Professional Documents
Culture Documents
This report has been prepared by WINDTECH Consultants Pty Ltd on behalf of our client and in accordance with relevant standards.
It takes into account our client's particular instructions and requirements. It is not intended for and should not be relied upon by a third party and
no responsibility is undertaken to any third party.
Page
1.0 Summary 3
5.3 Displacements 32
5.4 Accelerations 35
7.0 Conclusions 40
References 41
This report presents the results of the wind tunnel study of the wind induced
structural loads and load effects in relation to the proposed development known
as The Pad, located in the Dubai Business Bay precinct. This study is to estimate
the following aspects in relation to the proposed development;
• The distribution of equivalent static forces for the design of the structure.
Wind loads on the structure of the tower were measured using a 5 component
high-frequency base balance. This provides information on the moments about
the three principal axes of the structure and the total shear loads along the two
principal axes of the tower.
Measurements were carried out using a 1:300 scale model of the proposed
development, including the land topography and surrounding buildings for a
radius of approximately 375m. Testing was performed using Windtech’s
boundary layer wind tunnel, which has a 2.6m wide work section and has a fetch
length of 14m.
Design base moments, equivalent point load distributions and load cases have
been derived for the tower based on a 50 year return period wind speed and a
structural damping of 2.0% of critical. Conversion factors have been provided in
the report to convert these results to the 100 year return period wind speed.
The results of the study indicate that the proposed development benefits from
the shielding of the surrounding buildings. However, since this site is located in
an area which is currently being developed, it is important to assess if the
proposed surrounding buildings that were assumed in this study will be in place
at the time of completion of the construction of The Pad. If the surrounding
buildings are not in place at the time of completion of the subject
development then the base moments could be as much as 25% higher
than indicated in this report.
The 5 and 10 year return period building accelerations have been calculated for
the tower and are presented within the report. These have been calculated for
structural damping values of 0.5%, 1.0% and 1.5% of critical at the highest
occupiable level of the tower.
The results indicate that the peak combined accelerations are caused by the
rotational component of motion (about the Z-axis). The 5 year recurrence
standard deviation accelerations and the 10 year recurrence peak accelerations
satisfy the relevant occupant comfort criteria for all levels of structural damping
that were investigated.
The data presented in the following page, for the maximum means and gusts,
are based on an analysis of 31 years of data from the Dubai International Airport
(1974 to 2004) at a height of 10 metres in a flat terrain (the nearest houses to
the observation tower are located about 500 metres away).
• Synoptic winds - these dominate only the very short return period events
(for a return period of 1 year and less – plotted in blue in Figure 1).
If we ignore the fact that there are two systems, the results will be un-
conservative at the upper end of the range of return periods investigated.
The effect of a Shamal wind speed profile is not considered more critical for this
building than a Deaves and Harris (1978) profile, which is the basis of our
boundary layer modelling for this project. This is due to the following:
• Our analysis of the wind speed data partly accounts for the effect of a
Shamal wind in that the steeper profile is used in arriving at the design
windspeed.
• The Shamal wind speed profile tends to peak at heights greater than
200m, typically 300m to 400m. Below this height, the profile is somewhat
similar to that of a standard Deaves and Harris profile but with a more
linear variation.
• A sensitivity study carried out for a building taller than the one proposed
indicates that the Shamal wind speed profile did not govern the design
loads.
Based on this data, the maximum mean wind speeds for Dubai used for this
study, at a height of 10 metres in a flat terrain, are as follows;
Note: Divide by 0.6 to convert the reference mean wind speeds above into
reference gust wind speeds. For example, the reference gust design wind
speed for the 50 year recurrence is 25.5/0.6 = 42.5 m/s.
40
y = 11.85 + 3.773x
35
30
25
Gust (m/s)
20
15
10
0
-3 -2 -1 0 1 2 3 4 5 6 7 8
1yr 5yr 10yr 50yr 100yr
Reduced Variate
30
y = 6.62 + 2.945x
25
20
10minute mean (m/s)
15
10
0
-3 -2 -1 0 1 2 3 4 5 6 7 8
1yr 5yr 10yr 50yr 100yr
Reduced Variate
330 40 30
35
30
25
300 60
20
15
10
270 0 90
240 120
210 150
180
Testing was performed using Windtech’s blockage tolerant boundary layer wind
tunnel, which has a 2.6m wide work section and has a fetch length of 14m. The
model was placed in an open and suburban terrain boundary layer wind flows
based on the Deaves and Harris model (1978).
The mean wind speed and turbulence profiles, as well as the normalised power
spectral density function, were modelled in the wind tunnel match the full-scale
equivalent values for the terrains being modelled, as indicated in Appendix G.
The reference wind speeds were corrected for changes in the upstream land
topography. The upstream terrain category types and corresponding mean wind
speeds at the height of the tower are indicated in Table 2 for the 36 wind
directions tested.
The tower has a height of 107m above ground. The base of the tower, used for
the calculation of the base moments, structural loads and accelerations, is taken
to be the footing at the base of the basement level B2, 10.55m below the
ground level.
Measurements were carried out using a 1:300 scale model of the proposed
development, including the land topography and surrounding buildings, for a
radius of approximately 375 metres. The boundary layer wind flows matched the
model scale and the overall surrounding terrain characteristics beyond the 375m
radius. For the fetch beyond the extent of the surround model, the wind profiles
are simulated based on the Deaves and Harris model (1978).
The upstream terrains used in the analysis of this tower, as indicated in Table 2,
are summarised in Table 3. The corresponding gust terrain and height multiplier
M ( z ,cat ) and mean terrain and height multiplier M ( z , cat ) , based on the exposed
height of the tower, are also shown in Table 3.
M ( z ,cat ) at M ( z , cat ) at
Wind Direction Terrain Terrain
(Degrees) Type Description Building Building
Height Height
Figure 3 shows an aerial image of the site and the local surrounds within a 3km
radius of the proposed development. The fetch length of 3km is selected based
on the height of the proposed development, as outlined in the AS/NZS
1170.2:2002. The terrain type within this 3km zone defines the terrain category
type for each wind direction tested.
The figure above shows the section of terrain used in the determination of the
terrain category type for the 36 wind directions tested. The smaller circle
represents the extent of the physical surrounds model tested in the wind tunnel,
which for this study had a radius of 375m. The larger circle, which is 3km away
from the edge of the smaller circle, defines the extent of the fetch length of the
development.
The study model replicates the form of the proposed development to within
approximately 1m. Photographs of the model in the wind tunnel are shown in
Figures 4a and 4b. In addition to using a relatively large scale of 1:300, the
curved corners of the building model were artificially roughened to correctly
simulate the full-scale flow regime around the building (Reynolds Number
effect).
The model of this tower is attached to a high-frequency base balance rig. Figure
5 illustrates the base balance rig setup that was used for this study.
