1

Buffeting Response of Long-span Cable-stayed Bridge during Strong Typhoon

Shouqiang Wang1, Lin Zhao1, Yufeng Zhang3, Feng Yin2, Yaojun Ge1 1State Key Laboratory for Disaster Reduction in Civil Engineering, Tongji University, Shanghai, China 2Shanghai Changjiang Tunnel and Bridge Construction and Development Co., Ltd., Shanghai, China 3Jiangsu Transportation Research Institute, Nanjing, China email: [email protected], [email protected], [email protected]

ABSTRACT: In this paper, we first discuss the wind characteristics and then the bridge response is studied based on the wind analysis result. We study the acceleration responses of the deck over time, conclude that the maximum RMS of acceleration response is not synchronize with wind speed and has some relationship with wind fluctuation characteristics, such as turbulence intensity and turbulence integral scale. In addition, some other significant conclusions are also concluded. Following this, investigation via dynamic method is carried out for comparison.

KEY WORDS: Full scale measurement; Typhoon; Wind characteristics; Bridge acceleration response.

1 INTRODUCTION Typhoon is one of several major natural disasters in China, causing certain economic losses and casualties every year and threatening sustainable development of coastal areas. In China, the eastern coastal area is a powerful typhoon landing area, frequently. According to incomplete statistics, there are about 82 typhoons that had great influence on Shanghai from 1949 to 2005. But because of the particularity and complexity of typhoon, it is difficult for wind tunnel laboratory to make a precise simulation. So full scale measurement becomes an effective means for research [1]. Miyata [2] focused on power spectral density and spatial correlation of wind-speed fluctuation, and response of the deck, using full-scale data measured for the Akashi-Kaikyo Bridge during 9807 and 9918 typhoon. Zhu [3] compared the computed acceleration responses of the bridge deck and cable with the responses from Wind and Structural Health Monitoring System of Tsing Ma Bridge subjected to Typhoon Sam in 1999. Wang [4] also investigated the acceleration responses of the deck and cable of Sutong Bridge under Typhoon Fung-Wong. But the researches are far from enough. Typhoon Muifa (1109) was the ninth named storm, the third typhoon and the second super-typhoon in the 2011 Pacific typhoon season. On August 7, as the typhoon swept the coast of Shanghai, Shanghai Yangtze River Bridge was forced to close. Heavy rains and strong winds fluttered the bridge. Therefore, the research of this typhoon and the bridge response is very important and useful for the study of wind engineering. In this paper, we first discuss the wind characteristics and then the bridge response is studied based on the wind analysis result. We study the acceleration responses of the deck over time, conclude that the maximum RMS of acceleration response is not synchronize with wind speed and has some relationship with wind fluctuation characteristics, such as turbulence intensity and turbulence integral scale. In addition, some other significant conclusions are also concluded. Following this, investigation via dynamic method is carried out for comparison.

2 SENSORS AND TYPHOON MUIFA 2.1 Sensors Test data is collected from health monitoring system of Shanghai Yangtze River Bridge which stretches across the Changjiang Yangtze River amongst Chongming and Changxing island. According to the wind resistance design code in China, the surrounding terrain of the observation position should be classified into A class. There are four wind anemometers, 22 acceleration sensors for vertical component calibration and 7 acceleration sensors for lateral component calibration. In this paper, the anemometer and the acceleration sensor in mid-span are mainly discussed (as shown in figure 1). Test data is from health monitoring system of Shanghai Yangtze River Bridge which is located in the changjiang river between chongming and changxing island. According to the wind resistance design code in China, the surrounding terrain of the observation position should belong to A class. Four anemometers are as follows: two for two main channel bridge towers (number SWS1401, SWS1402 respectively, the level height is 215.7 m), two for the upward side and down side of the midspan

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 2

bridge deck (number SWS1201, SWS1202 respectively, the level height is 70.2 m). Through the Shanghai Yangtze waterway administration official website [5], water level changed over time, but kept in about 3.2 m. In order to simplify the calculation, water level is taken as a constant value (3.2m). The height of stormy waves changed with distance from typhoon and the strength of wind speed. This paper refers to the result of Xiao-bin Chen [6] on Yangtze River during period of typhoon Muifa (1109). In order to simplify the calculation, it is taken as 7m (fixed value). Thus the calculated height of the two points are 60 m and 205.5 m respectively (the following referred to the calculation height without special description) and the horizontal distance is 365 m, as shown in figure 1. The sampling frequency of anemometer is 4 hz, wind angle defined the north is 0°, clockwise rotation, so the east is 90°. According to this role, the direction of the bridge is 40°. The sampling frequency of the acceleration is 50 Hz.

