Professional Documents
Culture Documents
Monitoring Wind Characteristics and Structural Performance of A Supertall Building During A Landfall Typhoon
Monitoring Wind Characteristics and Structural Performance of A Supertall Building During A Landfall Typhoon
Abstract: Wind characteristics in the atmospheric boundary layer (ABL) and structural performance under extreme wind conditions are of
major concern in the design of tall buildings in tropical cyclone-prone regions. On August 22, 2008, Typhoon Nuri (0812) made landfall over
Hong Kong and then passed over its city center. This paper presents the analyzed results of the observations collected at two offshore
meteorological stations and by a structural health monitoring system installed in a 420-m-high supertall building in downtown Hong Kong
during the entire passage of Typhoon Nuri. The wind characteristics over open (sea) and urban (city) terrains at different stages of the typhoon
are presented and discussed. Moreover, the wind-induced responses of the supertall building during the typhoon are investigated. The struc-
tural dynamics properties and serviceability of the supertall building during a landfall typhoon are evaluated. This study aims to investigate
the wind characteristics in the ABL during a landfall typhoon, and its impact on high-rise buildings in a metropolis so as to provide useful
information for the wind-resistant design of supertall buildings in tropical cyclone-prone regions. DOI: 10.1061/(ASCE)ST.1943-541X
.0001564. © 2016 American Society of Civil Engineers.
Author keywords: Wind characteristics; Typhoon; Supertall building; Structural performance; Field monitoring; Wind effects.
alyzed results of the data recorded at three measurement sites in 1,299 MHz. The profiler system works in both low and high modes
Hong Kong during the entire passage of Typhoon Nuri: two off- simultaneously. The low mode range is set as 213–1,656 m with
shore meteorological stations and a 420-m-high supertall building a height resolution of 60 m; while the high mode range is set as
located in the city center. Both the meteorological stations are 459–5,371 m with a resolution of 204.5 m. Both the modes can
equipped with ground-based anemometers, while one station is also provide 25 gate records which are updated every 2 min. Due to
equipped with a Doppler radar profiler. A SHM system including adopting the overlapping average technique, all the gate records
measurement devices for wind velocities and structural responses stand for 10-min mean values. The accuracy of the profiler system
was installed in the supertall building. In the first part of this paper, is within 1 m=s and 10°. The observed data collected by the profiler
the wind characteristics of Nuri over offshore sites and a typical were adopted for analyzing the vertical wind structure of Typhoon
city terrain are investigated. Then, the structural responses of the Nuri. The other meteorological station is situated on the top (55.8 m
supertall building during Nuri are analyzed, and the serviceability AMSL) of Waglan Island (22°10’56′′N; 114°18’12′′ E) which is a
of the high-rise structure under typhoon conditions is evaluated small island with an area less than 0.1 km2 . This station, which is
based on the measurements by the SHM system. Since Typhoon called WGL station hereafter, is about 5 km off the southeast coast
Nuri passed over the city center of Hong Kong directly, compre- of Hong Kong and approximately 29 km east of the CCH station.
hensive information on the wind characteristics in the ABL and the Horizontal surface wind components are measured by a cup
wind effects on the supertall building at different storm stages is anemometer installed at height 26.9 AGL or 82.7 m AMSL at
presented and discussed. The observed data and analyzed results the WGL station.
presented in this paper are expected to provide useful information
for the wind-resistant designs of supertall buildings in TC-prone
SHM System Installed in a Supertall Building
regions.
For the purpose of investigating the typhoon-generated wind char-
acteristics over city terrain and the impact of a landfall typhoon
Instrumental and Meteorological Information on high-rise buildings in a metropolis such as Hong Kong, this
paper analyzes and discusses the data recorded from a SHM sys-
tem installed in a 420-m-high supertall building (22°17′7′′ N;
Offshore Meteorological Stations
114°9′33′′ E) in Hong Kong during the passage of Typhoon Nuri.