The axis convention adopted in this study, with regards to the definition of the
X, Y and Z axes for the tower, is illustrated in Figure 6.
The structural data used in this study is provided by the structural engineers
which includes three-dimensional mode shapes and the corresponding natural
frequencies for the various modes of vibration. The mass of each floor and
variation of the centre of mass with height for the tower is also included. The full
set of structural data for the tower is presented in Appendix A.
The first mode of the tower is pure translation along the X-axis (an about Y-axis
motion), with a full-scale natural frequency of 0.183Hz. The second mode is
mostly rotation about the Z-axis, with some coupling with the translational
motion along the Y-axis, with a full-scale natural frequency of 0.339Hz. The
third mode is mostly a translational motion along the Y-axis (an about X-axis
motion), with some coupling with the rotational motion about the Z-axis, with a
full-scale natural frequency of 0.347Hz. Note that the axis system referred to
here is based on that shown in Figure 6.
The spectra of the base moments about the X, Y and Z axes and the force
spectra for the primary modes of the tower are obtained directly from the wind
tunnel measurements. The responses are then derived to match the natural
frequencies and three-dimensional mode shapes of the prototype structure by
applying the modal analysis technique, described in Appendix F of this report.
The assumed structural damping of the prototype is 2.0% of critical for the 50
and 100 year return period levels. For serviceability levels, the results were
computed based on assumed structural damping levels of 0.5%, 1.0% and
1.5% of critical for the 5 and 10 year return periods.
Note that all results presented in this report for this tower are based on the axis
system and origin location shown in Figure 6. This is the aerodynamic
centreline axis of the tower that was used in the wind tunnel tests. Based on
the axis coordinate system from the structural engineer (shown in Figure A1 of
Appendix A), this is located at the X-coordinate of +0.0m and the Y-
coordinate of -13.0m. The location of this axis remains constant for the height
of the tower (unlike the centre of mass location, which changes for each
level).
The test procedures followed for the wind tunnel tests performed for this study
are based on the high-frequency force balance technique and generally adheres
to the guidelines set out in the Australasian Wind Engineering Society Quality
Assurance Manual (AWES-QAM-1-2001) and ASCE 7-02 (Section 6.6).
The design wind speeds used for the 50 and 100 year return period design
loads, as well as the 5 and 10 year serviceability design loads, are presented in
Section 2.0 of this report. These wind speeds are converted to mean building
height speeds in Table 2 of Section 3.0 of this report. The wind speeds used for
the determination of working stress and serviceability loads are based on the
exposed height of the tower above ground; 107m for this tower.
For example, the mean 50 year return design reference mean wind speed for the
tower height of 107m, for wind incident from the north, is calculated as follows;
where M ( z ,cat ) is the mean terrain and height multiplier for semi-urban terrain
(Terrain Category 3.5) at the exposed building height of 107m
above ground.
V50 is the basic gust reference wind speed for the 50 year return
period.
The maximum mean wind speed for the 50 year return period for the tower is
calculated to be 35.9m/s. This occurs for winds from the east (090 degrees), as
indicated in Table 2 of Section 3.0 of this report.
• between 174 and 288 minutes for the 5 year return period wind speeds
• between 155 and 260 minutes for the 10 year return period wind speeds
• between 126 and 218 minutes for the 50 year return period wind speeds
• between 117 and 202 minutes for the 100 year return period wind speeds
The mean and background response was derived from the time trace (Holmes et
al. 2003). The resonant component of the force was derived using a technique,
which accounts for modal coupling between the various axial components. This
technique involves the derivation of the time histories of the generalized forces
for each mode, by weighting the time histories of the measured base moments,
as shown in equations 4.2a to 4.2c as follows;
η1x M y (t ) η1 y M x (t )
F1 (t ) = + + η1θ M z (t ) (4.2a)
h h
η 2 x M y (t ) η 2 y M x (t )
F2 (t ) = + + η 2θ M z (t ) (4.2b)
h h
η3 x M y (t ) η3 y M x (t )
F3 (t ) = + + η3θ M z (t ) (4.2c)
h h
where η1x , η1 y , η1θ , η 2 x etc are correction factors that allow for the mode shape,
as well as the different contributions of each coordinate to the total modal
generalized force. η is equivalent to η1 in the notation of Boggs (1989, 1991).
Note that X-forces produce My base moments, and Y-forces produce Mx base
moments (by the Right-Hand Rule).
Then using the mode-shape correction factors for sway components derived in
Holmes (1987) and corrections for twist based on Tallin and Ellingwood (1985),
the above factors are given by the following;
⎛ 4 ⎞ ⎛ 4 ⎞⎟ ⎛ 1 ⎞
η1x 2 = ϕˆ1x 2 ⎜⎜ ⎟⎟ η1 y 2 = ϕˆ1 y 2 ⎜⎜ ⎟
η1θ 2 = ϕˆ1θ 2 ⎜⎜ ⎟⎟ (4.3a)
⎝ 1 + 3β1x ⎠ ⎝ 1 + 3β1 y ⎠ ⎝ 1 + 2 β1θ ⎠
⎛ 4 ⎞ ⎛ 4 ⎞⎟ ⎛ 1 ⎞
η 2 x 2 = ϕˆ2 x 2 ⎜⎜ ⎟⎟ η 2 y 2 = ϕˆ2 y 2 ⎜⎜ ⎟
η 2θ 2 = ϕˆ 2θ 2 ⎜⎜ ⎟⎟ (4.3b)
⎝ 1 + 3β 2 x ⎠ ⎝ 1 + 3β 2 y ⎠ ⎝ 1 + 2 β 2θ ⎠
⎛ 4 ⎞ ⎛ 4 ⎞⎟ ⎛ 1 ⎞
η3 x 2 = ϕˆ3 x 2 ⎜⎜ ⎟⎟ η3 y 2 = ϕˆ3 y 2 ⎜⎜ ⎟
η3θ 2 = ϕˆ3θ 2 ⎜⎜ ⎟⎟ (4.3c)
⎝ 1 + 3β 3 x ⎠ ⎝ 1 + 3β 3 y ⎠ ⎝ 1 + 2 β 3θ ⎠
β1x is the power law exponent of the mode shape for Mode 1 in the x direction,
as follows;
β1 x
⎛z⎞
ϕ1x = ϕˆ1x ⎜ ⎟ (4.4)
⎝h⎠
Then the spectral densities of each generalized force, S F 1 (η ) , S F 2 (η ) , S F 3 (η ) , are
calculated from their time histories.
Having obtained the spectral densities of the generalized forces, the mean
square values of the modal coordinates (time dependent but position
independent) in each mode can be obtained using standard random process
theory. Since we are dealing only with resonant response with low damping, the
standard white noise approximation can be used.
