Engineering Structures 51 (2013) 23–35

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Comparative study of major international wind codes and standards for

wind effects on tall buildings
Dae Kun Kwon ⇑, Ahsan Kareem
NatHaz Modeling Laboratory, 156 Fitzpatrick Hall, Department of Civil & Environmental Engineering & Earth Sciences, University of Notre Dame, Notre Dame, IN 46556, USA

a r t i c l e i n f o a b s t r a c t

Article history: Globalization of construction industry, burgeoning growth of tall buildings, and the recent focus on the
Received 17 October 2012 development of unified international codes/standards has increased the need to better understand the
Revised 6 December 2012 underlying commonalities and differences among the major international wind loading codes/standards,
Accepted 10 January 2013
which are also constantly being revised and updated. To address this need, a comprehensive comparison
of wind loads and their effects on tall buildings is conducted utilizing eight major international codes/
standards: ASCE 2010 (USA), AS/NZ 2011 (Australia and New Zealand), AIJ 2004 (Japan), CNS 2012
Keywords:
(China), NBCC 2010 (Canada), Eurocode 2010 (Europe), ISO 2009 and IWC 2012 (India). The key areas
Wind code and standard
Tall building
of comparison include the provisions for survivability design as well as the serviceability requirements
Wind loading in the alongwind and acrosswind directions. As most codes/standards utilize a common theoretical
Building response framework for modeling dynamic load effects, basic equations here are recast in a general format in order
Buffeting to compare the influence of individual parameters on the overall recommendations of codes/standards.
Alongwind Ó 2013 Elsevier Ltd. All rights reserved.
Acrosswind
Torsion

1. Introduction Indian Wind Code (IWC) [10,11]. Although all these codes/stan-
dards use a random-vibration-based gust loading factor approach
Globalization of construction industry, burgeoning growth of [12] for assessing the dynamic alongwind loads and their effects
tall buildings around the world and the development of unified on tall structures, different parameters are defined in each code/
international codes/standards, i.e. ISO [1], has increased the need standard, in addition, different provisions for the acrosswind and
to better understand the underlying commonalities and differences torsional loads are adopted [2]. The use of discordant terminology
among the features of major international wind codes/standards and different choices of expressing various factors make it hard for
related to tall buildings. Previous studies on the comparison of ear- designers to work in a global environment, such as a design firm in
lier versions of codes/standards have found that the varying defini- the United States designing a tall building in the Pacific-Rim coun-
tions of wind field characteristics, including mean wind velocity tries. For this reason, the basic equations are rewritten in a general
profile, turbulence intensity profile, wind spectrum, turbulence form in this study where the parameters involved are clearly iden-
length scale and wind correlation structure, are the primary con- tified. This will help designers decipher the nuances associated
tributors to the scatter in predicted response quantities [2,3]. As with different codes/standards and understand the resulting differ-
nearly every major building codes/standards has been updated in ences in the load effects.
some aspects in the last few years, there is a need to revisit and In this study, the key areas of comparison include the provisions
compare the updated versions of these codes/standards. in the alongwind and acrosswind directions for survivability design
To this aim, a comprehensive comparison of the wind loads and as well as serviceability requirements. The comparisons are carried
their effects on tall buildings is conducted utilizing eight major out for three example cases and differences in the predictions are
international codes/standards in this study: the American Society discussed and suggestions are made to further improve agreement
of Civil Engineers’ Minimum Design Loads for Buildings and Other among the codes/standards towards the goal of developing a global
Structures (ASCE) [4], the Australian and New Zealand Standard standard.
(AS/NZ) [5], the Architectural Institute of Japan Recommendations
(AIJ) [6], the China National Standard (CNS) [7], the National Build-
ing Code of Canada (NBCC) [8], the European Standard (EU) [9], 2. Wind characteristics and loads in codes and standards
International Organization for Standardization (ISO) [1] and the
In this section, all the relevant features needed for the assess-
⇑ Corresponding author. Tel.: +1 574 631 5380; fax: +1 574 631 9236. ment of wind loads according to the different codes/standards
E-mail addresses: [email protected] (D.K. Kwon), [email protected] (A. Kareem). are closely examined and compared to the scope of dynamically

0141-0296/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2013.01.008
24 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

 
sensitive regular-shaped buildings with flat roofs that are classified u z
VðzÞ ¼ ln V0 ð2Þ
as enclosed buildings. Note that all codes/standards recommend 0:4 z0
that extremely tall and irregular shaped structures be designed
using wind tunnel experiments. Nonetheless, even in cases where where u is the friction velocity; z0 is the roughness length. The vari-
a wind tunnel study is conducted, codes/standards are used for ables used in the wind profiles in Eqs. (1) and (2) (a, b, u and z0)
preliminary design and wind related input as basic wind speed define the exposure factor (Cexposure) given in Eq. (4) and are depen-
and its profile, terrain condition, turbulence intensity/scale, etc. dent on the exposure category defined in each standard. Each stan-
to determine wind tunnel test conditions. It is important that cau- dard uses between three and five exposure categories, however, for
tion be exercised as wind information based on a code/standard the sake of comparison, this study classifies them as six general
may often be different from the condition at the site. This may be- exposure categories (EC) shown in Table 2 [16]. Please note that
come a source of epistemic or model/load uncertainty to assess the basic wind velocity is always defined at 10 m height in the open
wind loading effects on a structures. In this regard, most codes/ terrain, which corresponds to EC4 in the definition used here for
standards do offer some provisions in their guidelines on wind tun- comparison. Based on general forms in Eqs. (1) and (2), the values
nel procedure regarding regional climate, complex terrain, etc. of variables a and b for ASCE, AIJ, CNS, NBCC and IWC are reported
in Table 3 and the friction velocity (u) and roughness length (z0) in
Eq. (2) for AS/NZ, EU and ISO in Table 4.
2.1. Averaging time
To compare the profiles, it is important to account for the differ-
ences in averaging time. Thus, the velocity profiles are compared
Averaging times for basic wind velocity (V0) and wind-induced
based on the averaging times used in the calculation of wind-in-
response vary among the codes/standards as well as within the
duced response in each code/standard, i.e., 1-h (ASCE, NBCC and
codes/standards themselves. Averaging times for basic wind veloc-
IWC) or 10-min (others) as described in Table 1. As shown in
ity (V0) and wind-induced response vary among codes/standards as
Fig. 1, it is observed that the codes/standards have similar profiles
well as within a code/standard. In particular, ASCE, AS/NZ, ISO and
in the EC2 (urban), EC3 (suburban) and EC4 (open terrain) with the
IWC standards define the basic wind speed as 3-s gust speed
exceptions of ISO in EC2 and AS/NZ and IWC for all three ECs that
whereas other standards define it as the mean wind speed (10-
show somewhat different profiles from others. Please note that it is
min mean in AIJ, CNS, EU or 1-h mean in NBCC). However, in the
still a subject of debate in the wind engineering community as to
calculation of wind-induced response, all codes/standards utilize
the definition of an urban area in terms of terrain condition and
a longer averaging period such as 10 min or 1-h (Table 1). Recently,
at this time each code/standard offers its own definition. ASCE 7
there is conversation regarding revisiting the method of defining
has gone to the extreme of even removing exposure A condition
the average interval which particularly concerns shorter duration
since ASCE 7-02 version due to the wide variability of the wind
peak gust wind speed, e.g., 3-s gust. In such cases, there are several
caused by local channeling and wake buffeting effects in such an
possible ways to define averaged wind with varying results [13–
environment. The discrepancies in the profiles of AS/NZ, ISO and
15]. In addition, the reference height (href) at which the gust load-
IWC in EC2 shown in Fig. 1 have lead authors to combining them
ing/effect factor and other parameters are calculated is different
into EC1 category instead (Table 2).
among the codes/standards, as summarized in Table 1. These dif-
ferences in averaging time and reference heights affect the inter-
mediate parameters in the calculations of gust loading/effect 2.3. Alongwind loads
factor and the resulting load effects, making a simple comparison
between the codes/standards challenging. Throughout this analy- The general expression for wind pressures (pz) on a building for
sis, the effect of differing averaging times in the definition of the all the codes/standards is:
basic wind speed among codes/standards has been minimized as pz ¼ qz GC p ð3Þ
much as possible. In this regard, a conversion factor is utilized to
compensate among different averaging times in the basic wind where qz is the velocity pressure at height z; G is the gust effect fac-
speeds, which can be found in Example 1 later. tor; CP is the pressure coefficient. The loads are then determined by
combining the wind pressures acting on a building surface and the
2.2. Wind velocity profile corresponding tributary areas. Moments are determined by multi-
plying the load at a given height by the corresponding height. Base
The wind velocity profile in each standard is described by either shear forces and moments are then determined by the summation
a power or a logarithmic law. AS/NZ, EU and ISO use the logarith- of the loads and moments at each level.
mic law, whereas the others use a power law. ISO also provides The velocity pressure (qz) can be generalized as:
power law fits to the profiles primarily for use in resonant response 1
calculations. The power law is described as: qz ¼ qV 20  C exposure  C topography  C direction  C importance  C other ð4Þ
2
 z a
VðzÞ ¼ b V0 ð1Þ where q is the air density; Cexposure is the velocity profile or exposure
10 factor; Ctopography is the topography factor; Cdirection is the wind direc-
where a and b are terrain variables, z is the height of interest, and V0 tionality factor; Cimportance is the building importance factor; Cother
is the basic wind velocity. The logarithmic law is described as: is a factor accounting for other things such as hurricane zone,

Table 1
Averaging times and reference heights.