This paper adopts the observations collected at two offshore As shown in Fig. 1(a), the supertall building is approximately
meteorological stations in Hong Kong during Typhoon Nuri. 17 km northeast of the CCH station, and approximately 19 km
Fig. 1. (a) Topographic map and (b) aerial photo of Hong Kong (image courtesy of the Government of the Hong Kong Special Administrative Region,
http://DATA.GOV.HK)
Fig. 2. Some sensors of the SHM system installed in a supertall building (images by authors): (a) propeller anemometer; (b) GPS antenna; (c) control
system; (d) locations of the accelerometers and orthogonal directions X and Y; (e) locations of the instrumented supertall building and the center of
Typhoon Nuri around 5:40 p.m. on August 22, 2008
northwest of the WGL station. Fig. 1(b) shows an aerial image Typhoon Nuri
of the area marked by a rectangle in Fig. 1(a). As indicated in
Nuri developed as a tropical depression in the east of Philippines
Fig. 1(b), the site of this supertall building is at the seashore in
on August 17, 2008, and edged towards northwest afterwards.
Hong Kong Island while Victoria Harbor is located in the north of
Hong Kong Observatory (HKO) issued the Strong Wind Signal
the building. A number of tall buildings with heights over 150 m
No. 3 at 8:40 a.m. HKT (Hong Kong Time) on August 21 when
are located adjacently to the building in a sector spanning clock-
Nuri was about 310 km southeast of Hong Kong. On the following
wise from east to west. The building section is convex-square day, August 22, the No.8 Northwest Gale or Storm Signal was is-
shaped and the building width tapers from 57 to 39 m along its sued at 07:40 a.m., and then replaced by Increasing Gale or Storm
height. A reinforced concrete central core is linked by steel beams Signal No. 9 at 1:40 p.m. As shown in Fig. 1(a), Nuri made landfall
and outriggers to eight exterior composite megacolumns. The mea- over the eastern part of Hong Kong at around 4:50 p.m., and then
suring devices of the SHM system that provided data for this study reorganized itself by forming a new center at Point B while the
are as follows. A Young 05103 propeller anemometer (R. M. original center dissipated rapidly at Point A. The center of Nuri
Young Company, Traverse City, Michigan) [Fig. 2(a)] installed continued to move west-northwestwards and passed over the city
approximately 14 m above the building’s roof at 420.55 m above center of Hong Kong with its track shown in Fig. 1(b). At approx-
Hong Kong Principle Datum (420.55 mPD), or 419.32 AMSL. imately 5:40 p.m., the distance between the site of the supertall
A high-precision Leica GMX902 GG GPS receiver and an Smart- building equipped with the SHM system and the center of Nuri
Track AX1202 GPS antenna (Leica Geosystems, Norcross, was less than 2 km [Fig. 2(e)], suggesting that the supertall building
Georgia) [Fig. 2(b)] mounted on the building’s roof at height of was entirely submerged in the typhoon eye region. In order to pro-
414.55 mPD. The GPS utilizes a continuously operating reference vide comprehensive information on the typhoon-generated charac-
station located on an island in Victoria Harbor with about 4 km teristics during the entire passage of Nuri over Hong Kong, this
away from the building site. The reference station established by paper adopts the data collected from the CCH station, the WGL
Survey and Mapping Office of Hong Kong Lands Department con- station, and the SHM system installed in the supertall building dur-
tains a high-precision Leica System GRX1200 Pro receiver (Leica ing a 36-h-long time period ranging from 12:00 a.m. on August 22
Geosystems, Norcross, Georgia) with Leica AT504 choke-ring to 12:00 p.m. on August 23. Then, the wind characteristics of Nuri
antenna installed on a low-rise building’s roof. Signals collected by at the two offshore locations and at the city center are analyzed and
the anemometer and the accelerometers were sampled at 20 Hz discussed. Furthermore, taking the supertall building under moni-
while those recorded by the GPS were sampled at 10 Hz, and all toring as an example to explore the impact of a landfall typhoon on
these signals were continuously and simultaneously recorded by a high-rise buildings in a typical metropolis, structural performance
data acquisition unit (DAU) during the passage of Typhoon Nuri. of the supertall building during the passage of Typhoon Nuri are
Fig. 2(c) shows a photo of the control system of the SHM system. detailed evaluated and discussed.
