See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228373230

e-Analysis of High-Rise Buildings Subjected to Wind Loads

Article  in  Journal of Structural Engineering · July 2008

DOI: 10.1061/(ASCE)0733-9445(2008)134:7(1139)

CITATIONS READS
58 1,207

3 authors:

Dae Kun Kwon Tracy Kijewski-Correa

University of Notre Dame University of Notre Dame
66 PUBLICATIONS   764 CITATIONS    54 PUBLICATIONS   780 CITATIONS   

SEE PROFILE SEE PROFILE

Ahsan Kareem
University of Notre Dame
490 PUBLICATIONS   10,275 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Application of shapelet transform to long-term monitoring networks View project

Full-scale Monitoring of Tall Buildings View project

All content following this page was uploaded by Tracy Kijewski-Correa on 03 June 2014.

The user has requested enhancement of the downloaded file.

 e-Analysis of High-Rise Buildings Subjected to Wind Loads
Dae-Kun Kwon1; Tracy Kijewski-Correa2; and Ahsan Kareem3

Abstract: The NatHaz Aerodynamic Loads Database 共NALD兲 共http://aerodata.ce.nd.edu兲 introduced in 2000 has served an important
first step in establishing an on-line experimental archive of high-frequency base balance 共HFBB兲 data for use in the preliminary design of
high-rise buildings subjected to wind loads. As a result, NALD was recently introduced in the Commentary of ASCE 7-05 共C6.5.8兲 as an
alternative means of assessing the dynamic wind load effects on high-rise buildings. This paper presents NALD version 2.0 共v. 2.0兲,
integrating the latest advances in data management and mining for interactive queries of aerodynamic load data and an integrated on-line
analysis framework for determining the resulting base moments, displacements, and equivalent static wind loads for survivability and
accelerations for serviceability 共habitability兲. The key feature of NALD v. 2.0 is the flexibility its analysis module offers: Users may select
not only the data from the on-line NatHaz aerodynamic loads database, but also may input desired power spectral density 共PSD兲
expression or wind tunnel-derived PSD data set obtained from a HFBB experiment for the evaluation of wind load effects on high-rise
buildings. Thus, it serves as a stand-alone analysis engine. Examples illustrate the capabilities of NALD v. 2.0 and provide comparisons
of response estimates to demonstrate the flexibility of the analysis engine to provide a platform that can be readily expanded and
supplemented to yield a comprehensive, simplified, and efficient avenue for e-analysis of high-rise buildings.
DOI: 10.1061/共ASCE兲0733-9445共2008兲134:7共1139兲
CE Database subject headings: Aerodynamics; Wind loads; Wind tunnels; Buildings, high-rise; Building design; Structural
response; Standards and cores; Internet; Information technology 共IT兲.

Introduction tive formats and simplifications, which often lead to tables and
plots that describe wind loads on structures. The level of accuracy
One of the major challenges in any engineering discipline is the inherent in codification information in this format and the uncer-
processing and archiving of large quantities of information. This tainty associated with interpolation or extrapolation of informa-
is no exception in the field of structural engineering, where such tion may compromise the overall accuracy in code-specified load
stores of data include those generated by wind tunnel studies, effects. This has led to database-assisted design procedures,
laboratory experiments, material testing, and even full-scale which offer convenient meshing with existing analysis software.
monitoring. Recent developments in information technology 共IT兲 Primarily, such databases rely on wind tunnel-derived data, which
offer attractive solutions to these challenges, allowing efficient may be couched in analysis portals to provide desired load ef-
means to collect, store, analyze, manage, and even share large fects. One such example is described below.
data sets with the worldwide community 共Kijewski et al. 2003; The NatHaz Aerodynamic Loads Database Version 1.0 共NALD
Kwon et al. 2005; Fritz and Simiu 2005兲. Not only do such ap- v. 1.0兲, established in the fall of 2000, is an example of Web-
proaches enable geographically dispersed researchers working on based archiving and distribution of wind tunnel test data for the
determination of alongwind, acrosswind, and torsional response
a similar topic to share data and findings, but it also provides a
共Zhou et al. 2003兲. This site has served as an important first step
venue in which this information can be disseminated to other
in establishing an on-line experimental database for use in the
members of the design community around the world.
preliminary design of high-rise buildings, which is being exten-
Most codes and standards traditionally have relied on reduc-
sively consulted by a number of firms 共McNamara 2005兲 and
1
individual researchers 共Chan and Chui 2006兲. This interactive da-
Postdoctoral Research Associate, Dept. of Civil Engineering and tabase provides users with the RMS base bending moment coef-
Geological Sciences, Univ. of Notre Dame, Notre Dame, IN 46556.
ficients and the nondimensionalized power spectra obtained from
E-mail: [email protected]
2
Rooney Family Assistant Professor, Dept. of Civil Engineering and high-frequency base balance 共HFBB兲 measurements on rigid
Geological Sciences, Univ. of Notre Dame, Notre Dame, IN 46556. building models of various aspect ratios and geometries, exposed
E-mail: [email protected] to two typical boundary layers. One attractive feature in this pack-
3
Robert M. Moran Professor, Dept. of Civil Engineering and Geologi- age was the use of JAVA-based applets to provide a specific spec-
cal Sciences, Univ. of Notre Dame, Notre Dame, IN 46556. E-mail: tral value at a given nondimensional 共reduced兲 frequency,
[email protected] negating potential errors associated with interpolation or curve
Note. Associate Editor: Kurtis R. Gurley. Discussion open until fitting of spectral data. However, the structure of this prototype
December 1, 2008. Separate discussions must be submitted for individual site was rather archaic in light of recent advances in data man-
papers. To extend the closing date by one month, a written request must
agement and mining. This paper discusses the use of advances in
be filed with the ASCE Managing Editor. The manuscript for this paper
was submitted for review and possible publication on December 1, 2006; the field of information technology to enhance, for the purposes
approved on December 2, 2007. This paper is part of the Journal of of analysis and design, the accessibility, organization, dissemina-
Structural Engineering, Vol. 134, No. 7, July 1, 2008. ©ASCE, ISSN tion, and utility of Web-archived wind tunnel data. The recently
0733-9445/2008/7-1139–1153/$25.00. redesigned NALD v. 2.0 serves as an example of the application