Thus, the accelerations at the top of the tower in the x, y and θ directions can be
obtained as follows;
Equations 4.5a to 4.5c imply that the three modes are well separated so that
their contributions to each response can be combined as the sum of squares.
⎡ πη1S F 1 (η1 ) ⎤
σ ax ,12 = ϕˆ1x 2 ⎢ ⎥
⎣ 4ζ 1G1 ⎦
2
1 1
⎡ πη S (η ) ⎤ 2 h ⎡ πη S (η ) ⎤ 2 ⎛ h ⎞
σM = ⎢ 1 F 1 2 1 ⎥ ∫ m( z )ϕ1x zdz = ⎢ 1 F 1 2 1 ⎥ ⎜⎜ ⎟⎟
y , R1
⎣ 4ζ G
1 1 ⎦
0
⎣ 4ζ G
1 1
λ
⎦ ⎝ 1x ⎠
G1
where
λ1x = (4.7)
⎛z⎞
∫0 m(z )ϕ1x ⎜⎝ h ⎟⎠dz
h
Hence the total mean square resonant base bending moment from the
background contribution and the resonant contributions from the three modes
can be obtained as follows;
⎛ h ⎞ ⎡ πη1S F 1 (η1 ) ⎤ ⎛ h ⎞ ⎡ πη 2 S F 2 (η 2 ) ⎤
2 2
σM = σ M y ,B + ⎜⎜ ⎟⎟ ⎢ ⎥ + ⎜⎜ ⎟⎟ ⎢
2 2
⎥
⎝ λ1x ⎠ ⎣ 4ζ 1 ⎦ ⎝ λ2 x ⎠ ⎣ 4ζ 2 ⎦
y
⎡ πη3 S F 3 (η 3 ) ⎤
2
⎛ h ⎞
+ ⎜⎜ ⎟⎟ ⎢ ⎥ (4.8)
⎝ λ3 x ⎠ ⎣ 4ζ 3 ⎦
Similar equations are applied to obtain the resonant components of Mx and Mθ.
Note that the parameters λ1θ , λ2θ , λ3θ , required to calculate the resonant
contributions to the base torque, are defined as follows;
G1
λ1θ = etc.
m(z )r 2ϕ1θ dz
h
∫0
The maximum and minimum peak moments about the X, Y and Z axes were
For serviceability loads, the peak factor, g, for resonant response, based on a 10
minute period, (T = 600 seconds) is given by the following;
g= 2 ln(ηoT ) (4.11)
Directional distributions of the peak, mean and RMS base moments are
presented in Appendix B of this report. These are the design base moments
based on the 50 year return period wind speeds and a structural damping of 2.0
percent of critical. A set of graphs showing the directional variations of the base
moments about the three axes, in both coefficient and full-scale loads format,
are also provided in Appendix B.
The base moment is defined here as the moment at the footing level. For this
tower, this is defined as the base of the basement level B2, which is 10.55m
below the ground level.
Table 4 presents the base moment results for the tower, based on the 50 year
return period wind speed. The wind direction which causes the peak base
moment is also indicated in this table. Note that the wind direction value listed
defines the direction that the wind is coming from. For example, a critical wind
direction of 0 degrees means wind coming from the north (ie: a northerly wind).
The maximum and minimum peak and mean base moments are presented in the
following table. For the maximum and minimum peak base moments, the
corresponding mean base moment for that particular wind direction is also
presented.
Note that to convert these base moments to 100 year return period base
moments a factor of 1.17 should be applied to the forces shown in Table 4.
Note also that the shielding of the surrounding buildings results in a reduction in
the design loads. Since this site is located in an area which is currently being
developed, it is important to assess if the proposed surrounding buildings that
were assumed in this study will be in place at the time of completion of the
construction of The Pad. If the surrounding buildings are not in place at
the time of completion of the subject development then the base
moments could be as much as 25% higher than indicated in this report.
Table 5 presents the base bending moments predicted by the Australian and
New Zealand Wind Loading Standard (AS/NZS 1170.2:2002) and the British
Loading for Buildings Standard (BS6399 Part 2, 1995), based on the 50 year
return period wind speed.
The Australian and New Zealand Wind Loading Standard estimates that motion
about the X and Y axes will be dominated by the along-wind response. The base
moments predicted by the Australian and New Zealand Standard are higher than
those measured from the wind tunnel tests. The base moments predicted by the
British Standard are also higher than those measured in the wind tunnel.
The design base moments (shown in Appendix B) have been converted into
series of point loads along the height of the tower by multiplying by the
generalised force distribution, as shown in Appendix C. The equivalent static
point loads in Appendix C are in full-scale values of the peak point loads (in kN
for X and Y and kNm for torsion). Along axis full-scale shear force diagrams for
the X and Y axes, based on the maximum base moment for the tower, are also
shown in Appendix D.
The point load distribution for the mean component is different to that of the
dynamic component of the peak load. This is with the exception of the torsional
base moments, where the mean component is distributed as per the dynamic
component. The mean base moments are distributed according to the distribution
of the net mean external force on the structure (combination of the external
windward and leeward pressures, as obtained from AS/NZS 1170.2 2002). In the
case of the dynamic component of the peak base moments, this is distributed
according to the inertial load distribution. The inertial moment contribution of
Floor i towards the dynamic component of the peak base moment is as follows;
2⎛ z ⎞⎟
M i , x = M b , x m(i )ψ (i, X )(2πη0, x ) ⎜⎜ ⎟ (5.1a)
⎝ M I ,x ⎠
⎛ z ⎞
M i , y = M b , y m(i )ψ (i, Y )(2πη0, y ) ⎜ ⎟
2
⎜M ⎟ (5.1b)
⎝ I ,y ⎠
2⎛ z ⎞⎟
M i , z = M b , z j (i )ψ (i, Z )(2πη0, z ) ⎜⎜ ⎟ (5.1c)
⎝ M I ,z ⎠
ψ (i, k ) is the mode shape value at Floor i for base moments about
axis k
Hence, from equations 5.1a to 5.1c, the equivalent sector shear or point load at
Level i become;
Fi , x = b , x (5.2a)
M I ,x
Fi , y = (5.2b)
M I ,y
Fi , z = b , z (5.2c)
M I ,z
Load envelopes have been derived for the tower based on the 50 year return
period wind speed. This is presented in Figure 7. From these load envelopes, 16
sets of load cases have been recommended for the tower. The recommended
load cases are presented in Table 7. The full set of point load distributions for the
tower are presented in Appendix C. Note that all point loads in the load case
must be applied at the same time.
The reference point load distributions for the tower are presented in Table 6. The
corresponding shear force distribution is presented in Table 8.
300
200
100
Mx (MNm)
0
-100
-200
o
Angle 100
-300
-100 -80 -60 -40 -20 0 20 40 60 80
My (MNm)
The wind directions resulting in a critical load combination (see Table 7) have
been highlighted in Figure 7 above for the 50 year return period wind speed.
