Zong et al.

/ J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 351

Journal of Zhejiang University-SCIENCE A (Applied Physics & Engineering)

ISSN 1673-565X (Print); ISSN 1862-1775 (Online)
www.zju.edu.cn/jzus; www.springerlink.com
E-mail: jzus@zju.edu.cn

Seismic response study on a multi-span cable-stayed bridge scale

model under multi-support excitations. Part I: shaking table tests*

Zhou-hong ZONG†1, Rui ZHOU2, Xue-yang HUANG3, Zhang-hua XIA3

(1School of Civil Engineering, Southeast University, Nanjing 210096, China)
2
( State Key Laboratory for Disaster Reduction in Civil Engineering, Tongji University, Shanghai 200092, China)
(3School of Civil Engineering, Fuzhou University, Fuzhou 350108, China)
†
E-mail: zongzh@seu.edu.cn
Received Oct. 16, 2013; Revision accepted Mar. 21, 2014; Crosschecked Apr. 22, 2014

Abstract: With the rapid development of bridge engineering in China, multi-span cable-stayed bridges have become one of the
main structures of modern highways and railways for crossing rivers or straits. In this study, a 1:100 scale model of a three-tower
cable-stayed bridge was tested using a shaking table array system. The mechanism associated with the seismic response of the
scale model under uniform and non-uniform excitations was clarified. The results from the tests indicated that: (1) the strong
vibration responses of the main girder and towers under four different horizontal earthquake wave excitations were identified, and
the seismic responses of the scale model were most evident under uniform Jiangxin (JX) wave excitation; (2) the seismic per-
formances of the main girder and towers of the scale model were adversely affected by traveling wave effects, especially when the
wave velocity exceeded 616 m/s, which suggests that traveling wave effects should be considered in the seismic design of
multi-tower cable-stayed bridges; (3) when the peak acceleration value of the El Centro (EC) wave reached 4.0 m/s2, shear failure
of the bearing of the middle tower first appeared. This kind of shaking table tests will help to improve our understanding of dy-
namic performance, and will be especially useful in the design process and numerical simulation of multi-span cable-stayed
bridges with large span subjected to spatially varying earthquake ground motions.

Key words: Multi-span cable-stayed bridge, Earthquake shaking table tests, Uniform excitations, Non-uniform excitations,
Traveling wave effect, Shear failure of bearing
doi:10.1631/jzus.A1300339 Document code: A CLC number: U448.27

1 Introduction Currently, more than ten multi-span cable-

stayed bridges with main span exceeding 400 m have
With the rapid development of bridge engi- been built or are planned to be constructed all over the
neering, multi-span cable-stayed bridges have be- world, including the famous Ting Kau Bridge in
come one of the main structures of modern highways Hong Kong and the Rion Antirion Bridge in Greece.
and railways for crossing large rivers or straits. They However, as the number of towers and the length
have advantages of design and construction, lower of the main span increase, the structural stiffness of
cost, and better landscape effects (Virlogeux, 2001). multi-span cable-stayed bridges degenerates signifi-
cantly. This type of bridge might be vulnerable to
seismic loading when different supports are subjected
*
Project supported by the National Natural Science Foundation of to significantly different seismic excitations (Gimsing
China (Nos. 51178101 and 51378112), the Doctoral Program Project and Georgakis, 2011).
of the Ministry of Education (No. 20110092110011), and the Na-
Multi-support excitations are often considered to
tional Key Laboratory of Civil Engineering Disaster Prevention of
Tongji University (No. SLDRCE08-HZ-02), China be one of the main contributors to the spatial variation
© Zhejiang University and Springer-Verlag Berlin Heidelberg 2014 of earthquake excitations. The assumption of
352 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

non-uniform excitation, which is used in estimating cable-stayed bridges by experimental study on a

