Upload
yogesh-billa
View
218
Download
0
Embed Size (px)
Citation preview
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 1/13
Engineering Structures 29 (2007) 2641–2653
www.elsevier.com/locate/engstruct
Wind load effects and equivalent static wind loads of tall buildings based onsynchronous pressure measurements
Guoqing Huang, Xinzhong Chen∗
Wind Science and Engineering Research Center, Department of Civil and Environmental Engineering, Texas Tech University, Lubbock, TX 79409, USA
Received 20 September 2006; received in revised form 9 January 2007; accepted 10 January 2007
Available online 1 March 2007
Abstract
This paper addresses the wind load effects and equivalent static wind loads (ESWLs) of tall buildings based on measured synchronous surface
pressures in a wind tunnel. The variations of the gust response factors (GRFs) associated with different alongwind responses are studied, pointing
out the deficiency of the traditional ESWL modeling based on the GRF associated with the top displacement. The higher mode contributions
to various building responses such as top acceleration, story shear force and bending moment at different building elevations are investigated,
which reveals the noticeable contributions of higher modes to some important building responses. The comparison of wind load effects with
those based on the wind load specification ASCE 7-05 is also conducted. Furthermore, the practical methodology of modeling ESWLs on tall
buildings involving the influence of higher mode contributions is proposed. Finally, using the spatiotemporally varying wind load information, the
mode shape correction factors required in the high frequency force balance (HFFB) technique are revisited to examine the efficacy of empirical
formulations adopted in current practice.c 2007 Elsevier Ltd. All rights reserved.
Keywords: Wind; Wind load; Tall building; Structural dynamics; Wind tunnel test; Aerodynamics
1. Introduction
Wind loads on tall buildings can be quantified through
multiple point synchronous scanning of pressures on a building
model surface in a wind tunnel, or by high frequency force
balance (HFFB) measurements, or by simplified wind loading
codes. The synchronous pressure measurements provide a
detailed description of spatiotemporally varying wind loads,
while the HFFB measurements offer an estimate of the
generalized forces of the fundamental modes of vibration. The
wind loading codes give simplified and often conservative
equivalent static wind loads (ESWLs) on isolated buildingswith simple geometric configurations. For tall buildings
with uncoupled mode shapes in primary directions, wind-
induced responses in each primary direction can be analyzed
independently often involving only the fundamental mode.
The fundamental mode contribution dominates global
building response such as the top displacement and base
∗ Corresponding author. Tel.: +1 806 742 3476x324; fax: +1 806 742 3446. E-mail address: [email protected](X. Chen).
bending moment. However, higher modes may have noticeable
contributions to some responses such as the top acceleration.
The studies by Simiu [19], Kareem [15] and Simiu and
Scanlan [20] using analytical loading models demonstrated that
the contributions from higher modes may reach to about 20%
of the top acceleration. More detailed discussions concerning
the contributions of higher modes to various building responses
using measured spatiotemporally varying wind loads have not
been adequately addressed in the literature.
Current design practice often requires the dynamic wind
loads to be represented in terms of the ESWLs. The gustresponse factor (GRF) approach proposed by Davenport [8]
has been widely used in design codes and standards worldwide
for modeling alongwind loading. This approach leads to an
ESWL given by the mean wind load multiplied by a GRF, often
associated with the top displacement. Studies have shown that
GRF may vary widely for different response (e.g., Holmes [12];
Chen and Kareem [4–6]). Therefore, using a single GRF
for all response components may lead to remarkable over-
or underestimates of building response. Moreover, the GRF
approach falls short in providing physically meaningful ESWLs
0141-0296/$ - see front matter c
2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2007.01.011
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 2/13
2642 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
for the acrosswind and torsional responses, which are typically
characterized by zero mean wind loading and response. It
should be mentioned that the Australian/New Zealand wind
loading standard AS/NZS 1170.2: [2] has adopted an advanced
ESWL modeling with a dynamic response factor which
increases with increasing height [2]. The advanced modeling
of ESWLs on tall buildings including background and resonantcomponents has been addressed extensively in the literature
(e.g., Kasperski [17]; Holmes [13]; Kareem and Zhou [16];
Chen and Kareem [4–6]; Repetto and Solari [18]).
The HFFB technique requires empirical mode shape
correction factors for estimating the fundamental generalized
force through measured base bending moment or base torque.
A number of empirical formulas have been suggested in the
literature (e.g., Vickery et al. [23]; Boggs and Peterka [3]; Xu
and Kwok [25]; Ho et al. [10]; Holmes [11]; Holmes et al. [14];
Chen and Kareem [4]). HFFB measurements have also been
used to identify spatiotemporally varying fluctuating wind loads
based on assumed loading models (e.g., Yip and Flay [26]; Xie
and Irwin [24]). The detailed spatiotemporally varying pressuredata provide an opportunity to verify the adequacy of these
empirical formulas used in practice.
This study addresses the wind load effects and ESWLs of tall
buildings based on measured surface pressures of 20- and 50-
story building models. The variations of the GRFs and higher
mode contributions for various building responses are studied in
detail. Furthermore, the comparison of building response with
that based on wind loads specified in ASCE 7-05 (ASCE [1])
is conducted. Finally, based on the detailed dynamic wind load,
the mode shape correction factors used in the HFFB technique
are revisited to examine the efficacy of empirical formulas
adopted in current practice.