ASCE AS/NZ AIJ CNS NBCC EU ISOa IWC

Basic wind velocity (V0) 3-s 3-s 10-min 10-min 1-h 10-min 3-s 10-min 3-s
Wind-induced response 1-h 10-min 10-min 10-min 1-h 10-min 10-min 1-h
Reference height (href) 0.6hb h h h h 0.6h h h
a
ISO provides two procedures for determining loads: one for peak response and one for mean response.
b
h = building height.
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 25

Table 2
Comparison of exposure categories.

General exposure categoryb Exposure categories defined in codes/standards

ASCE AS/NZ AIJ CNS NBCC EU ISO IWC
EC1 – 4 V – – – 4 4
EC2 Aa – IV D C IV – –
EC3 B 3 III C B III 3 3
EC4 C 2 II B A II 2 2
EC5 D 1 I A – I 1 1
EC6 – – – – – 0 – –
a
Although ASCE A is no longer defined since ASCE 7-02, it is used in this analysis for comparison purposes.
b
EC1 = city center with heavy concentration of tall buildings; EC2 = general urban area; EC3 = suburban area; EC4 = open terrain area where basic wind velocity is defined;
EC5 = open lake or land with minimal obstruction; EC6 = sea or coastal area.

Table 3
Power law coefficients for ASCE, AIJ, CNS, NBCC and IWC.

Category ASCE AIJ CNS NBCC IWC

3-s 1-h 10-min 10-min 1-h 3-s
a b a b a b a b a b a B
EC1 – – – – 0.35 0.394 – – – – 0.0683a 1.030a
EC2 0.20 0.64 0.33 0.30 0.27 0.576 0.30 0.52 0.36 0.426 – –
EC3 0.14 0.84 0.25 0.45 0.20 0.794 0.22 0.73 0.25 0.666 0.0850 0.980
EC4 0.11 1.00 0.15 0.65 0.15 1.000 0.15 1.00 0.14 1.000 0.0765 1.033
EC5 0.09 1.07 0.11 0.80 0.10 1.230 0.12 1.13 – – 0.0690 1.072
a
These coefficients are valid above 100 m height.

Table 4
Friction velocity and roughness length for profiles in AS/NZ, EU and ISO. where GLF is the gust loading factor; Gq is the gust factor for the
a a
wind velocity pressure [17,18]. The term gust effect factor (G) is
Category AS/NZ EU ISO
used in ASCE 7, while it is defined differently in international
u z0 u z0 u z0 codes/standards. For reference, AS/NZ, ISO and IWC describe it as
EC1 0.086–0.102 2 – – 0.1–0.109 3 dynamic response factor (Cdyn) and the EU as structural factor (CsCd).
EC2 – – 0.094 1 – – Note that the definition of G in ASCE has an additional multiplica-
EC3 0.075–0.083 0.2 0.086 0.3 0.081–0.089 0.3
tive factor of 0.925 introduced to adjust the prediction to match
EC4 0.064–0.070 0.02 0.076 0.05 0.069–0.074 0.03
EC5 0.055–0.061 0.002 0.068 0.01 0.059–0.064 0.003 previous versions of the standard. GLF, originally defined by [12],
EC6 – – 0.062 0.003 – – is generally expressed as:
a
Friction velocity for AS/NZ and ISO standards are estimated from fit of tabulated
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
values. GLF ¼ 1 þ r g 2B B þ g 2R R ð6Þ

shielding or mean recurrence interval (return period). Note that the where gB and gR are the peak factors for background and resonant
effects of topography, directionality, building importance and other responses, respectively; r is the turbulence intensity related vari-
factors are not considered in this study. able; B and R are the background and resonant factors, respectively.
The gust effect factor (G) for the codes/standards may be writ- Note that GLF in CNS [7] varies along the building height because it
ten in a general format as: is related to the inertial force.
The peak factor describes the peak of fluctuating response and is
GLF
G¼ ð5Þ defined as the ratio of the maximum of a quantity to the standard
Gq
deviation. Typically, the distribution is assumed to be Gaussian,

Fig. 1. Mean wind velocity profiles in EC2, 3, and 4: EC1 in the AS/NZ, ISO and IWC are used in EC2 plot for comparison purpose.
26 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

Table 5 values than those by other codes/standards. The rationale of this

Comparison of peak factors and parameters. choice is not clear.
gR gB gv Gq T (s) m The gust factor for the wind velocity pressure in Eq. (5), which is
ASCE
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3.4 3.4 1 + gvr 3600 f0
intended to compensate the difference in averaging times, e.g., be-
2 lnðmTÞ þ p0:577 ffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 lnðmTÞ tween basic wind velocity and wind-induced response calculation
AS/NZ 2 lnðmTÞ 3.7 3.7 1 + gvr 600 f0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (Table 1) or wind velocity pressure (qz), may be defined as:
AIJ 2 lnðmTÞ þ 1:2 gR – 1 600 f0C
CNS 2.5 2.5 – 1 600 f0 Gq ¼ 2Gv  1 ¼ 1 þ r  g v ð8Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NBCC 2 lnðmTÞ þ p0:577 ffiffiffiffiffiffiffiffiffiffiffiffi gR – 1 3600 f0C
2 lnðmTÞ
EU
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:6 gR 3.5 1 + gv r 600 f0C where Gv is the gust factor for wind velocity defined by V  s =V
 T ; s is
2 lnðmTÞ þ pffiffiffiffiffiffiffiffiffiffiffiffi
2 lnðmTÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the averaging time for gust velocity (e.g., 3-s); T is the averaging
ISO 2 lnðmTÞ þ p0:577 ffiffiffiffiffiffiffiffiffiffiffiffi 3.4 3.4 1 + gvr (peak) 600 f0
2 lnðmTÞ time for wind-induced response calculation (e.g., 10-min or 1-h);
1 (mean) gv is the peak factor for wind velocity. In AIJ, CNS, NBCC and the
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
IWC 2 lnðmTÞ 3.5 3.5 1 + gvr 3600 f0
mean method in ISO, the gust factor Gq is taken as unity because
f0 = natural frequency of a building [Hz]; C = a factor which is a function of the the mean wind velocity is used instead of the gust velocity. A sum-
background and resonant responses. mary of the peak and gust factors for the various codes/standards is
provided in Table 5.
The parameter r in Eq. (6) is related to the turbulence intensity
resulting in peak values between 3 and 4 (e.g., [3]). General expres- of wind in terms of the turbulence intensity profile and a multi-
sion for the peak factor is: plier. The general form of r is:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:5772 r ¼ 2Iz ð9Þ
g¼ 2 lnðmTÞ þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7Þ
2 lnðmTÞ
where Iz is the turbulence intensity profile. Note that ASCE uses a
where T is the averaging time and m is the up-crossing rate that is factor of 1.7 instead of 2, and AIJ uses a factor ranging from 2.13
approximately equal to a natural frequency (Hz) of a building to 2.418 based on the exposure category. ASCE, AIJ, AS/NZ, CNS,
assuming the process is narrow-banded Gaussian. All the codes/ NBCC and IWC have adopted a power law form for the turbulence
standards utilize Eq. (7) in different forms for the resonant peak fac- intensity profile, which can be expressed as:
tor (gR). As for the background peak factor (gB), some codes/stan-  d
10
dards (ASCE, AS/NZ, ISO and IWC) use a constant value, while Iz ¼ c ð10Þ
z
others (AIJ, NBCC and EU) use gR, however, they introduces a factor,
C, in the upcrossing rate (m) to account for the effects of background where c and d are variables related to the exposure category and z is
or resonant components (Table 5). Note that CNS utilizes a constant the desired height, while EU has adopted a logarithmic form defined
value of 2.5 for gB and gR, which would generally yield much lower as:

Table 6
Turbulence intensity profile parameters.

EC ASCE AS/NZa AIJ CNS NBCC EU ISO IWCb

c d c d c d c d c d z0 z0 c d
1 – – 0.45 0.30 0.53 0.40 – – – – – 3.000 0.47 0.30
2 0.45 0.167 – – 0.36 0.32 0.39 0.30 0.51 0.36 1.000 – – –
3 0.30 0.167 0.32 0.30 0.26 0.25 0.23 0.22 0.27 0.25 0.300 0.300 0.45 0.40
4 0.20 0.167 0.26 0.30 0.20 0.20 0.14 0.15 0.16 0.14 0.050 0.030 0.40 0.45
5 0.15 0.167 0.19 0.30 0.16 0.15 0.12 0.12 – – 0.010 0.003 0.35 0.48
6 – – – – – – – – – – 0.003 – – –
a
AS/NZ values fitted by Zhou et al. [27].
b
These coefficients are valid above 100 m height.

Fig. 2. Turbulence intensity profiles in EC2, 3, and 4: EC1 in the AS/NZ, ISO and IWC are used in EC2 plot for comparison purpose.
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 27

Table 7
Turbulence length scale parameters.