Two orthogonally [in X and Y directions as shown in Fig. 2(d),
respectively] mounted accelerometers placed at the top floor center
of the building (at height of 400.10 mPD). Wind direction of 0° Wind Characteristics
measured by the anemometer represents the north, while there is
an angle difference of 32° between direction X and the north as The wind characteristics during Nuri are analyzed utilizing the data
shown in Fig. 2(d). Fig. 2(e) shows the supertall building. measured at the two offshore meteorological stations, namely the
outskirt region (Stage OS2). Fig. 4 shows the profiles OS1, EW1,
Downloaded from ascelibrary.org by University College London on 08/16/16. Copyright ASCE. For personal use only; all rights reserved.
Wind Profiles Eye, EW2, and OS2, while the profiles of mean wind direction
Integrated wind profiles covering the entire TC boundary layer at these five stages are also presented in the figure. In Fig. 4(a),
(TBL) at the CCH station are obtained by synchronizing the data a more-distinct jet-like structure can be observed at Stages EW1,
from the two (low and high) modes of the Doppler radar profiler EW2, and OS2. The maximum wind, which may be regarded as the
system with those from the anemometer installed near the ground. gradient speed (Powell et al. 2003), occurred at a height of 1,000–
These integrated profiles are then averaged over each 30-min-long 2,000 m, which varied among the different stages. Stages EW2 and
segment starting from 12:00 a.m. on August 22 without overlap- OS2 basically correspond to an open sea exposure, and the jets are
ping. Fig. 3 illustrates the vertical profiles of 30-min mean horizon- around 1,000 m (EW2) and 1,500 m (OS2). These results reflected
tal wind speed (U) by filled color contours, which are overlaid by that more inner TC parts corresponded to lower jets, which are con-
[Ux , Uy ] wind vectors indicated by black arrows (downward for sistent with the observations over the ocean (Franklin et al. 2003;
north wind, leftward for east wind). The corresponding time history Vickery et al. 2009). As indicated in Figs. 3 and 4(b), Stages OS1
of air pressure, depicted by a dotted line in the upper part of Fig. 3, and EW1 corresponded to a hilly upwind terrain in the north of the
indicates the variation of storm-relative position during the passage CCH station (Fig. 1), while Stages EW2 and OS2 corresponded to
of Nuri. an open sea terrain in the south of the station. It is believed that
Fig. 3 reveals some typical features of a TC. The TC eye located because of these topographic effects, the jet in the mean wind speed
at central area around 18:00/22 was characterized by calm wind profile at Stage EW1 intended to be uplifted (centered 1,500 m)
indicated by the dark thick line; while the eyewalls surrounding compared to that at Stage EW2. Moreover, the topographic effects
the TC eye, centered at 12:00 and 10:00 p.m. on August 22, respec- were also reflected in the profiles of mean wind direction at these
tively, contained maximum winds colored by the light dashed typhoon stages. In Fig. 4(b), the wind directions with height below
line. Mean wind directions varied significantly from 0 to 200° after 2 km at Stages OS1 and EW1 increased around 30 and 50°, while
the passage of the TC eye. Within the regions of the two eye- those at Stages EW2 and OS2 rarely varied.