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1139

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
of these new technologies. These changes were achieved using a veniently quantifying generalized wind forces on tall buildings
combination of Web-based programming tools and popular engi- with uncoupled mode shapes 共Kareem and Cermak 1979; Tschanz
neering software, e.g., Apache Web servers, JAVA/JavaScript, hy- and Davenport 1983; Reinhold and Kareem 1986; Boggs and Pe-
pertext preprocessors 共PHP兲, structured query language databases terka 1989兲. The generalized forces are then utilized for estimat-
共MySQL兲, and MATLAB. The revised site offers more attractive ing building response with given structural characteristics. The
and user-friendly features to allow not only the retrieval of power HFBB technique generally requires mode shape corrections,
spectral values at specific reduced frequencies, but also the on- which are either based on empirical relationships or analytical
line determination of resulting base moments, displacements, and formulations derived on the basis of assumed wind loading mod-
equivalent static wind loads 共ESWL兲 for survivability and accel- els 共Vickery et al. 1985; Boggs and Peterka 1989; Xu and Kwok
erations for serviceability 共habitability兲 considerations. Thus, in 1993; Zhou et al. 2002; Holmes et al. 2003; Chen and Kareem
NALD v. 2.0, a dual purpose design aid is introduced: A database- 2004, 2005兲.
driven Web archive of HFBB data and a stand-alone analysis Since its inception a few decades back at Shimizu Corpora-
engine that can be used independently or in tandem for estimating tion’s wind tunnel laboratory 共Fujii et al. 1986; Kikuchi et al.
ESWL and building dynamic responses through a user-friendly 1997兲, synchronous pressure measurements 共SPM兲 on building
analysis interface. The latter feature will be particularly useful for surfaces have been increasingly implemented in wind tunnel prac-
those who may not be very familiar with the details of the random tice. This was largely facilitated by the availability of cheaper
vibration-based dynamic analysis procedure generally used in electronic pressure sensors and represented an advancement over
connection with HFBB measurements. the covariance-based integration methodology that involved sev-
eral configurations of limited pressure measurements over a
building surface 共Kareem 1982兲. SPM offers the added advantage
of providing more accurate estimates of generalized wind loads
Research to e-Analysis for buildings with nonlinear mode shapes, as approximate mode
shape corrections are not required. Nonetheless, the HFBB main-
Measurement of forces using HFBB and synchronous scanning of tains its attractiveness in cases where the mode shapes do not
pressures have become widely accepted techniques for wind tun- depart too far from linear.
nel studies of buildings and other structures. The translation of Individual researchers 共Chen and Kareem 2005; Huang and
wind tunnel data into ESWL and building response involves a Chen 2007兲 and wind tunnel laboratories 共Steckley et al. 1992;
random vibration-based analysis. Most wind tunnel study reports Ho et al. 1999兲 have their own favorite analysis format based on
do not provide details of this process with the exception of a either SPM or HFBB. Some groups 共Chen and Kareem 2004,
generic description in an appendix or a cited reference in the 2005兲 prefer to establish equivalent static wind loads from either
report. This practice has left designers largely in the dark sur- SPM or HFBB data for subsequent response analysis, while oth-
rounding the theory employed and completely unaware of the ers directly employ the data for calculating response components
many published advances in the procedures for predicting wind- 共Steckley et al. 1992; Tamura et al. 1996; Ho et al. 1999; Fritz
induced response. As such, they are generally not capable of re- and Simiu 2005兲.
peating these analyses in house for parametric investigations of The NALD consists of results from 162 different tests, derived
period and damping sensitivity that are essential when mitigation from nine cross-sectional shapes, three model heights, two expo-
of wind-induced motion is required. Instead, design offices often sure categories, and three response directions 共alongwind, across-
have to engage either a testing laboratory or an external expert to wind, and torsion兲, as shown by the NALD Web selection menu
conduct these additional parameter studies. To prevent these in Fig. 1. While a detailed description of the test procedures can
analysis procedures from languishing on the library shelves, the be found in Kareem 共1990兲, Kijewski and Kareem 共1998兲, and
NatHaz Modeling and DYNAMO laboratories at the University Zhou et al. 共2003兲, a brief summary is now provided. Each of the
of Notre Dame have mobilized their technology transfer using balsa wood models was tested in a boundary layer wind tunnel
information technologies. In this context, this paper chronicles the with a 3 m 共10 ft兲 ⫻ 1.5 m 共5 ft兲 cross section, of 18 m 共60 ft兲
development of an analysis portal that encompasses necessary length. The turbulent boundary layers simulated in this study were
features of random vibration analysis to predict building response generated by the natural action of surface roughness added on the
based on wind tunnel derived data, existing databases, or estab- tunnel floor and upstream spires. Two typical boundary layers
lished expressions for spectral loading, which does not require were simulated in this experiment, BL1 共␣ = 0.16, where
prior working knowledge of the subject by the user. First, a short ␣⫽power law exponent of the mean wind velocity profile兲 and
history of this development is presented, which is followed by the BL2 共␣ = 0.35兲, similar to the conditions of open 关Exposure C in
latest developments. the ASCE 7-05 共ASCE 2005兲兴 and urban 关Exposure A in ASCE
7-98 共ASCE 1998兲兴 flow environments, respectively. The output
NatHaz Aerodynamic Loads Database of the sensitive, multicomponent base balance was analyzed using
the fast Fourier transform 共FFT兲 to determine the spectral and
Aerodynamic loads on buildings may be derived through multiple cross-spectral density functions, which were later nondimension-
point synchronous scanning of pressures or by measured forces alized. This analysis was carried out for all 27 building models, in
on the model mounted on a high-frequency base balance. The both boundary layers, and at various angles of wind incidence,
simultaneously monitored pressure database offers great flexibil- though only the results from perpendicular approaching winds
ity in deriving mode generalized loads for buildings with mode 共zero degree angle of attack兲 were considered in the NALD v. 1.0.
shapes that depart from linear or exhibit coupling. However, for The authors plan to augment the 162 test cases currently housed
tall buildings with dominant resonant response, both the mean in the NALD with data for other building shapes and aspect ra-
and background components can be approximately quantified by tios, as they become available from other researchers and/or ad-
modal analysis using integrated wind loads derived from HFBB. ditional testing.
The HFBB measurements have been widely recognized for con- The reliability of the measured spectra within the NALD has

1140 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 1. NALD data selection menu

been established through verifications against datasets from other in this paper offer additional verification of NALD against se-
wind tunnel experiments. For example, the acrosswind spectra lected major studies, though these are by no means exhaustive or
have been compared to a model derived from earlier measure- meant to serve as a systematic comparison of HFBB data from
ments by Kareem 共1990兲. Results in the torsional direction were different laboratories, codes, and standards.
also compared to those derived from pneumatic averaging, to
overcome the uniform mode shape assumption inherent to the
HFBB-derived torsional loads 共Kareem 1990兲. More recently, Overview of NALD v. 2.0: from Theory to Practice
Zhou et al. 共2003兲 compared the NALD acrosswind loads with the
empirical expression suggested by the Architectural Institute of
Theoretical Background of NALD v. 2.0
Japan 共AIJ 1996; Tamura et al. 1996兲. In addition, nondimension-
alized base moment coefficients were compared to the empirical To account for the gustiness of turbulent boundary-layer winds on
expressions given by AIJ for acrosswind and torsional directions structures, most international codes and standards including
共Zhou et al. 2003兲. ASCE 7 have adopted the concept of gust loading factor 共GLF兲,
Since these previously reported comparisons, a number of new which was first introduced by Davenport 共1967兲 based on statis-
studies concerning HFBB and SPM have been published 共Liang tical theory of buffeting. This traditional GLF is based on the ratio
et al. 2002, 2004; Cheng and Wang 2004; Gu and Quan 2004; Ha of the maximum structural displacement to the mean displace-
et al. 2004; Lin et al. 2005; Flay and Bhat 2005兲. In particular, it ment 共Davenport 1967; Solari and Kareem 1998兲. Although the
is worth noting that Lin et al. 共2005兲 have provided an in-depth traditional GLF ensures an accurate estimation of the displace-
comparison of the NALD to their HFBB and SPM. They found ment response, it may fall short in providing a reliable estimate of
the NALD to be in close agreement with their studies with the other response components. To overcome this shortcoming, Zhou
exception of a few cases, stating: “With the linear mode shape and Kareem 共2001兲 proposed a new GLF format that is based on
assumption . . . integrated simultaneous point pressures and the ratio between the maximum base bending moment and the
HFBB agree for base force and moment spectra. The 关NALD兴 mean obtained from HFBB experiments, rather than the displace-
effectively provides the base moment spectra for preliminary de- ments utilized in the conventional approach.
sign and can be expanded on the Internet by the dataset here and This new GLF format associated with base moments has been
by the other experimental results in the future” 共Lin et al. 2005兲. introduced in ASCE 7-05 共ASCE 2005兲 as well as the AIJ 共2004兲
This speaks not only to the reliability of the NALD, but also the Recommendations for Loads on Buildings. Using the aerody-
robustness of its framework for future expansion. The examples namic base bending moment or base torque as the input, the

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1141

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
wind-induced response of a building can be computed using ran- gust loading factor approach 共Zhou and Kareem 2001; Kareem
dom vibration analysis as detailed in Zhou and Kareem 共2001兲. and Zhou 2003兲.
Utilizing the base bending moment, NALD v. 2.0 assists in evalu- Although the theoretical background adopted in NALD v. 2.0
ating the equivalent static wind loads and attendant response com- has been introduced in Zhou and Kareem 共2001兲, Zhou et al.
ponents. Due to relatively less sensitivity of the base moment to 共2003兲, Kareem and Zhou 共2003兲, and Tamura et al. 共2005兲, it is
mode shapes, the mode shape correction may not be necessary in briefly described here for completeness. Assuming the response is
this approach. Application of this framework for the alongwind a stationary Gaussian process, the expected maximum base bend-
response has proven effective in recasting the traditional gust ing moment response 共M̂兲 in the alongwind and acrosswind di-
loading factor approach into a new format. This procedure has rections or the base torque response can be expressed in the
been extended to the acrosswind and torsional response in a 3D following form:

M̂ = M̄ + g ⫻ ␴ M ⬇ M̄ + 冑M 2B + M 2R = M̄ + 冑 冉
共gB ⫻ ␴CM ⫻ M̄ ⬘兲 + gR ⫻ ␴CM ⫻ M̄ ⬘ ⫻
2 冑 ␲
4␨1
C M 共f r1兲 冊 2

共1兲

where M̄⫽mean moment; M B, M R⫽background and resonant

base moment or torque components, respectively; g, gB, gR⫽peak
1
P̄共z兲 = ␳ŪH2
2
z
H
冉冊 2␣
BCD⌬H = M̄
2 + 2␣ z
H2 H
冉冊 2␣
⌬H 共4兲
factors for total, background, and resonant moments, respect-
ively; ␴ M , ␴CM⫽RMS of the fluctuating base moment/torque Next, the background component for the alongwind and across-
response and base moment/torque response coefficient wind responses can be obtained by using the background GLF as
follows:
共=␴ M / M̄ ⬘兲; M̄ ⬘⫽reference moment or torque depending on

冉冊
response component; ␨1⫽building damping ratio in the first 2␣
mode; C M 共f r1兲⫽nondimensional moment coefficient at f r1 2 + 2␣ z
PB共D,L兲共z兲 = GMB共D,L兲 ⫻ P̄共z兲 = M B共D,L兲 ⌬H 共5兲
共=f r1 ⫻ S M 共f r1兲 / ␴2M 兲;f r1⫽reduced frequency according to f 1 H2 H
共=f 1B / ŪH兲; f 1⫽natural frequency of building in the direction of Similarly, the background component for the torsional response
motion; S M 共f兲⫽PSD of the fluctuating base moment or torque 共PB共T兲兲 is expressed as
response; f⫽frequency 关Hz兴; ŪH⫽mean wind velocity evaluated
at building height H. In addition, since ␴CM and C M 共f r1兲 are ob-
tained from the HFBB experiment, the mean, background, and
resonant base moments can be computed in the alongwind,
PB共T兲共z兲 = GMB共T兲 ⫻ P̄共z兲 = M B共T兲
1 + 2␣ z
H H
冉冊 2␣
⌬H 共6兲

acrosswind, and torsional directions using respective building where subscripts B, D, L, and T⫽background, alongwind, across-
properties. This has led to the introduction of a 3D GLF approach wind, and torsional components; ␳⫽air density; z⫽elevation
to facilitate evaluation of response in three directions 共Kareem above the ground; B⫽building width; CD⫽drag force coefficient;
and Zhou 2003兲. The gust loading factor G M associated with base ⌬H⫽floor-to-floor height of building; ␣⫽exponent of mean wind
moment can be described as the following form: speed profile defined in ASCE 7.
For the resonant components, the ESWL in sway modes is
given by
G M = M̂/M̄ ⬘ = Ḡ + 冑GMB
2 2
+ GMR 共2兲
m共z兲␸1共D,L兲
PR共D,L兲共z兲 = M R共D,L兲 共7兲
Thus, mean 共Ḡ兲, background 共GMB兲, and resonant 共GMR兲 GLF can 兺 m共z兲z␸1共D,L兲
be easily derived by comparing Eq. 共2兲 to Eq. 共1兲 共Kareem and
and in the torsional mode
Zhou 2003兲. Using Eqs. 共1兲 and 共2兲, the ESWL on a building in
the alongwind, acrosswind, and torsional directions can be com-
puted by distributing the base moments to each floor akin to the I共z兲␸1共T兲
PR共T兲共z兲 = M R共T兲 共8兲
manner in which base shear is distributed in earthquake engineer- 兺 I共z兲␸1共T兲
ing. The mean base moment 共M̄兲 has a relationship with the mean
component of the ESWL as follows: where subscript R⫽resonant component; m共z兲⫽mass per unit
height; ␸1⫽fundamental mode shape in the direction of motion
共=共z / H兲␤兲; ␤⫽mode shape exponent in the direction of motion,

冕 e.g., linear mode shape if ␤ = 1; I共z兲⫽mass moment of inertia per

z
M̄ = P̄共z兲 ⫻ zdz 共3兲 unit height 共=m共z兲 ⫻ ␥2兲; ␥⫽radius of gyration.
0
For the acceleration response, only the resonant component is
of interest. The peak accelerations for the three principle direc-
tions of motion, i.e., alongwind, acrosswind, and torsion, can be
where mean component of the ESWL 共P̄兲 is obtained by the following equations:

1142 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
• Alongwind and acrosswind 1
Y B共L兲共z兲 = ␳ ⫻ ŪH2 ⫻ D ⫻ gB ⫻ ␴CM共L兲
P* 2

冉冊
R共D,L兲
Ÿ Peak共D,L兲共z兲 = ␸1共D,L兲 ⫻ 共2␲f 1兲2 共2 + 2␣兲共2␤ + 1兲 z ␤
*
K共D,L兲 ⫻ 2 ⫻
m共2␣ + ␤ + 1兲共2␲f 1兲 H

冕
H
PR共D,L兲共z兲␸1共D,L兲dz
0
␸1共D,L兲 共9兲
冕
= H 1
2 Y R共L兲共z兲 = ␳ ⫻ ŪH2 ⫻ D ⫻ gR共L兲 ⫻ ␴CM共L兲
m共z兲␸1共D,L兲 dz 2

冑 冉冊
0
␤
␲ 共␤ + 2兲 z
• Torsion ⫻ C M共L兲共f r1兲 ⫻ 2 ⫻ 共13兲
4␵1 m共2␲f 1兲 H

冕
H
PR共T兲共z兲␸1共T兲dz Note that all parameters in Eqs. 共11兲–共13兲 are related to across-
*
PR共T兲 0 wind properties, e.g., f 1 here is natural frequency of building in
Ÿ Peak共T兲共z兲 = ␸1共T兲 ⫻ 共2␲f 1兲2 = ␸1共T兲
冕 the acrosswind direction. Note that the displacement response is
* H
K共T兲 2
I共z兲␸1共T兲 dz dictated by 50-year wind speeds, as this is the mean recurrence
0 interval 共MRI兲 for base moments and the ESWL 共survivability
共10兲 design兲, while the acceleration response is governed by the
10-year wind speed 共serviceability design兲.
where P*⫽generalized force; K*⫽generalized stiffness. The re-
sulting RMS acceleration can then be determined by dividing the
peak accelerations by the resonant peak factor gR. Note that the Database-Enabled Selection
angular accelerations in torsion may be resolved into the resultant
alongwind and acrosswind components at the corner of the build- NALD v. 1.0 共Zhou et al. 2003兲 provided users with wind tunnel
ing, and these lateral accelerations induced by torsion can be measurements of RMS base moment coefficients and the nondi-
combined with those generated by the sway motions to obtain the mensional power spectral values requisite for the above response
total lateral accelerations at the corner by the square root of the calculations for the 162 tests discussed previously. Upon entering
sum of the squares 共SRSS兲 or complete quadratic combination the database, the user stepped through a series of hypertext
共CQC兲 共Zhou et al. 2003; Chen and Kareem 2004, 2005兲. markup language 共HTML兲 links to identify the data of interest.
The displacement response calculation can be computed by a Once the desired test case and response component were selected,
modal analysis procedure. Assuming that building mass is uni- a JAVA applet retrieved the exact nondimensionalized power
formly distributed along the height, i.e., mass per unit height spectral value corresponding to a user-specified reduced fre-
关m共z兲兴 being a constant value 共m兲, the mean and maximum dis- quency. This automated process negates potential human errors
placements in the alongwind can be computed by the following that result from picking off values from hardcopy spectra and
expressions: eliminates the uncertainty associated with curve-fit expressions

