Research Article Structural Health Monitoring of a …downloads.hindawi.com/journals/tswj/2013/509350.pdfe Scientic World Journal (a) Direction 1 Direction 2 N (b) F : (a) Side-view

Hindawi Publishing CorporationThe Scientific World JournalVolume 2013 Article ID 509350 12 pageshttpdxdoiorg1011552013509350

Research ArticleStructural Health Monitoring of a Reinforced ConcreteBuilding during the Severe Typhoon Vicente in 2012

Sin-Chi Kuok and Ka-Veng Yuen

Department of Civil and Environmental Engineering University of Macau Macau

Correspondence should be addressed to Ka-Veng Yuen kvyuenumacmo

Received 8 August 2013 Accepted 7 September 2013

Academic Editors G Turk B Uy and H Wang

Copyright copy 2013 S-C Kuok and K-V Yuen This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

The goal of this study is to investigate the structural performance of reinforced concrete building under the influence of severetyphoon For this purpose full-scalemonitoring of a 22-story reinforced concrete building was conducted during the entire passageprocess of a severe typhoon ldquoVicenterdquo Vicente was the eighth tropical storm developed in theWestern North Pacific Ocean and theSouth China Sea in 2012 Moreover it was the strongest and most devastating typhoon that struck Macao since 1999 The overallduration of the typhoon affected period that lasted more than 70 hours and the typhoon eye region covered Macao for aroundone hourThe wind and structural response measurements were acquired throughout the entire typhoon affected periodThe windcharacteristics were analyzed using the measured wind data including the wind speed and wind direction time histories Besidesthe structural response measurements of the monitored building were utilized for modal identification using the Bayesian spectraldensity approach Detailed analysis of the field data and the typhoon generated effects on the structural performance are discussed

1 Introduction

Full-scale measurement is considered as a reliable approachfor the evaluation of wind effects on structural systemsAlthough there has been significant progress on the tech-niques of wind tunnel tests and numerical modeling in recentdecades full-scale monitoring is irreplaceable for the studyof some critical complicated phenomena such as dampingbehavior complex modes and nonlinear structural behav-ior under severe excitation Kijewski-Correa and Pirnia [1]chronicled the harvest of some representative full-scale mon-itoring projects conducted in various countries It demon-strated the importance of full-scale monitoring for the studyof the actual behavior of civil engineering infrastructuresTropical cyclone generates violent aerodynamic excitation oninfrastructures and causes considerable economic damageand human life loss to the affected coastal cities Numerousresearch efforts have been devoted to analyzing the in-situstructural response measurements under such severe windloading conditions Xu andZhan [2] and Li et al [3] presentedthe field measurements for some landmarked buildings dur-ing the passage of severe typhoons In [4] detailed analysis

was presented for the typhoon effects on the wind loadcharacteristics and structural responses of four super-tallbuildings In [5] the analytical and experimental modalanalysis of the Guangzhou New TV Tower (renamed as theCanton Tower in 2010) was carried out and comparison of thestructural performance was pursued under several typhoonsOther representative studies provided useful insights intowind-induced structural behavior under strong wind condi-tions [6ndash8]

The goal of this study is to investigate the structural per-formance of reinforced concrete buildings during the passageof severe typhoon since it is important for structural reliabil-ity evaluation [9ndash13] This information will be useful for theimprovement of structural design [14] For this purpose full-scale monitoring of a 22-story reinforced concrete buildingwas conducted One of the special features of this study isthat this building has an asymmetric L-shape floor plan sothe aerodynamic effects are more complicated than the onesstudied in the literature The monitoring period covered theentire passage process of the severe typhoon ldquoVicenterdquo It wasthe eighth tropical storm developed in the Western NorthPacific Ocean and South China Sea in year 2012 By passing

2 The Scientific World Journal

throughMacao with the shortest core distance of only 40 kmVicente is considered the strongest and most devastatingtyphoon for Macao since 1999 Since critical aerodynamicconditions were generated to the infrastructures in the city itprovided a valuable opportunity to investigate the structuralbehavior under harsh wind situation In order to illustrate thewind conditions the meteorological information statisticalproperties of the wind measurements and gust speed will bepresented Furthermore the wind characteristic propertiesincluding the turbulence intensity and gust factor will beanalyzed

On the other hand the structural acceleration responsesare analyzed and the Bayesian spectral density approach[15] is applied to identify the time-varying modal parame-ters Bayesian inference provides a rigorous framework forparametric identification and has been applied to variousengineering disciplines [8 10ndash13 16ndash18] The Bayesian spec-tral density approach requires only the structural responsemeasurement and it has been adopted for analyzing thestructural response of the Canton tower [19] One specialfeature of this frequency-domain approach is that it takes fulladvantage of Bayesian inference for the posterior uncertaintyquantification of the modal estimation As a result thestatistical uncertainties can be distinguished from the actualchanges of the identified parameters This is essential forreliable judgment about the structural integrity In this paperthe time-varying modal parameters as well as the associatedestimation uncertainties will be identified based on thestructural response measurements of the entire monitoringperiod

This paper is organized as follows In Section 2 theBayesian spectral density approach for modal identificationis briefly reviewed In Section 3 the instrumentation on themonitored building is described In Section 4 the aerody-namic properties of Vicente are presented Finally Section 5presents the results and discussions about the typhoon effectson the structural performance of the monitored building

2 Bayesian Spectral Density Approach forModal Identification

Consider a linear dynamical system with 119873119889 degrees offreedom

Mx + Cx + Kx = T0F (119905) (1)

where M C and K are the mass damping and stiffnessmatrix respectively and T0 is the force distributing matrixThe external excitation F can be modeled as zero-meanGaussian white noise with spectral intensity matrix S119865(120596) =S1198650 The measurement Y119873 = y(119899) 119899 = 1 2 119873

contains119873119900 channels of structural response corrupted by themeasurement noise 120576

y (119899) = L0q (119899) + 120576 (119899) (2)

where y(119899) isin R119873119900 is the measurement at the 119899th timestep q(119899) isin R119873119889 is the concerned structural response (egacceleration) at the same time step and L0 isin R119873119900times119873119889 is

the observation matrix comprised of zeros and ones Themeasurement noise 120576 is modeled as zero-mean Gaussianindependent and identical distributed (iid) process withcovariance matrix Σ

