Innovative Infrastructures - Toward Human Urbanism 1

Scaled Model Design and Verification of Shaking Table Test of Taizhou

Changjiang Highway Bridge

Tianbo PENG
PhD, Associate Professor, Master tutor
Research Division for Seismic Design of Bridges Department of Bridge Engineering
Tongji University ,1239 Siping Road ,Shanghai, China
ptb@tongji.edu.cn

Lei HAN Yi ZHAO Jukao YAN

Tongji University Tongji University Doctoral candidate
Shanghai,China Shanghai,China Tongji University
1130678hanlei@tongji.edu.cn yingchao224 @126.com Shanghai,China
yanjukao@163.com

Summary
Taizhou Changjiang Highway Bridge, which belongs to three-pylon two-span suspension bridge,
realizes the purpose of crossing the broad and open water area with super long span bridge,
attracting the attention in the field of the bridge around the world. However, the investigation on the
seismic performance of this kind of bridge is insufficient. Shaking table test of Taizhou Changjiang
Highway Bridge will be carried out to study in-depth the influences of some important factors on
the seismic performance of this bridge, including restraining system and non-uniform excitation.
During the process of designing the shaking table test model of Taizhou Changjiang Highway
Bridge, it would be difficult to manufacture the test specimens of the girder and pylons if the
similarity ratio is strictly followed. To overcome this problem, in this paper, the simplified scaled
model is adopted to consider the similarity ratio of flexural stiffness but ignore the similarity ratio
of the torsion and axial stiffness. Based on an FEM model, comparison of the internal forces and
displacements reveals that the differences between the simplified scaled model and the prototype
are acceptable. So, the simplified scaled model can be used to perform the shaking table test.
Keywords: three-pylon two-span suspension bridge; shaking table test; stiffness equivalence; FEM
verification.

1. Introduction
While china is conducting the world's largest transportation infrastructure, many large-span bridges
have been built or are under construction. Taizhou Changjiang Highway Bridge and Maanshan
Changjiang Highway Bridge , which belong to three-pylon two-span suspension bridge, realize the
purpose of crossing the broad and open water area with a super long span bridge, attracting the
attention in the field of the bridge engineering around the world. Compared to the conventional
two-pylon suspension bridge, the seismic performance of three-pylon suspension structure is very
different, on which many factors have significant influences , for example, horizontal stiffness ratio
of the middle pylon to side pylons, the type of infrastructure, pylons, girder pier support system.
Many challenges are encountered and many new requirements are brought out during the
construction of multi-pylon continuous-span suspension bridge.
The “Multi-functional Shaking tables Lab” of Tongji University was supported by the 985-project
program of ministry of education. The shaking tables testing system is composed of four shaking
tables in which the vertical load capacities of A and D are 30 tons and those of B and C are 70 tons,
and the lengths of two trenches are 70 and 30 meters respectively. All four tables are moveable
within the first 70m trench and work as a large linear shaking table array. Tables can work in same
way or with relative motion. The shaking tables testing system, with a total vertical load capacity of
2 18TH CONGRESS OF IABSE, SEOUL, 2012

200ton, will be among the largest and most capable multiple shaking tables systems in the word. A
world-class multi- functional earthquake simulation platform for the vibration and seismic testing
research of bridge engineering, building and spatial structure engineering will be developed.
Some researchers have done some shaking table tests to study the seismic performance of long-span
bridges, but most of the tests are small scaled models and the materials used are organic glass,
which can not reflect the characteristics of the prototype [1-5]. In this paper, a larger scaled model
and authentic material are used in shaking table test of Taizhou Changjiang Highway Bridge to
study in-depth the influences of some important factors on the seismic performance of this kind of
bridge, including restraining system and non-uniform excitation.

2. Taizhou Changjiang Highway Bridge

Taizhou Changjiang Highway Bridge, which belongs to three-pylon two-span suspension bridge,
and two main spans are 1080m. The height of the middle pylon with longitudinal herringbone shape
is 192m, and the transversal shape of middle pylon is gantry-type frame. The lower and upper
beams are trapezoidal steel beam and inverted K shaped beam, respectively. The height of each
reinforced concrete pylon is 180m. The bridge deck of the steel box girder is composed of
orthotropic plates, which is assembled into integration by bridge cover plates, longitudinal stiffeners
and diaphragms. Elastic cables at the middle pylon in the longitudinal direction are mounted to
restrict the longitudinal displacement of the girder. Wind-resistant bearings are set up on all the
three main pylons to restrict the lateral displacement of the girder. Vertical bearings on the lower
cross beams of the side pylons are also employed. And no vertical bearings are installed on the
middle pylon to reduce the negative bending moments of the girder.

Fig.1 Elevation of Taizhou Changjiang Highway bridge（unit:m）

Elevation of the Side view of Elevation of side Side view of

middle pylon the middle pylon pylons side pylons

Fig.2 Elevation and side view of the middle pylon and side pylons（unit:m）
 Innovative Infrastructures - Toward Human Urbanism 3

3. Similarity ratio and model design

3.1 Similarity ratio
Restricted to the laboratory, large-span structure, such as suspension bridges, must be a small scaled
model. During the process of designing the shaking table of Taizhou Changjiang, it would be
difficult to manufacture the test specimens of the girder and pylons if the similarity ratio of 1:40 is
strictly followed [6, 7]. To overcome this problem, in this paper, the simplified scaled model is
adopted to consider the similarity ratio of flexural stiffness but ignore the similarity ratio of the
torsion and axial stiffness. The geometry similarity ratio is set to be 1:40 when some absolutely
necessary conditions are taken into consideration, such as the size and bearing capacity of shaking
table, the elevation height of the laboratory etc. Acceleration similarity ratio is 1 to simulate the
gravity effect. Thus, the frequency similarity ratio is 6.324, which makes all important modal
frequencies within the frequency scope of the shaking table. Elastic modulus similarity ratio is 1:3.
To satisfy the modulus similarity ratio, the side pylon is manufactured by micro-concrete, the elastic
modulus of which is 1/3 times that of C50 concrete, the prototype material; the flexural stiffness of
the middle pylon is reduced to be 1/3 as the material used is steel; and the flexural stiffness of the
girder has to be increased as the material used is organic glass, the elastic modulus of which is
1/76.3 times that of steel[8,9]. Similarity ratios are listed in Tab 1.
Table.1 Similarity ratio
parameters side pylon middle pylon main girder
length 1/40 1/40 1/40
area 1/1600 1/4800 76.3/4800
elastic modulus 1/3 1 1/76.3
acceleration 1 1 1
density 13.3 40 1.908
displacement 1/40 1/40 1/40
strain 1 1 1
mass 1/4800 1/4800 1/4800
time 0.158 0.158 0.158
Because of the scale of 1/40, the Taizhou Changjiang Highway Bridge model length is 73.5m. But
the moveable trench of all four tables is totally 70m, so only one side span is modeled, the main
cable of the other side span is directly anchored on the bracket. The side pylon and the side span
are placed on a large shaking table, which is combined by two table B and C, the middle pylon is
placed on table A, and the side pylon is placed on table D. The distribution of the four shaking
tables is shown as Fig. 3.

Shaking table B, C Shaking table A Shaking table D

Fig.3 the test model (unit: m)

3.2 Model design

3.2.1 Side pylon

The material used in the side pylon is micro-concrete. Micro-concrete has some important
characteristics that make it suitable in the model, for example, by adjusting the mixture ratio, the
elastic modulus can be reduced to meet the requirement of the modulus similarity ratio. The cross
section of the upper beam of the prototype side pylon is shown in Fig.4 (a), the lower beam in Fig.4
4 18TH CONGRESS OF IABSE, SEOUL, 2012

(b); cross section of the upper beam of the simplified scaled model side pylon is shown in Fig.4 (b),
and the lower beam in Fig.4 (d).

(a) (b)

(c) (d)

Fig.4 Cross section of side pylon (unit: mm)

The longitudinal reinforcement and the stirrup are simulated by the galvanized wire, the elastic
modulus of which is approximate to the prototype. Longitudinal reinforcement and the stirrup are
designed based on the principle of flexural equivalence and principle of shear equivalence,
respectively [10].
Moment similarity ratio:
m m m
M m f y As h0 Asm
SM = = = S S
fy l (1)
Mp f py Asm h0p Asp

SM S
So that: Asm = Asp = σ Sl2 Asp (2)
S f y Sl S fy

Shear similarity ratio:

Asvm m
f yvm h0
Vm
Sm Sl Asvm
SV = = = S f yv (3)
Vp Asvp p S s Asvp
f yvm h0
Sp
SV S s p Sσ
So that: Asvm = Asv = Sl S s Asvp (4)
S f yv Sl S f yv

Where: p—prototype;
m —model;
fy —design strength of longitudinal reinforcement;
As —longitudinal reinforcement area;
h0 —sectional effective height;
fyv —design strength of stirrup;
Asv —stirrup area;
 Innovative Infrastructures - Toward Human Urbanism 5

S —stirrup spacing
Thus, by substituting the reinforcement area of prototype into formula (2) and (4), we can obtain the
reinforcement area of simplified scaled model. To obtain more reasonable model design, different
influential factors, i.e. different similarity ratios of concrete strength and reinforcement strength, as
well as spacing of stirrup and geometric similarity ratios should be well considered.

3.2.2 Main cable

The similarity ratio of axial stiffness of main cable must be strictly followed, as it has significant
influence on structure seismic response. Main cable wire cross section area of the simplified scaled
model is 6.8E-5m2; and the area of suspender is 2.3E-6m2.
3.2.3 Girder
The cross section of the prototype main girder is a flat thin-walled steel box. When the similarity
ratio is strictly followed, it would be difficult to manufacture the girder. As the girder is not the
focus of investigation, it is designed as a box section, seen in Fig.5

Fig.5 Cross section of main girder (unit: mm)

3.2.4 Middle pylon

The height of middle pylon stiffener is 500mm and the thickness is 40mm to 48mm, so it will be
difficult to manufacture middle pylon when the similarity ratio is strictly followed. The prototype
pylon section of the middle pylon is shown in Fig.6 (a), and the pylon section is shown in Fig.6 (c);
the pylon section of the simplified scaled model middle pylon is shown in Fig.6 (b), and the tower
section is shown in Fig.6 (d). The plate thickness is 1.5mm.

(a) (b)

(c) (d)

Fig.6 Cross section of middle pylon (unit: mm)

3.2.5 Additional masses

In the simplified scaled model, additional masses are attached to the model in order to augment the
density of the structure. The weight of all the additional mass is the difference between the mass
calculated from the similarity theory and the mass of the simplified scaled model. The total
additional mass is composed of 5000 pieces of cast iron, each of which is 5 kg. The additional
masses are required to be evenly distributed along the model, without inducing significant stiffness
variation in the model. Additional masses of the main cable are a pair of small steel blocks, which is
also used as a cable clamp.
6 18TH CONGRESS OF IABSE, SEOUL, 2012

4. FEM Verification
Without considering the impact of the soil stiffness, the foundation of the test model is completely
fixed. We impose the node mass to achieve the weight of the model. In order to verify the accuracy
of the simplified scaled model, SAP2000 was used to compare the dynamic characteristics and
response of the model and prototype. In order to facilitate comparison with the prototype, the
following chart of model data were normalized in accordance with the similar ratio. Error refers to
the (model value - prototype values) / prototype values.

South pylon

Middle pylon

North pylon

Side span

Fig.7 Sketch map of finite element model

4.1 Comparison of dynamic characteristics

Dynamic characteristics of the structure is the inherent characteristics of the structure, it is also the
important parameters to study the seismic response of structure [11]. The natural period of the
simplified scaled model must be consistent with that of the prototype, so that the dynamic
characteristics of the simplified scaled model could keep consistent with that of the prototype. Then
the simplified scaled model can be adopted to accurately reflect the dynamic response of the
prototype. The comparison of natural period between prototype and simplified scaled model are
listed in table 2.
Table.2 comparison of dynamic characteristics
Normalized
Mode Prototype period (s) simplified scaled Error Mode features
model period (s)
Girder first order anti-
1 13.279 13.248 -0.27% symmetric lateral bending
Girder first order anti-
2 13.169 13.059 -0.77% symmetric vertical bending
Girder first-order
3 11.067 11.091 0.16% symmetrical lateral
bending
Girder two order anti
4 8.778 8.774 -0.11% symmetrical vertical
bending
Main girder of first-order
5 8.627 8.634 0.02% symmetrical vertical
 Innovative Infrastructures - Toward Human Urbanism 7

bending
Girder two order
6 6.688 6.662 -0.26% symmetrical vertical
bending
Girder high order anti-
7 6.048 6.051 -0.04% symmetric vertical bending
Girder high order
8 4.628 4.512 -2.6% symmetrical vertical
bending girder
Middle pylon in the first-
9 4.433 4.416 -0.33% order lateral bending
Girder high order anti-
10 4.416 4.402 -0.35% symmetric vertical bending
It can be found from Table 2 that the errors of the natural period between the simplified scaled
model and prototype are very gentle, and the maximum error is only 2.6%.

4.2 Response spectrum analysis

In order to verify the simplified scaled model more accurately, SAP2000 was used to analyze the
reaction of simplified scaled model and prototype under the frequent earthquake ground motion, the
input acceleration response spectrum can be seen in Fig.8 (a), 8 (b).
8 8
ACCELERATION(m/s2)

ACCELERATION(m/s2)

7 7
6 6
5 100year10% 5 100year10%
4 4
3 3
2 2
1 1
0 0
0 5 10 15 20 0 0.5 1 1.5 2 2.5 3
PERIOD(s) PERIOD(s)

(a) Prototype (b) Model

Fig.8 response spectrum
The Table.3 and Table.4 show the comparison of the internal forces and displacements of the critical
section between simplified scaled model and the prototype under the acceleration response
spectrum, in both longitudinal+ vertical and transverse + vertical directions.
Table.3 Comparisons of forces of critical sections of main pylon
Normalized simplified
Prototype result Error
Working Section scaled model result
condition location axial force moment axial force moment axial
moment
KN KN-m KN KN-m force
bottom of
the north 2.127E+04 3.704E+05 1.846E+04 3.958E+05 -13.2% 6.9%
pylon
bottom of
longitudinal+
vertical
the south 2.124E+04 3.704E+05 1.792E+04 3.628E+05 -15.6% -2.1%
pylon
bottom of
the middle 2.862E+04 5.091E+04 3.337E+04 4.497E+04 16.6% 11.7%
pylon
bottom of
the north 2.827E+04 5.295E+05 2.776E+04 5.565E+05 -1.8% 5.1%
pylon
bottom of
lateral +
vertical
the south 2.868E+04 5.387E+05 2.523E+04 5.302E+05 -12.0% -1.5%
pylon
bottom of
the middle 4.102E+03 9.342E+04 4.014E+03 1.107E+05 -2.1% 18.4%
pylon
8 18TH CONGRESS OF IABSE, SEOUL, 2012

Table.4 Comparisons of displacements of critical points

Normalized simplified
Prototype result Error
scaled model result
Node location
Longitudinal lateral Longitudinal lateral
Longitudinal lateral
(m) (m) (m) (m)
top of the North
pylon
0.060 0.081 0.063 0.085 5.5% 4.6%
top of the middle
pylon
0.090 0.087 0.093 0.076 3.9% -12.6%
north end of the
main girder
0.090 0.036 0.091 0.037 1.4% 3.8%
middle of the
north main girder
0.082 0.204 0.083 0.204 0.8% 0.1%
Note: the comparisons of longitudinal displacements of the critical points under longitudinal input
and the comparisons of lateral displacements of the critical points under lateral input are listed in
this table.
From the chart above, though the maximum internal force error of the main pylon critical section
reaches 18.4%, and the maximum displacement error of middle pylon reaches 12.6%, according to
the test purpose, those errors caused by the simplified scaled model do not have much impact on
shaking table test results

5. Conclusions
Finite element model is successfully built through simplified scaled model method using the
similarity theory. Comparing finite element model to the prototype, we can see similar dynamic
characteristics, identical model type and small periodic error which is less than 2.6％.Meanwhile,
according to the response spectrum analysis , we can find that the errors of critical section internal
force and displacement are relatively small. So the errors would not affect the test result, which
shows that the simplified scaled model can be used for the shaking table test.

6. Acknowledgements
This work is undertaken with founding from the National Key Technology R&D Program
(No.2009BAG15B01).

7. References
[1] FAN LICHU, HU SHIDE, YE AIJUN., Seismic design of bridge with great span, Beijing:
China Communication Press, 2001.
[2] FAN KE., Experiment Research and Theoretical Analysis on Multi-support and Multi-
dimensional Earthquake of Large Span Structures, Beijing: Beijing University of
Technology, 2008.
[3] Yang CY., Seismic analysis of long span bridges including the effects of spatial variation of
seismic waves on bridges, Hong Kong University of Science and Technology,2007.
[4] LEI JUNQING, ZHENG MINGZHU, XU GONGYI., Suspension Bridge Design, Beijing:
China Communication Press, 2002.
[5] QU WEILIAN, SUN WUYI, ZHOU QIANG., “Shaking Table Test Research on
Insufficient-mass Model of Wanzhou Changjiang Bridge”, Journal of Wuhan University of
Technology, 2005, 45-49.
[6] ZHANG GUANYONG., Testing of Bridge Structure, Beijing: China Communication Press,
2002.
[7] XU TING., Similarity Theory and Model Test, Beijing: China Agricultural Machinery Press,
1982.
 Innovative Infrastructures - Toward Human Urbanism 9

[8] YANG XUDONG., the Studying of some Problems in Shaking Table Model Test, China
Academy of Building Research, 2005.
[9] GAO WENJUN,NIU SONGSHAN,LAN HAIYAN., “Test and Study on Model of Vibrating
Table in Cable-stayed Bridge with Single Cable Plane and Single Tower”, Journal of
Highway and Transportation Research and Development,2010（2）：45-48.
[10] ZHOU YING, LU XILIN, LU WENSHENG., “Shaking Table Test Model Design in
Different Structures”, Structural Engineers, 2006.22（4）：37-40.
[11] ANIL K. CHOPRA., Dynamics of Structures. (Theory and Application to Earthquake
Engineering), Tsinghua University Press，2009.