Exposure category ASCE AS/NZ AIJ EU ISO IWC

‘ (m) e ‘ (m) e ‘ (m) e ‘ (m) e ‘ (m) e ‘ (m) e

EC2 55 0.50 40 0.67

EC3 95 0.33 48 0.61
EC4 152 0.20 85 0.25 58 0.5 63 0.52 58 0.5 100 0.25
EC5 198 0.13 80 0.44
EC6 – – 96 0.38

CNS and NBCC adopt a constant length scale (CNS = 1200 m; NBCC = 1220 m).

1 Table 8
Iz ¼   ð11Þ
ln z Background response factor (B).
z0
Background response factor (B)
where z0 is the roughness length. Note that in the AS/NZ, ISO and
ASCE 1 0:63
IWC, the turbulence intensity profiles are provided in tabular forms. 1þ0:63 bþh
Lh
For the comparison of profiles, this study utilizes curve-fits made by ref

AS/NZ 1
[3] for AS/NZ and by the authors for IWC in accordance with Eq. 1þ
½ð0:26h2 Þþ0:46b2 0:5
Lh
ref
(10), while in the case of ISO it is noted that it fits a logarithmic form 
AIJ 4ð0:490:14aÞ2 0:07; h=b P 1
of Eq. (11). The parameters c and d are summarized in Table 6 to- 8  0:56 92 k¼
< p ffiffiffi = 0:15; h=b < 1
0:63 bh=Lh
gether with the roughness length parameter z0 for EU and ISO. 1þ
ref

: ðh=bÞk ;
The turbulence intensity profiles are compared in Fig. 2, where a
CNS –
significant deviation for the ‘‘urban’’ exposure category (EC2) is R 914=h h ih ih i
NBCC 4=3 1 1 x
dx
noted. This leads to variations in the resulting gust effect/loading xh
1þ457 xb
1þ122 ð1þx2 Þ4=3

factor [3]. EU B  1 0:63

bþh
The background response factor (B) for all codes and standards 1þ0:9 Lh
ref

is defined as the building dimensions with the turbulence length EU C 1

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2  2  2
scale (L) that can be written in the general format as: 1þ32 b
Lhref þ h
Lhref þ b h
Lhref Lhref
"  pffiffiffiffi 
2 
 e ISO 0:63ð bh=Lh Þ0:56 0:07; h=b P 1
h f1 þ 0:2ag 1þ ref
k¼
L¼‘ ð12Þ h k ðbÞ 0:15; h=b < 1
10 IWC 1
½36h2 þ64b2 0:5
1þ 2Lh
where h is the desired height, ‘ and e are defined in Table 7. While ref

ASCE and EU have length scale parameters that are dependent on

exposure categories, AIJ, ISO, and AS/NZ have uniform length scale
parameters. CNS and NBCC, however, have a constant length scale graphically, making it even dependent on the selections made by
at all heights. Overall, the length scale profiles are considerably dif- an individual user.
ferent among the codes/standards. The resonant response factor (R) is generally expressed as a
The definition of the background response factor (B) varies con- function of a size reduction factor (S), an energy factor (E) and a
siderably in formulation amongst the codes/standards, as shown in damping ratio of a building (f):
Table 8. In NBCC, the background response factor is provided

Table 9
Size reduction (S) and energy (E) factors.

Size reduction factor Reduced frequency Energy factor (E) Reduced frequency
ASCE 1 1 2gh 1 1 2gb 4:6f 0 E¼ 9:51N
(Kaimal) f0 Lh
S = RhRb(0.53 + 0.47Rd) Rh ¼ gh  ð1  e Þ Rb ¼ gb  ð1  e Þ g¼ ð1þ10:3NÞ5=3 N¼ ref
2ðghÞ2 2ðgbÞ2 Vh
ref Vh
ref
1 1 6:696gd
Rd ¼ 3:348gd
 ð1  e Þ
2ð3:348gdÞ2
AS/NZ 1
S ¼ ½1þ3:5gh½1þ4 f0 ð1þg v Ih Þ E¼ 4N
(Karman) f0 Lh ð1þg v Ih Þ
gb g¼ Vh
ref
ð1þ70:8N2 Þ5=6 N¼ ref
Vh
ref

ref ref
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AIJ S ¼ ð0:57  0:35a þ 2RB 0:053  0:042aÞSD g ¼ V hf0 E¼ 4N
(Karman) f0 Lh
ref
ref
ð1þ71:8N2 Þ5=6 N¼ Vh
0:9 1
SD ¼ 2 2 0:5
RB ¼ 1þ20bg
ref
½1þ6g h  ½1þ3gb
CNS – – E¼ N2
(Davenport)
f0 Lh
ref
ð1þN 2 Þ4=3 N¼ Vh
ref
h ih i
NBCC S ¼ p3 1 1 g ¼ Vh f0
E¼ N2
(Davenport)
f0 Lh
ref
1þ83gh 1þ10gb ref ð1þN 2 Þ4=3 N¼ Vh
ref

EU B S = RhRb g ¼ 4:6f 0 E¼ 6:8N

(Kaimal) f0 Lh
ref
Vh ð1þ10:2NÞ5=3 N¼ Vh
Rh ¼ g1h  1
ð1  e2gh Þ Rb ¼ g1b  1
ð1  e2gb Þ
ref
ref
2ðghÞ2 2ðgbÞ2
EU C S¼ 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi g ¼ V hf0
2 2 2
1þ ½Gy C y gb þ½Gz C z gh þ½p2 Gy C y Gz C z bhg2  ref

Gy, Gz = constants depending on mode shape

Cy = Cz = 11.5
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ISO S ¼ ð1 þ 0:6aÞ2 ð0:57  0:35a þ 2RB 0:053  0:042aÞSD g ¼ V hf0 E¼ 4N
(Karman) f0 Lh
ref
ref
ð1þ70:8N2 Þ5=6 N¼ Vh
0:9 1 ref
SD ¼ 2 2 0:5
RB ¼ 1þ20bg
½1þ6g h  ½1þ3gb
IWC S¼ 1 f0 ð1þg v Ih Þ E¼ 4N
(Karman) f0 Lh ð1þg v Ih Þ
½1þ4gh½1þ4gb g¼ Vh
ref
ð1þ70N2 Þ5=6 N¼ ref
Vh
ref

ref ref

EU provides two formulas for the size reduction factor found in Appendix B and C.
28 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

pSE Table 10
R¼ ð13Þ Alongwind acceleration.
4f
qhref GR K
Note that AS/NZ, AIJ, ISO and IWC also include a mode correc-
pffiffiffi
tion factor, which is equal to unity for a linear mode shape of vibra- ASCE 1
2 qV 2href 1:7g R Ihref R 1:65a
^

a^ þkþ1
pffiffi Rh
tion. The energy factor evaluates the normalized wind velocity AS/NZ 1
qV 2href 2g I R V 2 ðzÞzdz
2  R href  0
spectrum at the natural frequency of interest. As discussed in [2], 1þ2g v Ih
ref
V 2h h
2

the wind spectrum of the codes/standards takes one of the three pffiffiffi
AIJ 1
2 qV 2href g R Ihref R 1 - 0.4 ln k
forms: Karman (AIJ, AS/NZ, ISO and IWC), Kaimal (ASCE and EU) pffiffiffi h h i i
EU B 1
2 qV 2href 2g R Ihref R ð2kþ1Þ ðkþ1Þ lnðzh Þþ0:5 1
or Davenport (CNS and NBCC) (Table 9). The choice of the spectrum 0 
ðkþ1Þ2 ln h
impacts the value of the energy factor and the resulting resonant pffiffiffi 8
z0

response factor. The size reduction factors (S), while all incorporate EU C 1
2 qV 2href 2g R Ihref R >
>
1 uniform
< 3 linear
ky kz 2
the natural frequency, velocity, and building dimensions, vary con- umax ; ky ; kz ¼ > 5 parabolic
>
: 34
siderably in respective formulations. The size reduction factors are p sinusoidal
pffiffi Rh 2
summarized in Table 9. Details of background, size reduction factor IWC 1
2 qV 2href 2g R Ih
ref
R
0
V ðzÞzdz
ð1þ2g v Ih Þ 2
and energy factors are omitted here for the sake of brevity, but can ref V 2h h

be found in [16]. a; a^ = exposure exponents of wind velocity profile; k, ky, kz = mode shape expo-
As for the pressure coefficient (Cp) in Eq. (3), both external and nents; umax = maximum mode shape coefficient.
internal pressures are of great concerns for wind design. All codes/
standards considered in this study provide external pressure coef-
ficients as the windward and leeward directions. While the exter- fied load effects [19]. This is especially true in the cases of across-
nal pressure coefficients are reasonably consistent at building wind and torsional responses where wake-induced effects are
height, the internal pressure coefficients vary considerably among much more dominant than in the alongwind case. Thus, it is under-
the codes/standards. The external pressure coefficients can be di- standable why the results obtained through different codes/stan-
vided into global (overall) and local coefficients. While global coef- dards vary to a higher degree than in the case of the alongwind
ficient is suitable for the design of the overall building, these are direction, which will be shown in example sections later.
indeed the mean value over a surface of a building. For the design
of small elements and components, local pressure (or peak pres- 2.5. Acceleration responses
sure) may play an important role to ensure the integrity of compo-
nents such as cladding and roofs elements, joints, etc. High positive In addition to survivability design, the serviceability require-
or negative local pressures may be experienced at locations near ments in terms of acceleration responses have to be assessed to en-
the edges of walls and roofs due to flow separation, small (local) sure a level of human comfort in tall buildings. All codes/standards
areas of walls, etc. Note that some codes/standards consider the ef- except the CNS provide expressions for defining alongwind accel-
fects of local pressure, e.g., local pressure factor (Kl) in AS/NZ, peak erations, however, acrosswind and torsional accelerations are not
external pressure coefficients in AIJ, local pressure coefficients included in every standard. The alongwind acceleration can be
(Cpe,1) for loaded areas of 1 m2 or less in EU and tabular values generally expressed as:
for local coefficients in IWC.
qhref GR C fx bhK
^€xðzÞ ¼ /1 ðzÞ ð14Þ
2.4. Acrosswind and torsional loads m1