冉冊
* that tend to generalize spectral features. However, since the
Pmean共D兲 ␤
共2␤ + 1兲M̄ z NALD v. 1.0 could not support structured query language 共SQL兲,
Y mean共D兲共z兲 = ␸1 =
*
K共D兲 mH2共2␲f 1兲2 H the architecture associated with this prototype involved an expan-
sive hierarchy of directories with duplicate HTML files, requiring
Y max共D兲共z兲 = G M ⫻ Y mean共D兲共z兲 共11兲 the user to step through a sequence of at least five Web pages to
reach the desired JAVA applet.
Similarly, the maximum displacement in the acrosswind direc- To reduce the redundancy in the architecture, several Web-
tion is computed by only including background and resonant dis- based tools were utilized in NALD v. 2.0, now hosted by a dedi-
placements, since there is no mean displacement in this direction cated Apache Web server available to the public at http://

冉冊
* aerodata.ce.nd.edu. This hardware change now permits the use of
PB共L兲 M B共L兲共2␤ + 1兲共2 + 2␣兲z ␤
PHP, a kind of common gateway interface 共CGI兲 language, and
Y B共L兲共z兲 = ␸1 =
*
K共L兲 mH 共2␲f 1兲 共2␣ + ␤ + 1兲 H
2 2
MySQL for a database-oriented query to specify the desired test
data, replacing the archaic and sequential HTML structure of the

冉冊
* original site. This speeds the retrieval time and dramatically re-
PR共L兲 M R共L兲共␤ + 2兲 z ␤
Y R共L兲共z兲 = ␸1 = duces the number of HTML files, directories, and total file sizes
*
K共L兲 mH2共2␲f 1兲2 H by eliminating unnecessary redundancies on the server. It also
provides inherent scalability so the data archives can be readily
Y max共L兲共z兲 = 冑Y B共L兲
expanded. The new user-friendly interface was shown in Fig. 1
2
共z兲 + Y R共L兲
2
共z兲 共12兲
and allows the selection of a desired test case in only one step,
Alternatively, if the RMS moment coefficient 共␴CM共L兲兲 and nondi- which is then followed by the launch of the appropriate JAVA
mensional moment coefficient 关C M共L兲共f兲兴 in the acrosswind direc- applet from NALD v. 1.0 共Zhou et al. 2003兲, with the option for
tion, which can be obtained from NALD as well, are known for downloading data for further off-line analysis. It is worth noting
given building properties, the background and resonant displace- that NALD v. 2.0 has been introduced in the commentary of
ments in the acrosswind direction can be obtained from the fol- ASCE 7-05 关C6.5.8兴 共ASCE 2005兲 as an alternative means to
lowing expressions in which Eq. 共12兲 is expanded by using M B共L兲 assess the dynamic wind-induced loads on typical isolated build-
and M R共L兲 关see Eq. 共1兲兴: ings in the preliminary design stages.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1143

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 2. NALD v. 2.0 data selection menu for on-line analysis, including PSD option

On-Line Analysis of Wind Loads and Response category, and fundamental dynamic characteristics 共Fig. 3兲. Either
metric 共SI兲 or English units may be specified for the structural
In NALD v. 1.0, users would retrieve relevant spectral properties
for a given test case and then manually perform off-line calcula- inputs and calculated outputs. In addition, an on-line calculator is
tions to obtain the building base bending moments, ESWL, and provided for user-friendly unit conversion 共Fig. 4兲. It should be
accelerations based on the equations introduced previously 共Zhou noted that ASCE 7 recommends a 50-year mean recurrence wind
et al. 2003兲. To minimize the calculations required on the part of that is used in survivability design, e.g., ESWL evaluation,
the end user, an on-line analysis module was developed utilizing whereas, in serviceability design, a building’s acceleration is gen-
the theory presented in the previous section to supplement the erally based on a 10-year mean recurrence wind. Thus, it is re-
existing JAVA interface and provide these and other response quired to include a MRI factor to convert 50-year winds into
quantities automatically. 10-year winds for serviceability design. For convenience, wind
The new user interface developed for on-line analysis is shown speeds for both survivability 共50-year MRI兲 and serviceability
in Fig. 2. It is similar to the reorganized selection menu 共Fig. 1兲, 共10-year MRI兲 in the exposure of interest are calculated on-the-fly
but with additional options for specifying the input power spectral in NALD v. 2.0 共Fig. 5兲 based on the relationships in ASCE 7-05
density 共PSD兲. At present, three user options are available for 共ASCE 2005兲 utilizing the user-specified 3-sec gust 50-year ref-
prescribing a PSD for the analysis: PSD data from the NALD erence wind speed 共U10兲 in open terrain 共Fig. 3兲. Nondimensional
共default option兲, a user-specified PSD 共curve-fitted or analytical spectral values 关C M 共f兲兴 are then calculated on-the-fly for all di-
expression兲 or user-supplied PSD data 共X, Y data pairs兲. The user rections and mean recurrence intervals 共Fig. 5兲. Thus, the JAVA
selections are handled by a combination of PHP and MySQL as applets are no longer required in this new on-line analysis
inputs for the next stage in the process. After selecting these basic module.
inputs, the module requests additional inputs for the full-scale MATLAB provides an attractive programming framework for
system, including cross-sectional dimensions, height, exposure more complicated computations and can be easily extended to

1144 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 3. Interface for user-supplied structural inputs

Fig. 4. On-line unit conversion module

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1145

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Fig. 5. NALD v. 2.0 output from on-the-fly calculation of wind
speeds and nondimensional moment coefficients

more sophisticated numerical calculations due to many predefined

function capabilities. For these reasons, MATLAB 共version 6.5,
R13兲 is used as the computational framework for this study, and
the MATLAB Web server tool is internally utilized to supply user
inputs to the server-side MATLAB analysis. A MATLAB code,
running on the NALD server, determines the base moment/torque,
the structural displacements and accelerations, in addition, to the
ESWL for the mean, background, and the resonant components.
The following quantities are then displayed on the Web portal:
Nondimensional spectral base moment 共Fig. 6兲, RMS base mo-
ment coefficient, nondimensional moment coefficient, base mo-
ment and the maximum lateral displacements for survivability
design, 10-year RMS and peak lateral accelerations, correspond-
ing lateral accelerations induced by torsion, and total lateral ac-
celerations at the corner. All displacements and accelerations are
calculated at the roof level. All these quantities, i.e., base bending
moments, displacements, and accelerations are displayed for each
of the three response components 共Fig. 7兲. Finally, a plot of the
mean, background, and resonant components of the ESWL on the Fig. 7. Display of on-the-fly calculated survivability and
building are displayed for the end user, as shown in Fig. 8. An serviceability values
option is also available to download this information as a text file
for further off-line analysis and possible application to an existing
structural finite element model or a spreadsheet-based building The architecture of NALD v. 2.0 and the role of various Web-
analysis. As such, the NALD v. 2.0 can also be used to express based tools such as HTML/JAVA Script and PHP are summarized
wind loads, i.e., the ESWL, in three directions in terms of 3D gust in Fig. 9. It is basically operated by Apache Web server with two
loading factors, akin to the alongwind GLF 共Kareem and Zhou main processes, i.e., external process and internal process. The
2003; Tamura et al. 2005兲. external process includes user-friendly interfaces for the selection
of a desired analysis case in Fig. 2 共interface 1兲, additional inter-
faces for design inputs such as structural parameters of building
in Fig. 3 共interface 2兲, and display of analysis results for the
user-specified building. On the other hand, the internal processes
are server-side operations involving MySQL for database opera-
tions and MATLAB Web servers for the computational schemes
implicitly utilized in NALD v. 2.0. The MySQL database server
handles data transmissions between interfaces and if necessary,
transmits information stored in the database. The MATLAB Web
server functions as a numerical analysis engine for on-the-fly cal-
culations, as well as serving as the nexus between interface 2 and
the design results. The on-line analysis module performs the req-
uisite calculations and then generates meaningful figures such as
the nondimensional PSD and the ESWL, as well as the ESWL
text file.
It should again be emphasized that one unique feature pro-
vided by this on-line analysis module is the user’s nondimen-
sional PSD options. As mentioned earlier, NALD v. 2.0 provides
Fig. 6. Nondimensional base moment spectra display the user with three PSD options. Thus, the user can utilize not

