See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/278968534

Construction Geometry Control Of The 268m Span Prestressed Concrete Box

Girder Bridge Over The 2 nd Navigational Channel Of Sutong Bridge

Conference Paper · September 2007

CITATIONS READS

0 883

8 authors, including:

Carlos Wong
CBJ Ocean Platform Engineering Corp., China
15 PUBLICATIONS   15 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Hong Kong Stonecutters Bridge View project

Hong Kong Zhuhai Macao Bridge ; Pre-construction drawings phase View project

All content following this page was uploaded by Carlos Wong on 23 June 2015.

The user has requested enhancement of the downloaded file.

 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 1 of 10

Construction Geometry Control Of The 268m Span

Prestressed Concrete Box Girder Bridge Over The
2nd Navigational Channel Of Sutong Bridge

ZHOU, Xinya, China Zhongtie Major Bridge Engineering Group Co. Ltd.
WONG, Carlos, Ove Arup & Partners Hong Kong Ltd.
WEN, Wusong, China Zhongtie Major Bridge Engineering Group Co. Ltd.
LUI, James, Ove Arup & Partners Hong Kong Ltd.

Abstract

Sutong Bridge, crossing Chang Jiang (Yangtze River) between Nantong and Changshu,
comprises of a 1088m cable stayed bridge in the major navigation channel and a 268m
prestressed concrete box girder bridge at the 2nd navigation channel. Together with the
approach viaduct the length of the river section is about 8 km. Construction started in 2004.
When completed, the main bridge will become the world’s longest span cable stayed bridge
whereas the 268m concrete box girder bridge will be amongst the world’s top 5 longest span
of its kind. Balanced cantilever construction method of insitu concreting on traveler
formwork is adopted. This paper describes the plan, procedures and measures taken during
the construction to control the bridge profile so that it will match with the design road
alignment at the targeted year after the concrete having crept to its final value. Zhongtie
Major Bridge Engineering Group Co. Ltd. is the contractor. Ove Arup & Partners HK Ltd. is
the construction consultant. Client is the Sutong Bridge Construction Commanding
Department.

Keywords
Sutong Bridge, 268 m span, construction alignment control, concrete box girder, prestressed
concrete bridge, balanced cantilever.

Introduction
Sutong Bridge, crossing Chang Jiang (Yangyze River) between Nantong and Suzhou,
comprises of a 1088m cable stayed bridge in the major navigation channel and a 268m
prestressed concrete box girder bridge at
the 2nd navigation channel. Together with Nantong Suzhou
the approach viaduct the length of the river
section is about 8 km. Construction started
2004. When completed, the main bridge +2.0MWL

will become the world’s longest span cable

stayed bridge whereas the 268m concrete
box girder bridge will be amongst the
world’s top 5 longest span of its kind.
Figure 7 shows the advanced stage of the Figure 1 Span arrangement
bridge construction.
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 2 of 10

The span arrangement, see Figure 1, is 140m + 268m +140m. The bridge is divided into two
box girders, namely Bridge A and Bridge B and each one carries one carriageway of 3 lanes
plus a hard shoulder. The box girder varies in depth from 4.5m at the crown to 15m at the two
piers following a parabolic curve; see Figure 2 of the section. Piers are made up with two
hollow walled type rectangular columns 7.5m x 2.5m x 0.5m thick. They are cast with the
box girder to form a rigid frame structure. The road profile contains a longitudinal gradient of
1.5% forcing the deck to follow. Transverse fall of 2% is used. Bearings are used at the end
piers, which are also shared by the adjacent viaducts.

Figure 2 Cross Section

The deck is prestressed in three directions: longitudinal, transverse and vertical directions.
Longitudinal prestressing consists of slab tendons, web tendons and stitch tendons for the
closing segments. Transverse prestressing is for the top flange slab only. Vertical prestressing
is by stressing bars in the webs. Construction method is by balanced cantilever cast insitu on
traveler formwork for 31 segments. The construction specification requires that the deck
profile shall be matching with the design road profile at the 30th year after opening. It
demands an accurate assessment of the concrete creep effects so that a pre-camber can be set
on the formwork such that it compensates the time-dependent deflection at the 30th year. By
the time of writing this paper in April 2007, the bridge is at it final stage of casting the
closing segments at the side spans and at the mid-span.

Control Strategy
Control Target
The control target is to have the bridge structure profile to match with the design road profile
at the 30th year since opening of the bridge, i.e., the concrete bridge will creep from its
original pre-cambered profile to the road profile after 30 years, and this requires a reasonably
accurate estimate of the creep effect and the corresponding pre-camber values together with a
strict construction plan.

Target Control Tolerance

Global Vertical Alignment is within ±50mm.
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 3 of 10

Staging Vertical Alignment is within ±20mm

Section cross fall is within ±0.2％.