These load envelopes are based on the aerodynamic centreline axis of the
tower that was used in the wind tunnel tests (see Figure 6). Based on the axis
coordinate system from the structural engineer (shown in Figure A1 of Appendix
A), this is located at the X-coordinate of +0.0m and the Y-coordinate of -
13.0m. The location of this axis remains constant for the height of the tower
(unlike the centre of mass location, which changes for each level).
T.O.R 3 15 0 6 342
21ST B 89 95 7 45 1826
19TH 87 99 7 50 1915
18TH 87 92 7 47 1808
17TH 86 86 7 45 1678
16TH 85 79 7 42 1576
15TH 84 76 7 41 1497
14TH 83 69 7 39 1345
13TH 82 64 7 36 1237
12TH 82 57 7 33 1130
11TH 81 52 7 31 1022
10TH 80 47 7 28 916
9TH 79 40 7 25 811
8TH 78 35 7 23 702
7TH 76 29 7 20 592
6TH 75 26 7 17 539
5TH 73 20 7 14 423
4TH 72 17 7 12 305
3RD 70 11 7 9 249
2ND 69 9 7 7 189
1ST 67 6 7 6 141
PM 86 4 11 3 114
P 80 16 11 14 0
G 0 0 0 0 0
B1 0 0 0 0 0
B2 0 0 0 0 0
BASE 0 0 0 0 0
Note that to convert these point loads to 100 year return period point loads a
factor of 1.17 should be applied to the forces shown in Table 6.
Load Wind Mx My
Load Case Mz
Case No. Angle Mean Dynamic Mean Dynamic
Max Mz, Max My,
1 290 0.98 -- 0.02 0.90 1.00
Mean Mx
Max Mz, Min My,
2 300 1.00 -- -0.81 -1.00 0.81
Mean Mx
Max Mz, Mean My,
3 300 1.00 1.00 -0.81 -- 0.81
Max Mx
Max Mz, Mean My,
4 100 -0.58 -0.88 -1.00 -- 0.87
Min Mx
Max Mz and Maximum
5 resultant for Max My 300 1.00 0.71 -0.81 0.71 0.81
and Max Mx
Max Mz and Maximum
6 resultant for Min My 300 1.00 0.71 -0.81 -0.71 0.81
and Max Mx
Max Mz and Maximum
7 resultant for Min My 100 -0.58 -0.62 -1.00 -0.47 0.87
and Min Mx
Max Mz and Maximum
8 resultant for Max My 100 -0.58 -0.62 -1.00 0.47 0.87
and Min Mx
Min Mz, Max My,
9 290 0.98 -- 0.02 0.90 -0.42
Mean Mx
Min Mz, Min My, Mean
10 300 1.00 -- -0.81 -1.00 -0.47
Mx
Min Mz, Mean My,
11 300 1.00 1.00 -0.81 -- -0.47
Max Mx
Min Mz, Mean My, Min
12 100 -0.58 -0.88 -1.00 -- -0.26
Mx
Max Mz and Maximum
13 resultant for Max My 300 1.00 0.71 -0.81 0.71 -0.47
and Max Mx
Max Mz and Maximum
14 resultant for Min My 300 1.00 0.71 -0.81 -0.71 -0.47
and Max Mx
Max Mz and Maximum
15 resultant for Min My 100 -0.58 -0.62 -1.00 -0.47 -0.26
and Min Mx
Max Mz and Maximum
16 resultant for Max My 100 -0.58 -0.62 -1.00 0.47 -0.26
and Min Mx
The point loads for the remaining levels of the tower are calculated in a similar
fashion.
Note that these load cases are based on the aerodynamic centreline axis of
the tower that was used in the wind tunnel tests (see Figure 6). Based on the
axis coordinate system from the structural engineer (shown in Figure A1 of
Appendix A), this is located at the X-coordinate of +0.0m and the Y-
coordinate of -13.0m. The location of this axis remains constant for the height
of the tower (unlike the centre of mass location, which changes for each level).
T.O.R 18 7 417
Note that to convert these shear forces to 100 year return period shear forces a
factor of 1.17 must be applied to the forces shown in Table 8.
5.3 Displacements
The mean and peak deflections at the highest occupiable level are derived from
the base moments by dividing by the inertial base moment, MI, as defined by the
following equation;
Note that the calculation of displacements is approximate for both mean and
peak as it assumes an inertial load distribution (based on dynamic stiffness
rather than static stiffness).
The value of MI for moments about the X-Axis is 6,480 MNm per 1 metre
displacement at the highest occupiable level of the tower. The corresponding
values for moments about the Y-Axis and Z-Axis are 2,165 MNm/m and 29,223
MNm/rad, respectively. The resulting maximum displacements at the highest
occupiable level of the tower are presented in Table 9.
Note that the values of displacement indicated in Table 9 are only approximate
and based on the motion of the tower in the first mode of vibration for that axis.
More accurate predictions can be obtained from the structural engineer’s finite
element model (if available), after applying the point load distributions provided
in this report.
Note that to convert the displacements shown in Table 9 to 100 year return
displacements, a factor of 1.17 should be applied. Floor by floor deflections can
be obtained by multiplying the above values by the normalised mode shape
values presented in Table 10.
5.4 Accelerations
The standard deviation and peak accelerations have been derived as a function
of the resonant component of the standard deviation displacement at the highest
occupiable level in the tower. The relevant occupant comfort criteria are
presented in Table 11 for below.
ISO’s Criteria
5.3 4.1 4.1
(5 year standard deviation)
The maximum combined accelerations for the tower are presented in Table 12.
The full set of directional results is given in Appendix E of this report.
The relationship between the standard deviation acceleration and the standard
deviation displacement is given by the following;
φ x , Max is the value of normalised mode shape for the X-Axis at the
highest occupiable level in the tower
σ a ( x , y , z ) = σ ax 2 + σ ay 2 + σ az 2 (5.5)
The peak accelerations, shown in Appendix E, were obtained using the relation;
α ax = g xσ ax (5.6)
The combined peak acceleration is the maximum of the largest component and
the adjusted vector sum of the three components of the peak acceleration
obtained by the following;
⎛ ⎞
r
σ a ( x , y ,z ) = ⎜⎜ ⎟⎟ σ ax 2 + σ ay 2 + σ az 2 (5.7)
⎝ 1+ r ⎠
2
The greater of the “combined” result (Equation 5.4) and the highest acceleration
due to motion about any one axis has been adopted as the maximum.
