Field Load Tests and Numerical Analysis of Qingzhou

Cable-Stayed Bridge
Wei-Xin Ren1; You-Qin Lin2; and Xue-Lin Peng3

Abstract: A field load test is an essential way to understand the behavior and fundamental characteristics of newly constructed bridges
before they are allowed to go into service. The results of field static load tests and numerical analyses on the Qingzhou cable-stayed bridge
共605 m central span length兲 over the Ming River, in Fuzhou, China are presented in the paper. The general test plan, tasks, and the
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

responses measured are described. The level of test loading is about 80–95% of the code-specified serviceability load. The measured
results include the deck profile, deck and tower displacements, and stresses of steel-concrete composite deck. A full three-dimensional
finite-element model is developed and calibrated to match the measured elevations of the bridge deck. A good agreement is achieved
between the experimental and analytical results. It is demonstrated that the initial equilibrium configuration of the bridge plays an
important role in the finite-element calculations. Both experimental and analytical results have shown that the bridge is in the elastic state
under the planned test loads, which indicates that the bridge has an adequate load-carrying capacity. The calibrated finite-element model
that reflects the as-built conditions can be used as a baseline for health monitoring and future maintenance of the bridge.
DOI: 10.1061/共ASCE兲1084-0702共2007兲12:2共261兲
CE Database subject headings: Bridges, cable-stayed; Load tests; Bridge tests; Finite element method; Structural safety; China;
Monitoring.

Introduction Field load testing of a bridge gives a more accurate estimation

of its load-carrying capacity. Depending on the purpose of testing,
Field load testing is always the most essential way to understand the field load tests have been commonly used in the strength-
the behavior and fundamental characteristics of bridges for vari- evaluation activity of bridges 共Moses et al. 1994兲, experimental
ous purposes. Despite improved modern structural analysis with load distribution and rating of existing or deteriorated bridges to
computers, there are numerous cases where field load testing can determine their safe capacity 共Amer et al. 1999; Barker 2001;
not be completely avoided. The reasons for field testing include Huang et al. 2004兲, and comprehensively understanding the ef-
uncertainties in material and structural modeling of numerical fects of various design variables on bridge performance under
prototypes and concerns for serviceability limit states. Basically, static live loads 共Roberts-Wollmann et al. 2001兲. The information
the analytical bridge analysis requires accurate information about obtained from these tests is used to either refine and improve
material properties, support conditions, contribution of nonstruc- design procedures or bring the bridges up to the current standards
tural members, effect of deterioration, and many other factors. For of safety.
Cable-stayed bridges are appearing in various exotic forms
simplicity, conservative assumptions are often made to account
resulting in complex, efficient, and aesthetically pleasing struc-
for these uncertainties in numerical analysis. Therefore, it is com-
tures. The safety of cable-stayed bridges present an increasing
monly observed that during field load testing of bridges, the ac-
important concern in design, construction, and service. This spe-
tual load-carrying capacity is higher than what is predicted by
cial type of flexible, large span cable-stayed bridges makes the
analytical methods 共Bakht and Jaeger 1990; Saraf and Nowak
structural analysis more complex and difficult. A large span cable-
1998兲.
stayed bridge exhibits nonlinear characteristics under loading.
1
These geometrically nonlinear sources may come from the cable
Distinguished Professor, Dept. of Civil Engineering, Central South sag effect, the interaction between axial force and bending mo-
Univ., Changsha, Hunan Province 410075, People’s Republic of China
ment of bridge girder and tower, and the large displacement
共corresponding author兲. E-mail: renwx@mail.csu.edu.cn
2
Ph.D. Student, Dept. of Civil Engineering, Fuzhou Univ., Fuzhou, effect. Many investigators have presented different analysis meth-
Fujian Province 350002, People’s Republic of China. E-mail: ods to deal with such a nonlinear structural system 共Nazmy and
lyq@fzu.edu.cn Abdel-Ghaffar 1990; Wilson and Gravelle 1991; Adeli and Zhang
3
Research Assistant, Dept. of Civil Engineering, Fuzhou Univ., 1995; Ren 1999; Ren and Obata 1999兲. Some researchers disre-
Fuzhou, Fujian Province 350002, People’s Republic of China. E-mail: garded all sources of nonlinearities, whereas others included one
peng@fzu.edu.cn or more nonlinearity sources. Some nonlinear analyses of cable-
Note. Discussion open until August 1, 2007. Separate discussions stayed bridges have focused on either plane case 关two-
must be submitted for individual papers. To extend the closing date by
dimensional 共2D兲兴 or space case 关three-dimensional 共3D兲兴.
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- To propose cable-stayed bridges for very large spans of more
sible publication on June 26, 2005; approved on February 24, 2006. This than 1,000 m, particularly because of their special shapes, it is
paper is part of the Journal of Bridge Engineering, Vol. 12, No. 2, anticipated that more uncertainties are involved in the analysis
March 1, 2007. ©ASCE, ISSN 1084-0702/2007/2-261–270/$25.00. and design. The field load test is, therefore, even more important

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007 / 261

J. Bridge Eng. 2007.12:261-270.

 to exactly understand the actual performance of cable-stayed
bridges under live loads. Statically loaded field tests on newly
constructed cable-stayed bridges is now compulsory in China to
check both design and construction, and finally to determine
whether the bridge is allowed to open to traffic or not. There are
some guidelines to carry out the field load tests on the regular
bridges in several countries. However, there are no such guide-
lines available for the testing of cable-stayed bridges, and there is
a lack of reported work on field load tests on large span cable-
stayed bridges.
The field static or dynamic tests on large cable-stayed bridges
have been of great interest not only in investigating bridge fun-
damental behavior but also in calibrating a finite-element model.
Several results of field tests and correlated finite-element analyses Fig. 1. A photo of Qingzhou cable-stayed bridge
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

have been presented for large span cable-stayed bridges. Hulsey

and Delaney 共1993兲 reported the statically loaded test results and
comparison with a two-dimensional finite-element model for the 共41.13 m + 250 m + 605 m + 250 m + 40.21 m兲. The bridge was
Captain William Moore Greek cable-stayed bridge. Zhang et al. completed in 2000, but it was officially opened to the traffic in
共2001兲 described the field ambient vibration tests and three- 2002 due to the construction delay of approach spans. The two-
dimensional finite-element model updating for the Kap Shui Mun way roadway of the bridge deck is 29 m wide with six lanes. Fig.
Bridge, a 430 m main span length of double-deck cable-stayed 2 shows the general view of the Qingzhou cable-stayed bridge
bridge, in Hong Kong. Cunha et al. 共2001兲 implemented field with schematic plan, elevation, and typical cross section of the
ambient and free vibration tests for the Vasco da Gama Bridge steel-concrete composite deck. The main structural features of the
with 420 m central cable-stayed span length, showing a reason- bridge are briefly described as follows.
able correlation with the three-dimensional finite-element predic- The steel-concrete composite deck construction has an open
tions in terms of natural frequencies and mode shapes of the section consisting of two main I-type steel girders, steel floor
bridge. Fang et al. 共2004兲 presented the field static tests carried beams and 25 cm thick concrete slab. The slender steel girder is
out on the Kao-Ping-Hsi cable-stayed bridge 共330 m main span 2.45 m high and its maximum plate thickness is 80 mm. The ratio
length兲, the longest cable-stayed bridge in Taiwan, before it was of steel girder height to span length is about 1 / 250. There are, in
open to traffic. A three-dimensional finite-element analysis shows total, 257 steel floor beams with a spacing of 4.5 m. These steel
a very good agreement with the bridge measurements. Ren and floor beams are connected to both steel girders by M24 high
Peng 共2005兲 studied the baseline finite-element modeling through strength bolts. One steel stringer is designed in the middle of the
field ambient vibration tests for a newly constructed cable-stayed cross section. The precast concrete slab with cast-in-place joints
bridge with 605 m main span length in China. Some important and post-tensioning in both longitudinal and transverse directions
issues in the modeling of this type of complicated bridge—such is connected to the steel girders and floor beams by shear studs.
as the initial equilibrium configuration due to dead load, geo- The two diamond-shaped towers are of reinforced concrete.
metrical nonlinearities, concrete slab, the shear connection of The height of the towers is 175.5 m with 145.5 m above the
steel-concrete composite deck, and expansion joint effects—has bridge deck. The clear navigation is 43 m. The towers are erected
been clarified based on the test results. on a group of concrete-filled steel tube piles. The longest pile is
The objective of this paper is to present the procedure and 71.6 m with a diameter of 3.0 m.
results of statically loaded field tests that were carried out on the The cable arrangement is of fan type in both planes. There are,
Qingzhou cable-stayed bridge in Fuzhou, China before allowing it in total, 21⫻ 8 = 168 stay cables spaced at 13.5 m intervals along
to begin its service. Its 605 m main span length ranks it among each edge of the deck. The longest cable is over 312 m. The
the top twenty longest cable-stayed bridges in the world. The cables are composed of a number of strands varied from 27 to 85
general test plan and the measured responses are described. The in eight groups. One strand includes seven high strength steel
main tasks of the load tests include: 共1兲 measurement of the as- wires with the diameter of 5 mm 共7 ␾ 5 high strength wires兲. A
built deck profile; 共2兲 observation of the deck and tower displace- coextruded high density polyethylene 共HDPC兲 pipe for corrosion
ments under test truck loads; and 共3兲 checking the strains protection has been used which has a brilliant white outer layer
共stresses兲 of the steel-concrete composite deck of the key sec- eliminating the necessity to use a tape wrap.
tions. In addition to experimental studies, a full three-dimensional
finite-element model is developed and applied to calculate the
bridge responses. The comparison between analytical and test re- Objectives and Instrumentations of Field Load Tests
sults will indicate whether the real working condition of the
bridge is in accordance with the design expectation or not. The fundamental objective of the field static load tests on the
newly constructed Qingzhou cable-stayed bridge is to check if the
bridge load-carrying capacity satisfies the design requirements
Bridge Description and to determine if the bridge is allowed to go into service ac-
cordingly or not. More particular objectives of the load tests
The Qingzhou cable-stayed bridge is one of the bridges on Luo- include: 共1兲 Finding out the real bridge performance and load-
Chang Highway over the Ming River in Fuzhou, Fujian Province, carrying capacity under the planned static load conditions; 共2兲
China. A photo of the bridge just before its opening is shown in checking the quality and reliability of construction; 共3兲 verifying
Fig. 1. The bridge has a steel-concrete composite deck construc- the accuracy and rationality of design principle to facilitate the
tion consisting of five spans with an overall length of 1,186.34 m future design of similar types of bridges; 共4兲 calibrating a three-

262 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007

J. Bridge Eng. 2007.12:261-270.

Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. General view of Qingzhou cable-stayed bridge: 共a兲 plan view; 共b兲 elevation view; and 共c兲 typical cross section of composite deck

dimensional finite-element model to have a baseline model for lenses that were installed at the top of both towers. Two lenses on
further study of the bridge performance under other types of ser- one tower were symmetrically located on both sides of upstream
vice load conditions; and 共5兲 setting up the basic data for the and downstream. The data were taken by means of general total
health monitoring and future maintenance of the bridge. stations 共GTS兲, a kind of optoelectronic instruments.
The main measurement tasks of the load tests on the Qingzhou To study the stress responses of the steel-concrete composite
cable-stayed bridge include the deck deflections, tower horizontal deck under various load conditions, the strain measurements were
displacements, and strains 共stresses兲 of the key structural parts implemented at seven critical sections along the bridge length.
under the planned test loading. One of the important features of a They are Sections A–G as shown in Fig. 2共b兲. Twenty-eight strain
large span cable-stayed bridge is that the dead load 共self-weight兲 gauges were instrumented at A, C, and G sections, whereas 43
is dominant. The pretensions in the stay cables control the internal and 35 strain gauges were instrumented at B, E, D, and F sec-
force distribution and the bridge deck profile. The initial equilib- tions. The strain gauge instrumentation on the steel girders, floor
rium configuration of cable-stayed bridges is, therefore, the final beam, and concrete slab of E section 共span center兲 is shown in
elevation of the bridge deck, that is the bridge initial equilibrium Fig. 4. In addition, 40 and 36 strain gauges were instrumented on
position due to dead load, and tension forces in the stay cables. the towers close to the deck and cable anchorages, respectively, to
The initial equilibrium configuration is important in cable-stayed check the corresponding responses under the planned truck loads.
bridges since it is a starting position to perform the succeeding As a result, there were, in total, 316 strain gauges instrumented in
analysis. Just before the load tests, the as-built elevation of the the load tests.
bridge deck was surveyed by using the precision leveling
instruments.
To measure the deflection of the bridge deck, the water tubes
installed just beside the crashworthy balusters of both upstream Description of Field Load Tests
and downstream sides along the span length were used as shown
in Fig. 3共a兲. There were, in total, 2 ⫻ 21= 42 deflection measure- The field load tests on the Qingzhou cable-stayed bridge em-
ment stations spacing approximately at 60 m to take the data as ployed the heavily loaded four-axle dump trucks, each with ap-
shown in Fig. 3共b兲. At the same time, the deck deflections were proximately 300 kN weight, to simulate the design live loads of
simultaneously surveyed by using the precision leveling instru- the bridge 共Truck-20 and Trailer-120 loads specified in the Bridge
ments to supplement the deflection results. The horizontal dis- Design Code of China兲. Due to the difficulty to hire the same type
placements of both towers were measured by using four reflected of such heavy dump trucks in the area, there were, in total, 24

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007 / 263

J. Bridge Eng. 2007.12:261-270.

Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Measurement of deck deflections: 共a兲 water tube installed at main span; 共b兲 data taken from the measurement station of water tubes

trucks employed during the static load tests. The individual axle loadings of two critical load cases for the symmetric loading on
load and spacing of each dump truck were carefully measured at the main span 共LC6兲 and eccentric loading on the main span
the nearby weight station before it was moved to the bridge. The 共LC7兲 during the field tests are shown in Fig. 7.
average axle loads and axle spacing of the dump trucks used in The field load tests on the Qingzhou cable-stayed bridge were
the field tests are as shown in Fig. 5. carried out on November 20–22, 2002, prior to officially opening
Ideally, the applied test loads should be identical to the design the bridge. To reduce the disturbance of environmental tempera-
live loads of the bridge. Due to the limitation of actual test con- ture on the measurements, the load tests were performed during
ditions, however, the applied truck loads and their distribution 8:00 p.m. to 5:00 a.m. In each load case, the total truck loads
used in the tests might be different from those specified in the were proceeded on the deck in two steps as an incremental load-
design codes. The applied test loads are normally designated by ing. During each load case, the initial data were measured before
the static test load efficiency truck loading. In each loading, the data were collected after
10 min to obtain the stable strains and deformation of the bridge.
Ss After unloading of all trucks, the corresponding data were re-
␩= 共1兲 corded again 10 min later to check whether the strains and defor-
S共1 + ␮兲
mation can resume completely or not.
where Ss = most critical value of static deformation or resultant
force at the specified section under the planned static test loads;
S = most critical value of static deformation or resultant force at Three-Dimensional Finite-Element Modeling
the same specified section under the design live loads;
␮ = impact factor used in the design of the bridge. Creating a good three-dimensional finite-element model for large
To fully understand the loading performance of such a large span cable-stayed bridges is not an easy task. Many different
span cable-stayed bridge, nine different load cases were imple- modeling strategies 共i.e., which element types, how many degrees
mented during the field static load tests. The trucks were posi- of freedom, etc.兲 are possible. The choice of strategy depends on
tioned back-to-back at the designated arrangement for each test the skill and experience of the analysts and on the intended ap-
configuration as shown in Fig. 6. All test load efficiency ␩ values plication of the model. The established finite-element model often
are within 0.8–1.0, which demonstrates the validity of the stati- requires achieving a balance between full bridge description and
cally loaded tests on the bridge. The photos showing the field the degrees of freedom. There is no unique way to conclude that

Fig. 4. Strain gauge instrumentation at the Section E of main girders

264 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007

J. Bridge Eng. 2007.12:261-270.

 moved if the element goes into compression, thus simulating a
slack cable. No bending stiffness is included, whereas the preten-
sions of the cables can be incorporated by the initial strains of the
element. The stress-stiffening capability is needed for the analysis
of structures with a low or nonexisting bending stiffness as in the
case of cables. The cable sagging effect can be incorporated with
the stress stiffening capability. The element is nonlinear and re-
quires an iterative solution. Each stay cable is modeled by one
element, which results in 168 tension-only truss elements in the
model.
Two steel girders, central stringer, and the T-type concrete
Fig. 5. Typical dump truck used in the field tests beams of both side spans are modeled as the 3D elastic beam
elements 共BEAM4兲, since they are the structural members possi-
bly subjected to tension, compression, bending, and torsion. The
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

the model developed by one is the best. Aimed at establishing a

baseline finite-element model for the long-term health monitoring floor beams are of variable sections, and thus, they are modeled
of the Qingzhou cable-stayed bridge, a full three-dimensional by the BEAM44 elements. Towers consist of both equivalent and
finite-element model was developed in ANSYS 共1999兲. The ge- variable sections so they are discretized by both BEAM4 and
ometry and member details of the initial model are based on the BEAM44 elements. All piers and platforms are modeled by the
design information and design blueprints of the bridge. The main solid elements 共SOLID45兲. The concrete slab is divided into 508
structural members of the bridge are composed of stay cables, shell elements 共SHELL63兲. In addition, 210 concentrated mass
girders, floor beams, concrete slab and towers, all of which are elements 共MASS21兲 are used to include the mass of nonstructural
discretized by different finite-element types in current model. members such as equilibrium blocks, parapet, and anchorages.
Modeling of the stay cables is possible in ANSYS by employ- The modeling of bridge boundary conditions is very important.
ing 3D tension-only truss elements 共LINK10兲, and utilizing its Two types of bridge bearings are used in the Qingzhou cable-
stress-stiffening capability. With this element, the stiffness is re- stayed bridge. Fixed bearings are used on Pier 2, while expansion

Fig. 6. Truck configuration for each load case

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007 / 265

J. Bridge Eng. 2007.12:261-270.

 stay cables. Before the load tests, the elevation of bridge deck was
surveyed, which is the initial deformed equilibrium configuration
of the Qingzhou cable-stayed bridge. As a valid finite-element
model, the initial deformed equilibrium configuration of the
model should be consistent to the as-built geometry profile of the
bridge deck. This can be realized by manipulating the initial ten-
sion force in each stay cable that is specified as an input quantity
in the cable elements as information about actual cable tensions
after the construction is lack. To calibrate the initial cable forces
to match the as-built elevation of the bridge deck, the design
cable tensions are first applied to each stay cable and the static
analysis under dead load is carried out to compare the calculated
deck profile with the measured one. The cable tensions are then
adjusted until the best match is achieved. It is observed that the
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

adjustment of each cable tension is within 9% compared with the

design cable tensions, which can be acceptable considering the
actual construction of the bridge. The final calculated deck pro-
files of both upstream and downstream are compared with the
measured profiles as shown in Fig. 9. It can be observed that the
measured and calculated profiles of the bridge deck are in close
agreement. As a result, the developed finite-element model can be
used as the basis to perform the succeeding analysis of the Qing-
zhou cable-stayed bridge.

Load-Deflection Behavior of Bridge Deck

Among all nine truck load cases, the results of two critical load
cases 共LC6 and LC7兲, as shown in Figs. 10 and 11, respectively,
are presented herein to illustrate the load-deflection behavior of
bridge deck. In load case 6, 24 dump trucks on the main span
were used to symmetrically load the maximum positive bending
moment at the span center 共Section E-E兲. In load case 7, 12 dump
trucks on the main span were used to eccentrically load the maxi-
mum bending moment at the span center 共Section E-E兲.
It is found that the measured maximum static deck deflection
at the span center is 63.08 cm without including impact effect.
Fig. 7. Photos of two critical load cases: 共a兲 symmetric loading on According to AASHTO specifications 共1996兲, the deflection due
the main span 共LC6兲; 共b兲 eccentric loading on the main span 共LC7兲 to service live loads plus impact should not exceed 1 / 800 of the
span length when designing simple or continuous span bridges,
which results in a deflection of 76.25 cm. Considering an actual
bearings are used on the rest piers. In the current model, bridge
test load efficiency of 0.94 and impact effect, it can be anticipated
bearings are modeled by a set of rigid link elements connecting
that the main girder deflection of the Qingzhou cable-stayed
the superstructure and piers. To simulate the actual behavior, the
bridge will be closer to the deflection limits of the AASHTO
fixed and expansion bearings are simulated by coupling the cor-
共1996兲 specifications for regular bridges. The tentative design
responding translational and rotational degrees of freedom at both
specifications of highway cable-stayed bridge in China 共1996兲,
end nodes of the link elements. In addition, there are expansion
however, indicate that the maximum deflection limit is 1 / 400 of
joints at Pier 0 and Pier 5. Longitudinal springs 共COMBINE14兲
the steel girder span of cable-stayed bridges. Clearly, the mea-
are then applied to account for the restraining action in the lon-
sured maximum deck deflection of the Qingzhou cable-stayed
gitudinal direction. The actual spring stiffness coefficient is deter-
bridge is far from the limit specified in the cable-stayed bridge
mined by using the first measured longitudinal natural frequency
design specifications, which shows the adequate stiffness of
共Ren and Peng 2005兲. Subsequently, the developed full three-
bridge girders. Furthermore, the upstream and downstream de-
dimensional finite-element model of the Qingzhou cable-stayed
flections, as shown in Fig. 10, are consistent—showing a very
bridge is shown in Fig. 8. The complete model consists of 1,840
good symmetrical structural behavior under symmetrical loading.
nodes and 3,238 elements resulting in 9,193 active degrees of
The torsion effect of bridge deck is clearly observed under the
freedom 共DOFs兲. The model represents the bridge in its current
eccentric load case of LC7, as shown in Fig. 11, where the mea-
as-built configuration and structural properties.
sured maximum deck deflections are 38.25 cm at upstream and
26.23 cm at downstream.
Field Tests and Numerical Analytical Results Both Figs. 10 and 11 clearly demonstrate a good correlation
between the calculated and measured deflections of the bridge
deck. In addition, analytically predicted deck deflections have
Initial Equilibrium Configuration
been compared with measurements made on the bridge in all
The initial equilibrium configuration of cable-stayed bridges is other load cases and a good agreement was found. It is observed
the equilibrium position due to dead load and pretension forces in that the initial equilibrium configuration of the bridge, as dis-

266 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007

J. Bridge Eng. 2007.12:261-270.

Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 8. Three-dimensional finite-element model of the bridge; 共a兲 3D view; 共b兲 side view; and 共c兲 local view of tower and deck

cussed previously, plays an important role in the calculation of

deck deflections. Once the finite-element model was calibrated to
match the as-built measured profile of the bridge deck, the model
can predict a quite good deck deflection. In addition, all of the
deflections observed in the entire testing programs were elastic. It
is further shown that the deck deflections can resume completely
after all truck loads were removed. As a result, the load perfor-
mance of the Qingzhou cable-stayed bridge was shown to be
satisfactory in terms of stiffness of bridge girders.

Horizontal Displacements of Towers

A comparison of analytical horizontal displacements at the top of
towers with those measured from field load tests is given in Table
1. It can be seen that the agreements are generally good in all load
cases. The calculated displacements are slightly smaller than the
measured displacements at Pier 2, while the calculated displace-
ments are slightly larger than the measured displacements at
Tower No. 3 共Pier 3兲. These observations agree with the finite-
element modeling of the connections between the towers and the
bridge deck. Tower No. 2 was modeled as the fixed hinge while
Tower No. 3 was modeled as the expansion hinge. However, an
actual connection between the tower and the bridge deck might Fig. 9. Comparison of bridge deck profiles: 共a兲 upstream profile; 共b兲
not be completely either fixed or expansion hinge. downstream profile

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007 / 267

J. Bridge Eng. 2007.12:261-270.

Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Measured and calculated deflection of bridge deck under the Fig. 11. Measured and calculated deflection of bridge deck under the
load case of LC6 load case of LC7

Stresses of Steel-Concrete Composite Deck observed that the stress discrepancy at the bottom flange is less
than that at the top flange. The measured stress of Point 1 at the
The deck of the Qingzhou cable-stayed bridge is a type of steel-
top flange is positive, which further confirms that the actual cen-
concrete composite construction where steel girders and concrete
troid axis of the steel-concrete composite section is in the con-
slab work together to carry the loads. The bending stresses of a
crete slab. In general, the calculated stresses are less than the
steel-concrete composite beam are normally calculated by the
corresponding measured stresses. It is indicated that the calcula-
equivalent section method. To determine the equivalent section of
tion of the effective concrete width for the steel girder is not
a steel-concrete composite beam, the contribution of the concrete
suitable to such a large composite span length. The actual inter-
slab needs to be first identified. According to the Design and
action between steel girder and concrete slab is still an issue to be
construction specification for steel-concrete composite structures
addressed in the design and analysis of steel-concrete composite
in China 共1992兲, this contribution is approximately represented by
decks for large span cable-stayed bridges. In addition, it was ob-
the participating 共effective兲 width of the concrete slab
be = min兵be1,be2,be3其 共2兲
Table 1. Measured and Calculated Horizontal Displacements 共mm兲 at the
Choosing the most critical section at the span center under the Top of Towers
truck loading case of LC6, the participating width of the concrete
Measured 共calculated兲 displacements 共mm兲
slab of the Qingzhou cable-stayed bridge can be calculated by
be = min兵8.3, 3.8, 7.8其 = 3.8 m. Subsequently, the equivalent sec- Load case 2-1#a 2-2#b 3-1#c 3-2#d
tion of the composite deck, as shown in Fig. 12, can be achieved
LC1 0 共0兲 0 共0兲 3 共0兲 2 共0兲
and the centroid of the equivalent section can be determined ac-
LC2 −50 共−52兲 −51 共−52兲 1 共−5兲 0 共−5兲
cordingly. Once the equivalent section is determined, the bending
stress of a steel-concrete composite beam can be calculated like a LC3 −23 共−25兲 −25 共−25兲 4 共−2兲 3 共−2兲
normal beam LC4 2 共0兲 0 共0兲 2 共0兲 1 共0兲
LC5 37 共36兲 35 共36兲 0 共3兲 1 共3兲
M LC6 126 共124兲 125 共124兲 129 共150兲 130 共150兲
␴= y 共3兲
Iz LC7 65 共62兲 65 共62兲 67 共73兲 63 共73兲
LC8 2 共0兲 2 共0兲 1 共2兲 1 共2兲
where M = bending moment at the specified section that is ob-
LC9 −9 共−7兲 −8 共−7兲 −40 共−50兲 −41 共−50兲
tained from the finite-element calculation; Iz = inertia moment of
the equivalent section; and y = distance to the centroid of the Note: Positive displacement: Toward the main span; negative
section. displacement: outward the main span.
a
The calculated and measured stresses at points 1–8 共Fig. 12兲 of 2-1#: measurement station of upstream at Tower 2.
b
the mid-span section under the loading case of LC6 are compared 2-2#: measurement station of downstream at Tower 2.
c
in Table 2, where the sectional bending moment M = 11.8 MN m 3-1#: measurement station of upstream at Tower 3.
d
and inertial moment Iz = 0.5258 m4 are implemented. It can be 3-2#: measurement station of downstream at Tower 3.

268 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007

J. Bridge Eng. 2007.12:261-270.

 In addition to experimental studies, a full three-dimensional
finite-element model of the Qingzhou cable-stayed bridge is de-
veloped and calibrated to the measurements. An initial equilib-
rium configuration of the bridge model is achieved by correlating
the measured elevations of the bridge deck, which is a starting
position to perform the succeeding analysis. The analytical re-
sponses calculated from the calibrated finite-element model have
demonstrated a good agreement with those measured from field
load tests of the bridge. It is observed that the initial equilibrium
configuration of the bridge deck plays an important role in the
calculation of deck deflections. Once the finite-element model is
calibrated to match the as-built profile of the bridge deck, the
model can predict a quite good deck deflection. A calibrated
finite-element model that reflects the as-built conditions of the
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

bridge can be used as a baseline for the health monitoring and

further maintenance of the Qingzhou cable-stayed bridge.
The calculated stresses of the steel-concrete composite deck of
Fig. 12. Equivalent section of steel-concrete composite deck 共unit: the Qingzhou cable-stayed bridge, by using the equivalent section
meter兲 method, are different from the measured stresses. It is indicated
that the current calculation of the effective concrete width for the
steel girder is not suitable to such a large composite span length.
served that all of the strains in the entire testing program were The actual interaction between steel girder and concrete slab is
elastic and they can resume completely when all truck loads were still an issue to be addressed in the design of steel-concrete com-
removed. As a result, the performance of the Qingzhou cable- posite decks for large span cable-stayed bridges.
stayed bridge to load was shown to be satisfactory in terms of
strength.
Acknowledgments
Conclusions Support from the National Natural Science Foundation of China
共NSFC兲, under Grant No. 50378021, is greatly acknowledged.
The Qingzhou Bridge is one of the longest cable-stayed bridges The first writer also thanks the financial support from the Program
with steel-concrete composite deck construction. The truck loaded for New Century Excellent Talents 共NCET兲 in University, Minis-
field tests have been carried out on the Qingzhou cable-stayed try of Education, People’s Republic of China.
bridge just before it was officially opened to the traffic. Various
loading conditions to simulate the bridge design live loads were
implemented during the field tests. The results of measured strains References
共stresses兲 show that all values of measured strains satisfied the
requirements of design, and all strains varied linearly with the Adeli, H., and Zhang, J. 共1995兲. “Fully nonlinear analysis of composite
increase in the planned truck loads. This indicates that the bridge girder cable-stayed bridges.” Comput. Struct., 54共2兲, 267–277.
possesses an adequate strength to resist the loadings. The results Amer, A., Arockiasamy, M., and Shahawy, M. 共1999兲. “Load distribution
of measured deflections show that both bridge deck deflections of existing solid slab bridges based on field tests.” J. Bridge Eng.,
and horizontal tower displacements met the requirements of de- 4共3兲, 189–193.
sign, and all deformation can resume completely when the ap- AASHTO. 共1996兲. Standard specifications for highway bridges, 16th Ed.,
plied truck loads were removed. This demonstrates that the bridge Washington, D.C.
possesses an adequate stiffness to resist the deformation. As a ANSYS. 共1999兲. User’s manual, revision 5.6, Swanson Analysis System,
result, all field test results have indicated that the bridge works in Houston, Pa.
the elastic stage, and the bridge has an adequate load-carrying Bakht, B., and Jaeger, L. G. 共1990兲. “Bridge testing—A surprise every
capacity under the planned truck load conditions. All require- time.” J. Struct. Eng., 116共5兲, 1370–1383.
ments allowing the bridge to start its service have been satisfied. Barker, M. G. 共2001兲. “Quantifying field-test behavior for rating steel
girder bridges.” J. Bridge Eng., 6共4兲, 254–261.
Cunha, A., Caetano, E., and Delgado, R. 共2001兲. “Dynamic tests on large
cable-stayed bridge.” J. Bridge Eng., 6共1兲, 54–62.
Table 2. Measured and Calculated Stresses of Composite Deck
Design and construction specification for steel-concrete composite struc-
Point y Calculated stresses Measured stresses tures. 共1992兲. China Architectural Industry Press, Beijing 共in
No. 共m兲 共MPa兲 共MPa兲 Chinese兲.
1 −0.057 −1.28 2.52 Fang, I. K., Chen, C. R., and Chang, I. S. 共2004兲. “Field static load test on
Kao-Ping-Hsi cable-stayed bridge.” J. Bridge Eng., 9共6兲, 531–540.
2 0.143 3.21 —
Huang, H. X., Shenton, H. W., and Chajes, M. J. 共2004兲. “Load distribu-
3 −0.057 −1.28 3.36 tion for a highly skewed bridge: Testing and analysis.” J. Bridge Eng.,
4 2.063 46.30 68.30 9共6兲, 558–562.
5 2.343 52.58 67.40 Hulsey, J. L., and Delaney, D. K. 共1993兲. “Static live load tests on a
6 2.343 52.58 66.40 cable-stayed bridge.” Paper 930507, Transportation Research Record.
7 2.343 52.58 65.70 1393, Transportation Research Board, Washington, D.C.,
162–174.
8 1.103 24.75 42.80
Moses, F., Lebet, J. P., and Bez, R. 共1994兲. “Applications of field testing

JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007 / 269

J. Bridge Eng. 2007.12:261-270.

 to bridge evaluation.” J. Struct. Eng., 120共6兲, 1745–1762. Roberts-Wollmann, C. L., Breen, J. E., and Kreger, M. E. 共2001兲. “Live
Nazmy, A. S., and Abdel-Ghaffar, A. M. 共1990兲. “Three-dimensional load tests of the San Antonio ‘Y’.” J. Bridge Eng., 6共6兲, 556–563.
nonlinear static analysis of cable-stayed bridges.” Comput. Struct., Saraf, V., and Nowak, A. S. 共1998兲. “Proofload testing of deteriorated
34共2兲, 257–271. steel girder bridges.” J. Bridge Eng., 3共2兲, 82–89.
Ren, W. X. 共1999兲. “Ultimate behavior of long-span cable-stayed Wilson, J. C., and Gravelle, W. 共1991兲. “Modeling of a cable-stayed
bridges.” J. Bridge Eng., 4共1兲, 30–37.
bridge for dynamic analysis.” Earthquake Eng. Struct. Dyn., 20,
Ren, W. X., and Obata, M. 共1999兲. “Elastic-plastic seismic behavior of
long span cable-stayed bridges.” J. Bridge Eng., 4共3兲, 194–203. 707–722.
Ren, W. X., and Peng, X. L. 共2005兲. “Baseline finite-element modeling of Zhang, Q. W., Chang, T. Y. P., and Chang, C. C. 共2001兲. “Finite-element
a large span cable-stayed bridge through field ambient vibration model updating for the Kap Shui Mun cable-stayed bridge.” J. Bridge
tests.” Comput. Struct., 83共8–9兲, 536–550. Eng., 6共4兲, 285–293.
Downloaded from ascelibrary.org by Cornell University on 04/22/14. Copyright ASCE. For personal use only; all rights reserved.

270 / JOURNAL OF BRIDGE ENGINEERING © ASCE / MARCH/APRIL 2007

J. Bridge Eng. 2007.12:261-270.