120576

Use 120572 to denote the uncertain modal parameter vectorto be identified It includes the structural modal parameters(ie modal frequencies damping ratios and mode shapecomponents) of the contributing modes and the character-istic parameters of the spectral intensity of the excitationand the measurement noise To identify the uncertain modalparameter vector 120572 a discrete estimator of the power spectraldensity matrix is utilized [15]

S119910119873 (120596119896) =Δ119905

2120587119873

119873minus1

sum

1198991198991015840=0

y (119899) y (1198991015840)119879

exp [minus119894120596119896 (119899 minus 1198991015840) Δ119905]

(3)

where Δ119905 is the sampling time step the superscript 119879 denotesthe transpose of a vector Δ120596 = 2120587(119873Δ119905) is the frequencyprecision in the discrete Fourier transform and 120596119896 =

119896Δ120596 119896 = 0 1 INT(1198732)With 119873119904 independent sets of discrete time histories Y =

Y(119904)119873 119904 = 1 2 119873119904 the averaged spectral density matrix

estimator can be obtained

Savg119910119873

(120596119896) =1

119873119904

119873119904

sum

119904=1

S(119904)119910119873

(120596119896) (4)

where S(119904)119910119873

(120596119896) can be calculated using (3) with the measure-ment Y(119904)

119873 Given that119873119904 ge 119873119900 the averaged spectral density

matrix estimator follows the central complex 119873119900-variateWishart distribution with 119873119904 degrees of freedom With aproperly selected frequency range Ξ the averaged spectraldensity matrix estimators in Savg

Ξ= Savg119910119873

(120596119896) 120596119896 120598 Ξ areapproximately independent [20]

Using the Bayesrsquo theorem the posterior probability den-sity function (PDF) of 120572 given Savg

Ξis [16 21]

119901 (120572 | SavgΞ) =1198881119901 (120572) 119901 (Savg

Ξ| 120572) (5)

where 1198881 is a normalizing constant The prior PDF 119901(120572)

represents the prior information of the modal parametersin 120572 Throughout this study it is taken as a noninformativeprior distribution so it can be absorbed into the normalizingconstant The likelihood function 119901(Savg

Ξ| 120572) is given by the

product of Wishart distributions [15]

119901 (SavgΞ

| 120572)

= 1198882prod

120596119896isinΞ

10038161003816100381610038161003816119864 [S119910119873 (120596119896) | 120572]

10038161003816100381610038161003816

minus119873119904

times exp [minus119873119904 tr (119864 [S119910119873 (120596119896) | 120572]minus1

Savg119910119873

(120596119896))]

(6)

where 1198882 is a constant that does not depend on the modalparameters E[sdot] is the mathematical expectation | sdot | andtr(sdot) are the determinant and trace of a matrix respectivelyThe optimal modal parameter vector can be determined by

The Scientific World Journal 3

(a)

Direction 1 Direction 2

N

(b)

Figure 1 (a) Side-view and (b) typical floor plan of the East Asia Hall

maximizing its posterior PDF To enhance the computationalcondition the optimal modal parameters can be obtainedequivalently by minimizing the objective function definedas 119869(120572) = minus ln119901(Savg

Ξ| 120572) Consequently the covariance

matrix of the modal parameters is given by the inverse ofthe Hessian of 119869(120572) evaluated at 120572 = that is Σ

120572=

[H()]minus1

equiv [nabla119869(120572)nabla119879|120572=]minus1 [22] By employing the Bayesian

spectral density approachwith themeasured structural accel-eration response the modal parameters and the associatedestimation uncertainties can be determined The associateduncertainty is important for the subsequent step of damagedetection usingmodal parameters [23]TheBayesian spectraldensity approach can be applied also for model updating ofnonlinear dynamical systems [24]

3 Instrumentation of the East Asia Hall

Full-scale measurement of a residential building namelythe East Asia Hall under the passage of typhoon Vicente ispresented in this section This building was inaugurated in2005 for athletes lodging in the 4th East Asian Games hostedin Macao Thereafter it has been serving as a dormitory forthe University of Macau Figure 1 shows the side view and atypical floor plan of the building It is a 22-story reinforcedconcrete building of 6470m height and its floor layer is inL-shape with span lengths 5190m and 6175m

A Gill-type ultrasonic anemometer was utilized to recordthe wind speed and wind direction time histories It wasmounted on a 10m height mast at the top of the building(Figure 2) The resolution of the measured horizontal windspeed and wind direction was 001ms and 01∘ respectivelyFurthermore the sampling frequency was 32Hz On theother hand a biaxial state-of-the-art accelerometer wasinstalled on the 18th floor (corresponding to 535m heightfrom the ground out of the total building height of 647m)to capture the acceleration response of the building It wasplaced at the junction of the two spans and the measureddirections were indicated by the two perpendicular arrowsin Figure 1(b) The accelerometer was operated under the

Figure 2 Photo of anemometer and its supporting mast

standard exploration geophone spring-mass systemwith sen-sitivity of 50Vg and sampling frequency of 200Hz

4 Typhoon Vicente

41 Meteorological Information Typhoon Vicente developedas a tropical depression in the Northeast of Manila overthe Western North Pacific Ocean on 20 July 2012 On thenext day it entered the South China Sea and the StandbyTropical Cyclone Signal Number 1 was announced in Macaoby the Meteorological and Geophysical Bureau On 22 JulyVicente stalled over the Northeastern region of the SouthChina Sea and it was upgraded to a tropical storm On23 July it began to edge towards the West of Pearl RiverEstuary of the South China coast As the sustained windspeed became higher than 41 kmhr the Signal Number 3 wasreleased at 0630 Vicente strengthened rapidly and attainedto a severe typhoon Along with the increasing wind speed