Control Plan
Control plan is executed in the following sequence:
1. Pre-camber prediction calculation
2. Construction carried out according to the pre-camber prediction
3. Site measurement after casting, and feed-back results to geometry control unit
4. Error analysis and review of parameters, and prepare specs for correction (if any)
5. Correction (when required)
6. Move on to next stage of construction or to compute the next stage pre-camber values
based on the new parameters or new site measurement data.

Implementation of Control Plan

Pre-camber Prediction Calculation
The pre-camber prediction is based on the deflection calculation using space frame analysis.
Typical results are shown in Figure 3. Three programs in the forefront of this application are
used to cross check the results. They are the Midas, Ibdas and Tdv used by MBEC, COWI
and Arup respectively. The basic principle of all the three programs in this particular
application is that they all use space
frame with rigid offsets catering for 500
the varying C.G. of each element (or 450 Midas 78
the segment) and for the application 400 Midas-
Midas-78
of the tendons, see Figure 4. Each 350
program is believed to have some 300 TDV-
TDV-90
differences in the treatment of the
位 移 ( mm)

250

cables but the fundamental principle 200

150
is the same. The analysis follows the
100
casting sequence. The major Midas-
Midas-90
50
parameters used in the calculation are: 0
-50
Section Property 1 10 19 28 37 46 55 64 74 83 92 101 110 119 128 137
节点
Gross section of the concrete outline 武汉桥科院1 武汉桥科院2 Arup 设计院
without the contribution of
reinforcement, but with tendons
Figure 3 Comparison of using ceb78 and
activated according to the ceb90 in pre-camber calculation
construction sequence, i.e. the section
with voided ducts, the section with
un-bonded tendons and the section is
grouted with bonded tendons.

Concrete Property
The concrete grade is 60 with 20% Figure 4 Space frame calculation
PFA. The concrete property in the
calculation follows the ceb-fip78 model of the development of Young’s modulus, creep and
shrinkage and the concrete strength.

Tendon Property
The use of uPVC ducts in China is not popular, hence a study on its friction coefficient and
the performance under vacuum assisted pressured grout is carried out but before the test
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 4 of 10

results are available typical values as recommended in JTG D62-2004 [ref. 1]are used, i.e.
friction coefficient is taken as 0.15 and wobble coefficient is taken as 0.0015. Anchorage
draw-in is 6mm. The test results [ref. 2] recommend that the friction and wobble coefficient
be taken as 0.16 and 0.001, which are not far outside the assumed range.

Foundation Stiffness
Pile cap measured 33.2x49.6x7.0m is supported on 48 nos. 2.8m diameter bored piles. A
study to investigate the influence of the foundation stiffness upon the deflection concluded
that the foundation could be approximated by a fixed support at the pile cap top level, thus
simplify the modeling.

With the above assumed and measured parameters, the pre-camber values are calculated as
shown in Figure 3, which shows the difference between using the ceb-fip90 and ceb-fip78. It
is obvious that the pre-camber values as predicted by the latter are larger than the former one,
since the ceb-fip78 model gives greater creep value.

Construction Control
Traveler Formwork
Traveler formwork is the Doka system, see
Figure 5. The load-deflection characteristic
curve is obtained by test on site of the
formwork setup. The deflection is eliminated in
the setting of the formwork according to the
curve at the particular weight of the segment.
The centre of the gravity is also determined in
the test, which is taken as 1.0m from the face of
the already cast segment on the airside.

Temporary Construction Load

Machine and equipment are all weighed and
their existence in a segment construction period
is recorded with location.
Figure 5 Traveler Formwork
Section Dimension Control
The formwork is also used to make trial casting
(Figure 6) from which sectional dimension is
measured and the dimensional error can be
estimated. The error in dimension is used to
predict the error range in weight, which is in
the range of 1%. The influence of such error in
weight upon the pre-camber value is then
estimated, and is about 2mm.

Prestressing Control
While the stressing is done by force control, the
extension is used to check the tolerance, which
shall be within ±6% of the anticipated value.
Typical average deviation is 4.5% to -1.0%. Figure 6 Trial Casting of a Segment
This tells the fact that the friction and wobble
coefficients in the calculation have been over-estimated. Later tests prove that this is the case.
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 5 of 10

Site Measurement after Casting

Survey
The box girder has large area exposed to sunlight and wind. To minimize the influence of
temperature on the leveling survey, the work is carried out in the early morning before
sunrise. This is also the calm period when the wind speed is generally slowing down. The
bridge site is designated as the wind energy rich region; therefore it is not surprise to find out
the wind speed at site is very high (wind speed exceeds 10m/s 60% of the time in a year). The
strong wind could produce sufficient lift force acting on the cantilever deck causing
measurement error. A relatively calm climate is therefore targeted for surveying. Survey is
taken to the control points set in the segment covering the longitudinal profile and cross fall
measurements.

Environmental Data
The environmental data such as wind speed and temperature are continuously recorded and
monitored.

Strain/Stress Monitoring
Strain gauges are installed at the bridge piers and at the deck at the critical location. They are
used to monitor the stress in the concrete for 2 purposes: to provide another set of data to
cross-check with the calculation, and to give a warning sign should the stress in the concrete
become too high.

Error Analysis
Error analysis is conducted after obtaining the site measurements and the following are
identified in terms of pre-camber values:
• Surveying error including accidental, those due to wind load and system error: ±5mm
• Setting out error: ±5mm
• Concrete surface variation: ±5mm
• Traveler formwork deflection variation: ±2mm
• Segment weight deviation induced error: ±2mm
• Prestressing force deviation induced error: ±2mm

The maximum error when all the “misfortunes” happen at the same time would be about
±20mm, which is within the staging tolerance range.

Correction
The correction is done in several levels from the do-nothing Level 1 to the rectification work
in Level 4 and they are:
Level Action
1. No correction if the error is less than the staging tolerance (~20mm).
2. Half of the error is corrected in the remaining segments if the error is greater than
the staging tolerance but less than the correctable tolerance (~30mm).
3. Correct the target alignment if the error is greater than the correctable tolerance
but is less than the global tolerance (~50mm).
4. Carry out rectification work to rectify the error if it is greater than the global
tolerance.

Correction review is set at the Segment #19 when sufficient data are available to indicate a
need for the major correction of the pre-camber in the remaining segments. The major
correction so far is the support settlement due to foundation yielding, which record reads
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 6 of 10

25mm on the Pier 79 and 20mm on Pier 78. The deflection is then distributed evenly in the
remaining segments such that the Segment #31 will fully compensate the support settlement.
Method and procedures for the level correction are shown in Figure 8. Other corrections
carried out are for the revised Young’s modulus of the concrete, and the increase of the top
flange thickness by 25mm, as a result of design change.

Results of the Construction Geometry Control

Predicted Pre-camber
The predicted pre-camber is shown in Figure 9, which shows the three computer programs
produce practically identical results in the main span but not in the side spans. The difference
is due to the difference in interpretation of the closing procedures.

Measured Concrete Property

Experimental study [ref. 2] for the creep and shrinkage characteristic of the concrete used in
this project has been carried out. The study compares different codified shrinkage and creep
models with the tested values, as shown in Figure 10 for shrinkage and creep. The tests
indicate that the shrinkage predicted by ceb-fip78 is less than the tested values in magnitude
of 100 micron whereas the predicted creep factor expressed as the ratio of creep strain to
elastic strain under the sustained load is about 2 times of the tested value. The creep factor
calculated using ceb-fip78 is the highest amongst all the codified models, including ceb-
fip90, which is about 57% of the ceb-fip78 model. From calculation, the 10-year mid-span
deflection increment for ceb-fip90 and ceb-fip78 model is 39mm and 147mm respectively.

The average concrete cube strength exceeds 70 MPa; the tested Young’s modulus 49 MPa is
also higher than the codified value of 40 MPa. The increase of Young’s modulus has direct
effects on the pre-camber calculation, as such the pre-camber calculation is corrected midway
in the construction phase.

Vertical Alignment
The vertical alignment difference between the measured levels and the computed levels
(including pre-camber values) is shown in Figure 13 and 15 for the two cantilevers supported
by the Pier 78 and Pier 79 plotted as the level difference of the predicted verses the measured
data, after Segment #29 is cast. The maximum difference is 38mm at Segment #21. The
curves show that many of the corrections are done when the difference reaches 20mm except
for a few cases. The error analysis reveals that in most of the cases, the error came from the
setting of the formwork. After casting of segment #31, it is expected that there will be level
difference between the two cantilevers. As envisaged in the closing segment construction
method (Figure 11 and 12), the two cantilevers are to be pushed apart 60mm then to secure
the jacks and stabilize the two cantilevers. Clamper beams are installed between the two
cantilevers and stressing bars are installed for strand jacking to bring the two cantilevers into
leveling. Depending on the stiffness of the two cantilevers (one pier is taller than the other)
the level difference may not be equally shared by the cantilevers. However, the imposed
deflection at the cantilever tip (should be no more than 50mm) will not impose adverse stress
in the structure. The stress will become lock-in stress after the closing segment is cast and
cured and the removal of the clamping beams, and the jacks. It will be gradually dissipated
under the creep action of the concrete.

The built level and the computed level comparison are shown in Figure 14 and 16, and the
stress comparison between measured and computed values, at the maximum bending location
of Segment #2, is shown in Figure 17. It shows that the measured levels are greater than the
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 7 of 10

predicated levels echoing the fact that the actual creep is less than the predicated value. The
error analysis for the stress difference between the two indicates that the error of stress
difference could be 3 MPa, due to the uncertainty nature of the actual creep and the
computational creep. It can be seen from the plots that the max. deviation is less than 2MPa,
which is within the error tolerance range of ±3MPa. The max. stress is about -16MPa.

Discussions and Conclusions

The bridge has been successfully constructed up to the closing stage. The pre-cambered
bridge deck surveyed after casting of the Segment #29 shows that the vertical geometry
differs not more than 38mm from the calculated profile, and is less than the global tolerance
(50mm). The stress difference also shows good correlation of the two. The casting of the
remaining 2 segments is also expected to be within the limit. The careful preparation works
such as the load test of the traveler formwork to gain the reference load-deflection curve; the
trial casting; the skilful technicians who operate the traveler formwork; the dedicated survey
teams and the dedicated geometry control unit are the key factors to success. The experience
gained in this project could be used as a reference or a work example for similar bridge
construction.

The adoption of the ceb-fip78 creep and shrinkage model however, may over-estimate the
creep deflection, or the pre-camber values. The study carried out by the South East University
[ref. 2] shows that concrete property is close to the ceb-fip90 model, which reports a creep
value about half of that of ceb-fip78. One argument that supports the use of the ceb-fip78
model is that the traffic loads patterns in China differ substantially from the pattern of
European countries because the traffic load in China tends to be constantly running over the
bridge resemble a permanent load. A suggestion to take 1/3 of the live load as permanent load
for the determination of creep deflection has been proposed by members of the expertise
review panel [ref. 3]. The use of a greater creep factor would, to a certain degree, safeguard
the bridge from creeping below the design road profile, which is always psychologically
disturbance especially to the general public. Figure 7 shows the bridge in advanced stage.

ACKNOWLEDGEMENT
The authors sincerely thank the Sutong Bridge Construction Commanding Department for
their supports and kind permission of publishing this paper. Sincere thanks are also extended
to MBEC and Arup for their encouragement and supports on the writing of this paper.

Figure 7 Construction Progress on site sees the completion of 30 segment leaving

the last closing segment waiting to be cast
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 8 of 10

Figure 8 Procedures and rules used for setting out the traveler formwork
500
450 COWI (Ver. 1.5)
450 BSRI (2006.03.16)
400
ARUP (2006.04.07) Difference (BSRI-COWI)
400
BSRI (2006.03.16)
350
Theoretical Superelevation (mm)

350
Theoretical Superelevation [mm]

300
300
Main Pier Axis

Main Pier Axis

250
250

200 200

150 150

100 100

50
50

0
0

-50
-50
0 100 200 300 400 500 -100
X (m ) 0 100 200 300 400 500
- X - [m]

Figure 9 Comparison of pre-camber calculation between Arup-MBEC, and between

COWI-MBEC.

Figure 10 Comparison of tested Shrinkage and Creep data with codified values
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 9 of 10

Figure 11
Construction
sequence and
arrangement for the
Segment #31 before
closing up.

Figure 12
Construction
method proposed
for the closing
segment at the mid
span. Closing of
side spans is
similar. The side
spans will be closed
prior to the closing
of the mid span.

Figure 13 Level difference along the Figure 14 Predicted and measured level
alignment after Segment #29 is cast for cantilever supported by Pier 79, after
(plot for cantilever supported by Pier casting Segment #29 (note the max.
79) deviation is at the cantilever at the
midspan.)
 ZHOU, Xinya; WONG, Carlos, WAN, Wusong, LUI, James
Page 10 of 10

Figure15
Level difference
0.03
along the alignment
after Segment #29
is cast (plot for
cantilever supported
by Pier 78)

-0.03

Figure 16
Predicted and
58 measured level for
cantilever supported
57
by Pier 78, after
casting Segment
56
#29 (note the max.
deviation is at the
cantilever at the
midspan.)

Figure 17
Predicted and
measured stress at
Segment #2, after
casting Segment
#29
-10

-15

REFERENCES
[1] JTG D62-2004 “公路钢筋混凝土及预应力混凝土桥涵设计规范(Specification of design
of highway reinforced and prestressed concrete bridges and culverts)”, 中华人民共和国交通
部 (Ministry of Communication, People’s Republic of China, 2004.
[2] “苏通大桥预应力施工工艺及关键参数试验研究总结报告(Final Report of Study of
Prestressed Concrete Key Parameters in Application for Sutong Bridge”, 江苏省苏通大桥建
设 指 挥 部 、 中 港 集 团 第 二 航 务 工 程 局 、 东 南 大 学 ， (Sutong Bridge Construction
Commanding Depatment, China Harbour Engineering Company (Group), SouthEast
University), 2005.
[3] Sutong Bridge Expertise Review Panel Meeting, Nantong, 2005.

View publication stats