35000 73000 35000

SWS1201 SAH1101 SAH1102 SAH1103 SAH1104 SAH1105 SAH1106 SAH1107 SAV1101 SAV1102 SAV1103 SAV1104 SAV1105 SAV1106 SAV1107 SAV1108 SAV1109 SAV1110 SAV1111 SAV1201 SAV1202 SAV1203 SAV1204 SAV1205 SAV1206 SAV1207 SAV1208 SAV1209 SAV1210 SAV1211 Section1 Section2 Section3 Section4 Section5 Section6 Section7 Section8 Section9 Section10 Section11

Figure 1. Layout of sensors on the bridge (Unit: m). 2.2 Typhoon Muifa Typhoon Muifa, known in the Philippines as Typhoon Kabayan, was a large, strong and persistent typhoon which affected a number of countries in the Pacific, killing 22 and causing widespread damage worth US$480 million. It was the ninth named storm, third typhoon and the second super-typhoon of the 2011 Pacific typhoon season. The low-pressure area which became the typhoon originally formed on 23 July. It gradually drifted to the west, becoming a tropical depression. As it turned north and neared the Philippines it rapidly strengthened, becoming a Category 5 typhoon on the Saffir-Simpson Hurricane Scale (SSHS). In the Philippines, the storm claimed eight lives and caused much damage. The system brought down trees; the northeast Philippines experienced strong winds and heavy rains, leaving motorists stranded on several roads and expressways. Muifa also sank a Malay ship with 178 passengers. The system then drifted north, weakening steadily until it curved to the west and threatened Micronesia. The typhoon hit Okinawa, Japan with 41 inches of rain, flooding the small island and injuring 37 people in a 30-hour period. The storm disrupted air travel, leaving 13,630 people stranded on the island. The system then steadily drifted west, nearing Taiwan and prompting emergency warnings and high alerts; however, the storm missed the island. The typhoon moved further west towards mainland China, causing thousands to flee from their homes. A level-4 high-wave warning was issued, and some 11,000 rescue workers mobilized in 120 teams. A parade of low-pressure areas and tropical disturbances formed from the Intertropical Convergence Zone late on July 15. Late on July 23, one of the last low-pressure systems developed further to a weak tropical disturbance, which formed southeast of Chuuk in Micronesia. The system drifted to the west, and on July 25 the Joint Typhoon Warning Center (JTWC) upgraded the low-pressure area to a tropical depression. At that time, it was located approximately 505 nautical miles (935km; 581mi) west of Guam. At midnight that day, the JMA began monitoring the system as a tropical depression. Early on July 28, the JTWC upgraded the system to a tropical storm. A few hours later the JMA also upgraded the system to a tropical storm, naming it Muifa. Also on 28 July, the storm entered the Philippine Area of Responsibility (PAR); the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA) named it Kabayan. The storm drifted north over the next day, while maintaining tropical-storm strength. On the night of July 29, Muifa was upgraded to a severe tropical storm. Overnight, the storm strengthened rapidly and was upgraded to a typhoon the next morning. According to the JTWC, Muifa had strengthened from a tropical storm to asuper typhoon in less than 24 hours; it reported that the storm was reaching one-minute sustained wind speed of 140 knots (260 km/h; 160 mph). However, the typhoon weakened later in the day. According to the JTWC, on July 31 the typhoon encountered an upper-level trough and weakened to a category-4 typhoon on the SSHS. The system gradually moved north, then turned west and drifted towards Okinawa before turning northwest again (when it was finally downgraded to a tropical storm by the JTWC). Soon afterwards, the JMA downgraded Muifa to a severe tropical storm. After weakening to a tropical storm, Muifa made landfall at the estuary of the Yalu River on August 8 and the JTWC issued its final warning. Early on August 9, Muifa weakened to a tropical depression in northeast China and later became a low-pressure area. The maximum typhoon influence on observation position appeared at 8 on August 7, 8. At this time the typhoon center was located in north latitude 31.8° and 124.4° east longitude, the largest wind was level 13 (40m/s), minimum pressure was 960 mpa (see figure 2). Typhoon Muifa (1109) affected bridge up to 24 hours, causing traffic closed 11 hours. In order to compare the before and after the typhoon maximum and typhoon maximum features, a total of 42 hours data was selected, from 8:00 August 6 to 2:00 August 8, the path of typhoon Muifa was shown in figure 2.