where qhref is the velocity pressure at the reference height, Cfx is the
Although the expressions for alongwind loading in terms of drag force coefficient given by the absolute sum of the windward
conventional gust loading factor approach are fairly consistent, and leeward pressure coefficients, b is the building width, h is the
the treatment of the acrosswind and torsional loading differs building height, K is the mode shape correction factor, m1 is the
amongst the codes/standards. For example, ASCE, EU, and NBCC generalized mass in the first mode, and u1(z) is the first mode shape
utilize partial loading approach to account for acrosswind and tor- evaluated at height z that is approximated with a power form, e.g.,
sional loads, although ASCE provides an alternative method in the /1 ðzÞ ¼ ðhz Þk . Definitions of qhref , GR, and K are found in Table 10. Note
Commentary (C26.9) [4], i.e., the NatHaz Aerodynamic Loads Data- that both NBCC and ISO define the alongwind acceleration in terms
base (NALD) [19,20], which is a promising database-enabled design of displacement and may be written as:
procedure based on wind tunnel-derived data (non-dimensional
base bending moment spectra) utilizing high frequency base bal- ^€xmax ¼ GR
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2pf0 Þ2 xmax ð15Þ
ance. The partial loading approach simply applies different frac- 1 þ G2B þ G2R
tions of the alongwind pressures on selected portions of building
faces in different combinations. Torsion is introduced either by pffiffiffi pffiffiffi
where GB ¼ r  g B  B, GR ¼ r  g R  R, f0 is the natural frequency,
considering asymmetric loading, as in NBCC and EU, or by an ap- and xmax is the maximum displacement. An estimate for the maxi-
plied moment defined as a combination of the alongwind load mum displacement is not given in either NBCC or ISO, thus the
multiplied by a defined eccentricity [4]. NALD, AS/NZ, AIJ, CNS alongwind acceleration cannot be readily written in the general
and ISO provide procedures for determining the acrosswind and from of Eq. (14) for NBCC or ISO, thus those are omitted in Table 10.
torsional loads as a function of the base bending moment of the Different expressions to estimate the acrosswind acceleration
resonant response and IWC also offers a procedure to estimate are given in AS/NZ, AIJ, NBCC, ISO and NALD. The acceleration
the acrosswind loads. may be generally defined as:
Most codes/standards traditionally have relied on simplified
formats, often adopting tables and plots to describe wind loads. ^
^€ðzÞ ¼ M R K y u ðzÞ
y ð16Þ
However, the level of accuracy inherent to the use of this format m0 h
2 1

for codification of the information and the uncertainty associated

with the interpolation/extrapolation of information or simplified ^ R is the resonant component of the acrosswind moment de-
where M
expressions may compromise the overall accuracy in code-speci- fined in Table 11, Ky is the mode shape correction and, m0 is the
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 29

Table 11
Acrosswind and torsional accelerations.

^R
M Ky ^T
M Kt
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
NALD h dh2 4f
gR q p r2 C ðf Þ k+2 p r2 C ðf Þ
h dbh 4f
gR q k+1
cm m 2 cm m 3
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AS/NZ h bh
gR q
2 p r 2 C ðf Þ 0.72k + 2.28 – –
4f cm m 2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
AIJ h bh2 4fp r2 C 2 2k + 1 2 p 2 2 1.2k + 0.6
4f rcm C m;T ðf3 ÞC m;T ðf1 Þ ¼ k F T
gR q cm m;R ðf2 ÞC m;R ðf1 Þ ¼ k F L
h b h
gR q
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
IWC 2 p 0.72k + 2.28 – –
4f rcm C m ðf2 Þ
h bh
gR q 2

mass per height. For torsion, the acceleration may be generally de- and the spectrum method, e.g., Gv(10-min) = 1.46 and Gv(1-h) = 1.53
fined as: in ISO [1] which uses the spectrum method. The building is ana-
lyzed using first an urban exposure (EC2) then an open exposure
^€ ^ T Kt
M
hðzÞ ¼ / ðzÞ ð17Þ (EC4). Other sub-factors in Eq. (4) accounting for topography
I0 h 1 (Ctopography), wind directionality (Cdirection), building importance
^ T , Kt are the resonant component of the torsional moment (Cimportance) and other effects (Cother) are assumed to be 1. Note that
where M
EU standard does not provide windward and leeward pressure
and the mode shape correction, respectively, defined in Table 11; I0
coefficients when H/D > 5. In this case, the standard recommends
is the mass moment of inertia. As with the alongwind acceleration,
to use force coefficient (Cfx) for estimating total wind loading and
the accelerations in NBCC and ISO are not readily defined in terms
it is about 1.45 that includes consideration of the end-effects for
of the general definition. As opposed to the other codes and stan-
the example the building in this study (Section 7 in EU standard)
dards, the acrosswind acceleration for NBCC is defined by an empir-
[9]. Also, IWC offers drag force coefficients in a graphical format
ical formula rather than derived from a base moment. However, the
and the value is determined in terms of d/b and h/d, which may
empirical formula was derived from wind tunnel measurements
potentially lead to some interpolation errors. As mentioned earlier,
[21]. For ISO, the acrosswind and torsional accelerations are again
GLF in the CNS varies along the building height, thus GLF at the
based on the acrosswind displacement and angular deflection.
building top is indicated in all CNS results. It should be under-
scored that while the parameters are rewritten in a general format
2.6. Wind loads combination for the sake of comparison, the calculations are made as indicated
by the codes/standards. The comparison of the various parameters
There are many possible combinations of alongwind, across- reveals several areas of disagreement among the codes/standards,
wind and torsional wind loads associated with the total peak leading to differences in the wind loads and responses. Note that in
response. Indeed some codes/standards offer a combination rule, this comparison of the codes/standards, the analysis results by the
e.g., ASCE7 (Fig. 27.4-8) and NBCC (Fig. I-16) by partial loading NatHaz Aerodynamic Loads Database (NALD) [19,20] are also in-
approach (a load combination factor of 0.75); AS/NZ (Section 6.4) cluded as a reference. It is noted that some codes/standards (AS/
for a combination of alongwind and acrosswind loads; AIJ (Sec- NZ, AIJ, ISO and IWC) offer a provision of a nonlinear mode shape
tion A6.8) and ISO (Section G.3) for a combination of alongwind, by way of a mode shape correction factor while others do not offer
acrosswind and torsional loads using alongwind loads with a any provision. However, for the sake of comparison based on dif-
constant factor (0.4) without the information on the across- ferent codes/standards a linear mode shape was assumed. This
wind/torsional loads or consideration of a correlation coefficient assumption is not too far from the one implied in most high fre-
(qLT) between the acrosswind and torsional loads based on Asa- quency base balance studies in wind tunnels for buildings of this
mi [22]. For reference, Solari et al. [23] proposed a combination geometry and height. Kareem [26] discussed its impact on the
rule for the alongwind and acrosswind loads considering vecto- loading spectra and response. Zhou et al. [27] reported that that
rial effects, Chen and Kareem [24] revisited several combination the influence of a nonlinear mode shape is rather negligible for
rules. the structural displacement and the base bending moment, but
may not be insignificant for other load effects such as the general-
3. Example analyses of tall buildings in codes and standards ized wind loads, the base shear, and the acceleration response. It is
expected that response estimates obtained using linear mode
3.1. Example 1 shape may experience some discrepancies from the nonlinear
mode shapes. However, for the purpose of comparison the linear
To compare the international wind codes/standards considered mode shape based results should suffice.
in this study, an example building is analyzed using each standard. The survivability results are presented in Tables 12 and 13 for
The example building has a height (h) of 200 m with a square exposure EC2 and EC4, respectively. This example highlights
cross-section, width (b) and depth (d) of 33 m, interstory height important similarities and differences among the codes/standards.
of 4 m, natural frequency in the alongwind and acrosswind direc- For instance, both ASCE and AS/NZ determine the response using
tions of 0.2 Hz, damping ratio of 1% in both directions, linear mode peak velocity pressure which yields similar results despite varia-
shape, building density of 180 kg/m3, air density of 1.22 kg/m3, ba- tions in the intermediate parameters. It is noteworthy that the pre-
sic wind velocity (V0) for survivability design of 40 m/s (3-s) and vious version of AS/NZ standard [28] defined a combination factor
34 m/s (3-s) for serviceability design. To convert the velocity to dif- applied to the external pressure, Kc,e, as 1.0, while it is redefined as
ferent averaging times, the relationship developed and provided in 0.9 in the latest version [5]. This leads to a reduction in the overall
graphical form by Durst [25], which is based on the statistical anal- alongwind loads and accelerations. Considering Kc,e = 1.0, the base
ysis of meteorological wind velocity records, is utilized in this shears and moments in AS/NZ are much closer to ASCE loads in
study: V0(3-s) = Gv(10-min)V0(10-min) = Gv(1-h)V0(1-h) where Gv(10-min) = both EC2 and EC4. Note that although EU defines the basic wind
1.42 and Gv(1-h) = 1.51 estimated from the Durst graph. It is worth velocity as 10-min, the velocity pressure (qz) is based on gust re-
noting that an alternative method for compensating averaging sponse: for this reason, Gq in EU is not unity even if 10-min mean
time is a wind-velocity spectrum based method [17,18]. There wind velocity is used as a basic wind speed (Table 5). Among the
exists a slight difference in conversion factor between the Durst codes/standards on mean wind velocity-based procedure (AIJ,
30 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

Table 12
Alongwind survivability results for Example 1 (EC2).