2. Analysis framework
Consider the wind-excited response of a tall building in
one of three primary directions in which the building exhibits
a one-dimensional uncoupled mode shape of vibration. The
wind load acting at each floor level, denoted by the mean and
dynamic components P̄ j and P j (t ) ( j = 1, 2, . . . , N ; N is the
total number of building floors), are quantified from multiple
point synchronous scanning of pressures on the building model
surface in a wind tunnel. The mean component of a specific
response of interest, R, e.g., displacement, shear force, bending
moment and member force, is estimated by static analysis underthe mean wind load P̄ j and expressed as
¯ R = N
j=1
µ j P̄ j (1)
where µ j is the influence function of response R defined as the
response under a unit load acting at the j th floor.
The dynamic response, R(t ), is determined through modal
analysis:
R(t )
=
N
n=1
Bnqn(t ) (2)
M n[q̈n(t ) + 2ξnωn q̇n(t ) + ω2n qn(t )] = Qn(t ) (3)
where Bn is the modal participation coefficient; qn(t ) is the
generalized displacement; M n, ξn , and ωn are the generalized
mass, damping ratio and circular frequency, respectively;
Qn(t ) =
N j=1 Θ j n P j (t ) is the generalized force; and Θ j n is
the mode shape in terms of the j th floor motion.
The modal contributions to the response can be divided
into two parts: the first N d modes where the dynamic
effect is significant, and modes N d + 1 to N with natural
frequencies such that their response is essentially quasi-static.
Subsequently, the classical mode displacement superposition
method given by Eq. (2) can be replaced by the static correction
method (e.g., Chopra [7]):
R(t ) = N d n=1
Bnqn(t ) + N
n= N d +1
Bn qnb(t )
= N d n=1
Bnqnr (t ) + N
n=1 Bnqnb(t ) (4)
where qnb(t ) = Qn(t )/K n is the nth background (quasi-static)
generalized displacement and qnr (t ) = qn(t ) − qnb(t ) is the
nth resonant generalized displacement, i.e., the generalized
displacement excluding the quasi-static component; K n = M nω
2n is the nth generalized stiffness.
It is noted that the static correction method essentially
corresponds to the analysis approach of separating the dynamic
response into the background and resonant components. This
approach has traditionally been introduced in the frequency
domain, and widely used in the analysis of wind-induced
response of structures. The quantification of the backgroundresponse including all mode contributions is equivalent to the
quasi-static analysis in terms of the influence function, i.e.,
Rb(t ) = N
j=1
Bn qnb(t ) = N
j=1
µ j P j (t ). (5)
Response analysis can be carried out in either time or
frequency domain. In the time domain scheme, the response
time history is quantified by using the step-by-step integration
method, e.g., Newmark’s numerical method. Based on the
response time history, R(t ), the root-mean-square value (RMS),
σ R , and the peak value, Rmax, can be directly determined.The ensemble-averaged quantities are then computed based on
multiple samples. The peak factor and gust response factor
associated with R(t ) are given as g R = Rmax/σ R and G R =( ¯ R + Rmax)/ ¯ R, respectively.
Alternatively, when the frequency domain scheme is
utilized, the RMS value of response R(t ) is quantified by
combining the background and resonant components as
σ 2 R = σ 2 Rb+ σ 2 Rr
(6)
σ 2 Rb
=
N
j=1
N
k =1
µ jµk σ P j σ Pk r P jk (7)
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 3/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2643
σ 2 Rr =
N d m=1
N d n=1
Bm Bnσ qmr σ qnr r mnr (8)
where σ Rband σ Rr are the RMS values of the background and
resonant response components, respectively; σ P j is the RMS
value of P j (t ); r P jk is the correlation coefficient between P j (t )
and Pk (t ); σ qnr is the RMS value of the resonant generalizeddisplacement in the nth mode and can be estimated using the
closed-form formulation:
σ qnr =1
K n
π f n
4ξn
SQn ( f n) (9)
where SQn ( f ) is the power spectral density (PSD) function
of the generalized force Qn(t ); and r mnr is the correlation
coefficient between the mth and nth resonant modal responses,
which depends on not only the frequency ratio and damping
ratios, but also the coherence of the generalized forces, and can
be estimated as follows [9,5,6]:
r mnr = ρmnr αmnr (10)
ρmnr
= 8√ ξmξn(βmnξm + ξn)β
3/2mn
(1 − β2mn )
2 + 4ξmξnβmn(1 + β2mn ) + 4(ξ2
m + ξ2n )β
2mn
(11)
αmnr = Re[SQmn ( f )]/
SQm ( f )SQn ( f )
f = f m or f n
(12)
where βmn = f m/ f n; and Re denotes the real part of the
complex value.
The accuracy of estimating resonant response through
the closed-form formulations instead of using the numericalintegration of response PSD will be affected by the smoothness
of the PSD of the generalized force around the modal frequency.
