of this are the Bahrain
World Trade Center (Figure 5) and the Pearl River
Tower in China (Figure 6). The Bahrain World
Trade Centre Tower is formed to create a Venturi
effect, placing the horizontal axis turbines
between two wings of the building. This approach
clearly works for only a limited number of wind
directions, but may be useful in a location with a
very dominant prevailing wind direction.
Restricting the orientation of horizontal axis
turbines, however, severely limits the efficiencies
gained from using this type of turbine. In the Pearl
River Tower, slots through the tower are used to
relieve the pressure between the front and rear
faces of the tower with these slots being
aerodynamically shaped to increase flow through
them. Again, this approach is most efficient for
only a few wind directions, but has the advantages
accelerating the flow while likely reducing the
turbulence approaching the turbines.
Wind Energy Potential
A physical model of the Discovery Tower
and surroundings within a 450 m radius, shown in
Figure 2: Wind frequenc
y
distribution at Hobb
y
Airport, Houston TX.
Figure 3: Ali
g
nment of the Discover
y
Tower i n
Houston, TX.
S
S
S
S
E
E
S
S
W
W
Figure 4: Wind flow around a tall building.
WindPower 2008, Houston, Texas June 2-4, 2008 4
Figure 3, was placed in an atmospheric boundary
layer wind tunnel. The wind tunnel was used to
characterize the wind environment on the tower
roof, including any shielding or acceleration. The
analysis included vertical wind profiles at four
locations along the southwest end of the tower for
16 approach wind directions. Each profile was
collected using a five-hole probe, shown in Figure
8, which is capable of measuring the mean wind
vector and turbulence intensity values. The mean
wind speeds were combined with the wind
frequency distributions to determine the wind
resource at each measurement location. The
velocity vector and turbulence intensity values
were used to determine the appropriateness of
various wind turbine designs.
Figure 7 Photograph of the 1:300 scale model installed in
an atmospheric boundary layer wind.
The mean velocity vector within each profile
was used as the transfer value between the
reference velocity at 152 m above grade (upwind
and outside of the effects of the local buildings)
and the local wind speed. For profiles with little or
no local shear, the average value accurately
represents the wind environment across the entire
turbine rotor. In areas with high shear, this value
may be less valid. A more accurate wind speed is
Figure 6: Pearl River Tower (Image courtesy
Figure 5: Bahrain World Train Center (Ima
g
e
courtesy of www.bahrainwtc.com
)
WindPower 2008, Houston, Texas June 2-4, 2008 5
obtained by integrating the wind speed vertical
distribution. However, since information on the
impact of the vertical shear is not available from
most turbine manufacturers, the average wind
speed was used to characterize each profile.
Figure 9 shows the distribution of the normalized
mean wind speeds (the ratio of the local wind
speed with (U
mean
) and without (U
ref
) the presence
of the building) at each of the four measurement
locations as a function of the approach wind
speed.
Figure 8: Close-up of the roof showing the 5-hole probe
used to measure local wind velocities.
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
N NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW
Wind Dire ction
U
mean
/U
ref
Location 1 Location 2 Location 3 Location 4
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
N NNE NE ENE E ESE SE SSE S SSW SW WSW W WNW NW NNW
Wind Dire ction
U
mean
/U
ref
Location 1 Location 2 Location 3 Location 4
Figure 9: Mean longitudinal wind speed distribution vs.
approach wind direction
The results indicated that there is accelerated flow
(U
mean
/U
ref
greater than unity) on the roof of the
Discovery Tower at the proposed wind turbine
locations for wind directions from the SSW
through SW. For these wind directions the local
wind speeds are as much as 20 percent greater
than the reference wind speed. The results also
indicate that there is considerable sheltering of the
air flow for winds from E through SE and from W
through NW.
W
IND POWER DENSITY
In order to estimate how much energy a
specific turbine will be expected to produce at a
given location, the wind resource at that location
must be identified. A wind turbine works by
extracting kinetic energy out of the wind and
converting it to mechanical and then electrical
energy. The power that is available in the wind to
be converted to electrical energy is defined in the
following relationships:
AUP
w
3
2
1
ρ
= Equation 1
and
APPD
w
/
=
Equation 2
or
3
2
1
UPD
ρ
= Equation 3
Where:
PD -Power density, W/m
2
;
P
W
- Power available in the wind, W;
ρ - Air density, kg/m
3
;
U - Wind speed approaching the wind
turbine, m/s; and
A- Projected area of the turbine
perpendicular to the approaching wind,
m
2
.