ASCE NALD AS/NZ AIJ CNS NBCC EU ISO IWC

B C Peak Mean
V0 (m/s) 40.0 40.0 40.0 28.1 28.1 26.4 28.1 40.0 28.1 40.0
href (m) 120 200 200 200 200 200 120 200 200 200
 h (m/s)
V 27.5 32.6 46.4 36.4 35.9 33.1 31.5 50.8 30.8 50.8
ref

qh (kN/m2) 1.416 – 1.313 0.806 0.786 0.667 1.723 1.574 0.579 1.574
gR 3.787 3.787 3.094 3.209 2.5 3.755 3.207 3.281 3.281 3.628
r href 0.506 – 0.392 0.322 0.316 0.347 0.418 0.508 0.508 0.392
Lhref (m) 190 – 180 258 1200 1220 213 258 258 211
B 0.583 – 0.633 0.491 – 0.575 0.512 0.409 0.675 0.675 0.256
E 0.140 – 0.094 0.090 – 0.257 0.103 0.103 0.081 0.081 0.072
S 0.048 – 0.081 0.114 – 0.083 0.091 0.125 0.124 0.124 0.085
R 0.526 – 0.601 0.809 – 2.128 0.734 1.014 0.789 0.789 0.610
GLF 2.693 – 2.488 2.012 2.38 2.845 2.496 2.598 3.051 3.051 2.310
G 0.990 – 1.015 2.012 2.38 2.845 1.014 1.055 1.119 3.051 0.974
Cfx 1.30 1.30 1.30 1.30 1.40 1.30 1.45 1.30 1.30 1.46
Vbase (MN) 9.99 8.80 8.73 11.98 9.55 9.86 13.33 13.88 10.40 9.02 12.06
Mbase (MN-m) 1104 1111 957.0 1341 1211 1145 1506 1568 1251 1178 1373

V href = design wind velocity at href; qh = velocity pressure at building top (h); r href = turbulence intensity factor; Lhref = length scale at href; B = background response factor;
E = energy factor; S = size reduction factor; R = resonant response factor; Cfx = drag force coefficient; Vbase = base shear force; Mbase = base moment.

Table 13
Alongwind survivability results for Example 1 (EC4).

ASCE NALD AS/NZ AIJ CNS NBCC EU ISO IWC

B C Peak Mean
V0 (m/s) 40.0 40.0 40.0 28.1 28.1 26.4 28.1 40.0 28.1 40.0
href (m) 120 200 200 200 200 200 120 200 200 200
 h (m/s)
V 38.1 41.2 51.6 43.9 44.0 40.2 41.6 54.4 42.8 52.0
ref

qh (kN/m2) 1.836 – 1.624 1.177 1.181 0.984 2.206 1.805 1.117 1.649
gR 3.787 3.787 3.094 3.229 2.5 3.765 3.232 3.281 3.281 3.628
r href 0.225 – 0.214 0.245 0.179 0.210 0.257 0.212 0.212 0.214
Lhref (m) 251 – 180 258 1200 1220 230 258 258 211
B 0.624 – 0.633 0.528 – 0.575 0.524 0.428 0.636 0.636 0.256
E 0.144 – 0.116 0.102 – 0.290 0.115 0.115 0.101 0.101 0.084
S 0.078 – 0.122 0.159 – 0.108 0.130 0.170 0.182 0.182 0.113
R 0.889 – 1.107 1.272 – 3.140 1.175 1.535 1.439 1.439 0.944
GLF 1.854 – 1.939 1.969 2.00 2.349 2.083 2.170 2.013 2.013 1.844
G 1.051 – 1.082 1.969 2.00 2.349 1.097 1.143 1.170 2.013 1.054
Cfx 1.30 1.30 1.30 1.30 1.40 1.30 1.45 1.30 1.30 1.46
Vbase (MN) 14.93 14.19 12.61 17.97 14.22 15.94 19.85 20.62 14.74 14.15 14.63
Mbase (MN-m) 1599 1705 1336 1956 1707 1736 2168 2253 1667 1703 1598

NBCC and ISO Mean), AIJ has the largest velocity pressure and over- responses in the calculation of peak factors (AIJ, NBCC and EU),
all load for EC2 while NBCC is the largest for EC4. Somewhat large does not significantly affect the peak factor. The resulting loads
alongwind loads in IWC, especially in EC2 (Table 12), are likely due by CNS are generally smaller than other codes/standards that use
to the large drag force coefficient (Cfx = 1.46), however, as men- the same averaging time, i.e., 10-min (AIJ and EU), which this
tioned earlier, this may also include a possible interpolation error may be due to smaller peak factors (Table 5).
in estimating the force coefficient from a graphical format. The The scatter in the turbulence intensity factor (rhref ) is ex-
average coefficient of variation (CoV) for the alongwind base shear pected given the results of Fig. 2, especially for EC2, though
and moment in EC2 is about 0.151 and 0.145 in EC4 and for along- the variability is less significant for EC4. The variation of back-
wind RMS acceleration, 0.254 in EC2 and 0.297 in EC4. ground response factors (B) can be attributed to the large vari-
The scatter in the velocity profiles is evident in the mean veloc- ation in its definition [16]. The scatter is less significant for EC4,
ity at reference height. For example, both ASCE and EU define the where in particular ASCE, AS/NZ, and both ISO methods show
mean velocity at 0.6h using 1-h averaging time, yet the mean consistent values. AS/NZ and ISO have the highest background
velocity for EU is closer to the mean velocity at h for both NALD response factors while NBCC and EU C procedure have the
and NBCC. This causes the loading in EU to be larger than ASCE. smallest values. Note that the definition of background response
The effect of averaging time can also be noted for the resonant peak factor in IWC (Table 8) is similar to previous version of AS/NZ
factor (gR). AIJ, EU and ISO have consistent peak factors and use an standard [28], which produces much smaller value than the lat-
averaging time of 10-min (Table 5). Although AS/NZ also uses 10- est AS/NZ standard. However, it is observed that base shear and
min averaging time for the peak factor, the equation does not uti- moment in IWC are little influence by the different definition of
lize the second term found in other definitions (Table 5), which the background response factor.
yields a slightly smaller value than the rest. In addition, ASCE, The effect of using the Davenport wind spectrum in NBCC can
NALD, NBCC and IWC have higher peak factors due to the use of be observed in significantly larger energy factor (E) and resonant
1-h averaging time. The inclusion of an additional factor (C) (Ta- response factor (R), where the Davenport spectrum is considerably
ble 5), which accounts for the effects of background and resonant different from other wind spectra [3,16]. The energy factor is some-
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 31