Figure 3 Track of typhoon Vicente (provided by the Meteorologi-cal andGeophysical Bureau (httpwwwsmggovmowwwe indexphp) and the time corresponding to GMT +0800)

the Signal Number 8 was hoisted at 1900 on the same dayAt 0215 on 24 July the Signal Number 9 was issued due tothe strength of the gale gust and storm force wind Vicentepassed through Macao from the South-Southwest with theshortest core distance of 40 km and its maximum near-center10-minute sustained wind reached 150 kmhr AfterwardsVicente gradually moved away and made a landfall near theTaishan city in the Guangdong province The signal numberwas subsequently downgraded to Number 8 and Number 3 at0500 and 0930 respectively The local wind was weakenedduring the dissipation of Vicente Afterwards all tropicalcyclone signals were cancelled at 1620 The typhoon trackand the announced tropical cyclone signals of Vicente aresummarized in Figure 3 and Table 1 respectively From 1800on 21 July to 1620 on 24 July different tropical cyclone signalswere hoisted due to the relevant typhoon generated windsituations Full-scale monitoring was conducted for 70 hoursand 20 minutes to cover the entire tropical cyclone signalhoisting period

Vicente was the strongest and most devastating typhoonstruck Macao since the severe typhoon York in 1999 It wasalso the only typhoon with Signal Number 9 or above from2000 to 2013 It induced severe deterioration of the localweather The daily precipitation of 1432mm was the highestrecord of July since 2001 and the pressure of 9642 hPa wasthe lowest record since 1983 Moreover the near eyewallregion of Vicente coveredMacao and the monitored buildingexperienced the largest wind excitation in its entire history

42 Wind Measurement Thewind measurements includingthe wind speed and wind direction obtained from theanemometer are utilized for the wind analysis Figure 4shows the 10-minutemeanwind speed and the correspondingwind direction The dots represent the mean wind speedwhile the crosses represent the wind direction Herein thewind direction is expressed in the azimuth that is countingclockwise from the true North During the first 365 hoursof the monitoring period the Signal Number 1 was hoistedand the 10-minute mean wind speed was below 45 kmhrThereafter the wind speed was increasing for the next twenty

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

060

30

60

90

120

Mea

n w

ind

spee

d (k

mh

r)

0

90

180

270

360

Dire

ctio

n (d

eg)

Wind speed

Direction

Time (mmdd hh)

Figure 4 10-minute mean wind speed and wind direction

hours when Vicente was approaching The dominated winddirection was Northeast

From 1900 on 23 July to 0930 on 24 July the SignalNumber 8 or above was hoisted During this severe windperiod the 10-minute mean wind speed reached its maxi-mum 114 kmhr and the wind direction changed graduallyfrom Northeast to Southeast After the Signal Number 8was replaced by the Signal Number 3 the 10-minute meanwind speed became lower than 45 kmhr Meanwhile thewind direction remained in Southeast towards the end of themonitoring period and it merged to the background winddirection before the typhoon event

Figure 5 shows the gust speed under severe wind loadGust speed is commonly utilized to illustrate the instant windproperties and it is defined as the highest wind speed overeach 3-second interval The maximum gust speed 185 kmhroccurred at the midnight of 24 July On the other hand thegust speed exhibited amanifest drop for an hour around 0300on 24 July According to the typhoon track shown in Figure 1Vicente arrived at the location of the shortest distance toMacao at that time The measurements reveal that Macaoentered the typhoon eye region and hence the typhoongenerated wind suddenly dropped from over 100 kmhr tothe range from 20 kmhr to 60 kmhr As Vicente was leavingMacao no longer stayed inside the typhoon eye region and thegust speed recovered up to 140 kmhr The recovery was notcomplete because Vicente started to land on the Guangdongcoast at this stage

43 Turbulence Intensity Gust Factor and Wind SpectraTurbulence intensity and gust factor are widely adopted torepresent the statistical features of the atmospheric turbu-lence and wind gustiness [25 26] The turbulence intensity119868119906 is defined as the ratio between the standard deviation of


Table 1 Announced tropical cyclone signals of Vicente by the Macao Meteorological and Geophysical Bureau

Tropical cyclonesignal

Sustained wind speed(kmhr)

Gust speed(kmhr)

Hosting time(hrminute mmdd GMT +0800) Duration

1 lt41 mdash 1800 0721 36 hr 30min3 [41 62] 110 0630 0723 12 hr 30min8 [63 117] 180 1900 0723 7 hr 15min9 lt118 mdash 0215 0724 2 hr 45min8 [63 117] 180 0500 0724 4 hr 30min3 [41 62] 110 0930 0724 6 hr 50min0 lt41 mdash 1620 0724 mdash

0

20

40

60

80

100

120

140

160

180

200

Gus

t spe

ed (k

mh

r)

July

23

16

July

23

20

July

24

00

July

24

04

July

24

08

Time (mmdd hh)

Figure 5 Gust speed of under severe wind load

fluctuating wind 120590119906 and the mean wind speed 119880 calculatedfor each 10-minute interval [27]

119868119906 =120590119906

119880

(7)

On the other hand the gust factor 119866119906 is defined as theratio between the maximum 3-second sustained wind speedmax(1198803119904) and the mean wind speed119880 calculated for each 10-minute interval [27 28]

119866119906 =

max (1198803119904)

119880

(8)

Figure 6 shows the time histories of the turbulenceintensity and gust factor Throughout the entire monitoringperiod the turbulence intensity varied between 008 and 072and the gust factor varied between 115 and 402 Both ratioswere relatively more stable in the range of lower values asso-ciated with high wind speed In contrast their values at thebeginning and ending stages (ie corresponding to relativelycalm wind conditions) were larger and more fluctuatingFurthermore the relationship between the gust factor and

0

02

04

06

08

Turb

ulen

ce in

tens

ity

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

(a)

1

2

3

4

5

Gus

t fac

tor

Time (mmdd hh)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

(b)