1146 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 8. Display of on-the-fly calculated wind force components

only the PSD data offered by the NALD, but also any arbitrary face for user-supplied PSD data sets of X 共reduced frequency兲 and
PSD expression or data set for the on-line determination of wind Y 共nondimensional base moment PSD兲 pairs. Since this on-line
load effects on high-rise buildings. Depending on the selected analysis module mainly utilizes MATLAB, the aforementioned
PSD option specified in Fig. 2, additional inputs will be requested inputs should be MATLAB compatible. The “info link” displayed
following the prompt for structural inputs 共interface 2 in Fig. 3兲. in the top line provides the user with simple guidelines to mini-
Fig. 10共a兲 shows the supplemental interface for user-supplied mize unexpected input errors. The on-line analysis module also
PSD expressions, while Fig. 10共b兲 shows the supplemental inter- includes a simple error-detection scheme with pop-up error mes-

Fig. 9. Diagram of NALD v. 2.0 architecture

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1147

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 10. Supplemental interfaces for user-specified PSD with examples: 共a兲 interface for PSD expression; 共b兲 interface for PSD data

sages, which alert users if input values are beyond the NALD’s completeness and as a demonstration of the type of data an end
range of applicability. The provision for PSD options extends the user must input to analysis module: Building dimension perpen-
utility of NALD v. 2.0 beyond its predecessor by offering an dicular to oncoming wind B = 40 m; building dimension parallel
on-line dynamic analysis framework that can be utilized for esti- to oncoming wind D = 40 m; building height H = 200 m; natural
mating dynamic load effects on high-rise buildings with alterna- frequency in alongwind, acrosswind, and torsional directions, re-
tive input options, e.g., an independent wind tunnel study or spectively, f x = 0.2 Hz; f y = 0.2 Hz; f ␪ = 0.35 Hz; bulk density ␳B
empirical expression from any wind load standard. This versatil- = 250 kg/ m3; average radius of gyration ␥ = 18 m; damping ratio
ity provides users with a robust stand-alone, on-line analysis en- ␨ = 0.02; interstory height ⌬H = 4 m; air density ␳A = 1.25 kg/ m3;
gine that offers the flexibility of utilizing user-supplied custom drag force coefficient CD = 1.3; 3-sec reference wind speed at
spectral description or wind tunnel test 共HFBB兲 results to provide 10 m U10 = 63 m / s 共50-year MRI兲; mode shapes for all directions
final design estimates of wind load effects on buildings and per- are assumed to be linear and the building is assumed to be located
mit comparative studies of predictions from various sources. in an urban area. Thus, the NALD model best suited to this analy-
sis is: Shape 4 共D / B = 1兲, height= 20 in. 共H / 冑BD = 5兲 in terrain
category BL2 共Exposure A兲 共see Fig. 1兲. Based on these input
Examples conditions, analyses employing different PSD options are per-
formed to demonstrate the capabilities of the on-line analysis
The example building and wind environment used in Zhou et al. module. An additional example 共Example 4兲 is reported for a
共2003兲 are utilized again in this study for consistency. Note that building used extensively in comparative studies and also in the
NALD v. 1.0 provided users with the spectral amplitude at a Commentary for the Australian Standard 共Holmes et al. 1990兲.
specified reduced frequency only; thus, it was required to perform
manual calculation of the desired response components such as
Example 1—NALD Experimental Data „PSD Option 1…
base moments and accelerations of building following the proce-
dure provided on the NALD Web site 共Zhou et al. 2003兲. The new For the previously specified reference wind speed, the 50-year
version conducts all computations, including the ESWL calcula- and 10-year wind speeds at the building height are determined to
tions in all directions, automatically via the on-line analysis mod- be 51.30 m / s and 37.96 m / s, respectively, for survivability and
ule. The example building characteristics are summarized here for serviceability response estimates 共Zhou et al. 2003兲. The corre-

1148 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 Fig. 11. An example of error message for user-supplied inputs

sponding alongwind, acrosswind, and torsional loading spectra were the motivating factors behind the on-line database approach
based on the NALD experimental data were shown previously in represented by NALD v. 1.0. Despite the dependence on so many
Fig. 6 and the display of the corresponding response estimates variables, the NALD result shows relatively good agreement with
automatically computed by the on-line analysis module were pre- both empirical expressions with the exception of discrepancies in
viously demonstrated in Fig. 7. These include: Mean, peak back- the low-frequency range, which are not of concern given the lack
ground, peak resonant, and total peak base moments, and the of their practical significance for typical high-rise buildings. Note
maximum alongwind and acrosswind displacements at the roof also the high-frequency details lost in the empirical expressions.
level for survivability design, and peak and RMS accelerations at To perform the on-line analysis, the reduced frequency range
the roof level for serviceability design. The moment and accelera- of the NALD experimental data is imposed on the two acrosswind
tion values match those manually computed by Zhou et al. PSD expressions, i.e., f s = 0.0019; f e = 0.43; ⌬f = 0.0001. The RMS
共2003兲. The output also includes distributions of the equivalent base bending moment coefficients 共␴CM兲 are automatically calcu-
static wind load components: Mean, background, and resonant, lated from the respective empirical expressions 共Appendix兲. For
for all response directions, as shown previously in Fig. 8. These demonstrative purposes, the specifications of these empirical PSD
load distributions can be downloaded by the user for incorpora- expressions in a MATLAB compatible format are listed below:
tion into models developed using various commercial software AIJ 共1996兲
packages to allow for further analysis and design of structural
members. This exercise reaffirms that the real-time analysis mod- 4 * 0.85 * 共1 + 0.6 * 0.1688兲 * 0.1688/pi * 共f/0.0901兲∧2/
ule provides response estimates that are consistent with manual
共共1 − 共f/0.0901兲∧2兲∧2 + 4 * 0.1688∧2 * 共f/0.0901兲∧2兲
calculations presented previously by Zhou et al. 共2003兲.
AIJ 共2004兲
Example 2—User’s PSD Expression „PSD Option 2…
4 * 0.85 * 共1 + 0.6 * 0.2806兲 * 0.2806/pi * 共f/0.0901兲∧2/
As mentioned earlier, NALD v. 2.0 provides the user with an
opportunity to utilize the various types of empirical PSD expres- 共共1 − 共f/0.0901兲∧2兲∧2 + 4 * 0.2806∧2 * 共f/0.0901兲∧2兲
sions available. This permits comparative analyses to demonstrate
the impacts of generalized spectral expressions versus precise
spectral values drawn directly from PSDs of HFBB data. In this
example, acrosswind PSD expressions specified by AIJ 共1996,
2004兲 and Gu and Quan 共2004兲, detailed in the Appendix, are
considered.
The requisite inputs for this option were shown previously in
Fig. 10共a兲: C M 共f兲, S M 共f兲, ␴ M , in a MATLAB compatible format,
the reduced frequency range 共f s to f e兲, reduced frequency interval
共⌬f兲, and ␴CM. If the user leaves blank共s兲 for any loading direc-
tion, the analysis will default to the NALD experimental data for
that direction, and an error message will be displayed, as shown
in Fig. 11. Based on the aforementioned example parameters, a
comparison between the NALD v. 2.0 experimental PSD data and
other two aforementioned empirical PSD expressions was shown
in Fig. 12. It should be noted that AIJ 共1996, 2004兲 empirical
expressions are not a function of boundary layer condition, terrain
category, and building height, but are expressed mainly as a func-
tion of the side ratio 共D / B兲, whereas Gu and Quan 共2004兲 incor-
porate the preceding attributes in their empirical expression 共see
the Appendix兲. This demonstrates a major drawback of empirical
expressions: The need to incorporate an exhaustive set of vari- Fig. 12. Comparison of acrosswind spectra with empirical
ables in the expression in order to fully encompass various struc- expressions of AIJ 共1996, 2004兲 and Gu and Quan 共2004兲
tural and flow features influencing response. Such considerations 共D / B = 1.0, Exposure A兲