what larger for ASCE than the rest which are very close. Again, the coefficient of variation (CoV) for the acrosswind base shear and
scatter in the definitions of the size reduction factor (S) is evident moment in EC2 is about 0.332 and 0.235 in EC4 and for acrosswind
in the example. The differences in the energy and size reduction RMS acceleration, 0.317 in EC2 and 0.329 in EC4. Overall, the CoV
factors do not compensate in the resonant response factor. The in acrosswind responses has higher values than alongwind ones,
gust loading factor (GLF) for AIJ is the smallest in EC2, while ASCE which indicates more scatter in acrosswind responses.
and IWC are the lowest ones in EC4. The overall drag force coeffi- Some discrepancies are observed between AS/NZ and IWC even
cients (Cfx), defined as the windward external pressure coefficient though both standards are based on almost the same formulation
at building height minus the leeward external pressure coefficient, and graphical formats. Many parameters are involved in this re-
are the same (Cfx = 1.30) except for the higher values in CNS, EU gard, e.g., different wind velocity profiles, turbulence length scales,
and IWC. averaging time, etc., but for the alongwind loads (Tables 12 and
The ISO base shears are consistent with ASCE and AS/NZ, how- 13), the main factors appear to be the difference in the drag force
ever the moments are somewhat larger. In addition, the base mo- coefficients (Cfx = 1.30 vs 1.46) and the combination factor
ment calculated with NALD is moderately larger than that of (Kc,e = 0.9 vs 1.0). For the acrosswind loads, although turbulence
ASCE. Likely, the larger moments are due to the difference be- intensity profile is the same in both AS/NZ and IWC, wind velocity
tween calculating the response in terms of the building height at reference height makes a large difference because both stan-
as opposed to a reduced height of 0.6h. NALD and ASCE use dif- dards require interpolation of force spectrum coefficients
ferent distributions of the load, which NALD uses mode shape [r2cm C m ðf2 Þ] based on reduced frequency (e.g., Table 9) and the
based profile for the distribution of resonant responses (inertial loads are also proportional to the squared wind velocity at refer-
loads) instead of the wind profile [19,20]. Therefore, some differ- ence height. For this reason, the discrepancies in the acrosswind
ence in the base shear values between ASCE and NALD is loads in two standards are more prominent in EC2 (Table 14)
understandable. where more scatter in the wind velocity profile is observed
Despite different definitions of the parameters included in the (Fig. 2). In addition, different resonant peak factors (gR) due to dif-
gust loading factor, the resulting gust loading factors and loads ferent averaging time (10 min vs 1-h) and a smaller gR in AS/NZ due
are relatively consistent for both appendices of EU (EU B and C). to the ignorance of the second term in the general peak factor
Larger values of the base shear and moment in EU as compared expression (Eq. (7) and Table 5) also play a role as the acrosswind
to ASCE and AS/NZ can likely be attributed to the larger values of loads are proportional to gR.
the large velocity pressure (qh), drag force coefficient (Cfx) and a un- The serviceability results for Example 1 are presented in
ique definition of windward velocity pressure profile along the Tables 15 and 16. Note that estimates for ISO and NBCC in the
height when h > 2b. EU considers the profile as multiple parts, alongwind direction are not provided as the accelerations are
comprising: a lower part extending upwards from the ground by dependent on displacement, whereas no estimate of the displace-
a height equal to b; an upper part extending downwards from ment is provided in the standards. In addition, CNS is also excluded
the top by a height equal to b and a middle region, between the in the serviceability results because no provisions are provided.
upper and lower parts, which may be divided into horizontal strips Overall, the same general trends observed in the survivability re-
with a height (Fig. 7.4 in EU standard) [9]. The resulting loads for sults are observed in the serviceability results. For example, the ef-
AIJ are consistent with those of EU C. As discussed earlier, the def- fect of averaging time and velocity profile are noticeable. The
inition of the energy spectrum of NBCC differs considerably from alongwind RMS acceleration in EC2 is consistent for ASCE, NALD,
the other codes/standards, resulting in a much higher resonant fac- and AIJ. Variations in the peak acceleration are presumably due
tor. This is likely the cause of the larger base shear and moment, to discrepancies in the resonant peak factors (gR). For EC4, there
especially for EC4. is considerably more scatter in the RMS acceleration values. EU
The acrosswind survivability results for Example 1 are com- has the largest predictions of acceleration, and ASCE and AS/NZ
pared in Table 14. The large variability of the base moments can show lower acceleration values than others in the alongwind direc-
be attributed to the difference in the non-dimensional base bend- tion. Finally, the acrosswind accelerations for AIJ are considerably
ing moment spectrum values and the averaging time for the veloc- higher, as observed in the survivability results. A similar trend is
ity pressure. For both exposure categories (EC2 and EC4), NALD, observed in the IWC results, especially in EC2, due to large value
AS/NZ and CNS are relatively consistent, yet smaller than the loads of spectrum coefficient. NBCC accelerations are more consistent
in AIJ. It may be attributed to the use of empirical expressions in with NALD and AS/NZ.
AIJ, which mainly a function of the side ratio, i.e., the ratio of a
building width and depth. It ignores the terrain condition, building 3.2. Example 2
height, etc., which may yield over-/underestimated results [19].
Although AIJ and ISO have the same definitions for the acrosswind The analysis of Example 1 reveals the strong influence of veloc-
loads, the discrepancies in the velocity pressure and exposure ity profile, averaging time, turbulence profile and pressure/force
parameters result in discrepancies in the overall loads. The average coefficients on the resulting response. To reduce the scatter and

Table 14
Acrosswind survivability results for Example 1.

EC2 EC4
NALD AS/NZ AIJ CNS ISO IWC NALD AS/NZ AIJ CNS ISO IWC
Peak Mean Peak Mean
gR 3.787 3.094 3.283 2.5 3.281 3.281 3.628 3.787 3.094 3.283 2.5 3.281 3.281 3.628
qh (kN/m2) – 1.313 0.806 0.786 1.574 0.579 1.574 – 1.624 1.177 1.181 1.805 1.117 1.649
r2cm C m ðf2 Þ (103)* 1.80 6.08 3.66 – 2.46 2.46 7.79 1.52 4.75 6.34 – 5.92 5.92 5.32
Vbase (MN) 9.96 9.52 14.32 9.32 6.99 6.52 15.87 14.99 15.90 27.52 16.34 21.84 21.96 20.67
Mbase (MN-m) 1290 1283 1929 1212 999.0 951.9 2137 1898 2141 3706 2084 3021 3087 2784
*
rcm = the RMS base moment coefficient; C m ðf2 Þ = the acrosswind moment coefficient at the building’s natural frequency in the acrosswind direction (f2).
32 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

Table 15
Serviceability results for Example 1 (EC2).

ASCE NALD AS/NZ AIJ NBCC EU IWC

B C
Alongwind
V0 (m/s) 34.0 34.0 34.0 24.0 22.5 24.0 34.0
 h (m/s)
V 23.4 27.7 39.4 31.0 28.2 26.9 43.2
ref

qh (kN/m2) 1.023 – 0.949 0.588 0.484 1.257 1.137

r href 0.506 – 0.392 0.322 0.347 0.418 0.392
gR 3.787 3.787 3.094 3.210 3.744 3.185 3.628
R 0.370 – 0.433 0.595 1.525 0.538 0.763 0.441
GLF 2.548 – 2.403 2.066 2.621 2.364 2.441 2.172
G 0.937 – 0.981 2.066 2.621 0.960 0.991 0.916
K 0.502 – – 1.000 – 1.500 1.500 –
r€x (mg)* 3.44 3.81 2.79 3.96 – 5.06 6.02 4.36
^
x (mg)*
€ 13.03 14.44 8.65 12.72 – 16.11 19.19 15.81

Acrosswind
r2cm C m ðf2 Þ – 9.04  104 4.59  103 2.47  103 – – – 5.53  103
ry€ (mg)* – 6.42 9.85 12.75 7.19 – – 13.36
^
€ (mg)*
y – 24.30 30.48 40.92 26.92 – – 48.45

*
r€x ; r€y = RMS accelerations; ^€x, y^€ = peak accelerations at building height.

Table 16
Serviceability results for Example 1 (EC4).

ASCE NALD AS/NZ AIJ NBCC EU IWC

B C
Alongwind
V0 (m/s) 34.0 34.0 34.0 24.0 22.5 24.0 34.0
 h (m/s)
V 32.4 35.0 43.9 37.5 34.2 35.5 44.2
ref

qh (kN/m2) 1.333 – 1.173 0.858 0.714 1.609 1.192

r href 0.225 – 0.214 0.245 0.210 0.257 0.214
gR 3.787 3.787 3.094 3.229 3.757 3.217 3.628
R 0.638 – 0.814 0.941 2.285 0.881 1.189 0.693
GLF 1.766 – 1.868 1.952 2.181 1.980 2.051 1.749
G 1.001 – 1.043 1.952 2.181 1.042 1.080 1.000
K 0.501 – – 1.000 – 1.500 1.500 –
r€x (mg) 3.85 6.54 3.75 6.06 – 6.92 8.04 4.35
^
€
x (mg) 14.58 24.78 11.60 19.57 – 22.25 25.86 15.78

Acrosswind
r2cm C m ðf2 Þ – 1.09  103 2.31  103 3.98  103 – – – 2.53  103
ry€ (mg) – 11.34 13.20 24.11 13.67 – – 14.25
^
€ (mg)
y – 42.95 40.84 77.87 51.34 – – 51.68