The PSD of the generalized force calculated from wind loading
history with limited samples and time duration are often a
jagged function of frequency. To enhance the accuracy of the
predicted resonant response, a “smooth” estimation of the PSD
at the modal frequency can be used in the analysis by averaging
the spectra over the half-power bandwidth.
If the background response is quantified through modal
analysis, similar to the resonant response, the complete
quadratic combination (CQC) should be used:
σ 2 Rb=
N m=1
N n=1
Bm Bnσ qmbσ qnb r mnb (13)
σ qnb = 1
K n
∞0
SQn ( f n)d f = σ Qn
K n(14)
r mnb = r Qmn (15)
where σ qnb and σ Qn are the RMS vales of the nth background
generalized displacement and generalized force; r mnb and r Qmn
are the correlation coefficients between generalized background
displacements and between generalized forces in the mth and
nth modes.
The peak dynamic response including the background and
resonant components is:
Rmax =
g2bσ
2 Rb
+ g2r σ
2 Rr
(16)
where gb and gr are the peak factors for the background and
resonant components, respectively, usually ranging from 3 to 4.
The building acceleration is of interest for the building
habitability design, which can be determined through modal
analysis in either time or frequency domain. When the
frequency domain approach is employed, the RMS values of
the background and resonant components of the nth generalized
acceleration are determined as
σ ̈qnb = 1
K n
∞0
(2π f )4 SQn ( f )d f (17)
σ ̈qnr = ω2nσ qnr = 1
M n
π f n
4ξn
SQn ( f n). (18)
The background acceleration is generally negligibly small as
compared to the resonant component. When only the resonant
component is considered, the RMS value of the acceleration at
the j th floor level, i.e., a j (t ) = N n=1 Θ jn q̈nr (t ), is given by
σ 2a j=
N m=1
N n=1
Θ jm Θ j nσ ̈qmr σ ̈qnr r mnr . (19)
3. Building examples and associated wind loading
3.1. Building dynamic properties
Two shear buildings of 20 and 50 stories with the same
square cross section of 40 m × 40 m and story height of
4 m are chosen as examples. Both buildings are modeled
as a lumped-mass system with the floor mass of 1.228 ×106 kg that corresponds to a building density of 192 kg/m3.
Both buildings have uncoupled mode shape in each primary
direction, thus building response in each direction can be
analyzed independently. The fundamental frequencies in three
primary directions are assumed to be identical, whereas the
torsional frequency is generally larger than the translational
frequency. The fundamental frequencies for the 20- and 50-
story buildings are assumed as 0.4415 Hz and 0.2122 Hz,
respectively, according to the empirical formula for the natural
frequency of steel moment resisting frames suggested in
ASCE 7-05. Four different story stiffness distributions for each
building are considered. In the first three cases, the story
stiffness is assumed to vary over the building height which
is determined based on the prescribed fundamental frequency
and modal shape. The fundamental mode shape is assumed
to follow a power law with an exponent β = 1.0, 1.25, and
1.5, respectively. The last case corresponds to a uniformly-
distributed story stiffness which is determined from the given
fundamental frequency. Fig. 1 shows the mode shapes of the
first 5 modes of the 50-story building. The modal frequencies
of both buildings are found to be well separated at least in
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 4/13
2644 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
Fig. 1. Mode shapes of the 50-story building.
(a) 20-story building. (b) 50-story building.
Fig. 2. Experimental building models with pressure taps.
the first 10 modes. For the example of the 50-story building
with β = 1.25, the frequency ratios are f 2/ f 1 = 2.3 and
f 3/ f 2 = 1.6. The damping ratio for all modes of both buildings
is assumed to be 2%.
3.2. Wind loading
Wind loads on both buildings are derived from multiple
point synchronous scanning of pressures on the building model
surface in a wind tunnel [22]. These experiments were carried
out with a length scale 1/400 for a suburban terrain where thepower law exponent of the mean wind speed profile was 1/6.
Total 200 and 500 wind pressure taps shown in Fig. 2 were
uniformly distributed over the four wall surfaces of the models
of 10 cm × 10 cm × 20 cm and 10 cm × 10 cm × 50 cm for the
20- and 50-story buildings, respectively. Each wall of these twomodels has 10 and 25 layers of pressure taps, respectively. The
wind direction was set normal to the wall face.The sampling interval of wind pressure time history was
0.00128 s, and total 32 768 data were recorded during the time
(a) 20-story building. (b) 50-story building.
Fig. 3. Story force coefficients.
duration of about 42 s. As the mean wind speed at the building
top, U H , varies from 20 to 60 m/s, the full-scale duration of
the record ranges from 148 to 50 min for the 20-story building.
To maintain the same mean wind speed profile for the 50-story
building, the range of U H is set from 23.3 to 69.9 m/s and
the corresponding time duration is from 160 to 53 min. Theentire pressure data were divided equally into six independent
samples to compute the structural responses which are then
used for estimating their ensemble-averaged quantities.
The translational forces and torque at each floor level are
determined by integrating wind pressures within the tributary
area. The forces of the stories at which there is no pressure tap
located are determined by interpolating the story forces acting
on two adjacent stories. Fig. 3 shows the mean and RMS story
force coefficients in the three primary directions of the 20- and
50-story buildings. The translational force coefficient is defined
as the ratio of the force to the product of wind speed pressure
at the building top and the building frontal area of each story,
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 5/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2645
(a) Fundamental mode. (b) Second mode.