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1149

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Table 1. Acrosswind Analysis Results for Empirical PSD Expressions in
Example 2 共D / B = 1.0, H / 冑BD = 5兲
RMS Base bending
acceleration moment
␴CM 关mg兴 关106 kN m兴
NALD v. 2.0 0.1330 6.21 3.830
AIJ 共1996兲 0.1572 6.27 3.925
AIJ 共2004兲 0.1572 8.11 4.771
Gu and Quan 共2004兲 0.2122 7.34 4.692

Gu and Quan 共2004兲

1/0.2122∧2 * 共0.0396 * 0.1990 * 共f/0.0897兲∧1.8698兲/

共共1 − 共f/0.0897兲∧2兲∧2 + 0.1990 * 共f/0.0897兲∧2兲

The resulting RMS base bending moment coefficient 共␴CM兲,

total base moments 共M total兲, and RMS accelerations 共aRMS兲 de-
rived from the four different PSDs 关NALD default; user-specified
AIJ 共1996兲, AIJ 共2004兲 and Gu and Quan 共2004兲兴 are summarized
in Table 1. A quick review of the results suggests that those based
on the NALD experimental data and AIJ 共1996兲 compare well
with one another. The same can be said for the results based on
AIJ 共2004兲 and Gu and Quan 共2004兲, which take on slightly larger
values than the former pair. It is important to note that RMS base
bending moment coefficient 共␴CM兲 is quite sensitive to the ap-
proach flow characteristics. Thus, it becomes evident that reliable
estimates of the RMS coefficient and the spectral amplitude are
critical to the accurate evaluation of aerodynamic load informa-
tion. Though empirical fits to experimental data, such as those in
AIJ 共1996, 2004兲, provide compact representations for use in
codes and standards, they cannot accurately represent experimen-
tal data for all possible building configurations and flow condi-
tions, again motivating the on-line database philosophy of NALD
v. 2.0. However, in light of these factors, the results are in rea-
sonable agreement. Furthermore, it is also demonstrated here that Fig. 13. Comparison of alongwind and acrosswind between NALD
the on-line analysis module works effectively for user-supplied and Cheng and Wang 共2004兲 共D / B = 1.0, H / 冑BD = 5, Exposure C兲
PSD expressions.

Example 3—User-Specified PSD Data „PSD Option 3…

as Y values. Should the user inputs be in error, the user is alerted
Another PSD option offered by NALD v. 2.0 is the user-specified and NALD PSD data will be retrieved by default, as shown pre-
PSD data in the form of X, Y pairs. The prompt for this data viously by the error pop up in Fig. 11.
allows the user to implement his/her own PSD data derived from Fig. 10共b兲 shows an example of this PSD option, using the
a wind tunnel experiment. As shown previously in Fig. 10共b兲, download of one of the test cases archived in NALD v. 2.0. As
three inputs are requested in each direction: RMS base bending expected, the results exactly replicate the results obtained using
moment coefficient 共␴CM兲, X coordinate 关reduced frequency, PSD option 1 共Example 1兲 and presented by Zhou et al. 共2003兲
f ⫻ B / UH兲, and Y coordinate 共nondimensional power spectrum and, thus, are not repeated here. This demonstrates the accuracy
C M 共f兲兴. All X and Y input should be separated by a comma 共,兲 or of an on-line module in evaluating building response based on
single space, and the total number of X values should be the same user-specified spectral data values. Another example utilizing the
data by Cheng and Wang 共2004兲 共PSD option 3兲 is also compared
to the NALD experimental data for the case of D / B
= 1.0, H / 冑BD = 5 under BL1 共Exposure C兲 for the alongwind and
Table 2. Design Results for Data from NALD v. 2.0 and Cheng and
Wang 共2004兲 共D / B = 1.0, H / 冑BD = 5兲
acrosswind directions. Table 2 summarizes the resulting base
Alongwind Acrosswind bending moments and RMS accelerations. Since the NALD spec-
Base bending RMS Base bending RMS
tra and Cheng and Wang 共2004兲 data show a good agreement in
moment acceleration moment acceleration the alongwind and acrosswind directions, as shown in Figs. 13共a
关106 kN m兴 关mg兴 关106 kN m兴 关mg兴 and b兲, it is obvious that the response quantities correspondingly
show a good agreement in Table 2.
NALD v. 2.0 4.818 5.69 6.388 11.73 These last two examples demonstrate the utility of the various
Cheng and 5.158 6.29 6.639 11.17 PSD input options in this on-line analysis, providing the user with
Wang 共2004兲
the versatility to perform an automated on-line analysis of wind

1150 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Table 3. Comparison of NALD v. 2.0 with AS 1170.2 and ASCE 7-05 for CAARC Building
NALD v. 2.0
a
ASCE
AS exposure Exposure Exposure
Responses 1170.2 B Averageb C A
Case 1 Alongwind Peak 8.24c 6.97 8.60 7.67 9.52
acceleration
关mg兴
Base moment 2.80 2.42 2.31 2.16 2.46
关106 kN m兴
Acrosswind Peak 15.53c —d 11.52 11.22 11.81
acceleration
关mg兴
Base moment 1.30 —d 1.47 1.11 1.83
关10 kN m兴
6

Case 2 Alongwind Peak 5.47c 4.65 5.88 5.07e 6.69e

acceleration
关mg兴
Base moment 1.68 1.51 1.32 1.22 1.41
关106 kN m兴
Acrosswind Peak 17.44c —d 11.83 11.04e 12.62e
acceleration
关mg兴
Base moment 1.89 —d 1.88 1.71 2.04
关106 kN m兴
a
ASCE 7-05 共ASCE 2005兲. Peak accelerations are calculated multiplying RMS acceleration by resonant peak factor and total moments are integrated wind
loads determined by design wind pressure over full height of structure considering gust effect factor.
b
It is obtained from taking averages of Exposures C and A results.
c
Peak factor for 1 h is considered, instead of 10 min.
d
ASCE only treats the alongwind direction.
e
Extrapolated values, since ranges of reduced frequency are beyond acrosswind spectra.

effects on a high-rise building using his/her own data or an estab- calculated for a 5-year return period, and the peak factor was
lished expression for the base moment spectrum and evaluate evaluated for 10 min, instead of the 1-hour used in both NALD v.
various design options expediently. 2.0 and ASCE 7-05. On the other hand, NALD v. 2.0 observes the
standards set by ASCE 7: 50-year return period for base moments
Example 4—Comparison to Australian Standard 共survivability design兲, and 10-year return period for accelerations
共serviceability design兲. Thus, proper modifications to wind speed
It is of interest to examine how the NALD v. 2.0 analysis com- 共to account for differences in return period兲 and peak factor 共to
pares with the dynamic response estimates of major building account for differences in averaging interval兲 are required to com-
codes and standards. A Commentary on the Australian Standard pare AS 1170.2 with both NALD v. 2.0 and ASCE 7-05 results.
for Wind Loads reported a detailed procedure for the dynamic As such, AS 1170.2 RMS results are translated to peak accelera-
analyses in the alongwind and acrosswind directions in Appendix tions based on a peak factor calculated over 1 h. The design wind
C of that standard, using the Commonwealth Aeronautical Advi-
speed for NALD v. 2.0 and ASCE 7-05 are adjusted using the
sory Research Council 共CAARC兲 standard tall building 共Holmes
relationships in ASCE 7-05 for a 1,000-year return period in base
et al. 1990兲. This CAARC building is analyzed by AS 1170.2,
moment calculations and a 5-year return period in acceleration
NALD v. 2.0, and ASCE 7-05 共ASCE 2005兲. The CAARC build-
calculations, so that they may be compared to the results of AS
ing’s main characteristics are summarized here: Case 1.
B = 46 m; D = 30 m; Case 2. B = 30 m; D = 46 m 共Case 2 repre- 1170.2 directly. As shown in Table 3, AS 1170.2 responses show
sents a 90 deg angle of incidence for the same building in Case relatively higher values 共conservative兲 in comparison with NALD
1兲; H = 183 m; f x = f y = 0.2 Hz; ␳B = 160 kg/ m3; ␨ = 0.015 for ser- v. 2.0 and ASCE 7-05, except for the alongwind peak accelera-
viceability design and 0.050 for survivability design; CD = 1.3 for tions, which show good agreement. The discrepancies may in part
Case 1 and 1.19 for Case 2; mode shapes for all directions are be attributed to the measurement approach used to estimate aero-
assumed linear. In Holmes et al. 共1990兲, it was assumed that the dynamic loads. The data used in the Australian Standard are based
CAARC building was located in Brisbane 共terrain category 3兲, on an aeroelastic model, and load spectra are estimated by an
which corresponds to Exposure B in ASCE 7-05. Since NALD v. inverse approach, which may have inherent identification sensi-
2.0 handles Exposures A 共BL2兲 and C 共BL1兲 only, comparisons tivities. The other possible source may be the differences in the
are made for both exposures, as well as their average, as they approach flow conditions, which have been observed to have no-
should provide upper and lower limits for the CAARC building. It table influence on the acrosswind response. Another important
should be pointed out that base moments in the AS 1170.2 were advantage of NALD v. 2.0 is also underscored by this example; it
calculated for the ultimate limit state design, corresponding to provides a means to estimate the acrosswind response that ASCE
wind speed of 1,000-year return period, while accelerations were 7 does not provide, outside of its commentary.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1151