Since the potential power production is
proportional to the wind speed cubed, the annual
average wind power density cannot be defined by
strictly using the mean annual wind speed. Rather
some knowledge of the distribution of wind
speeds must be known to accurately estimate the
annual average wind power density (PD). This
can be achieved by applying Equation 3 to the
hourly recorded wind speed measurements
obtained throughout the year and then taking the
average of the hourly values, where U is the local
mean wind speed derived from the airport
anemometer and the normalized wind speed
distributions for each measurement location, using
the techniques described above.
WindPower 2008, Houston, Texas June 2-4, 2008 6
The results of this analysis are shown below in
Table 1. The table indicates that the wind power
density at the site in the absence of the Discovery
Tower and surrounding buildings is 472 W/m
2
.
When the impact of the Discovery Tower and
surrounding buildings is taken into account, the
wind power density values are reduced by
approximately 1/3
rd
and range from 114 W/m
2
to
153 W/m
2
.
Table 1
Average wind speed and wind power density
To determine the feasibility of wind power at a
site, the wind power density is often compared to
various classifications developed to describe a
site’s wind power potential. Table 2 lists the
classifications as a function of the predicted wind
power density and mean annual wind speed.
The Discovery site falls into Class 4. The
resource potential in a Class 4 environment is
considered “Good”. In the U.S. much effort has
been undertaken in the last few years to develop
wind turbines that are economically feasible in a
Class 3 environment, since this is the most
common wind class found in the U.S. Therefore,
these results suggest that the wind resource at the
Discovery Tower site is sufficient to be
considered feasible for modern wind turbines.
However, as currently configured, the wind
turbines on the roof of the Discovery Tower will
only experience a Class 1 to Class 2 wind
environment. In this environment it may be
economically difficult to justify wind turbine
installations.
Table 2
Wind Power Classifications
Wind
Power Resource
Class Potential
2 Marginal 200 - 300 5.6 - 6.40
3 Fair 300 - 400 6.4 - 7.00
4 Good 400 - 500 7.0 - 7.50
5 Excellent 500 - 600 7.5 - 8.00
6 Outstanding 600 - 800 8.0 - 8.80
7 Superb 800 - 1600 8.8 - 11.1
* - Wind classifications are typically base on the wind power density at 50m above grade
Wind Power
Density
*
(W/m
2
)
Annual Average
Mean Wind Speed
(m/s)
Wind Environment Characterization
In order to determine the appropriateness of
installing wind turbines at the site, it is not only
important to evaluate the wind power density, but
also to characterize the wind environment in terms
of flow vectors, gradients, and turbulence
intensities. The local wind speeds on the roof of
the Discovery Tower were measured using the
Aeroprobe 5-hole probe, described above. The 5-
hole probe provides measurements of wind speed
in each of the three coordinates, longitudinal (U),
lateral (V), and vertical (W) and can provide both
mean and fluctuating wind speeds. For this
analysis, each of the three components of the local
wind velocity were normalized by the mean
reference wind speed measured upwind of the
turntable in unobstructed flow at a full-scale
height of 152 m above the local grade. This results
in normalized wind speed values of U
mean
/U
ref
;
V
mean
/U
ref
; and W
mean
/U
ref
. The local turbulence
intensity values were calculated by normalizing
the fluctuating component of the wind speed by
the mean longitudinal wind speed (U
rms
/U
mean
;
V
rms
/U
mean
; and W
rms
/U
mean
). The lateral (θ
Y
) and
vertical (θ
z
) angles of attack were calculated as the
inverse tangent of the lateral to longitudinal
(V
mean
/U
mean
) and vertical to longitudinal
(W
mean
/U
mean
) velocity ratios. Finally, the
magnitude of the local velocity vector was
Annual Average Annual Wind
Wind Speed Power Density
Location (m/s)
(W/m
2
)
Reference
1
6.89 472
1 4.26 153
2 3.93 114
3 4.11 141
4 4.18 148
1) 152 m above local grade, outside of the influence of
nearby structures.
WindPower 2008, Houston, Texas June 2-4, 2008 7
calculated as the square root of the sum of the
squares of the three velocity components.
Figure 10 and Figure 11 show the vertical
distributions above roof level of mean velocity
and turbulence intensity at Location 2 for
approach wind direction of 135 degrees (the
prominent wind direction) and 202.5
(perpendicular to the broad side of the Discovery
Tower). The plot on the right side of each figure
shows the normalized mean velocity as a function
of the height above the local roof. The plot on the
left side shows the distributions of longitudinal
turbulence intensity as a function of the height
above the local roof.