walls, the radius of maximum wind (RMW) is found to tilt outward To quantitatively depict the obtained mean wind speed profiles,
with height, which is similar to those observed over deep oceans four frequently utilized models of wind speed profiles are em-
(Hawkins and Imbembo 1976; Giammanco et al. 2013). ployed to fit the results, including the log-law, power-law, Deaves
To further explore the mean wind profiles at varied stages during and Harris model (Cook 1997; hereafter referring to as DH model),
the passage of Nuri, five stages listed in Table 1 were extracted and the model presented by Vickery et al. (2009); hereafter referred
from the entire observation period: the first outskirt region (Stage to as the VWPC09 model
OS1), the first eyewall region (Stage EW1), typhoon eye region UðzÞ ¼ U ref ðz=zref Þα ð1aÞ
(Stage Eye), the second eyewall region (Stage EW2) and the second
u z 23 zd 15 zd 2
Uðzd Þ ¼ ln d þ −
κ z0 4 h 8 h
3 4
4 zd 1 zd
− þ ð1cÞ
3 h 4 h
n
zd z
Uðzd Þ ¼ ðu =κÞ ln − a d ð1dÞ
z0 H
Fig. 4. Mean wind profiles at five different stages (OS1, EW1, Eye, EW2, and OS2) during Nuri’s landfall at CCH station: (a) mean wind speed;
(b) mean wind direction
above open terrain (z0 ¼ 0.03 m) by using the standardization indicating that, as mentioned previously in section “Typhoon Nuri,”
method proposed by He et al. (2014) to eliminate the topographic the supertall building was entirely submerged in the TC eye region
effects. The fitting analysis of the measurements is then conducted and extremely adjacent to the center of Nuri during that period of
based on the records either in the lowest 1,000 m (for the log-law time. In Fig. 6(b), the 10-min mean wind direction measured atop
and the power-law) or in the lowest 3,000 m (for DH and VWPC09 the supertall building varied approximately 180°, and this shift in
models). The fitting results are indicated in Fig. 5 and Table 2. wind direction occurred more rapidly than those measured at the
two offshore meteorological stations. This again implied that the
supertall building was much closer to the track of Nuri’s center than
Mean Horizontal Wind the two meteorological stations.
Fig. 6 plots the variations of the 10-min mean wind speed and di-
rections at the CCH station, the WGL station and atop the supertall
building (denoted by Building in the figure). The AMSL heights of Fluctuating Horizontal Wind
each measurement points are also indicated in Fig. 6, while there Based on the measurements by the anemometer atop the supertall
are four measurement points with different heights above the CCH building, the turbulence integral length scales in longitudinal and
station (i.e., 26.7 m at the anemometer height, and 213, 333, and lateral directions are analyzed for the investigation of turbulence
454 m gate heights from the Doppler radar profiler). Due to the characteristics of wind during the passage of Nuri. Using the von
differences of the storm-relative locations among the CCH station, Karman spectra to fit the measured power spectral densities of
WGL station, and supertall building, the measurements at these fluctuating wind speeds, the turbulence integral length scales are
three sites showed noticeable phase differences around the typhoon determined by regarding them as the fitting parameters. Fig. 7 plots
eye region. The minimum mean wind speeds and the dramatic the power spectral densities (PSDs) of the measured wind velocities
shifts of mean wind directions reflected the arrivals of the typhoon in longitudinal and lateral directions and compares them with the
eye at the three measurement sites, and these records matched the von Karman spectra expressed by
timeline provided by HKO and plotted in Fig. 1(a). In Fig. 6(a),
a typical M-shaped distribution of the mean wind speed, which ex- nSε ðnÞ 4ðLUε nÞ
¼ ; ε¼4
hibits the classical double peak of an eyewall passage over a meas- σ2ε ½1 þ 70.8ðLUε nÞ2 5=6
urement site, can be observed with ease for all the cases, indicating
nSε ðnÞ 4ð Uε Þ½1 þ 755.2ð Uε Þ2
Ln Ln
that the three sites had experienced the five stages shown in Table 1. ¼ ; ε¼v ð2Þ
By comparing the measurements collected at the two offshore σ2ε ½1 þ 283.2ðLUε nÞ2 11=6
stations, apparently the wind strength measured above the CCH
station was much lower than that measured at the WGL station, where n = frequency; Lε ðε ¼ u; vÞ = estimated turbulence integral
although the gate records from the CCH station were measured at length scale; and Sε ðnÞjε¼u;v = power spectral density of the fluc-
much greater heights than that at the WGL station, suggesting that tuating wind components in the longitudinal and lateral direction,
the typhoon was rapidly weakened after it made landfall. On the respectively. Fig. 8 plots the turbulence integral length scales in
other hand, the mean wind strength atop the supertall building in longitudinal and lateral directions, denoted by Lu and Lv , respec-
the city center was significantly lower compared to that above the tively, against 10-min mean wind speeds. Clearly, the turbulence
CCH station at a similar height (i.e., 454 m gate records), or even integral length scales in both the longitudinal and lateral directions
to that at the WGL station at a height of only 87 m. The dramatic scattered significantly as the wind speed increased. An increasing
decline of the measured wind strength reflected the severe shield- trend of the longitudinal turbulence integral length scale with mean
ing effects caused by the local topography/terrain and surround- wind speed can be observed with ease. However, there is no clear
ing buildings around the supertall building. The lowest mean relationship between the lateral turbulence integral length scale and
wind speed measured atop the building was merely above zero, the wind speed.