dependence on velocity pressure, the second example analyzed the study. In CNS, additional peak factors using ISO values (gB = 3.4;
same building with modified velocity pressure at building top (qh) gR = 3.281) have also been utilized to investigate the effects of rel-
and the drag force coefficient (Cfx) for all the codes/standards for atively low peak factors defined in CNS (gB = gR = 2.5). From Tables
EC3. Note that NALD only offers urban and open exposures that 17 and 18, it is observed that the resulting loads of the CNS with
correspond to EC2 and EC4, respectively, therefore, averaged val- the ISO peak factors become more consistent with those in other
ues of these exposures are listed as a reference. In order to obtain codes/standards. In addition, the velocity pressures (qh) are now
a consistent value of velocity pressure across all codes/standards, more consistent among mean based codes/standards such as AIJ,
the basic wind velocity (V0) is adjusted so that the velocity pres- China, NBCC and ISO Mean procedure and the peak based codes/
sure at building top (qh) is the same. To account for different aver- standards such as ASCE, AS/NZ, EU, ISO Peak procedure and IWC.
aging times (3-s, 10-min and 1-h) in velocity pressure, the gust The resulting base shear and base moments for ASCE and AS/NZ
factor for wind velocity pressure (Eq. (8)) is utilized. If ISO conver- are essentially very similar and it may be because any variation
sion factors (Gv) (1.46 and 1.53 as mentioned earlier) are used in in the intermediary parameters was compensated in the overall re-
Eq. (8), Gq results in 1.92 and 2.06 for 10-min and 1-h, respectively. sponse estimate. The same observation can be made for both sur-
As shown in Table 17, the modifications made to the velocity pres- vivability (Tables 17 and 18) and serviceability results (Table 19).
sure (qh) in terms of basic wind velocity and pressure coefficients In Example 2, the average CoV of alongwind and acrosswind base
brought the estimated wind loads based on the various codes/stan- shear and moment is dramatically reduced to 0.049 (alongwind)
dards draw closer to each other. & 0.133 (acrosswind), whereas it was 0.151 (alongwind) & 0.332
It is clear that the reduction of velocity pressure and drag force (acrosswind) in EC2 and 0.145 (alongwind) & 0.235 (acrosswind)
coefficient for EU and IWC contributes to the reduction in the over- in EC4 in Example 1. For serviceability results, CoV of RMS acceler-
all loads compared to the other codes/standards. The discrepancies ation is also reduced to 0.111 (alongwind) & 0.193 (acrosswind),
can be attributed to variations of turbulence intensity and slightly while it was 0.254 (alongwind) & 0.317 (acrosswind) in EC2 and
different GLF procedures among the codes/standards used in this 0.297 (alongwind) & 0.329 (acrosswind) in EC4 in Example 1.
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 33

Table 17
Alongwind survivability results for Example 2 (EC3).

ASCE NALDa AS/NZ AIJ CNSb NBCC EU ISO IWC

B C Peak Mean
V0 (m/s) 40.0 40.0 42.1 26.0 26.6 25.8 25.8 39.2 27.6 41.1
href (m) 120 200 200 200 200 200 120 200 200 200
 h (m/s)
V 33.5 – 52.2 37.6 37.6 36.3 33.3 52.2 37.6 52.2
ref

qh (kN/m2) 1.659 – 1.659 0.864 0.864 0.805 1.659 1.659 0.864 1.659
gR 3.787 3.787 3.094 3.214 2.5 (3.281) 3.760 3.212 3.281 3.281 3.628
rhref 0.337 – 0.278 0.276 0.238 0.255 0.334 0.312 0.312 0.278
Lhref (m) 223 – 180 258 1200 1220 220 258 258 211
B 0.607 – 0.633 0.513 – 0.575 0.517 0.417 0.653 0.653 0.256
E 0.143 – 0.110 0.092 – 0.273 0.105 0.105 0.092 0.092 0.080
S 0.065 – 0.112 0.126 – 0.095 0.098 0.134 0.156 0.158 0.103
R 0.729 – 0.970 0.910 – 2.579 0.803 1.098 1.127 1.150 0.819
GLF 2.228 – 2.178 1.939 2.17 (2.59) 2.499 2.232 2.320 2.384 2.393 2.037
G 1.038 – 1.073 1.939 2.17 (2.59) 2.499 1.029 1.070 1.157 2.393 1.032
Cfx 1.30 1.30 1.30 1.30 1.30 1.30 1.30 1.30 1.30 1.30
Vbase (MN) 12.84 11.50 12.31 12.67 9.88 (11.55) 11.92 13.09 13.61 12.62 11.93 12.46
Mbase (MN-m) 1394 1408 1325 1399 1222 (1438) 1351 1403 1459 1460 1486 1378
a
Averaging values obtained from EC2 and EC4 using NALD.
b
Values in parentheses are obtained using ISO peak factors.

Table 18
Acrosswind survivability results for Example 2 (EC3).

NALDa AS/NZ AIJ CNSb ISO IWC

Peak Mean
gR 3.787 3.094 3.283 2.5 (3.281) 3.281 3.281 3.628
qh (kN/m2) – 1.659 0.864 0.864 1.659 0.864 1.659
r2cm C m ðf2 Þ – 5.13  103 4.02  103 – 3.96  103 4.07  103 5.52  103
Vbase (MN) 12.48 14.36 16.09 10.86 (14.77) 13.21 13.44 18.10
Mbase (MN-m) 1594 1933 2167 1397 (1900) 1849 1916 2438
a
Averaging values of EC2 and EC4 using NALD.
b
Values in parentheses is the results using ISO peak factors.

Table 19
Serviceability results for Example 2 (EC3).

ASCE NALD AS/NZ AIJ NBCC EU IWC

B C
Alongwind
V0 (m/s) 34.0 34.0 35.7 22.1 21.9 22.0 34.9
 h (m/s)
V 28.5 – 44.3 31.8 30.9 28.3 44.3
ref

qh (kN/m2) 1.199 – 1.199 0.624 0.582 1.199 1.199

r href 0.337 – 0.278 0.276 0.255 0.334 0.332
gR 3.787 3.787 3.094 3.193 3.751 3.191 3.628
R 0.519 – 0.710 0.654 1.851 0.586 0.824 0.598
GLF 2.111 – 2.093 1.845 2.311 2.119 2.187 1.922
G 0.984 – 1.032 1.845 2.311 0.977 1.008 0.974
K 0.501 – – 1.000 1.000 1.500 1.500 –
r€x (mg) 4.03 5.18 4.02 4.14 – 4.19 4.97 4.10
^
€
x (mg) 15.26 19.61 12.43 13.22 – 13.37 15.86 14.87

Acrosswind
r2cm C m ðf2 Þ – – 2.91  103 2.65  103 – – 3.08  103
ry€ (mg) – 8.88 12.85 13.77 9.74 – 13.51
^
€ (mg)
y – 33.63 39.79 45.20 36.55 – 49.01

3.3. Example 3 velocity pressure and pressure coefficients are also modified to
be consistent. The survivability results are presented in Tables
Finally, a third example is presented in which a rectangular 20 and 21 and serviceability results in Table 22. Note that NALD
building with h = 180 m, b = 30 m, d = 60 m, natural frequencies does not have the dataset for the dimensional ratio (b:d:h =
in both sway directions = 0.25 Hz, building density = 180 kg/m3, 1:2:6) in this example, the closest one (b:d:h = 1:2:6.67) is
damping ratio of 2%, and basic wind velocity of 60 m/s (3-s) utilized as a reference, which may lead to some discrepancy
for survivability and 45 m/s (3-s) for serviceability. The building with results by the codes/standards. Despite the higher basic
is assumed to be in EC3. Based on the discussions in Example 2, wind velocity, different building dimension and property, the
34 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35

Table 20
Alongwind survivability results for Example 3 (EC3).

ASCE NALDa AS/NZ AIJ CNSb NBCC EU ISO IWC

B C Peak Mean
V0 (m/s) 60.0 60 62.8 39.3 40.3 39.1 38.6 58.8 42.3 61.3
href (m) 108 180 180 180 180 180 108 180 180 180
 h (m/s)
V 48.9 – 77.1 55.6 55.5 53.7 50.0 77.1 55.6 77.1
ref

qh (kN/m2) 3.622 – 3.622 1.887 1.887 1.758 3.622 3.622 1.887 3.622
gR 3.845 3.845 3.166 3.270 2.5 (3.348) 3.810 3.265 3.348 3.348 3.688
rhref 0.343 – 0.287 0.283 0.244 0.262 0.340 0.323 0.323 0.287
Lhref (m) 216 – 175 245 1200 1220 206 245 245 206
B 0.617 – 0.651 0.519 – 0.625 0.523 0.427 0.660 0.660 0.521
E 0.159 – 0.124 0.106 – 0.302 0.118 0.118 0.105 0.107 0.090
S 0.085 – 0.152 0.177 – 0.135 0.137 0.178 0.218 0.224 0.140
R 0.530 – 0.740 0.741 – 2.036 0.637 0.825 0.902 0.937 0.793
GLF 2.153 – 2.159 1.923 1.97 (2.29) 2.433 2.195 2.241 2.361 2.377 1.988
G 0.994 – 1.047 1.923 1.97 (2.29) 2.433 1.003 1.024 1.125 2.377 0.992
Cfx 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10 1.10
Vbase (MN) 18.07 15.09 17.74 18.31 13.79 (15.76) 17.49 18.85 19.25 18.72 18.43 18.10
Mbase (MN-m) 1792 1637 1742 1853 1528 (1758) 1819 1835 1873 1938 2065 1807
a
Averaging values obtained from EC2 and EC4 using NALD.
b
Values in parentheses are obtained using ISO peak factors.

Table 21
Acrosswind survivability results for Example 3 (EC3).

NALDa AS/NZ AIJ CNSb ISO IWC

Peak Mean
gR 3.845 3.166 3.350 2.5 (3.348) 3.348 3.348 3.688
qh (kN/m2) – 3.622 1.887 1.887 3.622 1.887 3.622
r2cm C m ðf2 Þ – 5.80  103 7.45  103 – 7.43  103 7.72  103 6.07  103
Vbase (MN) 23.16 19.35 28.30 21.08 (28.23) 23.06 24.12 23.94
Mbase (MN-m) 2648 2348 3434 2442 (3270) 2909 3094 2810
a
Averaging values of EC2 and EC4 using NALD.
b
Values in parentheses is the results using ISO peak factors.