Maximum
Critical Wind
Combined Assumed Combined
Direction
Acceleration Damping Accelerations
(Degrees)
(milli-g)
The results indicate that the peak combined accelerations are caused by the
rotational component of motion (about the Z-axis). Hence the appropriate
criterion for the 5 year recurrence maximum combined standard deviation
accelerations is 4.1 milli-g, and 15.4 milli-g for the 10 year recurrence maximum
combined peak accelerations (as indicated in Table 11).
The natural frequencies of the three modes of vibration used for this study are
based on the structural data provided by the structural engineers which is
presented in Appendix A of this report. These are summarised in Table 13 below.
The sensitivity of the base moment results from the wind tunnel study, based on
a ±25% variation in the natural frequencies, is summarised in Table 15. Table
14 shows the values of the natural frequencies with a ±25% variation.
- 25 % 0% + 25 %
Mode Number
Variation Variation Variation
Note that to convert these base moments to 100 year return period base
moments a factor of 1.17 must be applied to the forces shown in Table 15. The
base moment is defined here as the moment at the footing level. For this tower,
this is defined as the base of the basement level B2, located 10.55m below the
ground.
A wind tunnel study has been carried out for the estimation of the structural
loads and building motion in relation to the proposed development known as The
Pad, located in the Dubai Business Bay precinct. Wind loads on the structure of
the tower were measured using a 5 component high-frequency base balance.
This provides information on the moments about the three principal axes of the
structure and the total shear loads along the two principal translational axes of
the tower.
Measurements were carried out using a 1:300 scale model of the proposed
development, including the land topography and surrounding buildings for a
radius of approximately 375m. Testing was performed using Windtech’s
boundary layer wind tunnel, which has a 2.6m wide work section and has a fetch
length of 14m.
Design base moments, equivalent point load distributions and load cases have
been derived for the tower based on a 50 year return period wind speed and a
structural damping of 2.0% of critical. Conversion factors have been provided in
the report to convert these results to the 100 year return period wind speed.
The results of the study indicate that the proposed development benefits from
the shielding of the surrounding buildings. However, since this site is located in
an area which is currently being developed, it is important to assess if the
proposed surrounding buildings that were assumed in this study will be in place
at the time of completion of the construction of The Pad. If the surrounding
buildings are not in place at the time of completion of the subject
development then the base moments could be as much as 25% higher
than indicated in this report.
The 5 and 10 year return period building accelerations have been calculated for
the tower and are presented within the report. These have been calculated for
structural damping values of 0.5%, 1.0% and 1.5% of critical at the highest
occupiable level of the tower.
The results indicate that the peak combined accelerations are caused by the
rotational component of motion (about the Z-axis). The 5 year recurrence
standard deviation accelerations and the 10 year recurrence peak accelerations
satisfy the relevant occupant comfort criteria for all levels of structural damping
that were investigated. Note that for a structure of this height and shape, it is
expected that the structural damping for a 5 or 10 year return event will be
approximately 1.0% of critical.
Structural Data
0.00E+00
G
22 B
P
D
SE
.R
21 A
PM
22 F
B1
B2
21 B
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
A
O
1S
9T
8T
7T
6T
5T
4T
3R
2N
D
ST
ST
O
BA
RO
20
19
18
17
16
15
14
13
12
11
10
N
N
T.
-1.00E-03
-2.00E-03
-3.00E-03
Phi (m)
-4.00E-03
-5.00E-03
-6.00E-03
-7.00E-03
-8.00E-03
-9.00E-03
Level Name
8.00E-03
6.00E-03
4.00E-03
Phi (m)
2.00E-03
0.00E+00
G
22 B
P
D
SE
.R
21 A
PM
22 F
B1
B2
21 B
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
A
O
1S
9T
8T
7T
6T
5T
4T
3R
2N
D
ST
ST
O
BA
RO
20
19
18
17
16
15
14
13
12
11
10
N
N
T.
-2.00E-03
-4.00E-03
-6.00E-03
Level Name
0.00E+00
G
22 B
P
D
SE
.R
21 A
PM
22 F
B1
B2
21 B
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
A
O
1S
9T
8T
7T
6T
5T
4T
3R
2N
D
ST
ST
O
BA
RO
20
19
18
17
16
15
14
13
12
11
10
N
N
T.
-1.00E-03
-2.00E-03
-3.00E-03
Phi (m)
-4.00E-03
-5.00E-03
-6.00E-03
-7.00E-03
-8.00E-03
-9.00E-03
Level Name
Directional Results of
Moments and Moment Coefficients
(50 Year Return Period)
400
300
200
Mx (MNm)
100
0
0 50 100 150 200 250 300 350
-100
-200
-300
1.00
0.80
0.60
0.40
CMx
0.20
0.00
0 50 100 150 200 250 300 350
-0.20
-0.40
-0.60
0.14
0.12
0.10
σMx Coefficient
0.08
0.06
0.04
0.02
0.00
0 50 100 150 200 250 300 350
80
60
40
20
My (MNm)
0
0 50 100 150 200 250 300 350
-20
-40
-60
-80
-100
0.20
0.15
0.10
0.05
0.00
CMy
-0.05
-0.10
-0.15
-0.20
-0.25
0.07
0.06
0.05
σMy Coefficient
0.04
0.03
0.02
0.01
0.00
0 50 100 150 200 250 300 350
50
40
30
20
Mz (MNm)
10
0
0 50 100 150 200 250 300 350
-10
-20
-30
0.20
0.15
0.10
0.05
CMz
0.00
0 50 100 150 200 250 300 350
-0.05
-0.10
-0.15
-0.20
0.04
0.04
0.03
σMz Coefficient
0.03
0.02
0.02
0.01
0.01
0.00
0 50 100 150 200 250 300 350
Results of Full-Scale
Point Load Distributions
(50 Year Return Period)
120
100
80
40
20
0
G
P
SE
H
PM
B
22 A
.R
T
B2
B1
F
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
21 H
22 B
21 A
O
1S
4T
5T
6T
7T
8T
9T
T
2N
3R
D
ST
ST
O
BA
10
11
12
13
14
15
16
17
18
19
20
RO
N
T.
Floor
Mean (kN) Dynamic (kN)
Point Load Distribution for the Maximum Peak Moments
About the X Axis (2.0% damping)
250
200
100
50
0
G
P
SE
H
PM
B
22 A
.R
T
B2
B1
F
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
21 H
22 B
21 A
O
1S
4T
5T
6T
7T
8T
9T
T
2N
3R
D
ST
ST
O
BA
10
11
12
13
14
15
16
17
18
19
20
RO
N
T.
Floor
Mean (kN) Dynamic (kN)
Point Load Distribution for the Torsional Loads (Expressed as a Moment)
About the Z Axis (2.0% damping)
4500
4000
3500
3000
2000
1500
1000
500
0
G
P
SE
B
PM
22 A
.R
T
F
B2
B1
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
21 H
22 B
21 A
O
1S
4T
5T
6T
7T
8T
9T
T
2N
3R
D
ST
ST
O
BA
RO
10
11
12
13
14
15
16
17
18
19
20
T.