the time lag factor of the wave propagation effects, 1:150 scale model. Shoji et al. (2008) carried out
should be taken into consideration when conducting shaking table tests of a pre-stressed concrete cable-
seismic design and analysis of long span bridges, such stayed bridge and clarified the mechanisms associated
as multi-span cable-stayed bridges (Mylonakis et al., with the seismic responses of a long-period structure
2001). when subjected to a long-period seismic excitation.
Many previous studies concentrated on the dy- Yang and Cheung (2011) conducted the shaking table
namic mechanisms associated with the seismic re- tests of a double-deck concrete cable-stayed bridge
sponses of cable-stayed bridges under earthquake with a scale ratio of 1:120 subjected to non-uniform
excitations, using theoretical deduction and numerical excitations, and discussed the effects of non-uniform
simulation. Ren and Makoto (1999) investigated the excitations on two types of tower-deck connection.
elastic-plastic seismic behavior of a long span cable- However, there have been few experimental studies
stayed steel bridge under three strong earthquake focusing on the whole structural dynamic behavior
records by considering the geometric and material of multi-span cable-stayed bridges under multi-
nonlinearities. Abdel-Ghaffar and Khalifa (1991) excitations.
conducted a 3D nonlinear static analysis and dynamic The seismic response characteristics of multi-
analysis of cable-stayed bridges to explore the span cable-stayed bridges subjected to multi-support
mechanisms associated with the seismic response. excitations have not been studied sufficiently from
Their study showed the contribution of cable vibra- theoretical and experimental perspectives. To explore
tions to the seismic response of cable-stayed bridges. their sophisticated seismic characteristics, a 1:100
Gattulli and Lepidi (2007) systematized the problem scale model of a three-tower cable-stayed bridge was
showing the importance of the avoidance of 1:1 selected to conduct shaking table tests under uniform
modal resonances from global and local modes in the and non-uniform excitations in this paper. In Section
dynamic performance of cable-stayed bridges. Ra- 2, the elaborate design of the scale model and the
heem et al. (2011) discussed the effects of spatial details of the shaking table tests are introduced. In
variability on the feasibility and efficiency of seismic Section 3, three kinds of experimental results are
control systems for controlling the vibration of cable- compared. The conclusions are given in Section 4.
stayed bridges. Fang et al. (2011) explored the in- The experimental results from the shaking table tests
fluence of traveling-wave effects on the seismic re- provide valuable evidence which may be useful for
sponse of a multi-span cable-stayed bridge with dif- the seismic design and analysis of this type of multi-
ferent constraint conditions manners among the tow- tower cable-stayed bridge.
ers and the main girder.
Compared with these numerous theoretical and
numerical advances, the experimental study of seis- 2 Shaking table tests
mic responses of cable-stayed bridges has lagged
2.1 Prototype of Wuhan Erqi Yangtze River
behind the demands of practical engineering. In re-
Bridge (WEYRB)
cent years, the shaking table array system has become
one of the most favorable techniques to perform The Wuhan Erqi Yangtze River Bridge
earthquake simulation tests of long-span cable-stayed (WEYRB) (Fig. 1) was selected as the prototype
bridges under multi-support excitations (Huang et al., structure of a three-tower and double-cable-plane
2008). For example, the dual tri-axial shaking table cable-stayed bridge. This bridge, connecting Wu-
system in the Chongqing Transportation Research chang and Hankou Towns in Wuhan city, was opened
Institute was used to conduct earthquake shaking to traffic in December, 2011. The main span is a
table tests of a scale Xiamen-Zhangzhou Bay Bridge steel-concrete composite girder, while two side spans
(Zhang, 2009). Kitazawa and Fujita (1989) completed are pre-stressed concrete girders. The arrangement of
seismic tests on a 1:100 scale model of a cable-stayed the spans along the bridge is 90 m+160 m+
bridge under 3D earthquake excitations. Kawashima 616 m+616 m+160 m+90 m, with an overall length of
et al. (1993) evaluated the damping mechanism of 1732 m. The total width of the bridge deck is 31.4 m
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 353

with six traffic lanes and two pedestrian walkways. 2008); (4) The ratios between the scale model and the
There are total 132 cables, of which the longest has a prototype bridge for the tensile stiffness and bending
length of 336.2 m, connecting the towers and main stiffness of the main girder, towers and cables, and the
girder, and the typical distance between two cables on mass of these components, should be similar. Detailed
the bridge deck is 13.5 m. The cables are composed of derivations and descriptions of dimensional analysis
a number of high-strength strands (1860 MPa level) and modeling theory can be found in (Harris and
with 15.2 mm in diameter. The height of the inverted Sabnis, 1999). Based on these requirements, the sim-
H-shaped reinforced concrete tower is 205 m with a ilarity ratios of the scale model with respect to the
clear navigation height of 45 m. Two vertical bearings prototype bridge were determined (Table 1).
and longitudinal spacing blocks are located among
Table 1 Similarity ratios of the scale model to the proto-
each of the main towers and the main girder, and two type bridge
two-way activity bearings and horizontal wind- Similarity
resistant bearings are installed on each side of the Physical quantity
ratio
towers among the towers and the main girder. Modulus of elasticity, E 1/12.81

2.2 Scale model of Wuhan Erqi Yangtze River Material Strain, İ 1/1
property Stress, Eİ 1/12.81
Bridge (WEYRB)
Equivalent quality density, U 1/1.432
In the process of designing the whole scale Length, l 1/100
model of the WEYRB based on dynamic similitude Area, l2 1/1.0×104
laws, the following basic requirements were consid- Geometry
Moment of inertia, l4 1/1.0×108
property
ered: (1) The scale model should be simple in terms of Flexural rigidity, El2 1/1.28×105
its fabrication and construction; (2) The geometric Tensile rigidity, El4 1/1.28×109
scale of 1:100 was selected according to the ar- Mass, U l 3 1/1.43×106
rangement of the shaking table array system and the Time, l ( U E 1 )1/ 2 1/0.043
length of the prototype; (3) The scale model should be Dynamic
Frequency, (l 2 U E 1 ) 1/ 2 1/23.35
as large as possible to eliminate the effects of irregu- property
Velocity, ( E / U )1/ 2 1/0.233
larities in structural configuration and variation
caused by the properties of the materials (Shoji et al., Horizontal acceleration, E / ( U l ) 1/5.45

Fig. 1 Wuhan Erqi Yangtze River Bridge (WEYRB): (a) elevation (unit: m); (b) prototype
354 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