Fig. 5. Mean wind profiles fitted by different models at three typhoon stages (Log: logarithmic law; Power: power law; DH: Deaves and Harris model;
VWPC09: model proposed by Vickery et al. 2009)
Table 2. Fitting Results of Mean Speed Profiles at Three Stages during Nuri’s Passage via Four Models
Model
Power-law Log-law DH VPWC09
Parameter
Stage (m=s) α m=s (m) (m) m=s (m) (m) (m) (m) a n
EW1 13.8 0.18 1.57 0.38 0.19 0.96 0.026 0 1600 1629 0.4 2
EW2 — — 1.87 0.54 4.5 — — — — 965 0.4 2
OS2 9.28 0.21 1.36 0.85 0.45 0.84 0.79 1.86 1600 1641 0.4 2
Note: For VPWC09 model, parameter values obtained in the Log-law model case are adopted; a and n are set as 0.4 and 2; for the EW2 case, the power-law
and DH model fail to provide acceptable fitting results.
Turbulence intensity, I, and gust factor, GF, are two important Gust factor is often defined as the ratio of the peak gust speed
parameters in determination of fluctuating wind actions and wind- over a shorter duration τ to the mean wind speed U in a relatively
induced dynamic responses of tall buildings. The turbulence inten- longer duration T
sities in the longitudinal and lateral directions can be determined by
the following equation: ûi ðτ Þ
GFi ðτ ; TÞ ¼ ; i ¼ u; v ð4Þ
σ UðTÞ
Ii ¼ i ; i ¼ u; v ð3Þ
U
where τ and T ¼ 3 s and 10 min, respectively, in the following
where σi = root-mean square (RMS) value of the fluctuating com- discussions. The variation of the gust factor with 10-min mean
ponent in direction i. wind speed and that with turbulence intensities are illustrated in
Fig. 6. Mean horizontal wind speed and direction measured atop the Fig. 8. Variations of the turbulence integral length scales with 10-min
supertall building and above the CCH station mean wind speed
Fig. 7. Power spectral density (PSD) of longitudinal and lateral wind speed components
Fig. 9. Variations of gust factors with (a) 10-min mean wind speed and (b) turbulence intensities
calm responses at Stage Eye, the acceleration responses in both Displacement Responses
directions enlarged again at Stage EW2, reaching 5.0 and
Apart from the acceleration responses of the supertall building,
3.7 cm=s2 in X and Y directions, respectively. Furthermore, 10-min
the building’s displacement responses, whose time histories are
mean and RMS values of wind speed are also plotted in Fig. 10,
illustrated in Fig. 13, were obtained by the GPS and adopted in
allowing the dependence of acceleration responses upon wind
this paper to comprehensively investigate the structural perfor-
strength and fluctuation to be observed with ease.
mance during the typhoon. A 2-h-long time segment starting at
The 10-min RMS of the acceleration responses Ax and Ay,
9:00 a.m. on August 22, which corresponded to the commencement
denoted by σAx and σAy respectively, are plotted in Fig. 11, and
their peaks are observed at Stage EW1 and EW2 as expected. of Stage EW1, was selected for the discussions of the building’s
Fig. 12 compares these RMS acceleration responses with 10-min displacement responses. Given that the supertall building is located
mean wind speed and with RMS values of wind speed, along with in a complex urban terrain and unfavorable weather conditions
regression curves expressed in power laws as follows: during the typhoon would inevitably lower the quality of the GPS
signals, the displacement data required validation to ensure the ac-
curacy of the outputs from the GPS. The performance of the GPS
σ a ¼ c 1 U c2 ð6Þ
installed in this building has been discussed in detail by Yi et al.