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 3

Figure 2. Path of typhoon Muifa.

3 RESULTS 3.1 Wind characteristics 3.1.1 Mean wind speed and wind direction The data collected from anemometer is under the polar coordinate wind speed S and the azimuth β. Decomposing to the rectangular coordinate system, the north-south direction wind speed time history 푢푥 and the east-west direction wind speed time history 푢푦 can be getten. By 10 min as the basic interval, calculating average wind speed, first calculate the average wind speed

푢̅푥、푢̅푦, and then calculate the average velocity U, namely

2 2 U = √푢̅푥 + 푢̅푦 （1） According to the relationship between the original rectangular coordinate system, the main wind direction Angle can be obtained by formula: u̅x arccos u̅y > 0 ∅ = { U （2） u̅ 3600 − arccos x u̅ < 0 U y Figure 3 and figure 4 are 10-min average wind speed changing with time and the main direction changing with time, respectively.

35 60m 30 205.5m 25

20 maximum impact 15

10

wind speed（m/s） wind 5

0 8:00 16:00 0:00 8:00 16:00 0:00 2011.8.6 2011.8.6 2011.8.7 time 2011.8.7 2011.8.7 2011.8.8

Fig. 3 10 min-mean wind speeds during typhoon Muifa

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 4

360

300

240

180 maximum impact 60m 205.5m 120

60 wind direction(°) wind

0 8:00 16:00 0:00 8:00 16:00 0:00 2011.8.6 2011.8.6 2011.8.7 time 2011.8.7 2011.8.7 2011.8.8

Fig. 4 10 min-mean wind direction during typhoon Muifa

From figure 3, 10-min average wind speed for 60m and 205.5m reach to the maximum at the same time in, 27.0m/s and 33.5m/s, respectively, and the main wind direction angle 332° and 323°, respectively. For the small differences of wind direction, the main reason for this difference is caused by different measuring point of the horizontal position and height. 3.1.2 The fluctuation component of the wind 3.1.2.a Turbulence intensity Turbulence intensity is the characterization of the flow fluctuation index. Because the fluctuating wind has 3-d feature, the evaluation index of turbulent intensity has a total of three indexes, including along, horizontal and vertical turbulence intensity. Due to along wind direction and horizontal direction of the wind characteristics is more important to the bridge structure, along and horizontal turbulence intensity are canculated according to the formula (3), shown in figure 5. σ (z) σ (z) I = u ; I = v （3） u U(z) v U(z)

Here, σu(z)，σv(z) are the along and horizontal wind speed standard deviation of z place, respectively. Below figure shows the changing situation between wind direction and along and horizontal wind turbulence intensity.

0.25 maximum impact 0.20 60m 205.5m

0.15

Iu 0.10

0.05

0.00 maximum impact 0.20 60m 205.5m 0.15

Iv 0.10

0.05

0.00 8:00 16:00 0:00 8:00 16:00 0:00 2011.8.7 2011.8.8 2011.8.6 2011.8.6 time 2011.8.7 2011.8.7 Fig. 5 Turbulence intensity along with time From figure 5, turbulence intensity decreases with the increase of height, but horizontal wind turbulence intensity changes less significantly than along direction. Turbulence intensity is more scattered before and after the typhoon maximum speed, and is relatively stable near maximum speed. Because of the existence of this phenomenon, the most unfavorable situation may not appear at the biggest wind speed when study the wind-induced vibration of a bridge.