Floor
Appendix D
Results of Full-Scale
Shear Force Distributions
(50 Year Return Period)
4500
3500
3000
2500
2000
1500
1000
500
0
G
P
SE
B
PM
22 A
.R
T
F
B2
B1
TH
TH
TH
TH
TH
TH
TH
TH
TH
TH
21 H
22 B
21 A
O
1S
4T
5T
6T
7T
8T
9T
T
2N
3R
D
ST
ST
O
BA
10
11
12
13
14
15
16
17
18
19
20
RO
N
T.
Floor
About X Axis (kN) About Y Axis (kN)
BA
SE
B2
B1
P
PM
1S
T
2N
D
3R
D
4T
H
5T
H
6T
H
7T
H
8T
H
9T
H
10
TH
11
TH
12
Floor
TH
13
TH
14
TH
15
TH
16
TH
17
TH
18
TH
19
TH
20
T
21 H
ST
Distribution of Accumulated Moment About the Z Axis
21 A
ST
22 B
N
D
22 A
N
D
B
RO
O
F
T.
O
.R
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
(kNm)
Cumulative Torque for the 50 Year Return Period
Appendix E
Accelerations Results
(5 and 10 Year Return Periods)
2.5
Accelerations (milli-g)
2.0
1.5
1.0
0.5
0.0
0 50 100 150 200 250 300 350
1.6
1.4
Accelerations (milli-g)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0 50 100 150 200 250 300 350
1.4
1.2
Accelerations (milli-g)
1.0
0.8
0.6
0.4
0.2
0.0
0 50 100 150 200 250 300 350
10.0
Accelerations (milli-g)
8.0
6.0
4.0
2.0
0.0
0 50 100 150 200 250 300 350
7.0
6.0
Accelerations (milli-g)
5.0
4.0
3.0
2.0
1.0
0.0
0 50 100 150 200 250 300 350
6.0
5.0
Accelerations (milli-g)
4.0
3.0
2.0
1.0
0.0
0 50 100 150 200 250 300 350
Relevant Literature
ABSTRACT: The paper discusses mode shape corrections and reviews processing
methodologies for the determination of the overall wind loading and response of tall buildings
using the high-frequency base balance technique. It is concluded that mode shape correction
factors currently used for twist modes, are conservative. The effect of cross-correlations
between base moments is found to be significant when calculating the response for coupled
modes.
1 INTRODUCTION
The high-frequency base balance (HFBB) technique is now more than twenty years old [1], and
has become the standard wind-tunnel method by which overall wind loads and responses such as
accelerations, displacements and velocities are determined for tall buildings at the design stage.
Essentially the mean (time-averaged) and quasi-static background base bending moments and
torques are determined by direct measurement, but the resonant dynamic components are
computed from the recorded time histories or spectral densities of the base moments. The
simplicity of the wind-tunnel models, the rapidity at which tests can be carried out and the ease
by which changes in basic dynamic properties such a frequency and damping can be
incorporated greatly outweigh the disadvantage of the neglect of aeroelastic effects, generally
regarded as negligible for the majority of habitable tall buildings.
However, although this is a well established technique, there are apparently still
significant differences in the methods in use for dealing with :
§ mode shape corrections for both sway and torsional, or twisting, dynamic
modes
§ coupled modes involving simultaneous sway and twist motions
§ the redistribution of the final base moments as effective static wind forces
over the height of the structure
This paper discusses the first two aspects of HFBB methodologies.
In the following, it will be assumed that mode shapes in sway can be fitted by a power function
of the form :
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
ß
z
µ(z) = (1)
h
A number of authors have considered the theoretical corrections required to correct the
spectra of linearly weighted base bending moments to those for generalized forces. Holmes [2]
derived the following correction factor assuming low correlation between the fluctuating
sectional forces at any pair of height levels on the building. The spectral density of fluctuating
sectional forces were assumed to be invariant with height.
2
1 3
SFx (n) = SMy (n) (2)
h 1 + 2ß
Holmes [2] also derived the corresponding limit for full correlation of the fluctuating
sectional forces, and proposed the following as an intermediate correction factor between the low
and high correlation limits.
2
1 4
SFx (n) = SMy (n) (3)
h 1 + 3ß
Boggs and Peterka [3] only considered the full correlation limit, but assumed that the
fluctuating forces varied with height as a power law. Vickery et al. [4] made some direct
measurements of the mode shape correction, using time histories of sectional forces based on
pressure measurements. Xu and Kwok [5] modified the low and high limits of Holmes [2] to
account for other variations of spectral density with height.
Table 1 summarizes the theoretical correction factors for various values of the mode
shape exponent β, and of the exponent, γ, used to describe the variation of spectral density of
fluctuating sectional forces with height in the form:
Table 1. Correction factors for spectral densities of generalized forces in sway modes
Mode- Low correlation High correlation Recommended Measured*
shape 3 + 2? 2+?
2
4 Vickery et al
exponent 1 + 2? + 2ß 1 + 3ß [4]
β 1+ ? + ß
γ=0 γ = 0.25 γ = 0.5 γ=0 γ = 0.25 γ = 0.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 -
1.25 0.86 0.88 0.89 0.79 0.81 0.83 0.84 0.86
1.5 0.75 0.78 0.80 0.64 0.67 0.69 0.73 0.71
2.0 0.60 0.64 0.67 0.44 0.48 0.51 0.57 0.61
The measurements of Vickery et al [4] were averages over the reduced frequency range
of interest for the resonant response of tall buildings in urban exposures. For the mode shape
exponents in the table, which cover the typical range for tall buildings, the measured factors
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
generally fall in between the high and low correlation limits, as expected. Note that assuming
high correlation is unconservative for β > 1.0.
For practical purposes, a simple function, that does not require knowledge of the
variation of spectral density with height, is desirable. This is provided by Equation (3), i.e. the
function 4/(1+3β), suggested by Holmes [2]. As shown in Table 1, this fits the experimental data
of Vickery et al [4] well. Another alternative would be the slightly more conservative low
correlation limit for γ = 0, Equation (2), which also matches the experimental data well.
To determine the acceleration in this mode, the spectral density of the generalized force is
factored by the transfer function for generalized acceleration. This involves the square of the
generalized mass in the denominator. This is the mass per unit height each height level
multiplied by the square of the mode shape, and integrated over the height of the structure. Then
the effective correction factor for mean square acceleration (to correct from the assumption of a
linear mode shape, β =0) is:
2
4 1 + 2ß
ηa =
2
(5)
1 + 3ß 3
To determine the mean square resonant base moment, the moment due to the inertial
forces arising from mass times acceleration at each height level is calculated. The resulting
mode shape correction factor to the mean square resonant base bending moment is then :
2
4 1 + 2ß
ηM =
2
(6)
1 + 3ß 2 + ß
For a value of β of 1.5, the resultant correction factor for the mean square resonant base
moment is 0.95. As the resonant component is usually about one half the total peak base
bending moment the total error in neglecting mode shape corrections is typically only about 1-
2%. Corrections to base bending moments are commonly ignored.