Fig. 4. PSDs of the generalized forces (50-story building, β = 1.25).
Fig. 5. RMS values of the generalized forces (50-story building, β = 1.25).
whereas the torque coefficient is calculated by further dividing
the building width. It is observed that the mean alongwind
force coefficient increases along the building elevation, whilethe mean acrosswind story force and torque coefficients are
negligibly small. The RMS force coefficients in three primary
directions are almost constant over the building height. Figs. 4
and 5 show the PSDs and RMS values of the generalized forces
associated with the fundamental and second modes in three
directions. The mode shapes are normalized according to unity
generalized masses. It is noted that the acrosswind dynamic
load is greater than the alongwind and torsional loads. The
fundamental modes correspond to larger loads than those of the
higher modes.
4. Results and discussions
4.1. Mean, dynamic responses and GRF
As the buildings with four different story stiffness
distributions show similar characteristics of wind load effects,
only the building with the power law exponent of the
fundamental mode shape β = 1.25 is discussed here. Fig. 6
shows the alongwind displacement, shear force and bending
moment at different building elevations of the 50-story building
at U H = 46.6 m/s calculated from the time history analysis.
The RMS and peak values of the acrosswind and torsional
responses follow similar variations along the building elevation
as the alongwind response. The peak factor of the response
ranges between 3.5 and 4.0. The analysis in the frequency
domain is also carried out which leads to agreeable RMS
response with an error of less than 10%.
Fig. 7 shows the RMS values of the alongwind background,resonant and total dynamic top displacement of both buildings
at varying wind speeds. The building response increases with
wind speed approximately following a power law. The power
law exponent of the background response is 2.0. The resonant
responses of the 50-story building in alongwind, acrosswind
and torsional directions have an exponent of about 2.9, 3.5 and
2.2, respectively. Consequently, the total dynamic responses in
three directions correspond to an exponent of about 2.4, 3.0, and
2.1, respectively. The top accelerations in three directions have
a power law exponent of 2.5–3.4. The acrosswind acceleration
is significantly greater than those in the other two directions.
The power law exponent of resonant response is greater than
2.0, because in addition to the wind load proportional to the
wind speed square, the decrease in the modal reduced frequency
as wind speed increases introduces extra wind load. As the
resonant response increases with wind speed at a faster rate
than the background response, the ratio of the background
to resonant response decreases with increasing wind speed. It
should be noted that the level of modal damping ratio strongly
influences the resonant response and thus the ratio of the
background to resonant component.
The variations of the GRFs associated with different
alongwind responses of both buildings obtained from the time
domain analysis are shown in Fig. 8. The GRFs for the top
displacement, base shear force and base bending moment arealmost same and about 2.0. The GRFs for the shear forces and
bending moments at higher floor levels are remarkably larger.
Moreover, GRFs for the shear force and bending moment at
the same level are almost identical. As pointed in Holmes [12]
and Chen and Kareem [4], the alongwind building response at
higher floor levels will be significantly underestimated using
the ESWL given as the mean wind load multiplied by the
GRF associated with the top displacement or base bending
moment as suggested in most current wind loading codes and
specifications. The variations of GRFs over building height
follow similar features at different wind speeds. As the dynamic
response grows with wind speed at a faster rate than the static
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 6/13
2646 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
Fig. 6. Alongwind displacement, shear force and bending moment (50-story building, β = 1.25,U H = 46.6 m/s).
(a) 20-story building. (b) 50-story building.
Fig. 7. Alongwind top displacement at varying wind speeds (β = 1.25).
(a) 20-story building
(U H = 40 m/s).
(b) 50-story building
(U H = 46.6 m/s).
Fig. 8. GRFs for different alongwind responses (β = 1.25).
response, the GRF slightly increases with increasing wind
speed as shown in Fig. 9.
4.2. Higher modal contributions
Tables 1 and 2 show the influence of higher mode
truncation on the base and top responses of the 50-story
building at U H = 46.6 m/s. Figs. 10 and 11 show the
influence of higher mode truncation on the 50-story building
response expressed as the ratiosσ 2 R1b
+ σ 2 R1r /
σ 2 Rb
+ σ 2 Rr
and
σ 2 Rb
+ σ 2 R1r /
σ 2 Rb
+ σ 2 Rr . These ratios for the 20-story
building are shown in Figs. 12 and 13. The difference between
Figs. 10 and 11, and between Figs. 12 and 13, is in the
estimation of the background response. Figs. 10 and 12 show
the influence of truncating higher mode contributions to both
background and resonant components, while Figs. 11 and 13
are those by truncating higher mode contributions only to the
resonant component.As the background modal response may have noticeable
negative correlation, considering only the fundamental mode
may not necessarily underestimate background response. When
the frequency domain modal analysis procedure is followed,
the CQC scheme should be used for estimating the background
response especially for lower buildings where the background
component is more dominant. In such cases, the analysis of
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 7/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2647
(a) 20-story building. (b) 50-story building.
Fig. 9. GRFs at varying wind speeds (β = 1.25).