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 5

3.1.2.b Turbulence integral scale Fluctuating wind speed can be taken as a superposition of different frequency and cycle component, or can be intuitive thought as combination of different size of vortexes whose average value are known as turbulence integral scale. In practical application, the expression of turbulence integral scales are as follows: ∞ r 2 La = ∫ Ra(τ)dτ/σa ； τ = r/U （4） 0 2 Among them, a = u, v, w.R = x, y, z. σa is the square deviation of wind speed. Ra(τ) is autocorrelation function of different time at same point. Due to the three dimensional nature of fluctuating wind, the three directions of fluctuation component corresponds to a total of nine integral scales. The along and horizontal 10-min turbulence integral scale of the fluctuating wind x x in the x direction (Lu and Lv) are calculated, as shown in figure 6.

500

maximum impact 60m 400 205.5m

300

(m) u

x 200 L

100

5000 maximum impact 60m 400 205.5m

300

(m) v

x 200 L

100

0 8:00 16:00 0:00 8:00 16:00 0:00 2011.8.7 2011.8.8 2011.8.6 2011.8.6 2011.8.7 time 2011.8.7 Fig. 6 Turbulence scale along with time x x x x From figure 6, at 60m, the average of Lu、Lv are 143.2m and 85.4m, respectively, and the relation between Lu and Lv is x x x x x x Lv = 0.596Lu. At 205.5m, the average of Lu、Lv are 211.8m and 121.7 m, respectively, and the relation between Lu and Lv is x x Lv = 0.596Lu. 3.1.2.c Gust factor

Wind intensity of fluctuation can also use gust factor to represent. Gust factor G(tg) is defined as the ratio between maximum average wind speed for time of duration tg (wind duration time is generally taken as 3s in the structure wind engineering, so in this article tg is taken as 3s) and the average wind speed for basic time interval: 푚푎푥(푢̅(푡 )) G (푡 ) = 1 + 푔 （5） 푢 푔 푈

Here, 푢̅(푡푔) is the average value of along fluctuating wind for a set period of time푡푔.

Through calculation, Gu are in average of 1.254 and 1.192 at 60m and 205.5m, respectively, with good fitting to the results of wang [7]. Study shows that there is a certain difference for the gust factor before and after the peak average wind speed, as shown in figure 7.

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 6

1.4 1.5 before maximum before maximum after maximum after maximum fitting before maximum fitting before maximum 1.4 1.3 fitting after maximum fitting after maximum 1.3 1.2

1.2 gust factor gust gust factor gust 1.1 1.1

1.0 1.0 5 10 15 20 25 10 15 20 25 30 35 wind speed（m/s） wind speed（m/s） (a) 60m (b) 205.5m Fig. 7 Variation of gust factor at different speed

From figure 7, we can see that different rules between Gu and wind speed exist for two different Gu: before maximum wind −0.137 speed, Gu decreases with the increase of wind speed, and in line with the simulated formulation G푢 = 1.904(1 + U) at

60m; but Gu keeps in the vicinity of 1.235, decreasing slightly with the increase of wind speed at 205.5m. After maximum wind speed, Gu keeps in the vicinity of 1.183, decreasing slightly with the decrease of wind speed at 60m; Gu decreases with the 0.086 decrease of wind speed, and in line with the simulated formulation G푢 = 0.879(1 + U) at 205.5m. This rule is unique to typhoon Muifa, and further research is needed for this conclusion. Using the statistical method to establish the relation between turbulence intensity and gust factor is concerned by wind engineering field, and Ishizaki [8] and Choi [9] suggested that the functional form is as flowing 푇 G(t) = 1 + 푘 × 퐼푘2푙푛 （6） 1 푢 푡 Here, T is average time interval, t is the gust wind continuous interval, and 푘1、푘2 are two undetermined parameters. 1.7 measured value 1.40 measured value 1.6 Ishizaki 1.35 Choi Ishizaki 1.5 Cao 1.30 Choi fitting result Cao 1.4 1.25 fitting result 1.20

1.3

Gu Gu 1.15 1.2 1.10 1.1 1.05 1.0 1.00 0 2 4 6 8 10 12 14 16 18 20 22 24 26 0 2 4 6 8 10 12 14 Iu(%) Iu(%) (a)60m (b)205.5m Fig. 8 Variation of gust factor at different turbulence intensity

Table 1 Comparison of 푘1、푘2fitting result

author height k1 k2 Ishizaki - 0.5 1.0 Choi - 0.62 1.27 Cao 15m 0.5 1.15 20m 0.2 0.63 Xu Wang 30m 0.39 0.96 40m 0.49 1.1 60m 0.2998 0.8081 This research 205.5m 0.4336 0.9821