Figure 6 Time histories of (a) turbulence intensity and (b) gustfactor

turbulence intensity is presented in Figure 7 The fluctuationof this severe typhoon is illustrated by the fact that thereare a number of data points with gust factors over 20 andturbulence intensity over 05 Furthermore it demonstratesclearly the strong positive correlation between these twoquantities This relationship can be well approximated aslinear [27 29] with the high coefficient of determination of1198772= 08544 Moreover it is observed that the data points

on the top right corner which is corresponding to the regionof high gust factor and high turbulence intensity are morescattering in comparison with the low-value region

Figure 8 shows the spectra of wind speed square atdifferent tropical cyclone signals The wind speed square isused for demonstration because it is proportional to theinduced wind pressure that affects the structural response


0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45

Turbulence intensity

Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544

Figure 7 Gust factor versus turbulence intensity

It can be seen that most of the frequency contents occurredbelow 01Hz Moreover the frequency distribution varies fordifferent tropical cyclone signals

5 Structural Performance

51 Acceleration Response The acceleration responses of thebuilding were measured in the two orthogonal directionsshown in Figure 1(b) To illustrate the global trend of thestructural response the root-mean-square (RMS) accelera-tion is calculated for every 10-minute segment of the accelera-tionmeasurements and it is presented in Figure 9These RMSaccelerations lay within the interval [1023 times 10minus4 5322 times

10minus3]ms2 and [9554 times 10minus5 3654 times 10minus3]ms2 for direction1 and 2 respectively It was found that the RMS accelerationunder severe wind load could be more than 10 times ofthe magnitude under Signal Number 1 inducing more than100 times of wind load to the structure Furthermore itis observed that the acceleration responses in direction 1were generally higher than that in direction 2 Due to theasymmetrical configuration of the building the windwardareas of the two spans are different Moreover the wind forcecontribution to the structural response depends highly onthe attacking angle It turns out that the difference is moresignificant when the strength of the wind is higher and thewind direction changes from Northeast to Southeast

In order to visualize the structural response during thesevere wind period the acceleration responses measuredfrom 1900 on 23 July to 0930 on 24 July (ie the periodwhen Signal Number 8 or above was hoisting) are shownin Figure 10 As expected the amplitude of the accelerationresponse in direction 1 was generally higher than that indirection 2 Moreover a notable reduction of the accelerationresponses occurred for an hour around 0300 on 24 July whenthe eye region of the typhoon enteredMacao Since the severewind load was temporarily removed the amplitude of thestructural response was similar as that under Signal Number

1 only It is worth noting that the observations are consistentwith the conclusions drawn from Figure 5

52 Modal Identification Results The acceleration responsesof the entire monitoring period are utilized for the modalidentification For each 10-minute segment of the accelerationresponses the Bayesian spectral density approach is appliedto estimate the modal frequencies and damping ratios of thebuilding and the spectral intensities of the wind excitation aswell as their associated covariance matrices Therefore thevariations of the identified modal parameters as well as theestimation uncertainties can be traced

Spectral intensity of the modal force describes the excita-tion energy for a mode The time histories of the identifiedspectral intensities of the modal forces for the first threemodes are shown in Figure 11 with the semilogarithmic scaleIt is found that themaximum values of the spectral intensitiesof the modal forces were thousand times higher than thecorresponding values under Signal Number 1 Furthermorethe overall variations showed a similar trend as the 10-minutemean wind speed presented in Figure 4 For instance theresults continuously increased when Vicente was approach-ing Macao A sudden drop of the excitation energy occurredwhen Macao was covered by the eye region of the typhoonThen the structural response decreased rapidly during thedissipation of the typhoon effect

Figure 12 shows the identified modal frequencies anddamping ratios of the building with the associated estimationuncertainties The identified modal frequencies are shownin the left column while the identified damping ratios areshown in the right column The solid lines represent theidentified values and the dotted lines represent the plus andminus three standard derivations confidence intervals (ieplusmn3120590 of the estimates) which yield a probability of 997Since the building may not behave linearly under the severewind excitation the identified structural modal parametersare referred to the corresponding quantities of the equivalentlinear system

Although the trends of the identified modal frequenciesand damping ratios were opposite all their peak valuesoccurred approximately at the same timewhen themaximumvalues of the identified spectral intensities of the modalforceswere achieved Considerable reduction of the identifiedmodal frequencies was observed during the severe windperiodThedifferences between themaximumandminimumof the three concerned modes were 835 590 and333 respectively Moreover the standard deviations ofthe estimates were less than 05 for all the three modesIt turns out that the confidence intervals were sufficientlynarrow comparedwith the variations of estimates throughoutthe monitoring period This statistical evidence confirmedthat instead of the statistical uncertainty there was notablereduction of the modal frequencies due to the severe windload subjected to the building

On the other hand the identified damping ratios of thethree concerned modes were in the range of [050 373][019 359] and [017 143] respectively The asso-ciated maximum standard derivations were 080 048


0

500

1000

15002100 July 22 (number 1)

10minus2 10minus1 100 101

S WS2

(a)

0

2

40300 July 24 (number 9)


times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)

Frequency (Hz)10minus2 10minus1 100 101

S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)

Figure 8 Wind speed square spectra under different tropical cyclone signals

and 023 respectively The fluctuation of the estimates andthe estimation uncertainties were significantly larger thanthose for the modal frequencies Nevertheless it is sufficientto conclude that the damping ratios were higher during thesevere wind period

53 Investigation of Permanent Effect Next the identifiedstructural modal parameters at the beginning and the endof the monitoring period are compared From Figure 12it is realized that some of the modal frequencies werenot fully recovered even after the typhoon generated windexcitation was dissipated The largest difference is found tobe 469 for the identified modal frequency of the firstmode The two major sources of this phenomenon are thedramatic change of the environmental conditions and the

nonlinear behavior of the structure Due to the atmosphericmechanism of tropical cyclones the scorching weather at theearly stage of the typhoon event changes to the cool andshowering condition afterwards Hence there were notabledifferences in the corresponding environmental conditions ofthe temperature and relative humidity For instance the dailyaverage temperature and relative humidity on 21 July and 24July were (308∘C 74) and (260∘C 92) respectively Pre-vious studies revealed that the structural modal frequenciesdepend on the environmental conditions [30 31] Thereforedirect comparison between the modal frequencies shouldbe conducted under same environmental conditions Takingthis into account the identified structural modal parametersobtained on 7 August 2012 which was two weeks after thetyphoon event is utilized for comparisonThe environmental