The HFFB measures a base torque uniformly weighted from the local torques per unit height
over the full height of the building model. Since the mode shapes for the lowest twist mode of
vibration of typical tall buildings increase monotonically with height from zero at the base, to a
maximum value at or near the top of the building, large corrections are required when converting
the measured fluctuating base torque to the fluctuating generalized force in the lowest twist mode
of vibration. This contrasts with the situation with the lowest sway modes, for which the
corrections to the measured base bending moments are small, and often neglected.
As for the sway modes, it will be assumed that mode shape can be fitted by a power
function of the form:
ßt
z
µt (z) = (7)
h
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
where β t is the mode shape exponent for the twist mode shape described by a power law.
As for the sway modes, a power law variation of the spectral density of fluctuating
sectional torque with height can be assumed :
Based on a uniform distribution (γ=0), and low correlation, the following correction
factor, is obtained :
1
S Ft (n) = S Mz (n) (9)
1 + 2ß t
Alternative corrections, derived by Boggs and Peterka [3], allowed for variation of the
fluctuating torque with height, but assumed full correlation of these fluctuating sectional torques
over height separations. The theoretical correction factors obtained with assumptions of both
high and low correlation, are shown in Table 2, together with the analysis of Tallin and
Ellingwood [6], based on measurements of Reinhold [7].
Table 2. Correction factors for spectral densities of generalized forces in twist modes
βt Low correlation High correlation Recommended Measured
1 + 2? 1+ ?
2 1 Tallin &
1 + 2? + 2ß t + Ellingwood [6]
1+ ? + ß 1 2ßt
t
Table 2 shows that a correction factor, based on low correlation and uniform fluctuating
torques with height (Equation (9)), fits the data analysed by Tallin and Ellingwood well. This is
not surprising as the correlations between fluctuating torques at various heights were computed
by Tallin and Ellingwood, and were found to be very low.
At several wind-tunnel laboratories, a simple factor of 0.7 is used to convert the
fluctuating measured base torque, Mz (t), to the fluctuating generalized force in the twist mode,
irrespective of the mode shape. This corresponds to a factor on the spectral density of (0.7)2 =
0.49. As shown in Table 2, this is considerably greater than that obtained by Tallin and
Ellingwood [6] from Reinhold’s data, and appears to be over-conservative.
The correction factors for angular acceleration and resonant base torque are obtained in a
similar way to the sway modes, giving Equations (10) and (11) respectively.
1 1 + 2ß t
2
ηat =
2
(10)
1 + 2ß t 1
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
2
1 1 + 2ß t
ηT =
2
(11)
1 + 2ß t 1 + ß t
Many modern tall buildings have dynamic modes that involve simultaneous sway and twist
motions. This often results from differences between the centre of mass and centre of stiffness
(shear centre) of the cross-section. Several methods are available to process the outputs of the
HFFB to make reasonable predictions of the resonant contributions from the coupled modes of
tall buildings. These are summarized in the following.
Method 1. In this method, the spectra (i.e. auto spectral densities) of the three output
signals proportional to Mx , My and Mz, are determined. Then the spectra of the generalized
forces for the lower modes (usually three in number) are determined by linear weighting and
summing of the resulting spectral densities. The relevant equations for three modes are as
follows.
SF1 (n) = η1x 2 (1/h2 ) SMy(n) + η1y2 (1/h2 ) SMx(n) + η1θ2 SMz(n)
SF2 (n) = η2x 2 (1/h2 ) SMy(n) + η2y2 (1/h2 ) SMx(n) + η2θ2 SMz(n)
SF3 (n) = η3x 2 (1/h2 ) SMy(n) + η3y2 (1/h2 ) SMx(n) + η3θ2 SMz(n) (12)
Weighting terms like η1x , η1y etc. allow for the contribution of each component of base
moment in the generalized force, as well as the mode shape corrections as discussed earlier.
Method 2. Errors in neglecting any correlation between the three measured moments in
Method 1, are avoided by directly forming the time histories of the generalized forces for each
mode, by weighting the time histories of the measured base moments, and then calculating the
spectral densities from the new time series.
F1 (t) = η1x (1/h) My (t) + η1y (1/h) Mx (t) + η1θ Mz(t)
F2 (t) = η2x (1/h) My (t) + η2y (1/h) Mx (t) + η2θ Mz(t)
F3 (t) = η3x (1/h) My (t) + η3y (1/h) Mx (t) + η3θ Mz(t) (13)
Method 3. The correlations between the measured base moments, Mx , My and Mθ, can be
incorporated in Method (1) by including the additional cross-spectral density terms in equations
for the spectral densities of the generalized forces for the coupled modes. This approach was
described by Irwin and Xie [8]. For example, the equation for the generalized force in Mode 1
can be written.
SF1 (n) = η1x 2 (1/h2 ) SMy(n) + η1y2 (1/h2 ) SMx(n) + η1θ2 SMθ(n)
+ 2η1x η1y (1/h2 ) SMyMx(n) + 2η1x η1θ (1/h) SMyMθ(n) +2η1y η1θ (1/h) SMxMθ(n) (14)
This method should give similar results to Method 2.
Method 4. The Yip and Flay [9] approach is a sophisticated frequency domain method
designed to account for the shortcomings of the previous methods – namely neglect of the cross-
spectral terms in Method 1, and also the reliance on empirical or theoretical mode shape
corrections in all three methods. The cross-spectral densities (both real and imaginary
components) are represented by low-order polynomial expressions in two space variables with
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
unknown frequency-dependent coefficients. The full set of auto-spectra and cross spectra from a
5-component base balance test, are used to determine the unknown coefficients. With this
information, the unknown spectra of the generalized forces of the coupled modes can be
determined. This method has apparently not been used in commercial wind-tunnel practice,
and will not be discussed further in this paper.
As an example to illustrate the differences in results obtained from Methods 1 to 3, the results of
processing HFBB data from a tall building with significant coupling in two modes of vibration
are presented. The proposed building was approximately 200 metres in height with the cross
section shown in Figure 1. Due to an eccentricity in the lift core, the building has significant
coupling between sway in the x-direction and twist in both Modes 1 and 3. Wind-tunnel tests
were performed on a tall building model (scale 1:250) placed in Windtech’s boundary-layer wind
tunnel with a blockage-tolerant test section. The tower section of the model was attached to a
high-frequency base balance. The axis convention adopted for the tests is shown in Figure 1. A
sample rate of 512 samples per second was used to sample the three base bending moments, with
a sample time of 64 seconds. The full-scale building had frequencies in the first three modes
between 0.15Hz and 0.20Hz. Time histories and spectral densities of generalized forces for three
modes were calculated, using the mode shape corrections for sway and twist described in
Sections 2.1 and 2.2, respectively. The resonant response contributions to the total base
moments, and the accelerations at the top of the building, were calculated using standard
procedures, based on the white noise approximation.