3.2 Bridge response Acceleration responses of the Shanghai Yangtze River Bridge under Typhoon Muifa were recorded by the structure health monitoring system at 11deck sections and here we study the three deck sections as shown in figure 9. The lateral acceleration response of the bridge deck was directly calibrated by the lateral accelerometer. The vertical acceleration response of the bridge

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 7

deck was obtained by averaging the accelerations recorded by the two vertical accelerometers. The torsional acceleration response was calculated according to the difference between the two vertical accelerations divided by 33.4m, the horizontal distance between the two vertical accelerometers. From figure 9 we can conclude that the maximum RMS response of vertical acceleration is not coordinate with wind speed on August 7th at about 9:20, when the yaw angle is about 15.6°. At this time, the wind characteristics, such as the turbulence intensity and turbulence integral scale which are relatively smaller than other time, plays an important role. The RMS response of lateral, vertical and torsional accelerations of this time is listed in table 2. In order to study the relationship between response and wind speed, figure 10 is plotted.

30

25 wind speed 20

15

10 wind speed(m/s) wind

5

1200

90

60

30 yaw angle

0 yaw angle(°) yaw -30

-60

0.05-90 lateral acceleration

) 0.04 2

0.03

RMS(m/s 0.02

0.01

0.100.00

) 0.08 vertical acceleration 2

0.06

RMS(m/s 0.04

0.02

0.00 0.20 tortional acceleration

0.16

) 2

0.12

0.08 RMS(ms 0.04

0.00 8:00 16:00 0:00 8:00 16:00 0:00

2011.8.6 2011.8.6 2011.8.7 2011.8.7 2011.8.7 2011.8.8 time

Figure 9. Wind speed, yaw angle and RMS of vertical response. Table 2. Wind characteristics. yaw time speed(m/s) gust factor TI TIS angle(°) 8:00 27.0 1.22 0.095 417.1 22.3 8:10 25.2 1.28 0.095 286.5 21.1 8:20 26.6 1.32 0.086 210.1 19.6 8:30 25.1 1.24 0.080 185.5 19.4 8:40 25.0 1.16 0.071 118.7 17.7 8:50 25.3 1.19 0.075 157.3 19.1 9:00 26.0 1.19 0.072 155.9 16.5 9:10 26.0 1.17 0.073 164.1 17.0 9:20 26.5 1.17 0.066 129.8 15.6 9:30 25.3 1.15 0.077 209.2 14.1 9:40 25.8 1.22 0.084 223.7 12.5 9:50 24.4 1.21 0.078 207.2 11.5 Table 3. RMS of acceleration responses. Section 1 Section 2 Section 4 Section 6 Section 8 Section 10 Section 11 Lateral(m/s 2) 0.0271 0.0303 0.0439 0.0439 0.0420 0.0267 0.0254 Vertical(m/s2) 0.0541 0.0668 0.0656 0.0783 0.0766 0.0638 0.0539 Torsional*33.4(m/s2) 0.3000 0.2107 0.1313 0.2048 0.2162 0.2021 0.3275 The RMS responses of acceleration are calculated. From the table, we can see the maximum RMS responses of lateral and vertical acceleration, which are 0.0439, 0.0783m/s2, respectively, occur at the mid-main span of the bridge. But the maximum

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 8

RMS responses of torsional acceleration which is 0.3275m/s2, occur at the middle of side pier and assistant pier. For torsional acceleration, the RMS responses in some sections, such as section 1, section 2, section 10, section 11, are larger than those in the mid-main span.

0.10 0.05 0.24

before maximum of wind speed ) ) before maximum of wind speed 2

) before maximum of wind speed 2 simulation 2 simulation 0.20 simulation 0.08 after maximum of wind speed 0.04 after maximum of wind speed after maximum of wind speed simulation simulation simulation 0.16 0.06 0.03 0.12 0.04 0.02 0.08

0.02 0.01

0.04

RMS of lateral acceleration(m/s lateral of RMS

RMS of vertical acceleration(m/s vertical of RMS RMS of torsional acceleration(m/s torsional of RMS 0.00 0.00 0.00 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 wind speed(m/s) wind speed(m/s) wind speed(m/s) Figure 10. Bridge response along with wind speed. In the figure, x-axis value represents the orthogonal component of wind speed. From the figure, we can conclude that there is a big difference between the acceleration before and after the maximum of wind speed: before the maximum of wind speed, the RMS value of vertical, lateral and torsional acceleration grows little by little along with wind speed which agrees with the results of Zhu; but the RMS value decreases sharply along with wind speed after the maximum and when the wind speed is less than 23m/s, the RMS value remains a relatively small value, especially for lateral and torsional acceleration. Because the yaw angle changes all the time during this period, it may be one important reason for the above finding.