0

1

2

3

4

5

6

Direction 1Direction 2

times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

Figure 9 10-minute RMS structural acceleration response

002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)

Figure 10 Acceleration response during the severe wind period

conditions of this day returned to subtropical climatewith hotand humid summer weather and the environmental condi-tions were similar to the conditions at the beginning of thetyphoon event In particular the daily average temperatureand relative humidity were 295∘C and 79 respectivelyTherefore indication of possible permanent effects on the

July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)

Figure 11 Time histories of the identified spectral intensities of themodal forces

building can be achieved by comparing the modal frequen-cies of this day with the beginning of the typhoon Againthe modal frequencies and damping ratios were identified forevery 10-minute response measurement on 7 August and theshaded rectangles in Figure 12 enclose the ranges for eachmodal parameter It is found that they covered the rangesof the early stage of the typhoon event so it indicates thatall the modal frequencies and damping ratios recovered totheir original levels when similar environmental conditionsare encountered after the typhoon In otherwords it confirmsthat no permanent effect was induced by Vicente to thebuilding

In order to demonstrate the hysteretic behavior of thestructure the relationships between the identified structuralmodal parameters and the identified spectral intensities ofthe corresponding modal forces are presented in Figure 13The data points under the loading and releasing process arerepresented by the dots and crosses respectively Herein theloading process is referred to the stage with increasing 10-minute RMS structural response while the releasing processis referred to the latter stage with decreasing 10-minute RMSstructural response The left column shows the identifiedmodal frequencies versus the identified spectral intensitiesof the corresponding modal forces For all three modes theidentified modal frequencies had a decreasing trend with thespectral intensities This indicates certain nonlinear behaviorof the building Reduction of the equivalent linear stiffnessindicates the nonlinear behavior of this reinforced concretebuilding under severe wind load As a result downwardtrends can be observed in all these figures in the left columnMoreover it is observed that the modal frequencies weregenerally lower in the releasing process when the spectral


13

135

14

145

15

1205961

(Hz)

July 21 18 July 22 18 July 23 18Time (mmdd hh)

(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)

Figure 12 Time histories of the identified structural modal parameters

intensities were high It is noted that the ambient temperatureand relative humidity were stable during the severe windperiod so the variation of the structural properties due tothe environmental influences were negligible at this stage

One possible explanation is that the hysteretic behaviorof the structure for the loading and releasing process wasdifferent and this is commonly observed for reinforced con-crete structures [32 33] When the spectral intensities were


13

135

14

145

15

1205961

(Hz)

10minus16 10minus12 10minus8

S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)

Figure 13 Identified structural modal parameters versus the identified spectral intensity of the modal force

decreasing the modal frequencies in the releasing processwere lower than the corresponding values with the same levelof excitation energy in the loading process

The identified damping ratios versus the identified spec-tral intensities of the corresponding modal forces are shownon the right column in Figure 13 For all the three modesincreasing trends were observed as the excitation energyincreased This observation reconfirms the nonlinear hys-teretic behavior of the building As a result energy wasdissipated more efficiently through the hysteretic loops sothe equivalent damping ratios were significantly increasedunder severe wind condition Moreover it is found thatthe identified damping ratios of the releasing process wereslightly larger than those of the loading process when the

identified spectral intensities of the modal forces were highThe temporary increase of the damping ratios vanished whenthe excitation energy returned to the calm wind conditions

Figure 14 shows the identified modal frequencies versusthe 10-minute mean wind speed during the loading processFor all the three concerned modes decreasing trends areobserved It is found that the data points are distributedmore closely along the trend lines for the low wind speedregion When the wind speed was higher than 60 kmhr thedata points became more scattering It turns out that thecoefficients of determination (1198772) of the three modes are08012 08448 and 07405 respectively and they providestrong evidence of the nonlinear behavior of the buildingduring strong wind load


0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012

10min mean WS (kmhr)

(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)

Figure 14 Identified modal frequencies versus 10-minute mean wind speed

6 Conclusion

In this study the typhoon effects on the structural perfor-mance of a reinforced concrete building were investigatedWind and structural response measurements were acquiredthroughout the passage of the severe typhoon Vicente Itwas the strongest typhoon for Macao since 1999 During themonitoring period Macao was covered by the typhoon eyeregion for around an hour The measured data were utilizedto evaluate the wind characteristics as well as the structuralproperties throughout themonitoring periodThemagnitudeof the acceleration response under Signal Number 9 could bemore than 10 times higher than that under Signal Number1 When the monitored building was inside the typhooneye region of Vicente both the wind speed and structuralresponse measurements were temporarily decreased for asignificant amount Furthermore the structural responseswere utilized for modal identification using the Bayesianspectral density approach As the spectral intensities of themodal forces increased the modal frequencies decreasedbut the damping ratios increased During the severe windexcitation nonlinear hysteretic behavior of the structure wasobserved but no permanent effect was induced

Acknowledgments

This work was supported by the Research Committee ofUniversity of Macau under Research Grant MYRG081 (Y1-L2)-FST13-YKV and the Science and Technology Develop-ment Fund of the Macau SAR government under ResearchGrant 0122013A1 These generous supports are gratefullyacknowledged

References

[1] T Kijewski-Correa and J D Pirnia ldquoDynamic behavior of tallbuildings under wind insights from full-scale monitoringrdquoTheStructural Design of Tall and Special Buildings vol 16 no 4 pp471ndash486 2007

[2] Y L Xu and S Zhan ldquoField measurements of Di Wangtower during typhoon Yorkrdquo Journal of Wind Engineering andIndustrial Aerodynamics vol 89 no 1 pp 73ndash93 2001

[3] Q S Li Y Q Xiao J Y Fu and Z N Li ldquoFull-scalemeasurements of wind effects on the Jin Mao buildingrdquo JournalofWind Engineering and Industrial Aerodynamics vol 95 no 6pp 445ndash466 2007