Figure 2 shows the variation of spectral density of the generalized force for Mode 1 at the
natural frequency of that mode, as a function of wind direction. Methods 2 and 3 give similar
predictions as expected – any differences are due to statistical sampling, or numerical errors in
the calculations. However, values from Method 1 are generally greater than those from Methods
2 and 3. This can be attributed to negative correlations between the moments My and Mz. This
contributes to the largest cross-spectral term in Equation (14), and hence reduces the predicted
generalized force from that given by Method 1, which ignores the cross-spectral terms.
Yy
My, +ve
70000
60000
50000
Method 1
40000
SF1(n1 )
Method 2
30000
Method 3
20000
10000
0
0 50 100 150 200 250 300 350
Wind Direction (degrees)
Figure 2. Comparison of the spectral density of generalized force (at frequency, n1 ) for Mode 1 by three different
methods
Table 3 gives the results of some final peak response predictions for the building.
Coefficients of the base moments are defined in Equation (15). h is the building height, b is the
maximum breadth, andUh is the mean wind speed at the top of the building.
Mx My Mz
C Mx = 2 2
C My = 1 2 2
C Mz = 2 2 (15)
1 1
2
?U h
bh 2
?U h bh 2
? U h b h
The maximum coefficient of the base bending moment, Mx , which is not affected by the
coupling, is similar by all three Methods. However, Method 1 gives significant overestimates
for My and Mz due to neglect of the cross-coupling terms in calculating the resonant
contributions, as discussed previously. The resultant acceleration is also slightly overestimated
by Method 1. However, neglect of any coupling (i.e. assuming that Modes 1, 2 and 3 are pure x-
sway, y-sway and twist, respectively), results in even greater overestimates for My and Mz, but
the resultant acceleration, in this case, is quite accurate.
Table 3. Comparison of response predictions for a building with two coupled modes
5yr. Std. Devn. Coefft. of peak Coefft. of peak Coefft. of peak
Acceleration Base bending Base bending Base torque
(mg) moment M x moment M y Mz
Method 1 3.0 1.30 0.64 0.40
Method 2 2.8 1.33 0.58 0.31
Method 3 2.7 1.31 0.56 0.32
Uncoupled 2.7 1.33 0.70 0.41
11th International Conference on Wind Engineering, Texas, June 1-5, 2003
4 CONCLUSIONS
Mode shape corrections for the high-frequency base balance for both sway and twist modes have
been reviewed. It is concluded that mode shape correction factors currently used for twist
modes, are conservative, and a new function, based on the low correlation limit, is proposed.
Four methods of processing the recorded base moments to deal with resonant modes that
have significant components of both sway and twist, have been discussed. The effect of cross-
correlations between base moments was found to be significant when calculating the response
for coupled modes. In this case, this resulted in overestimation of the resonant contributions to
two of the base moments, when those terms were neglected (Method 1). This may not be a
general conclusion for all buildings, however.
ACKNOWLEDGEMENTS
The permission of Bovis-Lend Lease and The Arup Group to present results from their building
in this paper, is acknowledged. However, the opinions and interpretations expressed are solely
those of the authors. Acknowledgements are also due to Dr. Daryl Boggs (CPP Inc.) and Dr.
Eric Ho (University of Western Ontario) for valuable comments on the coupling methodologies.
REFERENCES
1 T. Tschanz, Measurement of total aerodynamic loads using elastic models with high natural frequencies,
International Workshop on Wind Tunnel Modeling Criteria and Techniques in Civil Engineering Applications,
Gaithersburg, Maryland, 1982
2 J.D. Holmes, Mode shape corrections for dynamic response to wind, Engineering Structures, 9 (1987) 210-212.
3 D.W. Boggs and J. A. Peterka, Aerodynamic model tests of tall buildings. Journal of Engineering Mechanics,
ASCE, 115 (1989) 618-635.
4 P.J. Vickery, A. Steckley, N. Isyumov, and B.J. Vickery, The effect of mode shape on the wind-induced
response of tall buildings, 3rd. U.S. National Conference on Wind Engineering, Lubbock, Texas, November 6-8,
1985.
5 Y.L. Xu, and K.C.S. Kwok, Mode shape corrections for wind tunnel tests of tall buildings, Engineering
Structures, 15 (1993) 387-392.
6 A. Tallin, and B. Ellingwood, Analysis of torsional moments on tall buildings, Journal of Wind Engineering &
Industrial Aerodynamics, 18 (1985) 191-195.
7 T.A.Reinhold, Measurement of simultaneous fluctuating loads at multiple levels on a tall building in a simulated
urban boundary layer, Ph.D. Thesis Virginia Polytechnic Institute and State University, 1977.
8 P.A. Irwin and J.Xie, Wind loading and serviceability of tall buildings in tropical cyclone regions, 3rd. Asia-
Pacific Symposium on Wind Engineering, Hong Kong, 1993.
9 D.Y.N. Yip and R.G. J. Flay, A new force balance data analysis method for wind response predictions of tall
buildings, Journal of Wind Engineering & Industrial Aerodynamics, 54/55 (1995) 457-471.
Appendix G
Dear Adam,
This is to confirm that I have reviewed the results for the overall
wind load study for the above project, dated 30 July 2007 and found them
to be acceptable, subject to confirmation from the local pressure study to be
carried out later. As far as I can determine, the experimental and analysis
procedures for the load study were consistent with the Quality Assurance
Manual (QAM-1-2001) of the Australasian Wind Engineering Society.
Regards
J. D. Holmes
(Ph.D., F.I.E.Aust., C.P.Eng.)
Velocity and Turbulence Profiles
1:300 Scale, Terrain Category 2
350
300
250
200
Height (m)
150
100
50
0
0 0.2 0.4 0.6 0.8 1 1.2
0.25
Normalised Spectral Density, nS(n)/VAR
0.2
0.15
0.1
0.05
0
0.00001 0.0001 0.001 0.01
Reduced Frequency, f/U
Velocity and Turbulence Profiles
1:300 Scale, Terrain Category 3
350
300
250
200
Height (m)
150
100
50
0
0 0.2 0.4 0.6 0.8 1 1.2
Normalised Velocity, Local Turbulence Intensity
0.25
Normalised Spectral Density
0.2
0.15
0.1
0.05
0
0.0001 0.001 0.01
Reduced Frequency, f/U