Table 22
Serviceability results for Example 3 (EC3).

ASCE NALD AS/NZ AIJ NBCC EU IWC

B C
Alongwind
V0 (m/s) 45.0 45.0 47.1 29.5 29.4 29.0 46.0
 h (m/s)
V 36.7 – 55.3 44.7 40.3 36.7 57.8
ref

qh (kN/m2) 2.038 – 2.038 1.061 0.989 2.038 2.038

r href 0.343 – 0.283 0.251 0.262 0.340 0.287
gR 3.845 3.845 3.166 3.228 3.789 3.223 3.688
R 0.301 – 0.437 0.425 1.164 0.375 0.516 0.607
GLF 2.005 – 2.045 1.788 2.168 2.038 2.063 1.828
G 0.926 – 0.992 1.788 2.168 0.931 0.942 0.912
K 0.501 – – 1.000 1.000 1.500 1.500 –
r€x (mg) 2.42 2.98 2.51 2.83 – 2.67 3.13 2.59
^
€
x (mg) 9.30 11.45 7.93 9.14 – 8.61 10.09 9.57

Acrosswind
r2cm C m ðf2 Þ – – 2.96  103 4.56  103 – – 3.09  103
ry€ (mg) – 6.15 8.39 10.83 6.94 – 8.76
^
€ (mg)
y – 23.65 26.56 36.28 26.29 – 32.31

results in this example are similar in trend to those noted in The analysis in these three examples reveals that, although the
Example 2, being reasonably consistent for alongwind responses, intermediate parameters may vary, the codes/standards used in
while more scatter is observed in acrosswind responses: CoV this study will yield reasonably consistent results if the basic wind
of 0.07 (alongwind) & 0.128 (acrosswind) for base shear and velocity is modified to match the velocity pressure at building top
moment and 0.09 (alongwind) & 0.220 (acrosswind) for RMS and if force/pressure coefficients are the same. Thus, while some
acceleration . discrepancies can be attributed to the turbulence intensity, the
 D.K. Kwon, A. Kareem / Engineering Structures 51 (2013) 23–35 35

velocity profile is the primary cause for discrepancies between the References
overall loads and standards.
[1] ISO. 4354 Wind actions on structures. Switzerland: International Organization
for Standardization (ISO); 2009.
[2] Tamura Y, Kareem A, Solari G, Kwok KCS, Holmes JD, Melbourne WH. Aspects
4. Concluding remarks of the dynamic wind-induced response of structures and codification. Wind
Struct 2005;8:251–68.
[3] Zhou Y, Kijewski T, Kareem A. Along-wind load effects on tall buildings:
This paper examines the differences and similarities in the wind Comparative study of major international codes and standards. J Struct Eng,
loads and their effects on tall buildings based on major interna- ASCE 2002;128:788–96.
tional wind codes/standards such as ASCE, AS/NZ, AIJ, CNS, NBCC, [4] American Society of Civil Engineers (ASCE). Minimum design loads for
buildings and other structures. Reston (VA): ASCE; 2010.
EU, ISO and IWC and a database-enabled design procedure, NatHaz [5] Joint Technical Committee. AS/NZS 1170.2:2011 Structural design actions –
Aerodynamic Loads Database (NALD). The aim was to compare not Part 2: Wind actions. Australian/New Zealand Standard (AS/NZS): Joint
only the overall building loads but also to identify the role of var- Technical Committee BD-006, Australia/New Zealand; 2011.
[6] Architectural Institute of Japan (AIJ). RLB recommendations for loads on
ious parameters that contribute to the overall analyses. In order to buildings. Tokyo (Japan): Structural Standards Committee, Architectural
make the comparisons meaningful and consistent, the equations Institute of Japan; 2004.
have been first rewritten in a general format. While significant dis- [7] National Standards Committee. Load code for the design of building structures,
China National Standard (CNS). GB 50009-2012, Beijing (China): National
crepancies are apparent in the comparison of parameters, the over- Standards Committee; 2012.
all loads are reasonably consistent in the alongwind response but [8] National Research Council (NRC). National building code of Canada. Ottawa
more scatter is observed in the acrosswind response. (Canada): Associate Committee on the National Building Code, National
Research Council; 2010.
The parameters associated with the wind velocity characteris-
[9] European Committee for Standardization (CEN). Eurocode 1: Actions on
tics contribute the most towards apparent differences in the result- structures – Part 1–4: General actions – wind actions. EN 1991-1-4:2005/
ing wind responses such as base shear/moment and peak/RMS AC:2010 (E). Europe: European Standard (Eurocode), European Committee for
Standardization (CEN); 2010.
acceleration. Additional examples (Examples 2 and 3) illustrated
[10] Indian Wind Code (IWC). IS: 875 (Part 3): Wind loads on buildings and
the significant effects of the variation in velocity profile descrip- structures – proposed draft & commentary. Document No: IITK GSDMA-Wind
tions by ensuring a consistent velocity pressure at the reference 02-V 50. India; 2012.
height. With a few modifications to specific parameters, the dis- [11] Halder L, Dutta SC. Wind effects on multi-storied buildings: a critical review of
Indian codal provisions with special reference to American standard. Asian J
crepancies among the codes/standards were further reduced. Civ Eng (Build Housing) 2010;11:345–70.
While discrepancies with other parameters such as peak factors, [12] Davenport AG. Gust loading factors. J Struct Div, ASCE 1967;93:11–34.
turbulence intensity, energy factors, etc. contribute to some differ- [13] McCullough M, Kwon DK, Kareem A, Wang L. Efficacy of averaging interval for
non-stationary winds. In: 2012 Joint conference of the engineering mechanics
ences in the overall loads, the standardization of wind loading institute and 11th ASCE joint specialty conference on probabilistic mechanics
codes/standards may be achievable by eliminating differences in and structural reliability (EMI/PMC 2012). Notre Dame, IN; 2012.
the velocity profiles. This is especially true for the alongwind load [14] Holmes JD, Allsop A. Averaging times and gust durations for codes and
standards. In: 10th UK Conference on wind engineering (WES-2012).
effects as all are based on a gust loading/effect factor approach. Southampton (United Kingdom); 2012. p. 207–10.
Regarding the acrosswind and torsional directions where wake-in- [15] Gurley K, Kareem A. Gust loading factors for tension leg platforms. Appl Ocean
duced effects are more dominant than the buffeting effects in the Res 1993;15:137–54.
[16] Bashor R. Dynamics of wind sensitive structures. USA: University of Notre
alongwind direction, a database-enabled design framework such
Dame; 2010.
as NALD is a more promising design procedure for better assess- [17] Solari G. Gust Buffeting. 1. Peak wind velocity and equivalent pressure. J Struct
ment of such loading effects on tall buildings. Like ASCE 7 adding Eng, ASCE 1993;119:365–82.
[18] Solari G, Kareem A. On the formulation of ASCE7-95 gust effect factor. J Wind
such provisions in the supplementary documents of a standard,
Eng Ind Aerod 1998;77–8:673–84.
e.g., a commentary may help to further supplement empirical [19] Kwon DK, Kijewski-Correa T, Kareem A. E-analysis of high-rise buildings
expressions based on reductive format and simplifications with subjected to wind loads. J Struct Eng, ASCE 2008;134:1139–53.
more realistic information. [20] Zhou Y, Kijewski T, Kareern A. Aerodynamic loads on tall buildings: interactive
database. J Struct Eng, ASCE 2003;129:394–404.
[21] Boggs D. Some personal observations on the development of motion comfort
criteria. In: 2002 ASCE structures congress. Denver (CO); 2002.
Acknowledgements [22] Asami Y. Combination method for wind loads on high-rise buildings. In: 16th
National symposium on wind engineering. Tokyo (Japan); 2000. p. 531–4.
[23] Solari G, Reinhold TA, Livesey F. Investigation of wind actions and effects on
The support for this paper was in part provided by the Grant # the Leaning Tower of Pisa. Wind Struct 1998;1:1–23.
CMMI 0601143 from the National Science Foundation and Global [24] Chen XZ, Kareem A. Dynamic wind effects on buildings with 3D coupled
modes: application of high frequency force balance measurements. J Eng
Center of Excellence (GCOE) at Tokyo Polytechnic University Mech, ASCE 2005;131:1115–25.
funded by MEXT, Japan. The authors also wish to specially [25] Durst CS. Wind speeds over short periods of time. Meteorol Mag 1960:181–7.
acknowledge Dr. Rachel Bashor for her initial calculations and Dr. [26] Kareem A. Model for predicting the acrosswind response of buildings. Eng
Struct 1984;6:136–41.
Shouying Li, Hunan University, China, for his contribution pertain-
[27] Zhou Y, Kareem A, Gu M. Mode shape corrections for wind load effects. J Eng
ing to China National Standard and input regarding different stan- Mech, ASCE 2002;128:15–23.
dards received from Drs. Y. Tamura, J. Holmes, G. Solari, T. [28] SAA. AS1170.2 Part 2: Wind forces, Australian standards AS1170.2:2002.
Sydney (Australia): Standards Association of Australia; 2002.
Stathopoulos, A. K. Mittal, S. Arunachalam.