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 These examples demonstrate the capabilities and accuracy of 03-24331, CMMI 06-01143, and CBET 07-42191, and a Postdoc-
NALD v. 2.0, providing a user-friendly procedure to reliably es- toral Fellowship from Korea Science and Engineering Foundation
timate building dynamic responses. The writers envision this ca- 共KOSEF兲.
pability to be particularly useful for those who may not be very
familiar with the details of the dynamic analysis procedure typi-
cally employed in response estimation for wind-sensitive struc- Appendix. Acrosswind PSD Expressions Used
tures. In addition, the robust framework presented here is in Example 3
conveniently amenable to including additional data for other
building cross sections and flow conditions. • AIJ Recommendations for Loads on Buildings 共1996, 2004兲
f ⫻ S M 共f兲
C M 共f兲 =
Concluding Remarks ␴M2
N
4K j共1 + 0.6␤ j兲␤ j 共n0/nsj兲2
The rapid development of information technologies has revolu-
tionized many engineering applications. This study discusses the
= 兺
j=1 ␲ 兵1 − 共n0/nsj兲2其2 + 4␤2j 共n0/nsj兲2
use of these advances to enhance, for the purposes of analysis and
where

再
design, the accessibility, organization, dissemination, and utility
of wind tunnel data. The second version of the NatHaz Aerody- 1, D/B ⬍ 3
namic Loads Database 共NALD v. 2.0兲 共http://aerodata.ce.nd.edu兲 N= K1 ⬇ 0.85 K2 ⬇ 0.02
2, D/B ⱖ 3
integrates these technologies for the purpose of wind-induced re-
sponse prediction. NALD v. 2.0 offers more attractive and user- f⫻B 0.12 0.56
friendly features to allow on-line determination of not only the n0 = ns1 = ns2 =
base moments, displacements, and the equivalent static wind UH 兵1 + 0.38共D/B兲2其0.89 共D/B兲0.85
loads for survivability design, but also accelerations for service-
ability 共habitability兲 design. Several Web-based tools such as PHP 共D/B兲4 0.12
␤1 = + 共AIJ 1996兲
and MySQL are fused with MATLAB to create efficient yet com- 1.2共D/B兲 − 1.7共D/B兲 + 21 共D/B兲
4 2

putationally robust interfaces that process, convert, and analyze

wind tunnel data on-the-fly with minimal user effort. The attrac- 共D/B兲4 + 2.3共D/B兲2
tive feature of this on-line processing approach is that no user ␤1 =
2.4共D/B兲 − 9.2共D/B兲3 + 18共D/B兲2 + 9.5共D/B兲 − 0.15
4
intervention is expended in the determination and display of wind
loads and response quantities for the preliminary design of high- 0.12
+ 共AIJ 2004兲
rise buildings. Moreover, this on-line analysis module provides 共D/B兲
the flexibility to utilize not only the NALD experimental PSD
data, but also user-specified PSD expressions or data sets. This ␤2 = 0.28共D/B兲−0.34
versatility provides users with a robust stand-alone, on-line analy-
sis engine for high-rise buildings using various data sources. Fur- ␴CM = 0.0082共D/B兲3 − 0.071共D/B兲2 + 0.22共D/B兲
ther, the architecture used in this study permits easy extensions to
• Gu and Quan 共2004兲
more sophisticated numerical analyses by employing the many
predefined function capabilities of MATLAB operating on the f ⫻ S M 共f兲 1 S p␤共n0/f p兲␣
server side. While, the analysis capabilities offered by NALD v. C M 共f兲 = = 2 ⫻
␴M 2
␴CM 兵1 − 共n0/f p兲2其2 + ␤共n0/f p兲2
2.0 are not necessarily intended to replace customized wind tun-
nel testing in the final design stages, they do provide users with an where
efficient means to approximate the complete 3D response of n0 = fB/UH
buildings in the early design stage, which has not been fully
treated in most codes and standards. Additionally, the analysis
f p = 10−5共191 − 9.48␣w + 1.28␣hr + ␣hr␣w兲共68 − 21␣db + 3␣2db兲
engine built into NALD v. 2.0 offers the option of utilizing user-
supplied custom spectral description or wind tunnel test results to ␣w
obtain final design estimates of wind load effects on buildings. S p = 共0.1␣−0.4
w − 0.0004e 兲共0.84␣hr − 2.12 − 0.05␣hr兲
2

It should be noted that the Web-based tools used to establish db − 0.08␣db 兲

⫻共0.422 + ␣−1 −2

the interface and analysis modules described in this study are

continuously updated as evolving security and vulnerability issues ␤ = 共1 + 0.00473e1.7␣w兲共0.065 + e1.26−0.63␣hr兲e1.7−3.44/␣db
are identified. Due to this constant updating, the interfaces are
likely to experience some cosmetic changes since the publication ␣ = 共− 0.8 + 0.06␣w + 0.0007e␣w兲共− ␣hr
0.34
+ 0.00006e␣hr兲
of this manuscript.
⫻共0.414␣db + 1.67␣−1.23
w 兲

Acknowledgments ␣w = 1共A兲, 2共B兲, 3共C兲, 4共D兲 A,B,C,D : Terrain categories

The writers wish to acknowledge Yin Zhou, formerly of Notre ␣hr = H/冑BD
Dame, Chii-Ming Cheng, Tamkang University, Taipei, and Rachel
Bashor, NatHaz Modeling Laboratory for their respective contri- ␣db = D/B
butions. The writers are also grateful for the financial support in
part by the National Science Foundation via Grant Nos. CMMI ␣ht = H/T 关T = min共B,D兲兴