Turbulence Intensity. Turbulence intensity
is the ratio of the fluctuating velocity to the mean
velocity and is a measurement of the gustiness of
the wind. Although all three components of
turbulence intensity are calculated, typically only
the longitudinal turbulence is relevant because it is
most readily available from field measurements.
Therefore, for the purpose of this analysis, the
focus was placed on the longitudinal turbulence
intensity, even though the lateral and vertical
turbulence intensity values can be significant
In an unobstructed open field environment
turbulence intensity values are typically in the
range of 10% to 15% at 30 m above grade and
decrease at higher elevations. Utility scale wind
turbines are typically designed for maximum
turbulence intensity values around 17% to 18%.
Smaller turbines, particularly those designed to be
integrated into buildings, must be able to
withstand much higher turbulence intensities.
Figure 10 indicates that the turbulence
intensity values at Location 2 for a southeast wind
direction are consistently above 60% throughout
the entire profile. This is likely due to vortex
shedding occurring at the upwind corner of the
Discovery Tower for this wind direction. The
turbulence intensity values for a southwest wind
direction, shown in Figure 11 are substantially
lower, particularly at 5 m above the roof and
higher. In this region, the turbulence intensity
values are within the range that one would expect
to find in an open field environment.
Wind Shear. The wind shear is a description of
the rate of change in wind speed along the vertical
profile. It is defined by the exponent, n, in the
power law equation. Wind shear is important,
particularly on large turbines, because it can
create unequal wind loading along the vertical
axis of the wind turbine. In an unobstructed open
environment the power law exponent typically
ranges between 0.1 and 0.2. Utility scale wind
turbines are typically limited to operating in
environments with wind shear values less than 0.2
to 0.23. Once again, building integrated wind
turbines must be designed to handle the higher
wind shear values that are commonly present on
and around physical structures.
WindPower 2008, Houston, Texas June 2-4, 2008 8
Figure 10: Wind shear and turbulence intensity ratios (in the plane of the rotor) for a wind direction of 135 degrees
(SE).
Figure 11: Wind shear and turbulence intensity ratios (in the plane of the rotor) for a wind direction of 202.5 degrees
(SW).
Wind shear values on the Discovery Tower roof
tend to follow the same trends as the turbulence
intensity values. They are often low in areas of
accelerated flow, whereas, they are typically
higher in areas exposed to either vortex
shedding or flow separation near the roof top.
In areas where sheltering exists, the wind shear
values tend of be fairly low throughout the
profile.
Angle of Attack. The angle of attack of the
local wind vector was defined in two
components, the lateral angle of attack, θ
y
, and
the vertical angle of attack, θ
z
. The lateral angle
of attack is only important if it varies
significantly with height. If the angle is
consistent over the height of the entire rotor, the
wind turbine will respond to it as a change in
the mean wind direction. A lateral angle of
attack that varies within the profile can be more
troublesome, particularly to horizontal axis
wind turbines (HAWT). If a HAWT rotor
experiences different wind directions across its
span, some portion of the rotor will always be
exposed to wind forces that are not
perpendicular to its plane. This will, at best,
WindPower 2008, Houston, Texas June 2-4, 2008 9
result in inefficient power production because
portions of the blades will not be creating lift.
At worst, it could destroy the blade due to
excessive loading. Because most vertical axis
wind turbines (VAWT) are omni-directional,
the lateral angle of attack will likely have little
or no influence on the turbine behavior, even if
the angle of attack is varying along the axis of
the turbine. It may alter the torque profile of the
turbine through the rotation as different
portions of the rotor enter and exit the
maximum power production at different
segments of the rotation. A large vertical angle
of attack can also be responsible for decrease in
turbine performance due to less than optimum
lift on the blades and can create destructive
loads if the turbines are not designed to
sufficiently handle the vertical component of
the velocity.
Wind Turbine Selection
Process
The wind power density calculations
described above were combined with
manufacturer published power production
curves for four different wind turbines to
determine annual energy production (AEP)
values. All four of the turbine evaluated can be
described as vertical axis Darrieus wind
turbines. VAWTs are expected to have a better
chance of withstanding the strong wind shear
predicted to exist on the roof of the Discovery
Tower.
The first wind turbine evaluated, shown in
Figure 12, is the UK Quiet Revolution, QR5.
The QR5 turbine is 5 m tall and 3.1 m in
diameter, and it is rated at 10kW at 11 m/s.
Based on the manufacturer’s supplied power
curve, ten of these turbines placed on the
Discovery Tower roof are predicted to produce
approximately 77 MWh per year.
Figure 12: Quiet Revolution twisted Darrieus wind
turbine
The second turbine evaluated is the PAC
Wind Delta II 10kW H-Darrieus, shown in
Figure 13. Eight of the 10 kW units are
expected to have an AEP of approximately 70
MWh. (It is the Author’s opinion that this value
is inflated due to the manufacturer overstating
their turbine performance, which indicates
efficiencies in excess of 40%.)
Figure 13: PAC Wind H-Darrieus wind turbine
The third arrangement evaluated consisted
of fifteen 2.5 kW Turby twisted Darrieus wind
turbines (Figure 14). The calculated AEP for
the fifteen units is 35 MWh per year.
WindPower 2008, Houston, Texas June 2-4, 2008 10
Figure 14: Turby twisted Darrieus wind turbine
The last turbine evaluated is the 3 kW
Eurowind H-Darrieus. Twelve of the 2.2 m tall
by 2.5 m wide turbines are expected to have an
AEP of 52 MWh per year.
Figure 15: Eurowind H-Darrieus Wind Turbine
It should be noted that at the time of this
publication, none of these turbines have
undergone testing to show their performance
and/or reliability in a wind environment similar
to that found on the roof of the Discovery
Tower.
Conclusions
When assessing the merit of building
integrated wind turbines, it is important to
consider that wind conditions near the building
surface will be very different from the general
wind conditions in the region, due to both the
influence of neighboring structures and the
effects of the building itself. The winds will
typically be more gusty (turbulence intensity)
and uneven across the turbine blades (wind
shear), which can significantly affect the
turbine’s performance. Improperly located, a
wind turbine in this environment may be
subjected to an inadequate wind resource,
resulting in less than optimum power
production, and/or an environment that the
turbine is not designed to withstand.
Through the use of an atmospheric
boundary layer wind tunnel, the building design
team is able to identify the wind resource and
wind flow characteristics at the proposed
turbine location(s) during the design process so
that an accurate assessment can be made of the
potential power performance and survivability
of the wind turbine before the building is
constructed.
WindPower 2008, Houston, Texas June 2-4, 2008 11
Table 3
Estimated Annual Average Energy Production
REFERENCES
ABEL, C. (2003). Sky High
. Academy Publications. London.
ASCE, American Society of Civil Engineers (2003), Minimum Design Loads for Buildings and Other
Structures (ASCE 7-02).
ASCE, American Society of Civil Engineers (1999), Wind Tunnel Model Studies of Buildings and
Structures (ASCE Manual of Practice Number 67).
Cermak, J.E. (1971), “Laboratory Simulation of the Atmospheric Boundary Layer,” AIAA Jl., Vol. 9,
September.
Cermak, J.E. (1975), “Applications of Fluid Mechanics to Wind Engineering,” A Freeman Scholar
Lecture, ASME Journal of Fluids Engineering, Vol. 97, No. 1, March.
Cermak, J.E. (1976), “Aerodynamics of Buildings,” Annual Review of Fluid Mechanics, Vol. 8, pp. 75
– 106.
Derickson, R.G. and Peterka, J.A. (2004), Development of a Powerful Hybrid Tool for Evaluating
Wind Power in Complex Terrain: Atmospheric Numerical Models and Wind Tunnels, Paper
No. AIAA-2004-1005, AIAA/ASME Conference on Wind Power, Reno, NV.
ESDU (1993a) Strong winds in the atmospheric boundary layer, Part 1: mean hourly wind speeds,
ESDU Report 82026, ESDU International.
ESDU (1993b) Strong winds in the atmospheric boundary layer, Part 2: discrete gust speeds, ESDU
Report 83045, ESDU International.
WOOD, A. (2003). Pavements in the Sky: The Use of the Skybridge in Tall Buildings. Architectural
Research Quarterly. Cambridge University Press, UK. vol. 7. Nos. 3 & 4. 2003. pp. 325 -333.
YEANG, K. (1999). The Green skyscraper.
Prestel Verlag. Muni
Potential
Wind Height Width Area Number of
Turbine (m) (m)
(m
2
)
Turbines Location 1 Location 2 Location 3 Location 4 Total
Pacwind Delta II 4.3 4.0 7.4 8.0 9,196 8,013 8,770 8,925 69,808
Turby 2.9 2.0 5.7 15.0 2,509 2,130 2,383 2,419 35,404
QR5
5.0 3.1 15.5 10.0 8,189 6,976 7,782 7,973
77,300
Eurowind 3 kW 2.2 2.5 5.5 12.0 4,645 3,881 4,389 4,511 52,278
Rotor
Estimated Annual Average Energy Production
(kW hrs/yr)
(assumes 100% availablity)