Table 1
Influence of higher mode truncation on translational response (50-story building, β = 1.25,U H = 46.6 m/s)
Top displacement Shear force Bending moment Top acceleration
Top Base Top Base
Alongwind
σ R1b/σ Rb
1.09 2.47 0.76 2.47 0.96 –
σ R1r /σ Rr 1.00 0.81 0.99 0.81 1.00 0.81
σ Rb/σ Rr 0.73 0.26 1.04 0.26 0.83 –
σ 2 R1b
+ σ 2 R1r
/σ 2
Rb+ σ 2
Rr 1.03 1.00 0.87 1.00 0.99 0.81
σ 2 Rb
+ σ 2 R1r
/σ 2
Rb+ σ 2
Rr 1.00 0.82 0.99 0.82 1.00 0.81
Acrosswind
σ R1b/σ Rb
1.16 3.94 0.72 3.94 0.96 –
σ R1r /σ Rr 1.00 0.93 1.00 0.93 1.00 0.92
σ Rb/σ Rr 0.44 0.12 0.70 0.12 0.50 –σ 2
R1b+ σ 2
R1r /σ 2
Rb+ σ 2
Rr 1.03 1.03 0.92 1.03 0.99 0.92
σ 2 Rb
+ σ 2 R1r
/σ 2
Rb+ σ 2
Rr 1.00 0.93 1.00 0.93 1.00 0.92
Table 2
Influence of higher mode truncation on torsional response (50-story building, β = 1.25,U H = 46.6 m/s)
Top rotation Torque Top angular acceleration
Top Base
σ R1b/σ Rb
1.15 1.99 0.72 –
σ R1r /σ Rr 0.99 0.64 0.96 0.64
σ Rb/σ Rr 0.39 0.15 0.61 –
σ 2 R1b
+ σ 2 R1r
/σ 2
Rb+ σ 2
Rr 1.01 0.69 0.91 0.64
σ 2 Rb
+ σ 2 R1r
/σ 2
Rb+ σ 2
Rr 0.99 0.65 0.97 0.64
the background response directly using the influence function
is computationally more efficient and thus recommended. For
very tall buildings, the background response becomes less
important and thus can be approximately estimated by the
modal analysis including only the fundamental mode.
In the case of resonant response, as the modal frequencies
of both buildings are well separated, the correlation between
resonant modal responses is negligible. Accordingly, neglecting
the higher mode contributions always results in lower resonant
response. Table 3 summarizes the contributions of first five
modes to the alongwind resonant responses of the 50-story
building, expressed in terms of the ratio of the RMS modal
response to the total RMS response. It is observed that the
higher mode contributions have different influence on different
responses. That is attributed to their different values of modal
participation coefficients. Compared to the top displacement,
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 8/13
2648 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
(a) Alongwind. (b) Alongwind. (c) Alongwind. (d) Torsional.
Fig. 10. Influence of higher mode truncation in both background and resonant responses (50-story building,β = 1.25,U H = 46.6 m/s).
(a) Alongwind. (b) Alongwind. (c) Alongwind. (d) Torsional.
Fig. 11. Influence of higher mode truncation in the resonant response (50-story building, β = 1.25,U H = 46.6 m/s).
Table 3
Mode contributions to alongwind resonant response (50-story building, β = 1.25,U H = 46.6 m/s)
Mode number Top displacement Shear force Bending moment Top acceleration
Top Base Top Base
1 1.00 0.81 0.99 0.81 1.00 0.81
2 0.09 0.43 0.15 0.43 0.03 0.43
3 0.02 0.28 0.06 0.28 0.01 0.28
4 0.01 0.18 0.03 0.18 0.00 0.18
5 0.00 0.13 0.02 0.13 0.00 0.13
and base shear force and bending moment, the story forces
at upper floor levels are more influenced by the higher mode
contributions. For the top acceleration of the 50-story building,
considering only the fundamental mode leads to an error of
about 19% in alongwind direction, 8% in acrosswind direction
and 36% in torsion. These results are consistent with those
reported in the literature [19,15,20]. It should be emphasized
that the significance of higher mode contributions depends
on both structural dynamics and wind loading characteristics,
particularly on the PSDs of the generalized forces at the
modal frequencies. In the case of the 20-story building,
considering only the fundamental mode leads to the predicted
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 9/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2649
(a) Alongwind. (b) Alongwind. (c) Alongwind. (d) Torsional.
Fig. 12. Influence of higher mode truncation in both background and resonant responses (20-story building, β = 1.25,U H = 40 m/s).
(a) Alongwind. (b) Alongwind. (c) Alongwind. (d) Torsional.
Fig. 13. Influence of higher mode truncation in the resonant response (20-story building,β = 1.25,U H = 40 m/s).
top acceleration with an error of about 30%, 37% and 39% in
alongwind, acrosswind and torsional directions, respectively.
Figs. 11 and 13 also demonstrate that story forces at upper
floor levels will be significantly underestimated when only
the fundamental mode is involved in the resonant responseand the background response is estimated by the quasi-static
analysis. The influence of higher mode truncation on resonant
component is almost insensitive to the variation of wind speed
as shown in Fig. 14. When both background and resonant
response components are estimated by only considering the
fundamental mode as shown in Figs. 10 and 12, the dynamic
response may be over- or underestimated, particularly for
lower buildings where the background component has more
significant influence on the total dynamic response. It is
noted that the top displacement and base bending moment are
less sensitive to higher mode contributions. In particular, for
buildings with a linear fundamental mode shape, the higher
modes will provide no contribution to the base bending moment
due to the orthogonality between the higher mode shape and the
influence function, although they would affect other response
components.