[4] Q S Li Y Q Xiao J R Wu J Y Fu and Z N Li ldquoTyphooneffects on super-tall buildingsrdquo Journal of Sound and Vibrationvol 313 no 3ndash5 pp 581ndash602 2008

[5] W H Chen Z R Lu W Lin et al ldquoTheoretical and exper-imental modal analysis of the Guangzhou new TV towerrdquoEngineering Structures vol 33 no 12 pp 3628ndash3646 2011

[6] A C Khanduri T Stathopoulos and C Bedard ldquoWind-induced interference effects on buildingsmdasha review of the state-of-the-artrdquo Engineering Structures vol 20 no 7 pp 617ndash6301998

[7] J Chen and Y L Xu ldquoOn modelling of typhoon-induced non-stationary wind speed for tall buildingsrdquo The Structural Designof Tall and Special Buildings vol 13 no 2 pp 145ndash163 2004

[8] S K Au and P To ldquoFull-scale validation of dynamic windload on a super-tall building under strong windrdquo Journal ofStructural Engineering vol 138 no 9 pp 1161ndash1172 2012

[9] J N Yang Y Lei S Lin and N Huang ldquoHilbert-Huangbased approach for structural damage detectionrdquo Journal ofEngineering Mechanics vol 130 no 1 pp 85ndash95 2004

[10] K V Yuen and L S Katafygiotis ldquoAn efficient simulationmethod for reliability analysis of linear dynamical systems using


simple additive rules of probabilityrdquo Probabilistic EngineeringMechanics vol 20 no 1 pp 109ndash114 2005

[11] C Papadimitriou C P Fritzen P Kraemer and E NtotsiosldquoFatigue predictions in entire body of metallic structures froma limited number of vibration sensors using Kalman filteringrdquoStructural Control andHealthMonitoring vol 18 no 5 pp 554ndash573 2011

[12] C Papadimitriou E Ntotsios D Giagopoulos and S Natsi-avas ldquoVariability of updated finite element models and theirpredictions consistent with vibrationmeasurementsrdquo StructuralControl and Health Monitoring vol 19 no 5 pp 630ndash654 2012

[13] Y Lei Y Su andW Shen ldquoA probabilistic damage identificationapproach for structures under unknown excitation and withmeasurement uncertaintiesrdquo Journal of Applied Mathematicsvol 2013 Article ID 759102 7 pages 2013

[14] S Kitipornchai W Kang H F Lam and F Albermani ldquoFactorsaffecting the design and construction of Lamella suspen-domesystemsrdquo Journal of Constructional Steel Research vol 61 no 6pp 764ndash785 2005

[15] L S Katafygiotis and K V Yuen ldquoBayesian spectral densityapproach for modal updating using ambient datardquo EarthquakeEngineering and Structural Dynamics vol 30 no 8 pp 1103ndash1123 2001

[16] J L Beck and L S Katafygiotis ldquoUpdating models and theiruncertainties I bayesian statistical frameworkrdquo Journal ofEngineering Mechanics vol 124 no 4 pp 455ndash461 1998

[17] K V Yuen and H Q Mu ldquoA novel probabilistic methodfor robust parametric identification and outlier detectionrdquoProbabilistic Engineering Mechanics vol 30 pp 48ndash59 2012

[18] K V Yuen P F Liang and S C Kuok ldquoOnline estimation ofnoise parameters for Kalman filterrdquo Structural Engineering andMechanics vol 47 no 3 pp 361ndash381 2013

[19] S C Kuok and K V Yuen ldquoStructural health monitoring ofCanton tower using Bayesian frameworkrdquo Smart Structures andSystems vol 10 no 4-5 pp 375ndash391 2012

[20] K V Yuen L S Katafygiotis and J L Beck ldquoSpectral densityestimation of stochastic vector processesrdquo Probabilistic Engi-neering Mechanics vol 17 no 3 pp 265ndash272 2002

[21] K V Yuen and S C Kuok ldquoBayesian methods for updatingdynamic modelsrdquo Applied Mechanics Reviews vol 64 no 1Article ID 010802 18 pages 2011

[22] K V Yuen Bayesian Methods for Structural Dynamics and CivilEngineering John Wiley amp Sons New York NY USA 2010

[23] H F Lam K V Yuen and J L Beck ldquoStructural healthmonitoring via measured Ritz vectors utilizing artificial neuralnetworksrdquo Computer-Aided Civil and Infrastructure Engineer-ing vol 21 no 4 pp 232ndash241 2006

[24] K V Yuen and J L Beck ldquoUpdating properties of nonlineardynamical systemswith uncertain inputrdquo Journal of EngineeringMechanics vol 129 no 1 pp 9ndash20 2003

[25] P J Vickery and P F Skerlj ldquoHurricane gust factors revisitedrdquoJournal of Structural Engineering vol 131 no 5 pp 825ndash8322005

[26] S S Law J Q Bu X Q Zhu and S L Chan ldquoWindcharacteristics of TyphoonDujuan asmeasured at a 50m guyedmastrdquoWind and Structures vol 9 no 5 pp 387ndash396 2006

[27] Q S Li Y Q Xiao C K Wong and A P Jeary ldquoFieldmeasurements of typhoon effects on a super tall buildingrdquoEngineering Structures vol 26 no 2 pp 233ndash244 2004

[28] L SongQ S LiWChen PQinHHuang andYCHe ldquoWindcharacteristics of a strong typhoon in marine surface boundarylayerrdquoWind and Structures vol 15 no 1 pp 1ndash15 2012

[29] J Y Fu J R Wu A Xu Q S Li and Y Q Xiao ldquoFull-scalemeasurements of wind effects onGuangzhouWest TowerrdquoEngineering Structures vol 35 pp 120ndash139 2012

[30] J F Clinton S C Bradford T H Heaton and J Favela ldquoTheobserved wander of the natural frequencies in a structurerdquoBulletin of the Seismological Society of America vol 96 no 1pp 237ndash257 2006