The span arrangement of the scale model was Table 2 Parameters of materials used in the scale model
0.90 m+1.60 m+6.16 m+6.16 m+1.60 m+0.90 m, bridge
with a total length of 17.32 m, and the height of the Material E (MPa) ȡ (kg/m3) İ ı (MPa)
main tower was 2.05 m, excluding the foundation PMMA 2.69×10 3
1180 0.391 126
piles. The middle tower of the scale model was lo- 4
Aluminum 7.53×10 2700 0.326 187
cated in the middle of a fixed table (No. 2 table) and
Steel cable 1.95×105 7850 0.300 1330
the two side towers were located on two mobile
side-tables (No. 1 and No. 3 tables) (Fig. 2). The clear
distance between the No. 2 table and No. 1 table (or
No. 3 table) was 3.46 m, and steel plates were adopted
to simulate a rigid foundation. The supplemented
mass of the scale model was attached to the corre-
sponding components after calculation.
Sensor

(a)

No. 3 table

No. 2 table
Fixation
No. 1 table

Fig. 2 Three-towerr cable-stayed bridge model test

The main towers and side-span concrete girder
were made of polymethyl methacrylate (PMMA) (b)
material, and the cables were made of high-strength
steel wire with a diameter of 0.6 mm (Fig. 3). Based
Bearing
on the similar stiffness relations, the steel-concrete
composite girder was simplified to composite com-
ponents, including a groove aluminum alloy stringer
of 1 mm thickness, a small angle-type aluminum alloy
stringer of 0.8 mm thickness, an aluminum alloy
transom of 0.8 mm thickness, and the PMMA bridge
deck of 0.4 mm thickness. The free and the fixed (c)
support conditions of the longitudinal, transverse, and
rotational movements among the towers and the main
girder were idealized by combining a slider and a
bearing. The detailed parameters of the materials of
the scale model are listed in Table 2 (where E repre- Artificial mass

sents the modulus of elasticity, ȡ is the density, İ is the

Poisson’s ratio, and ı is the yield strength). These
materials had good fabricating performance in ma-
chining and bonding, and also helped the reduction of
the total weight of the scale model within the maxi- (d)
mum payload capacity of the shaking table array
Fig. 3 Detail construction of the scale model: (a) accelera-
system. Additional structurally uncoupling mass (ar- tion sensor; (b) fixation of the middle tower; (c) construc-
tificial mass simulation) was attached to the model to tion of strand anchor and bearing; (d) artificial mass ar-
augment the mass density. rangement of sided-span
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 355

2.3 Input ground motions To unify the intensity, the EC wave from the Pacific
Earthquake Engineering Research Center (PEER)
A total of 34 channels of accelerometers and 32
Strong Motion Database was also employed to excite
channels of strain gauges were instrumented in the
the shaking tables, but was modified by compressing
scale model to measure the vertical and transverse
the peak acceleration values. The time history curves
accelerations of the main girder surface, the trans-
verse and longitudinal accelerations of each tower, of these earthquake waves are given in Fig. 4. The
and the strain level in the key cross-sections of the frequency band distribution of the JX wave was the
main towers. The data acquisition system Dewesoft widest among the four waves and the prime energy of
was selected to obtain the test data, and the sampling the EC wave was concentrated on the lowest fre-
frequency was 200 Hz. Unfortunately, the laser dis- quency area through the fast Fourier transform (FFT).
placement transducers could not be instrumented to The interval between each pair of steps was 0.02 s
measure the superstructure dynamic displacements of when conducting dynamic time history analysis on
the scale model. the scale model.
Before each formal earthquake excitation, Tables 3 and 4 list the typical cases for the
white-noise scanning was conducted to track the dy- shaking table tests. Uniform excitations of the four
namic properties of the scale model. Three synthetic earthquake waves and non-uniform excitations of
earthquake waves with a duration of 20.5 s, including only the EC wave were conducted in both the X and Y
the Wuchang (WC), Jiangxin (JX), and Hankou (HK) directions with the same ground motion. Due to the
waves, were selected as the input excitations based on conflict between the acceleration similarity rule of the
an earthquake safety evaluation report provided by input excitations and the experimental load condition,
the Seismological Bureau of Hubei Province. The the peak acceleration values of all input excitations
acceleration peak values of these waves were were adjusted to the integer multiple of 1 m/s2. This
1.836 m/s2, 1.65 m/s2, and 1.826 m/s2, respectively. was done to compare the seismic responses of the

2.0 2.0
1.5 1.5
1.0 1.0
Acceleration (m/s2)

Acceleration (m/s2)

0.5 0.5
0.0 0.0
-0.5 -0.5
-1.0 -1.0
-1.5 -1.5
(a) (b)
-2.0 -2.0
0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21
Time (s) Time (s)
2.0 2.0
1.5 1.5
1.0
Acceleration (m/s2)

1.0
Acceleration (m/s2)

0.5 0.5
0.0 0.0
-0.5 -0.5
-1.0 -1.0
-1.5 -1.5
(c) (d)
-2.0 -2.0
0 3 6 9 12 15 18 21 0 5 10 15 20 25 30 35 40 45 50 55
Time (s) Time (s)
Fig. 4 Time history curves of four earthquake waves: (a) WC wave; (b) JX wave; (c) HK wave; (d) EC wave
356 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

Table 3 Multi-support uniform excitations cases

Case Wave Acceleration peak (m/s2) Input direction
1 EC 1 X+Y
2 WC 1 X+Y
3 JX 1 X+Y
4 HK 1 X+Y
8 EC 2 X+Y
9 EC 3 X+Y
10 EC 4 X+Y