4.3. Modeling of ESWLs
When only the fundamental mode contribution is consid-
ered, the ESWLs in three primary directions can be defined by
distributing the base bending moment or base torque follow-
ing the fundamental inertial force distribution (e.g. Chen and
Kareem [4]). This concept can also be used in the cases with
consideration of higher mode contributions by further intro-
ducing modification factors as adopted in current seismic load-
ing specifications (e.g., Chopra [7]). However, considering the
distinct influence of higher modes on the background and reso-
nant responses, it is more convenient and physically meaningful
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 10/13
2650 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
(a) 20-story buildings. (b) 50-story buildings.
Fig. 14. Influence of higher mode truncation in the resonant response at varying wind speeds (β = 1.25).
to model the background and resonant ESWLs (BESWL and
RESWL) separately, especially when the background response
is noticeable.
The BESWL can be modeled using the load response
correlation (LRC) approach [17], which leads to a different
spatial distribution of ESWL for different response. In thisstudy, the gust loading envelope (GLE) approach [4] is adopted.
According to this approach, the BESWL is given as the
gust loading envelope scaled by a background factor. The
background factor for a given response is defined as the
ratio of the background response to the response under the
gust loading envelope and represents a reduction effect of
partially correlated wind loads on the background response.
The background factors for the displacements at different floor
levels of both buildings are almost constant as 0.83. The shear
force and bending moment at the same floor level have almost
same background factor, which varies over the building height
following an almost linear function from 0.82 at the baseto 0.97 at the top for both buildings. The background factor
is insensitive to individual response, which demonstrates the
advantage of the GLE approach for modeling BESWL.
The RESWL is given as the fundamental mode inertial
load by distributing the resonant base bending moment or
base torque with additional amplification factors to reflect
the higher mode contributions to story forces at upper floor
levels. The higher mode contributions can also be modeled
more conveniently by introducing a concentrated force at the
building top as adopted in current seismic loading specifications
(e.g., Chopra [7]). The background and resonant responses are
then quantified separately and combined by using the square-
root-of-sum-of-squares (SRSS) rule for the total dynamic
response.
4.4. Comparison with ASCE 7-05
The alongwind responses of the 20- and 50-story buildings
based on the wind loads specified in ASCE 7-05 are computed
for the purpose of comparison. The terrain exposure category C,
i.e., open country terrain, is chosen according to the conditions
of the wind tunnel experiments. The wind directionality factor,
importance factor and topographic factor defined in ASCE 7-
05 are taken as unity. The wall pressure coefficients are 0.8
on the windward wall and -0.5 on the leeward wall. Like
(a) 20-story building. (b) 50-story building.
Fig. 15. Comparison of the mean responses between ASCE 7-05 and measured
pressure (β = 1.25).
the Australian/New Zealand Standard, ASCE 7-05 uses a 3 s
gust envelope distribution multiplied by a dynamic response
factor (referred to as the gust effect factor in ASCE 7-05) todefine the alongwind ESWL. Generally, 3 s gust envelope is
different from the mean load distribution. In order to make
a meaningful comparison, the gust effect factor defined in
ASCE7-05 but excluding the reduction factor 0.925 and thegust factor 1 + 1.7gv I ̄ z (where gv is the peak factor and I ̄ z is
the intensity of turbulence at 60% building height) is regarded
as the traditional GRF [21], and compared to that from the
pressure measurements. Consequently, the ESWL specified inASCE 7-05 divided by the gust effect factor that excludes the
factors 0.925 and 1 + 1.7gv I ̄ z can be regarded as an equivalent
mean load profile. Regardless of this treatment, the ESWL
and resulting response by ASCE 7-05 remain unchanged.Fig. 3 shows the equivalent mean force coefficient profiles of
both buildings by ASCE 7-05 as compared to the measured
pressure data. For the 20- and 50-story buildings, 1 + 1.7gv I ̄ zare taken as 1.89 and 1.76, respectively. The comparison of the resulting mean responses is shown in Fig. 15, which is
expressed as the ratio of the response by ASCE 7-05 to that
by measured pressures. It is seen that ASCE 7-05 provides
slightly conservative mean load and response for the 20-story
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 11/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2651
(a) 20-story building (U H = 40 m/s). (b) 50-story building (U H = 46.6 m/s).
Fig. 16. Comparison of the peak dynamic and total responses between ASCE 7-05 and measured pressure (β = 1.25).
building, but almost identical ones for the 50-story building.
Fig. 9 shows the comparison of the GRFs. It is noted that
the GRF defined in ASCE 7-05, which is associated with theGRF of the top displacement, is close to that derived from the
pressure measurements.Fig. 16 compares the peak dynamic and total (including
mean) responses, which are expressed as the ratio of the
response by ASCE 7-05 to that through pressure measurements.