[31] K V Yuen and S C Kuok ldquoAmbient interference in long-termmonitoring of buildingsrdquo Engineering Structures vol 32 no 8pp 2379ndash2386 2010

[32] Y Tamura and S Y Suganuma ldquoEvaluation of amplitude-dependent damping and natural frequency of buildings duringstrong windsrdquo Journal of Wind Engineering and IndustrialAerodynamics vol 59 no 2-3 pp 115ndash130 1996

[33] H Sohn C R Farrar F M Hemez D D Shunk D WStinemates and B R Nadler ldquoA review of structural healthmonitoring literature 1996ndash2001rdquo Los Alamos National Labo-ratory Report LA-13976-MS 2003

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom


throughMacao with the shortest core distance of only 40 kmVicente is considered the strongest and most devastatingtyphoon for Macao since 1999 Since critical aerodynamicconditions were generated to the infrastructures in the city itprovided a valuable opportunity to investigate the structuralbehavior under harsh wind situation In order to illustrate thewind conditions the meteorological information statisticalproperties of the wind measurements and gust speed will bepresented Furthermore the wind characteristic propertiesincluding the turbulence intensity and gust factor will beanalyzed

On the other hand the structural acceleration responsesare analyzed and the Bayesian spectral density approach[15] is applied to identify the time-varying modal parame-ters Bayesian inference provides a rigorous framework forparametric identification and has been applied to variousengineering disciplines [8 10ndash13 16ndash18] The Bayesian spec-tral density approach requires only the structural responsemeasurement and it has been adopted for analyzing thestructural response of the Canton tower [19] One specialfeature of this frequency-domain approach is that it takes fulladvantage of Bayesian inference for the posterior uncertaintyquantification of the modal estimation As a result thestatistical uncertainties can be distinguished from the actualchanges of the identified parameters This is essential forreliable judgment about the structural integrity In this paperthe time-varying modal parameters as well as the associatedestimation uncertainties will be identified based on thestructural response measurements of the entire monitoringperiod

This paper is organized as follows In Section 2 theBayesian spectral density approach for modal identificationis briefly reviewed In Section 3 the instrumentation on themonitored building is described In Section 4 the aerody-namic properties of Vicente are presented Finally Section 5presents the results and discussions about the typhoon effectson the structural performance of the monitored building

2 Bayesian Spectral Density Approach forModal Identification

Consider a linear dynamical system with 119873119889 degrees offreedom

Mx + Cx + Kx = T0F (119905) (1)

where M C and K are the mass damping and stiffnessmatrix respectively and T0 is the force distributing matrixThe external excitation F can be modeled as zero-meanGaussian white noise with spectral intensity matrix S119865(120596) =S1198650 The measurement Y119873 = y(119899) 119899 = 1 2 119873

contains119873119900 channels of structural response corrupted by themeasurement noise 120576

y (119899) = L0q (119899) + 120576 (119899) (2)

where y(119899) isin R119873119900 is the measurement at the 119899th timestep q(119899) isin R119873119889 is the concerned structural response (egacceleration) at the same time step and L0 isin R119873119900times119873119889 is

the observation matrix comprised of zeros and ones Themeasurement noise 120576 is modeled as zero-mean Gaussianindependent and identical distributed (iid) process withcovariance matrix Σ

120576

Use 120572 to denote the uncertain modal parameter vectorto be identified It includes the structural modal parameters(ie modal frequencies damping ratios and mode shapecomponents) of the contributing modes and the character-istic parameters of the spectral intensity of the excitationand the measurement noise To identify the uncertain modalparameter vector 120572 a discrete estimator of the power spectraldensity matrix is utilized [15]

S119910119873 (120596119896) =Δ119905

2120587119873

119873minus1

sum

1198991198991015840=0

y (119899) y (1198991015840)119879

exp [minus119894120596119896 (119899 minus 1198991015840) Δ119905]

(3)

where Δ119905 is the sampling time step the superscript 119879 denotesthe transpose of a vector Δ120596 = 2120587(119873Δ119905) is the frequencyprecision in the discrete Fourier transform and 120596119896 =

119896Δ120596 119896 = 0 1 INT(1198732)With 119873119904 independent sets of discrete time histories Y =

Y(119904)119873 119904 = 1 2 119873119904 the averaged spectral density matrix

estimator can be obtained

Savg119910119873

(120596119896) =1

119873119904

119873119904

sum

119904=1

S(119904)119910119873

(120596119896) (4)

where S(119904)119910119873

(120596119896) can be calculated using (3) with the measure-ment Y(119904)

119873 Given that119873119904 ge 119873119900 the averaged spectral density

matrix estimator follows the central complex 119873119900-variateWishart distribution with 119873119904 degrees of freedom With aproperly selected frequency range Ξ the averaged spectraldensity matrix estimators in Savg

Ξ= Savg119910119873

(120596119896) 120596119896 120598 Ξ areapproximately independent [20]

Using the Bayesrsquo theorem the posterior probability den-sity function (PDF) of 120572 given Savg

Ξis [16 21]

119901 (120572 | SavgΞ) =1198881119901 (120572) 119901 (Savg

Ξ| 120572) (5)

where 1198881 is a normalizing constant The prior PDF 119901(120572)

represents the prior information of the modal parametersin 120572 Throughout this study it is taken as a noninformativeprior distribution so it can be absorbed into the normalizingconstant The likelihood function 119901(Savg

Ξ| 120572) is given by the

product of Wishart distributions [15]

119901 (SavgΞ

| 120572)

= 1198882prod

120596119896isinΞ

10038161003816100381610038161003816119864 [S119910119873 (120596119896) | 120572]

10038161003816100381610038161003816

minus119873119904

times exp [minus119873119904 tr (119864 [S119910119873 (120596119896) | 120572]minus1

Savg119910119873

(120596119896))]

(6)

where 1198882 is a constant that does not depend on the modalparameters E[sdot] is the mathematical expectation | sdot | andtr(sdot) are the determinant and trace of a matrix respectivelyThe optimal modal parameter vector can be determined by