(2013). In this paper, the displacement data recorded by the GPS
were bandpass filtered about the first mode and then double differ-
σa ¼ c3 σcU4 ð7Þ enced to derive GPS estimate of accelerations. Then, these results
are further compared to similar bandpass filtered accelerations mea-
where σa = RMS, or standard deviation, of the acceleration re- sured by the accelerometers. It can be observed with ease that in
sponse; U and σU = mean and RMS of 10-min wind speed atop Fig. 14, the time histories of the acceleration data derived from the
the supertall building, respectively; c1 , c2 , c3 and c4 = coefficients displacement responses of the GPS show a good agreement with
describing the sensibilities of these relationships. those measured by the accelerometers. Fig. 15 demonstrates the
extent of correlation between the accelerations obtained by the
Fig. 12. Variation of RMS acceleration responses with 10-min mean and RMS wind speeds
accelerometers and the GPS, with the indications of regression during severe windstorms such as typhoons. According to Yi et al.
coefficients (slope and intercept) as well as correlation coefficients. (2013), the raw data of displacement responses from the GPS
Structural dynamic responses of a building can be divided into installed in the supertall building were high-pass filtered with a
resonant component and quasi-static background component. From cut-off frequency of 0.02 Hz to eliminate multipath effects along
the viewpoint of structural designs, the resonant component is of with other residual effects bedded at the frequency of 0 Hz. These
primary concern for evaluations of human comfort at upper floors filtered displacement data were then spectrally analyzed, and the
of tall buildings. Previous field measurement studies without em- PSDs within the bandwidth associated with the resonant peak,
ployment of GPS mostly focused on the discussions of the resonant which corresponded to the resonant component of the building dis-
component, which is essentially the only component that can be placement, were extracted using a narrow bandpass filter, while the
measured by accelerometers. However, it was reported that the remaining PSDs were regarded as the background component. The
background contributions are noteworthy for the design of tall entire GPS measurement was divided into 10-min-long segments
buildings (Yi et al. 2013). Therefore, the estimation of the back- for the calculation of RMS resonant responses and RMS back-
ground component, which can be acquired by GPS, is a necessity ground responses of each segment. Fig. 16(a) presents the original
in the investigations of wind-induced responses of tall buildings RMS resonant responses and RMS background responses of each
Fig. 13. Time histories of displacement responses in X and Y Fig. 14. Time histories of acceleration responses measured by accel-
directions erometers and derived from displacement responses by GPS
Fig. 15. Correlation between the acceleration responses measured by accelerometers and derived from the displacement responses by GPS
segment against 10-min mean wind speed, indicating that both Therefore, although the dynamic responses of the supertall building
the resonant responses and the background responses increased were dominated by the resonant responses at high wind speeds, the
exponentially with the wind speed. Fig. 16(b) presents the averaged contributions by the background responses should not be neglected
proportion of total responses, namely the averaged individual con- since these background responses occupied certain proportions of
tributions by the RMS resonant responses and the RMS back- the total responses.
ground responses, against 10-min mean wind speed. It can be
intuitively recognized that the resonant responses rapidly increased
after the wind reached a high speed, say 10 m=s, while the back- Modal Identification
ground responses increased with a much lower rate. Fig. 16(b) also
indicates that the averaged background responses contributed ap- Natural Frequencies
proximately 1=3 and 1=5 of the total responses in direction X The peak-picking (PP) method and the random decrement tech-
and Y, respectively, when the mean wind speed reached 20 m=s. nique (RDT) are used herein to identify the modal parameters of
Fig. 16. Relative contributions of resonant and background RMS displacement responses during the passage of Nuri: (a) original; (b) averaged
Damping Estimation
Damping ratio describes the efficiency of a structure to dissipate the
vibration energy transferred by dynamic loads such as typhoons
and earthquakes. As modern tall buildings are getting taller and
more flexible, reliable information on the damping ratios measured
from prototype supertall buildings are of great value since similar
Fig. 17. Variation of fundamental frequencies in X and Y directions
measurements are still very limited in the literature. In this study,
with vibration amplitudes
the damping ratios of the supertall building are estimated by the
Fig. 18. Power spectral density (PSD) of accelerations by accelerometers and by GPS in X and Y directions
Fig. 19. Comparison of amplitude-dependent fundamental frequencies identified by the outputs of accelerometers and GPS
Fig. 22. Comparison of amplitude-dependent damping ratios estimated by the outputs from accelerometers and GPS