The peak dynamic response based on pressure measurementsis given as the RMS response multiplied by the peak factor.
The peak factors are chosen as 3.8 and 3.5 for the 20- and
50-story buildings, respectively. It is seen that ASCE 7-05
underestimates the shear forces and bending moments at upperfloor levels, which is primarily attributed to the fact that
the single GRF associated with building top displacement is
adopted in the ESWL and used for the quantification of allresponses. As discussed earlier, the actual GRFs for building
story forces at upper floor levels are much higher than that
for the top displacement. The differences in story forces at thelower floor levels are primarily attributed to the differences in
the mean loads.The estimation of the building response at upper floor levels
by ASCE 7-05 can be improved through modifying the ESWL
distribution associated with the peak dynamic response. Insteadof following the load distribution similar to mean load, it
can be defined to follow the fundamental modal inertial force
through distributing the base bending moment. Clearly, the
modified ESWL results in the same base bending momentas the original ESWL defined in ASCE 7-05, but different
estimation of other responses. Fig. 17 shows the ratio of the
peak dynamic and total response under the modified ESWL tothat with measured pressures. It is observed that the modified
EWSL leads to improved estimation of the response. These
results are consistent with those shown in Figs. 10 and 12. That
the response ratios are less than 1.0 for the 50-story buildingconcerning the total response is attributed to the fact that the
based bending moment determined by ASCE 7-05 is less than
that from the pressure measurements.
4.5. Mode shape correction factors
The modal shape correction factor is defined as η x, y( f ) = H 2 SQ x , y ( f )/S M x, y ( f ) for the translational mode, and ηθ =SQθ ( f )/ST ( f ) for the torsional mode, where S M x, y ( f ) and
ST ( f ) are the PSDs of the external loading in terms of the base
bending moment and base torque, and SQ x , y ( f ) and SQθ ( f )
are the PSDs of the generalized forces. The empirical formulas
suggested by Holmes [11] and Holmes et al. [14] are chosen as
an example in this study, which are η x, y = √ 4/(1 + 3β) and
ηθ = √ 1/(1 + 2β) where β is the power law exponent of the
fundamental mode shape.Fig. 18 compares the mode shape correction factors
determined from the measured pressures with those based on
Holmes’ empirical formulas over the reduced frequency range
f B/U H = 0.1–6 in the cases of both 20- and 50-story
buildings. The fundamental mode shape is assumed to follow
a power law with an exponent β = 1.0, 1.25 and 1.5. It can
be observed that the empirical formulas match well with the
correction factors based on the measured pressure data. The
factor 0.7 which is widely used in practice for the mode shape
correlation associated with the torsional mode is conservative.
5. Concluding remarks
The wind load effects of 20- and 50-story buildings in
three primary directions were analyzed using detailed dynamic
pressure data measured in a wind tunnel. The results of this
study reconfirmed some of the findings of previous studies
using simplified loading models, and presented some new
results that helped to better understand and quantify wind-
induced response of tall buildings.
This study highlighted the variation of the GRFs associated
with different responses. The GRFs for the alongwind top
displacement, base shear force and base bending moment are
close to each other. However, use of a single ESWL as the mean
wind load multiplied by the GRF associated with the building
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 12/13
2652 G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653
(a) 20-story building (U H = 40 m/s). (b) 50-story building (U H = 46.6 m/s).
Fig. 17. Comparison of the peak dynamic and total responses between modified ESWL and measured pressure (β = 1.25).
(a) 20-story building. (b) 50-story building.
Fig. 18. Comparison of the mode shape correction factors.
top displacement or base bending moment led to noticeable
underestimates of the story forces at upper floor levels.
Advanced modeling of alongwind ESWL can be achieved by
using the GRF, which varies along building elevation, or by
using physically more meaningful load distributions.
The conventional analysis practice involving only the
fundamental mode of vibration offered sufficiently accurate
predictions of the building top displacement, top rotation
and base bending moment. However, neglecting the highermode contributions to both background and resonant responses
may noticeably under- and overestimate other responses.
The analysis involving only the fundamental mode did not
necessarily lead to lower response due to negative correlation
between the background modal responses. It was recommended
to use quasi-static analysis in terms of the response influence
function for accurately estimating the background response,
which is equivalent to including all mode contributions. When
only the higher mode contributions to the resonant response
were neglected, building story forces at higher floor levels
were markedly underestimated. The acrosswind acceleration of
taller buildings was shown to be insensitive to higher mode
contributions as compared to the accelerations in the other two
directions and those of lower buildings. As the significance
of higher mode contributions depends on building dynamics
and wind loading characteristics, the perception that higher
mode contributions to building acceleration in any direction of
vibration and under any wind condition are negligible may not
be taken as granted.
Modeling the background and resonant ESWLs separately
was more convenient and physically meaningful especiallywhen the background response is significant. The results
demonstrated the advantage of the gust loading envelope
approach for modeling the background loading. The resonant
loading can be modeled as the fundamental modal inertial
load with additional amplification factors and/or a concentrated
force at the building top to take into account the contributions of
higher modes. The background and resonant responses are then
quantified separately and combined by using the SRSS rule for
the total dynamic response.
The adequacy of the alongwind load specified in ASCE 7-05
was also examined. As the ESWL in ASCE 7-05 was essentially
defined as the mean load multiplied by the GRF associated with
8/8/2019 2007_ES_TallBuilding
http://slidepdf.com/reader/full/2007estallbuilding 13/13
G. Huang, X. Chen / Engineering Structures 29 (2007) 2641–2653 2653
the top displacement, building response at upper floor levels
was markedly underestimated. The modified ESWL following
the fundamental inertial load over the building height can
significantly improve the accuracy of response prediction.
Finally, using spatiotemporally varying wind loads, the
adequacy of the empirical mode shape correction factors used in
the current force balance technique was confirmed for both thetranslational and torsional modes, while the factor 0.7 adopted
for torsional mode was conservative.
Acknowledgements
The support for this work provided in part by the new
faculty startup funds of the Texas Tech University is gratefully
acknowledged. The writers are also grateful to Professor Yokio
Tamura of Tokyo Polytechnic University in Japan for providing
the valuable wind pressure data used in this study.
References
[1] American Society of Civil Engineers (ASCE). Minimum design loads for
building and other structures. ASCE 7-05, New York; 2006.
[2] AS/NZS 1170.2: 2002. Standards Australia/Standards New Zealand,
Structural design actions, Part 2: Wind actions; 2002.
[3] Boggs DW, Peterka JA. Aerodynamic model tests of tall buildings. J Eng
Mech 1989;115(3):618–35.
[4] Chen X, Kareem A. Equivalent static wind loads on tall buildings: New
model. J Struct Eng 2004;130(10):1425–35.
[5] Chen X, Kareem A. Coupled dynamic analysis and equivalent static wind
loads on building with 3-D modes. J Struct Eng 2005;131(7):1071–82.
[6] Chen X, Kareem A. Dynamic wind effects on buildings with three-
dimensional coupled modes: Application of high frequency force balance
measurements. J Eng Mech 2005;131(11):1115–25.
[7] Chopra AK. Dynamics of structures: Theory and applications to
earthquake engineering. 2nd ed. Prentice-Hall; 2000.[8] Davenport AG. Gust loading factors. J Struct Div ASCE 1967;93(1):
11–34.
[9] Der Kiureghian A. Structural response to stationary excitation. J Eng
Mech 1980;106(6):1195–213.
[10] Ho TCE, Lythe GR, Isyumov N. Structural loads and responses from
the integration of instantaneous pressures. In: Larsen, Larose, Livesey,
editors. Wind engineering into the 21st century. Rotterdam: Balkema;
1999. p. 1505–10.
[11] Holmes JD. Mode shape corrections for dynamic response to wind. Eng
Struct 1987;9:210–2.
[12] Holmes JD. Along-wind response of lattice towers: Part I – Derivation of
expressions for gust response factors. Eng Struct 1994;16:287–92.
[13] Holmes JD. Effective static load distributions in wind engineering. J Wind
Eng Ind Aerodyn 2002;90:91–109.
[14] Holmes JD, Rofail A, Aurelius L. High frequency base balance
methodologies for tall buildings with torsional and coupled resonant
modes. In: Proc., 11th int conf on wind eng. Lubbock (TX); 2003.
p. 2381–7.
[15] Kareem A. Wind-excited response of buildings in higher modes. J Struct
Div ASCE 1981;107(4):701–6.
[16] Kareem A, Zhou Y. Gust loading factor: Past, present, and future. J Wind
Eng Ind Aerodyn 2003;91(12–15):1301–28.
[17] Kasperski M. Extreme wind load distributions for linear and nonlinear
design. Eng Struct 1992;14:27–34.
[18] Repetto MP, Solari G. Equivalent static wind actions on vertical
structures. J Wind Eng Ind Aerodyn 2004;92(5):335–57.
[19] Simiu E. Equivalent static wind loads for tall building design. J Struct Div
ASCE 1976;102(4):719–37.
[20] Simiu E, Scanlan RH. Wind effects on structures: Fundamentals and
applications to design. 3rd ed. New York: John Wiley & Sons, Inc.; 1996.
[21] Solari G, Kareem A. On the formulation of ASCE 7-95 gust effect factor.
J Wind Eng Ind Aerodyn 1998;77–78:673–84.
[22] Tamura Y, Suganuma S, Kikuchi H, Hibi K. Proper orthogonal
decomposition of random wind pressure field. J Fluids Struct 1999;13:
1069–95.
[23] Vickery PJ, Steckley A, Isyumov N, Vickery BJ. The effect of mode shape
on the wind-induced response of tall buildings. In: Proc., 5th U.S. national
conf on wind eng, Lubbock (TX): Texas Tech. Univ.; 1985. p. 1B-41-1B-
48.
[24] Xie JX, Irwin PA. Application of the force balance technique to a building
complex. J Wind Eng Ind Aerodyn 1998;77–78:579–90.
[25] Xu YL, Kwok KCS. Mode shape corrections for wind tunnel tests of tallbuildings. Eng Struct 1993;15(5):387–92.
[26] Yip DYN, Flay RGJ. A new force balance data analysis method for wind
response predictions of tall buildings. J Wind Eng Ind Aerodyn 1995;
54–55:115–23.