(a)


N

(b)





120572=

[H()]minus1








4 Typhoon Vicente







July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

060

30

60

90

120

Mea

n w

ind

spee

d (k

mh

r)

0

90

180

270

360

Dire

ctio

n (d

eg)

Wind speed

Direction

Time (mmdd hh)










Gust speed(kmhr)



0

20

40

60

80

100

120

140

160

180

200

Gus

t spe

ed (k

mh

r)

July

23

16

July

23

20

July

24

00

July

24

04

July

24

08

Time (mmdd hh)



119868119906 =120590119906

119880

(7)


119866119906 =

max (1198803119904)

119880

(8)


0

02

04

06

08

Turb

ulen

ce in

tens

ity

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

(a)

1

2

3

4

5

Gus

t fac

tor

Time (mmdd hh)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

(b)






0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45


Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544














0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















(a)


N

(b)





120572=

[H()]minus1








4 Typhoon Vicente







July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

060

30

60

90

120

Mea

n w

ind

spee

d (k

mh

r)

0

90

180

270

360

Dire

ctio

n (d

eg)

Wind speed

Direction

Time (mmdd hh)










Gust speed(kmhr)



0

20

40

60

80

100

120

140

160

180

200

Gus

t spe

ed (k

mh

r)

July

23

16

July

23

20

July

24

00

July

24

04

July

24

08

Time (mmdd hh)



119868119906 =120590119906

119880

(7)


119866119906 =

max (1198803119904)

119880

(8)


0

02

04

06

08

Turb

ulen

ce in

tens

ity

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

(a)

1

2

3

4

5

Gus

t fac

tor

Time (mmdd hh)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

(b)






0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45


Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544














0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





















July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

060

30

60

90

120

Mea

n w

ind

spee

d (k

mh

r)

0

90

180

270

360

Dire

ctio

n (d

eg)

Wind speed

Direction

Time (mmdd hh)










Gust speed(kmhr)



0

20

40

60

80

100

120

140

160

180

200

Gus

t spe

ed (k

mh

r)

July

23

16

July

23

20

July

24

00

July

24

04

July

24

08

Time (mmdd hh)



119868119906 =120590119906

119880

(7)


119866119906 =

max (1198803119904)

119880

(8)


0

02

04

06

08

Turb

ulen

ce in

tens

ity

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

(a)

1

2

3

4

5

Gus

t fac

tor

Time (mmdd hh)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

(b)






0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45


Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544














0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of




















Gust speed(kmhr)



0

20

40

60

80

100

120

140

160

180

200

Gus

t spe

ed (k

mh

r)

July

23

16

July

23

20

July

24

00

July

24

04

July

24

08

Time (mmdd hh)



119868119906 =120590119906

119880

(7)


119866119906 =

max (1198803119904)

119880

(8)


0

02

04

06

08

Turb

ulen

ce in

tens

ity

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)

(a)

1

2

3

4

5

Gus

t fac

tor

Time (mmdd hh)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

(b)






0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45


Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544














0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















0 01 02 03 04 05 06 07 0805

1

15

2

25

3

35

4

45


Gus

t fac

tor

Gu = 27637 Iu + 08283

R2 = 08544














0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















0

500

1000

15002100 July 22 (number 1)


S WS2

(a)

0

2



times105

(b)

0

2000

4000

6000

80001200 July 23 (number 3)


S WS2

(c)

0

5



times104

(d)

0

5000

10000

150002100 July 23 (number 8)


S WS2

(e)

Frequency (Hz)

0

2

4

6



times104

(f)






0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















0

1

2

3

4

5

6


times10minus3

RMS

acce

lera

tion

(ms2)

July

21

18

July

22

06

July

22

18

July

23

06

July

23

18

July

24

06

Time (mmdd hh)


002

0

0

004

0

July

23

19

July

23

23

July

24

03

July

24

07

minus004

002

minus002

004

0

Direction 1

Direction 2

minus004

minus002

July

23

19

July

23

23

July

24

03

July

24

07Ac

cele

ratio

n (m

s2)

Acce

lera

tion

(ms2)

Time (mmdd hh)

Time (mmdd hh)



July

21

18

July

22

18

July

23

18

July

24

06

July

23

06

July

22

0610minus16

10minus15

10minus14

10minus13

10minus12

10minus11

10minus10

10minus9

10minus8

S(11)f0

S(22)f0

S(33)f0

S(mm

)f0

Time (mmdd hh)





13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















13

135

14

145

15

1205961

(Hz)


(a)

0

2

4

6

120577 1(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(b)

155

16

165

17

1205962

(Hz)


(c)

0

2

4

6

120577 2(

)

July 21 18 July 22 18 July 23 18

Time (mmdd hh)

(d)

175

18

185

1205963

(Hz)


(e)

July 21 18 July 22 18 July 23 180

1

2

3

120577 3(

)

Time (mmdd hh)

(f)





13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















13

135

14

145

15

1205961

(Hz)


S(11)f0

(a)

0

1

2

3

4


S(11)f0

120577 1(

)

(b)

120577 2(

)

155

16

165

17

1205962

(Hz)


S(22)f0

(c)

0

1

2

3

4


S(22)f0

(d)

175

18

185

1205963

(Hz)


S(33)f0

(e)

0

1

05

15

2

120577 3(

)


S(33)f0

(f)







0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















0 40 80 120132

134

136

138

14

142

144

1461205961

(Hz)

y = minus000094744x + 14406

R2 = 08012


(a)

0 40 80 120158

16

162

164

166

168

17

1205962

(Hz)

R2 = 08448

y = minus00006655x + 16698


(b)

0 40 80 120176

177

178

179

18

181

182

183

184

1205963

(Hz)

y = minus000046939x + 18172

R2 = 07405


(c)


6 Conclusion


Acknowledgments


References









































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of














































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





















of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of
















Documents

Research Article Structural Health Monitoring of a …downloads.hindawi.com/journals/tswj/2013/509350.pdfe Scientic World Journal (a) Direction 1 Direction 2 N (b) F : (a) Side-view