Table 4 Multi-support non-uniform excitations cases

Acceleration Input Apparent wave No. 1 table No. 2 table No. 3 table
Case Wave
peak (m/s2) direction velocity (m/s) (s) (s) (s)
1 EC 1 X+Y Infinite 0 0 0
5 EC 1 X+Y 1232 0 0.5 1
6 EC 1 X+Y 616 0 1 2
7 EC 1 X+Y 308 0 2 4

scale model conveniently under four earthquake curves of the largest response points in the two di-
waves and at different strength amplitudes of the EC rections are shown in Figs. 5c and 5d.
wave without reflecting the prototype properties. The acceleration amplification coefficients of
During the non-uniform excitation cases, different the main girder under different excitations show the
input waves were applied to different tables. The same variation trends as a whole. The largest values
wave propagation effects were explored by changing of the vertical amplification coefficients in the right
the wave passage velocity of the EC wave to 308 m/s, main span under the EC, WC, JX, and HK waves are
616 m/s, 1232 m/s or infinite. The distances between 3.94, 4.75, 8.47, and 6.25, respectively; and the cor-
the two main towers were used to calculate the responding transverse amplification coefficients in
difference in the arrival time for different tables. the left main span are 2.94, 4.28, 6.25, and 4.95, re-
Only the longitudinal input was considered in this spectively. Therefore, for the scale cable-stayed
study. bridge, the seismic response of the main girder is the
greatest under the JX wave excitation, followed by
the HK wave. The EC wave gives the slightest re-
3 Experimental sponse due to the different frequency spectrum
3.1 Seismic responses of the scaled model under characteristics of those earthquake excitations.
uniform excitations The largest values of the vertical acceleration
response in the right main span and left main span
3.1.1 Acceleration responses of the main girder
under 1 m/s2 acceleration of JX wave excitation are
When the acceleration peak values were set to 8.47 m/s2 and 6.25 m/s2, respectively. Strong vibra-
1.0 m/s2 of the above four earthquake waves, the tion appeared in the main girder under horizontal
acceleration amplification coefficient was defined as earthquake excitations, which may cause strong vi-
the ratio of the measured acceleration value of the bration of cable strands under strong earthquake
structural component to the input acceleration value excitations.
of the shaking tables. The results of the acceleration
3.1.2 Acceleration responses of the main towers
amplification coefficients of the main spans in the
vertical and transverse directions are compared in Similarly, the results for the acceleration ampli-
Figs. 5a and 5b, and the corresponding time history fication coefficients of the main towers in the
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 357

longitudinal and transverse directions are compared tower in the longitudinal and transverse directions are
in Fig. 6. The acceleration sensors were located at the 7.93 and 6.13, respectively, when the acceleration
top and in the middle of each tower. The amplification peak value of the input JX wave is 1.0 m/s2. This
coefficients at the top are larger than those in the phenomenon suggests that some technical measures
centre of the main tower, and the values of the middle should be taken to increase the stiffness of the middle
main tower are the highest among the three main towers to improve the seismic performance of the
towers. The coefficients at the top of the middle main three-tower cable-stayed bridge.

9 7
EC (b) EC

Transverse amplification coefficient

(a)
Vertical amplification coefficient

8 WC WC
JX 6 JX
7 HK HK
5
6
5 4
4 3
3
2
2
1 1
Left side Left main right
Rightmain
main right
Rightside
side Left side Left main right
Right main
main right
Right side
side
Mid-span location Mid-span location
10 (c) 8
Transverse acceleration (m/s2)

EC (d) EC
8 6
Vertical acceleration (m/s2)

WC WC
6 JX JX
HK 4 HK
4
2 2
0 0
-2
-2
-4
-6 -4
-8 -6
0 3 6 9 12 15 18 21 0 3 6 9 12 15 18 21
Time (s) Time (s)
Fig. 5 Comparison of the seismic responses of the main girder under uniform excitations: (a) vertical amplification coef-
ficients; (b) transverse amplification coefficients; (c) vertical time history in the right main span; (d) transverse time
history in the left main span

9
(a) 7
EC (b)
Transverse amplification coefficient
Longitudinal amplification coefficient

EC
8 WC WC
JX 6 JX
7 HK HK
5
6
4
5
3
4

3 2

2 1
Left top Left center Middle topMiddle centerRight top Right center Left top Left center Middle topMiddle centerRight top Right center
Tower location Tower location

Fig. 6 Comparison of the seismic responses of the main towers under uniform excitations: (a) longitudinal amplification
coefficients; (b) transverse amplification coefficients
358 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

On the other hand, under the earthquake excita- seismic performance of multi-span cable-stayed
tion, the vibration of stayed-cable will aggravate the bridges under multi-support excitations will not be
vibration of the main towers. Because the cables of discussed in detail here.
the scale model are too small, the forces and defor-
3.2.1 Acceleration responses of the main girder
mations of the stayed cables are difficult to be meas-
ured during the tests. So the interaction of the cables Three wave velocities of the EC waves, such as
and the main towers were not considered in this study. 1232 m/s, 616 m/s, and 308 m/s, were adopted to
study the seismic responses of the scale model. The
3.1.3 Strain responses of the main towers
acceleration amplification coefficients of the main
Strain responses in the bottom sections of three- girder in the vertical and transverse directions are
towers and two side piers under four different earth- shown in Figs. 7a and 7b, and the corresponding time
quake excitations with the 1 m/s2 input acceleration history curves of the largest response points in the
are summarized in Table 5. All the strain responses vertical and transverse directions under non-uniform
under the EC wave are smaller than those under the excitations are shown in Figs. 7c and 7d.
other three waves, but the range of these values under All acceleration amplification coefficients of the
different earthquake excitations is not so large. All the main girder under non-uniform excitations in two
strain responses of the main towers are larger than directions are larger than those under uniform excita-
those of the side piers, and the responses of the middle tions. The largest ratios of the vertical acceleration
tower are the largest. The characteristics of the amplification coefficients under different velocities of
measured strain responses are similar to those of the 1232 m/s, 616 m/s, and 308 m/s, compared with those
acceleration responses of the scale model. from the uniform excitations, are 14.95, 8.13, and
However, the strain responses of the main girder 7.58, respectively. The corresponding largest ratios in
the transverse direction are 4.96, 3.76, and 3.43, re-
in the scale model could not be measured during the
spectively. Therefore, the seismic responses of the
tests. If possible, the dynamic strain of the main girder
main girder of the scale model become larger as the
should be included.
wave velocity increases. The results reveal that the
3.2 Seismic responses of the scaled model under seismic performance of the main girder of a
non-uniform excitations three-tower cable-stayed bridge should be markedly
affected by traveling wave effects, especially when
The wave velocity depends on the type of soil on
the velocity exceeds 616 m/s. It is suggested that the
which the bridge foundations are constructed, and the
apparent wave velocity should not been neglected
size of the wave velocity has different effects on the during the seismic design of multi-span cable-stayed
various structural systems of cable-stayed bridges bridges with large span. Under non-uniform excita-
(Nazmy and Abdel-Ghaffar, 1990a; 1990b; 1992). tions, the stiffness of the main girder of a multi-span
Therefore, the influence of the traveling wave on the cable-stayed bridge should be enhanced.
seismic responses of the main girder and towers of the
scale three-tower cable-stayed bridge was compared 3.2.2 Acceleration responses of the main towers
using three wave velocities in this study. Because of In the same way, the acceleration amplification
the limitations of the tests data, the principles of coefficients of the main towers considering traveling

Table 5 Strain responses of the bottom sections at the acceleration peak of 1.0 m/s2
Left Left Left Middle Right Right Right
Wave
outside pier inside pier tower tower tower inside pier outside pier
EC 345.31 357.43 920.66 1059.41 972.28 345.78 326.20
WC 352.19 357.59 926.09 1064.65 989.69 353.39 327.68
JX 353.20 358.47 930.14 1070.36 992.20 354.83 330.28
HK 352.95 358.27 927.46 1054.05 971.57 353.59 329.04
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 359

wave effects are also compared with those from uni- in the longitudinal direction are larger than those in
form excitations (Fig. 8). As the apparent wave ve- the transverse direction. For example, the longitudinal
locity increases, the acceleration responses of the and transverse acceleration coefficients at the top
main towers become obviously larger, and the values position in the middle tower under a velocity of

16 6.0
Uniform Uniform
(a) 1232 m/s
(b) 1232 m/s

Transverse amplification coefficient

5.5
14
Vertical amplification coefficient

616 m/s 616 m/s

308 m/s 5.0 308 m/s
12
4.5
10 4.0

8 3.5
3.0
6
2.5
4
2.0
2 1.5

0 1.0
Left side Left main Right main Right side Left side Left main Right main Right side
Mid-span location Mid-span location

16 (c) 6
Uniform (d)
Transverse acceleration (m/s2)

Uniform
Vertical acceleration (m/s2)

12 1232 m/s
4 1232 m/s
616 m/s 616 m/s
8 308 m/s 308 m/s
2
4
0
0
-2
-4
-4
-8
-6
0 5 10 15 20 25 30 35 40 45 50 55 0 5 10 15 20 25 30 35 40 45 50 55
Time (s) Time (s)

Fig. 7 Comparison of the seismic responses of the main girder under non-uniform excitations: (a) vertical amplification
coefficients; (b) transverse amplification coefficients; (c) vertical time history in the right main span;(d) transverse time
history in the left main span

20 14 Uniform
(a) Uniform (b)
1232 m/s 1232 m/s
Longitudinal amplification coefficient

18
Transverse amplification coefficient

616 m/s 616 m/s

12 308 m/s
16 308 m/s
14 10
12
8
10
8 6

6 4
4
2
2
0 0
Left Left Middle Middle Right Right Left Left Middle Middle Right Right
top center top center top center top center top center top center
Tower location Tower location

Fig. 8 Comparison of seismic responses of the towers under non-uniform excitations: (a) longitudinal amplification co-
efficients; (b) transverse amplification coefficients
360 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

1232 m/s are 18.58 and 13.23, respectively. The re- 16

(a) 1.0 m/s2
sults suggest that the seismic performance of the main 14 2.0 m/s2

Vertical amplification coefficient

3.0 m/s2
towers of the three-tower cable-stayed bridge is also 12 4.0 m/s2
markedly influenced by traveling wave effects. So the 10
stiffness and strength of the main towers of the 8
three-tower cable-stayed bridge should also be im- 6
proved to withstand strong earthquakes. 4

3.3 Seismic responses of the scaled model under 2

different intensities uniform excitations 0

Left side Left main right
Rightmain
main right side
Right side
Mid-span location
As strong earthquakes have occurred worldwide
8 1.0 m/s2
in recent years, the characteristic and mechanism of

Transverse amplification coefficient

(b)
7 2.0 m/s2
failure mode of long span bridges under strong 3.0 m/s2
earthquake have attracted more attention (Chen et al., 6 4.0 m/s2

2008). If the intensity of input seismic waves grows, 5

local failure of the long span cable-stayed bridge 4
structure may occur, even causing progressive col-
3
lapse. In this study, a scale model of a three-tower
2
cable-stayed bridge was subjected to the amplitude of
the EC wave with different intensities from 1.0 m/s2 1
Left side Left main right main
Right main right side
Right side
up to 4.0 m/s2, and the seismic performance of the Mid-span location
scale model was obtained by comparing the acceler- 12
(c) 2.0 m/s2
ation responses of the main girder and towers. When 3.0 m/s2
8
the acceleration peak value of the EC wave reached
Vertical acceleration (m/s2)

4.0 m/s2
4.0 m/s2, shear failure of the bearing of the middle 4
tower occurred (Fig. 9). The phenomenon of a broken 0
wire or cable was not observed during the tests.
-4

-8

-12

-16
0 5 10 15 20 25 30 35 40 45 50 55
Time (s)
8
(d)
2.0 m/s2
6
Transverse acceleration (m/s2)

3.0 m/s2
4 4.0 m/s2
Fig. 9 Shear failure of the bearing in middle tower 2
0
3.3.1 Acceleration responses of the main girder -2

Acceleration amplification coefficients of the -4

main girder under different excitation intensities are -6
compared (Figs. 10a and 10b), and the corresponding -8
0 5 10 15 20 25 30 35 40 45 50 55
time history curves of the largest response points of Time (s)
the main girder in vertical and transverse directions Fig. 10 Comparison of seismic responses of the decks
are described (Figs. 10c and 10d). The largest values under different amplitudes: (a) vertical amplification
coefficients; (b) transverse amplification coefficients;
of vertical amplification coefficients under ampli- (c) vertical time history in the right main span; (d) trans-
tudes of 2.0 m/s2, 3.0 m/s2, and 4.0 m/s2 are 4.81, 9.23, verse time history in the left main span
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 361

and 14.12, respectively. Similarly, the largest values transverse acceleration peak values at the top of the
of transverse amplification coefficients under the middle tower are 19.06 m/s2 and 13.81 m/s2, respec-
amplitudes of 2.0 m/s2, 3.0 m/s2, and 4.0 m/s2 are 3.56, tively. This may cause strong vibration of the cable
4.89, and 7.39, respectively. The results indicate that strands and increase the vibration of the main girder.
the seismic responses of the main girder become
larger with the intensity’s growth of input earthquake 3.3.3 Strain responses of main towers
wave. It can be predicted that the shear failure of Fig. 12 shows the strain time history at the bot-
bearing components located in the middle tower may tom of the middle tower under four input wave in-
occur first under a strong earthquake, and then cable tensities. The peak values of strain under the ampli-
may rupture. Local component failure may have a tudes of 2.0 m/s2, 3.0 m/s2, and 4.0 m/s2 are 1096.82,
negative impact on seismic performance, and may
1147.05, and 1188.32, respectively. The strain re-
even cause progressive collapse of the whole bridge.
sponse law at the bottom section of the middle tower
3.3.2 Acceleration responses of main towers is consistent with the acceleration response of the
middle tower. During the process of this test, the local
The acceleration responses of towers under dif-
failure did not appear in the main towers.
ferent earthquake excitation intensities are compared
(Fig. 11). The seismic responses of the main towers 1200 2.0 m/s2
Strain response at the bottom section

also become larger as the acceleration intensity in- 1150

3.0 m/s2
4.0 m/s2
creased. For example, when the acceleration of the
1100
input wave reaches 4.0 m/s2, the longitudinal and
1050

1000
20 950
(a) 2.0 m/s2
Longitudinal amplification coefficient

18 3.0 m/s2
4.0 m/s2 900
16
850
0 5 10 15 20 25 30 35 40 45 50 55
14
Time (s)
12
Fig. 12 Strain responses of the middle tower under dif-
10
ferent amplitudes
8

4
4 Conclusions
Left Left Middle Middle Right Right
top center top center top center
Tower location In this paper, a 1:100 scale model of a three-
16 tower cable-stayed bridge was designed and con-
(b) 2.0 m/s2
structed for shaking table tests under multi-excitation.
Transverse amplification coefficient

14 3.0 m/s2
4.0 m/s2
The following conclusions have been drawn from the
12
study:
10
1. Under the same intensities of uniform excita-
8 tions, the seismic responses of the main girder and
6 towers under the JX wave excitation are the largest,
4
followed by the HK wave and WC wave excitation.
The responses from the EC wave are the smallest
2
Left Left Middle Middle Right Right because of different frequency spectrum characteris-
top center top center top center
Tower location tics of input earthquake waves. The strongest vibra-
Fig. 11 Seismic responses of towers under different tion responses, including acceleration and dynamic
amplitudes: (a) longitudinal amplification coefficients; displacement in the main girder and towers under
(b) transverse amplification coefficients horizontal excitations, were identified. The internal
362 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363

force responses at the bottom of the side piers are cable-stayed bridge dynamics. Computer & Structures,
much smaller than those of the middle tower. These 85(21-22):1661-1668. [doi:10.1016/j.compstruc.2007.02.
016]
rules reveal that some necessary technical measures
Gimsing, N.J., Georgakis, C.T., 2011. Cable Supported
should be adopted to improve the stiffness of the main Bridges: Concept and Design, 3rd Edition. Wiley,
girder and the middle tower to enhance the seismic Chichester, UK, p.150-160.
performance of multi-tower cable-stayed bridges. Harris, H.G., Sabnis, G.M., 1999. Structural Modeling and
2. Under non-uniform excitations, all accelera- Experimental Techniques, 2nd Edition. CRC Press,
tion amplification coefficients of the main girder and p.121-124.
Huang, B.F., Lu, W.S., Zong, Z.H., 2008. Study on model
main towers are larger than those from uniform ex-
experimental methodology utilizing the multiple earth-
citations. The seismic performance of the main girder quake simulation shake table system. China Civil Engi-
and main towers of the scale cable-stayed bridge were neering Journal, 41(3):46-52 (in Chinese). [doi:10.3321/
obviously affected by traveling wave effects, espe- j.issn:1000-131X.2008.03.007]
cially when the wave velocity is larger than 616 m/s. Kawashima, K., Unjoh, S., Tunomoto, M., 1993. Estimation of
It is suggested that traveling wave effects should be damping ratio of cable-stayed bridges for seismic design.
ASCE Journal of Structural Engineering, 119(4):1015-
considered in the seismic design of multi-tower
1031. [doi:10.1061/(ASCE)0733-9445(1993)119:4
cable-stayed bridges. (1015)]
3. Under different intensities of uniform excita- Kitazawa, M., Fujita, K., 1989. Seismic Test on Higashi-Kobe
tions, the seismic responses of the main girder and Bridge: Cable Stayed Bridge-using Scale Models.
main towers become larger with the growth of inten- Earthquake Behavior of Buried Pipelines, ASME Pres-
sity of the input earthquake wave. When the acceler- sure Vessels and Piping Conference, 162:131.
Mylonakis, G., Papastamatiou, D., Psycharis, J., et al., 2001.
ation peak value of the El-Centro wave arrived at
Simplified modeling of bridge response on soft soil to
4.0 m/s2, shear failure of the bearing in the middle non-uniform seismic excitation. ASCE Journal of Bridge
tower occurred. Local component failure may have a Engineering, 6(6):587-597. [doi:10.1061/(ASCE)1084-
negative impact on the seismic performance of the 0702(2001)6:6(587)]
multi-tower cable-stayed bridge. Nazmy, A.S., Abdel-Ghaffar, A.M., 1990a. Three-dimensional
4. This kind of shaking table tests will help to nonlinear static analysis of cable-stayed bridges. Journal
of Computers and Structures, 34(2):257-272. [doi:10.
improve our understanding of the dynamic perfor-
1016/0045-7949(90)90369-D]
mance of multi-tower cable-stayed bridges subjected Nazmy, A.S., Abdel-Ghaffar, A.M., 1990b. Nonlinear
to spatially varying earthquake ground motions. In the earthquake-response analysis of long-span cable-stayed
future, numerical simulation and progressive collapse bridges: theory. International Journal of Earthquake En-
will be studied for comparison with experimental gineering and Structural Dynamics, 19(1):45-62. [doi:10.
results. 1002/eqe.4290190106]
Nazmy, A.S., Abdel-Ghaffar, A.M., 1992. Effects of ground
motion spatial variability on the response of cable-stayed
References bridges. Earthquake Engineering & Structural Dynamics,
Abdel-Ghaffar, A.M., Khalifa, M., 1991. Importance of cable 21:1-21. [doi:10.1002/eqe.4290210101]
vibration in dynamics of cable-stayed bridges. Journal of Raheem, S.E.A., Hayashikawa, T., Dorka, U., 2011. Ground
Engineering Mechanics, 117(11):2571-2589. [doi:10. motion spatial variability effects on seismic response
1061/(ASCE)0733-9399(1991)117:11(2571)] control of cable-stayed bridges. Earthquake Engineering
Chen, Y., Feng, M., Soyoz, S., 2008. Large-scale shake table and Engineering Vibration, 10(1):37-49. [doi:10.1007/
test verification of bridge condition assessment methods. s11803-011-0045-5]
Journal of Structural Engineering, 134(7):1235-1245. Ren, W.X., Makoto, O., 1999. Elastic-plastic seismic behavior
[doi:10.1061/(ASCE)0733-9445(2008)134:7(1235)] of long span cable-stayed bridges. ASCE Journal of
Fang, Y., Li, J.Z., Peng, T.B., et al., 2011. Influence of Bridge Engineering, 4(3):194-203. [doi:10.1061/(ASCE)
traveling-wave effect on seismic response of a long-span 1084-0702(1999)4:3(194)]
multi-span cable stayed bridge. Journal of Vibration and Shoji, G., Kitahara, J., Kojima, A., et al., 2008. Mechanism of
Shock, 29(10):148-152 (in Chinese). seismic response of a PC cable-stayed bridge subject to a
Gattulli, V., Lepidi, M., 2007. Localization and veering in long-period seismic excitation. Structural Engineering/
 Zong et al. / J Zhejiang Univ-Sci A (Appl Phys & Eng) 2014 15(5):351-363 363

Earthquake Engineering, 25(2):982-1001. [doi:10.2208/ Engineering, 14:931-938. [doi:10.1016/j.proeng.2011.07.

jsceja.64.982] 117]
Virlogeux, M., 2001. Bridges with multiple cable-stayed spans. Zhang, C.L., 2009. Shaking Table Test for the Effect of Ground
Structural Engineering International, 11(1):61-82. Motion Input to Seismic Response of Long-span Bridge.
[doi:10.2749/101686601780324250] Chongqing Jiaotong University, Chongqing, China,
Yang, C.Y., Cheung, M., 2011. Shake table test of cable-stayed p.25-40 (in Chinese).
bridge subject to non-uniform excitation. Procedia

₼㠖㰑尐᧶

㦻㠖欧䥽˖ཊ⛩◰࣡лཊ䐘ᯌ᣹ẕ㕙ቪ⁑රⲴൠ䴷૽ᓄ⹄ウˊ䜘࠶ 1˖ᥟࣘਠ䈅傼
Seismic response study on a multi-span cable-stayed bridge scale model under multi-support
excitations. Part I: shaking table tests
䪣䴅䥽䤓˖Ѫཊຄᯌ᣹ẕ൘а㠤◰઼࣡䶎а㠤◰࣡лൠ䴷૽ᓄⲴ⹄ウᨀ‫׋‬ൠ䴷⁑ᤏᥟࣘਠ‫ޘ‬ẕ䈅傼ᯩ
⌅ˈᒦᴹࣙҾ࣐ᕪཊຄᯌ᣹ẕᣇ䴷ᙗ㜭઼ᣇ䴷䇮䇑Ⲵ䇔䇶DŽ
⒪㠿尐䍈˖ᔪ・Ҷൠ䴷⁑ᤏᥟࣘਠཊਠ䱥䈅傼ᯩ⌅ˈᒦ俆⅑ሶަᓄ⭘Ҿཊຄᯌ᣹ẕ‫ޘ‬ẕ∄ֻ㕙ቪ⁑
රⲴൠ䴷૽ᓄ࠶᷀ѻѝ˗䙊䗷ൠ䴷⁑ᤏᥟࣘਠ䈅傼ˈ᨝⽪Ҷཊຄᯌ᣹ẕ൘ཊ⛩а㠤◰࣡
઼ཊ⛩䶎а㠤◰࣡Ⲵൠ䴷૽ᓄ⢩ᙗᒦᨀࠪҶཊຄᯌ᣹ẕ൘ᕪ䴷лⲴ⹤ൿ⁑ᔿDŽ
䪣䴅㡈㽤˖สҾࣘ࣋⴨լ⨶䇪ˈ䇮䇑оࡦ֌Ҷཊຄᯌ᣹ẕⲴ‫ޘ‬ẕ∄ֻ㕙ቪ⁑ර˄മ 2˅
˗࡙⭘ൠ䴷
⁑ᤏᥟࣘਠཊਠ䱥䈅傼ˈṩᦞ൘н਼ൠ䴷⌒֌⭘л઼㘳㲁㹼⌒᭸ᓄਾཊຄᯌ᣹ẕ⁑රⲴѫ
ằ઼ѫຄ૽ᓄˈ࠶᷀Ҷཊຄᯌ᣹ẕ൘ཊ⛩а㠤◰઼࣡ཊ⛩䶎а㠤◰࣡Ⲵൠ䴷૽ᓄ⢩ᙗ
˄മ 5~മ 8˅
˗ṩᦞ൘ൠ䴷⌒н਼ᕪᓖ֌⭘лཊຄᯌ᣹ẕ⁑රⲴѫằ઼ѫຄ૽ᓄˈ࠶᷀Ҷ
ཊຄᯌ᣹ẕ൘ᕪ䴷лⲴ⹤ൿ⁑ᔿ઼ൠ䴷૽ᓄ˄മ 9˅ DŽ
摜尐兢幉˖⊏ᗳ⌒֌⭘л䈕ཊຄᯌ᣹ẕⲴൠ䴷૽ᓄབྷҾަԆй⿽ൠ䴷⌒˗ཊຄᯌ᣹ẕ൘ᣇ䴷䇮䇑
ᰦ䴰㾱㘳㲁㹼⌒᭸ᓄⲴᖡ૽˗൘ El-Centro ⌒Ⲵ࣐䙏ᓖጠ٬儈䗮 4.0 m/s2 ᰦˈཊຄᯌ᣹ẕ
⁑රࠪ⧠Ҷ᭟ᓗ⹤ൿDŽ
␂枽幜兓 ཊຄᯌ᣹ẕ ൠ䴷⁑ᤏᥟࣘਠਠ䱥䈅傼 ཊ⛩◰࣡ 㹼⌒᭸ᓄ ⹤ൿ⁑ᔿ