4 INVESTIGATION VIA DYNAMIC METHOD With reference of Wind resistance design guidelines of highway bridge [10], the horizontal wind spectrum is chosen as Simiu spectrum: 푛푆 (푛) 200푓 푢 （ ） 2 = 5/3 7 푢∗ (1 + 50푓) The vertical wind spectrum is chosen as Panofsky spectrum: 푛푆 (푛) 6푓 푤 （ ） 2 = 2 8 푢∗ (1 + 4푓) 푛푍 푓 = （9） 푈(푍) 퐾푈(푍) 푢 = （10） ∗ 푍 − 푍 푙푛 푑 푍0

푍푑 = 퐻̅ − 푍0/퐾 （11）

Where 푆푢(푛) and 푆푤(푛) donate the power spectrum of the horizontal and vertical spectrum, respectively; n is the frequency of the fluctuating wind speed component(Hz); 푢∗ is air friction speed; K is dimensionless constant, K ≈ 0.4; 퐻̅ donates the average height of the surrounding building; 푍0 is the ground roughness height. The approximate expression for the spatial correlation function Coh푖푗(푓) of two points with the distance 푟푗 between the direction of j (j = x, y, z) and the fluctuating wind i component (i= u, v, w) is, 푓푟푗 Coh푖푗(푓) = exp (−λ푖푗 ) （12） 푈푍

Where λ푖푗 is dimensionless attenuation factor, the range of λ푖푗 is7~21, and in wind resistance design it is generally taken as 7. Conversion for circular frequency expressions is: 푤푟푗 Coh푖푗(푤) = exp (−λ푖푗 ) （13） 2휋푈푍

Assume that the distribution of the atmospheric boundary layer wind speed along the vertical height obey the power rate: 푈 푍 푍 = ( )훼 （14） 푈10 10

Where 푈푍 donates the wind speed of height Z (m/s); 훼 is dimensionless power considering the effects of the surface roughness. Considering the wind spectrum defined in the guide is heterogeneity square gauge a unilateral power spectrum, the wind spectrum form can be expressed in circular frequency conversion by comparing with the actual RMS of the wind speed:

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 9

100푍푢2 푆 (푛) = ∗ （15） 푢 25푤푍 휋푈(푍)(1 + )5/3 휋푈(푍) 3푍푢2 푆 (푛) = ∗ （16） 푤 2푤푍 휋푈(푍)(1 + )2 휋푈(푍) Using Deodatis’s method to simulate the nodes’ wind speed time history of the Yangtze River Bridge, the main parameters are shown in table 4. Table 4.Main parameters of the bridge span of the bridge 1430m Height of the main girder 64.7m height of the tower 211.3m ground roughness height 0.01m site classification I wind profile index 0.12 Correlation coefficient 10 number of simulation points 108 Cut-off frequency 15.7 dividing number of frequency 1024 sampling interval 0.2s sampling time 600s The results of the field measurement, wind tunnel test and calculation method are listed in table 5. Table 5.Comparation of the RMS of acceleration during different wind speed Analytical method calculation results field measured results wind tunnel result aerodynamic admittance 1 Sears Vertical(m/s2) 0.033 --- 0.125 0.071 middle span Torsional*33.4(m/s2) 0.121 --- 0.502 0.305 15m/s Vertical(m/s2) 0.029 --- 0.063 0.044 quarter span Torsional*33.4(m/s2) 0.125 --- 0.456 0.218 Vertical(m/s2) 0.061 0.085 0.181 0.101 middle span Torsional*33.4(m/s2) 0.151 0.337 0.810 0.418 20m/s Vertical(m/s2) 0.054 0.048 0.081 0.056 quarter span Torsional*33.4(m/s2) 0.061 0.253 0.675 0.305 Vertical(m/s2) 0.081 0.201 0.412 0.257 middle span Torsional*33.4(m/s2) 0.205 0.739 1.311 0.848 25/s Vertical(m/s2) 0.077 0.162 0.305 0.182 quarter span 2 Torsional*33.4(m/s ) 0.216 0.596 1.019 0.658

From the table, we can conclude that with the growing of the wind speed the acceleration of vertical and torsional direction are growing sharply and most of the time quarter span is much smaller than the middle span, but when the wind speed is 25m/s, the torsional acceleration of the quarter span is about 8 percent bigger than the acceleration of the middle span. Among the three methods, the calculation results are the biggest and the field measured results are the smallest. There are many reasons for this phenomenon, but the appropriate reason for this maybe aerodynamic admittance is different from the field measurement which from the table can be approximate drawn.

5 CONCLUSIONS Full scale measurement method based on the Bridge Health Monitoring System is becoming a more and more popular way to study wind resistant of structure, especially for strong typhoon. Typhoon Muifa (1109) is a large, strong and persistent typhoon which affected a number of countries in the Pacific, caused the Shanghai Yangtze River Bridge to be closed for traffic for more than 11 hours from about 4:30 to 15:30 on August 7th, 2011. The 10-min mean wind speed for Muifa is about 27m/s and the minimum range from the bridge is about 250KM and the effect of traffic can be ignored, so it is a typical typhoon for research and the results are relatively valuable. Compared with the RMS value with the wind speed, the maximum of RMS value is not synchronized with the wind speed, which means the wind characteristics affect the bridge response greatly. The maximum RMS responses of torsional acceleration occur at the middle of the side pier and the assistant pier, which is different from the lateral and vertical accelerations. After the maximum wind speed, the RMS value decreases sharply with wind speed, the yaw angle and wind characteristic are two important reasons for this phenomenon. We point out that the aerodynamic admittance is one of the reasons that lead to the difference of field measured results, wind tunnel test and calculation method, but more work need to be done to prove this.

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015 10

ACKNOWLEDGEMENT The authors gratefully acknowledge the support of National Key Basic Research Program of China (i.e.973 Program) (2013CB036300), Ministry of transport application foundation research project (2013319822070) and the National Natural Science Foundation of China (91215302, 51222809 and 51178353), and the support of Program for New Century Excellent Talents in University.

REFERENCES: [1] Yao, Y.J., Wind Resistance of Long Span Suspension Bridges. China Communications Press, 2011. ( in Chinese) [2] Miyata, T., et al., Full-scale measurement of Akashi--Kaikyo Bridge during typhoon. Journal of wind engineering and industrial aerodynamics, 2002. 90(12): p. 1517--1527. [3] Xu, Y.L. and Zhu, L.D., Buffeting response of long-span cable-supported bridges under skew winds. Part 2: case study. Journal of Sound and Vibration, 2005. 281(3): p. 675--697. [4] Wang, H., et al., Full-scale measurements and system identification on Sutong cable-stayed bridge during Typhoon Fung-Wong. The Scientific World Journal, 2014. 2014. [5] Shanghai Yangtze river waterway administration. http://www.shhdc.com/. 2014. ( in Chinese) [6] Chen, X.B., Zhou, L., Shi, W.L., et al. Characteristics of wave and surge fields of typhoon Muifa. Advances in marine science. 2013(01): 22-30. ( in Chinese) [7] Wang Xu, Huang Peng, Gu Ming. Field measurement on wind characteristics near ground during typhoon‘Muifa’[J]. China civil engineering journal. 2013(02): 54-61. ( in Chinese) [8] Ishizaki H. Wind profiles, turbulence intensities and gust factors for design in typhoon-prone regions[J]. Journal of Wind Engineering and Industrial Aerodynamics. 1983, 13(1–3): 55-66. [9] Choi E C C. Wind Loading in HongKong—Commentary on the Code of Practice on Wind Effects HongKong[R]. HongKong: HongKong Institute of Engineers, 1983. [10] Xiang Haifan, et al. Wind resistance design guidelines of highway bridge[S]. China Communications Press, 1996. ( in Chinese)

14th International Conference on Wind Engineering – Porto Alegre, Brazil – June 21-26, 2015