1152 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008

Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
 ␴CM = 共0.002␣2w − 0.017␣w − 1.4兲共0.056␣2db − 0.16␣db + 0.03兲 wind loads of tall buildings based on synchronous pressure measure-
ments.” Eng. Struct., 29共11兲 2641–2653.
⫻共0.03␣2ht − 0.622␣ht + 4.357兲 Kareem, A. 共1982兲. “Fluctuating wind loads on buildings.” J. Engrg.
Mech. Div., 108共6兲, 1086–1102.
Kareem, A. 共1990兲. “Measurement of pressure and force fields on build-
ing models in simulated atmospheric flows.” J. Wind. Eng. Ind. Aero-
References dyn., 36共1兲, 589–599.
Kareem, A., and Cermak, J. E. 共1979兲. “Wind tunnel simulation of wind-
Architectural Institute of Japan 共AIJ兲. 共1996兲. “Recommendations for structure interactions.” ISA Trans., 18共4兲, 23–41.
loads on buildings.” Architectural Institute of Japan. Kareem, A., and Zhou, Y. 共2003兲. “Gust loading factor—Past, present and
Architectural Institute of Japan 共AIJ兲. 共2004兲. “Recommendations for future.” J. Wind. Eng. Ind. Aerodyn., 91共12–15兲, 1301–1328.
loads on buildings.” Architectural Institute of Japan. Kijewski, T., and Kareem, A. 共1998兲. “Dynamic wind effects: A compara-
American Society of Civil Engineers 共ASCE兲. 共1998兲. “Minimum design tive study of provisions in codes and standards with wind tunnel
loads for buildings and other structures.” ASCE 7-98, ASCE, Reston, data.” Wind Struct., 1共1兲, 77–109.
Va. Kijewski, T., Kwon, D., and Kareem, A. 共2003兲. “e-technologies for wind
American Society of Civil Engineers 共ASCE兲. 共2005兲. “Minimum design effects on structures.” Proc., 11th Int. Conf. on Wind Engineering,
loads for buildings and other structures.” ASCE 7-05, ASCE, Reston, 2217–2224.
Va. Kikuchi, H., Tamura, Y., Ueda, H., and Hibi, K. 共1997兲. “Dynamic wind
Boggs, D. W., and Peterka, J. A. 共1989兲. “Aerodynamic model tests of tall pressures acting on a tall building model—Proper orthogonal decom-
buildings.” J. Eng. Mech., 115共3兲, 618–635. position.” J. Wind. Eng. Ind. Aerodyn., 69–71, 631–646.
Chan, C.-M., and Chui, J. K. L. 共2006兲. “Wind-induced response and Kwon, D., Kijewski-Correa, T., and Kareem, A. 共2005兲. “e-analysis/
serviceability design optimization of tall steel buildings.” Eng. Struct.,
design of tall buildings subjected to wind loads.” Proc., 10th Ameri-
28共4兲, 503–513.
Chen, X., and Kareem, A. 共2004兲. “Equivalent static wind loads on build- cas Conf. on Wind Engineering 共CD-ROM兲, Paper No. 123, AAWE.
ings: New model.” J. Struct. Eng., 130共10兲, 1425–1435. Liang, S., Li, Q. S., Liu, S., Zhang, L., and Gu, M. 共2004兲. “Torsional
Chen, X., and Kareem, A. 共2005兲. “Dynamic wind effects on buildings dynamic wind loads on rectangular tall buildings.” Eng. Struct.,
with 3-D coupled modes: Application of high frequency force balance 26共1兲, 129–137.
measurements.” J. Eng. Mech., 131共11兲, 1115–1125. Liang, S., Liu, S., Li, Q. S., Zhang, L., and Gu, M. 共2002兲. “Mathematical
Cheng, C.-M., and Wang, J. 共2004兲. “Wind tunnel database for an inter- model of acrosswind dynamic loads on rectangular tall buildings.” J.
mediate wind resistance design of tall buildings.” Proc., 1st Int. Symp. Wind. Eng. Ind. Aerodyn., 90共12–15兲, 1757–1770.
Lin, N., Letchford, C., Tamura, Y., Liang, B., and Nakamura, O. 共2005兲.
on Wind Effects on Buildings and Urban Environment, 10–25.
“Characteristics of wind forces acting on tall buildings.” J. Wind. Eng.
Davenport, A. G. 共1967兲. “Gust loading factors.” J. Struct. Div., 93共3兲,
Ind. Aerodyn., 93共3兲, 217–242.
11–34.
McNamara, R. J. 共2005兲. “Some current trends in high rise structural
Flay, R. G. J., and Bhat, J. 共2005兲. “Cross-wind force spectra for building
design.” Structure Magazine, September, 19–23.
geometries commonly proposed for New Zealand cities.” Proc., 2nd
Reinhold, T. A., and Kareem, A. 共1986兲. “Wind loads and building re-
Workshop on Regional Harmonization of Wind Loading and Wind
sponse predictions using force-balance techniques.” Proc., 3rd ASCE
Environmental Specifications in Asia-Pacific Economies (APEC-WW),
Engineering Mechanics Specialty Conf. on Dynamic Response of
19–30.
Structures, ASCE, New York, 390–397.
Fritz, W. P., and Simiu, E. 共2005兲. “Probabilistic description of tall build-
ing response to wind: Database-assisted design, dynamics, and wind Solari, G., and Kareem, A. 共1998兲. “On the formulation of ASCE 7-95
gust effect factor.” J. Wind. Eng. Ind. Aerodyn., 77–78, 673–684.
directionality effects.” Proc., 9th Int. Conf. on Structural Safety and
Steckley, A., Accardo, M., Gamble, S. L., and Irwin, P. A. 共1992兲. “The
Reliability 共CD-ROM兲.
use of integrated pressures to determine overall wind-induced re-
Fujii, K., Hibi, K., and Ueda, H. 共1986兲. “A new measuring system using
sponse.” J. Wind. Eng. Ind. Aerodyn., 42共1–3兲, 1023–1034.
electronically scanned pressure sensors 共ESP兲 and some applications
Tamura, Y., Kareem, A., Solari, G., Kwok, K. C. S., and Holmes, J. D.
of the ESP system to a square building shape.” Proc., 9th National 共2005兲. “Aspects of the dynamic wind-induced response of structures
Symp. on Wind Engineering, 313–318 共in Japanese兲. and codification.” Wind Struct., 8共4兲, 251–268.
Gu, M., and Quan, Y. 共2004兲. “Across-wind loads of typical tall build- Tamura, Y., Kawai, H., Uematsu, Y., Marukawa, H., Fujii, K., and Tani-
ings.” J. Wind. Eng. Ind. Aerodyn., 92共13兲, 1147–1165. ike, Y. 共1996兲. “Wind load and wind-induced response estimations in
Ha, Y. C., Kim, D. W., and Kil, Y. S. 共2004兲. “Characteristics of the the ‘Recommendations for loads on buildings,’ AIJ 1993.” Eng.
across-wind fluctuating force and spectral density of rectangular high- Struct., 18共6兲, 399–411.
rise buildings with various side ratios.” Proc., CTBUH 2004, 978– Tschanz, T., and Davenport, A. G. 共1983兲. “The base balance technique
982. for the determination of dynamic wind loads.” J. Wind. Eng. Ind.
Ho, T. C. E., Lythe, G. R., and Isyumov, N. 共1999兲. “Structural loads and Aerodyn., 13共1–3兲, 429–439.
responses from the integration of instantaneous pressures.” Wind En- Vickery, P. J., Steckley, A. C., Isyumov, N., and Vickery, B. J. 共1985兲.
gineering into the 21st Century: Proc., 10th Int. Conf. on Wind Engi- “The effect of mode shape on the wind-induced response of tall build-
neering, A. Larsen, G. L. Larose, F. M. Livesey, eds., Balkema, ings.” Proc., 5th U. S. National Conf. on Wind Engineering, Vol. 1B,
Rotterdam, The Netherlands, 1505–1510. 41–48.
Holmes, J. D., Melbourne, W. H., and Walker, G. R. 共1990兲. “A commen- Xu, Y. L., and Kwok, K. C. S. 共1993兲. “Mode shape corrections for wind
tary on the Australian standard for wind loads: AS 1170 part 2.” tunnel tests of tall buildings.” Eng. Struct., 15共5兲, 387–392.
Australian Wind Engineering Society. Zhou, Y., and Kareem, A. 共2001兲. “Gust loading factors: New model.” J.
Holmes, J. D., Rofail, A. W., and Aurelius, L. J. 共2003兲. “High frequency Struct. Eng., 127共2兲, 168–175.
base balance methodologies for tall buildings with torsional and Zhou, Y., Kareem, A., and Gu, M. 共2002兲. “Mode shape corrections for
coupled resonant modes.” Proc., 11th Int. Conf. on Wind Engineering, wind load effects.” J. Eng. Mech., 128共1兲, 15–23.
2381–2388. Zhou, Y., Kijewski, T., and Kareem, A. 共2003兲. “Aerodynamic loads on
Huang, G., and Chen, X. 共2007兲. “Wind load effects and equivalent static tall buildings: Interactive database.” J. Eng. Mech., 129共3兲, 394–404.

JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JULY 2008 / 1153

View publication stats Downloaded 18 Nov 2008 to 129.74.250.206. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright