Medway Bridge Design

Proceedings of the Institution of
Civil Engineers
Bridge Engineering 157
March 2004 Issue BE1
Pages 27^36
Paper 13512
Received 29/09/2003
Accepted 16/01/2004
Keywords:
David A. Smith Chris R. Hendy
beams & girders/bridges/concrete
Assistant Group Engineer, Special Structures Group
structures
Special Structures Group, Atkins Manager, Atkins Highways
Highways and Transportation, and Transportation, Epsom,
Epsom, Surrey, UK Surrey, UK
Design of the New Medway Bridge, England

D. A. Smith and C. R. Hendy
The New Medway Bridge comprises 12 continuous spans construction of two separate boxes with a common deck slab.
of twin-cell box girder with an overall length of 953 m and, This gave greater design efficiency due to the superior load-
at 152˝4 m has the longest externally post-tensioned span sharing properties but, by minimising the box width and
in the UK. It carries four London-bound lanes and a hard maximising the deck cantilever lengths, also created the
shoulder of the M2 and was constructed as part of the A2/ appearance of a lighter structure. The viaduct approaches, with
M2 widening project in Kent for the Highways Agency. spans up to 83 m, were constructed span-by-span with a
The need for external post-tensioning was imposed by the maximum span to depth ratio of 24 which adds to the
client for durability and maintenance reasons. External structure’s light appearance. They are supported on twin hollow
post-tensioning can lead to a loss in structural efficiency piers that were designed to be as slender as possible.
when compared with internal post-tensioning, particu-
larly with respect to anchor zones and overall flexure. The superstructure is monolithic with the river piers (piers 6
There is also an additional cost associated with the provi- and 7), which are of twin cell box construction. The viaducts
sion of deviators. The bridge was built under a Design and have fixed bearings at piers 3, 4 and 5 on the west viaduct and
Build contract, which made the need for economic and at pier 8 on the east viaduct. The remaining bearings are guided
buildable details all the more acute. This paper identifies longitudinally. The pier concrete was grade 50 to give adequate
the various ways in which a cost-efficient durable design axial capacity from the relatively slender box walls. The
was produced and the way detailing problems specific to viaduct twin piers share a common pilecap foundation.
such a large externally post-tensioned structure were
overcome. Examples include consideration of composite The site geology generally consists of made ground overlying
action between formwork and deck in construction, opti- terrace gravels and upper chalk. In the river, there was soft
misation of the permanent flexural design through overall
alluvial clay and silt of up to 3 m in depth. The west abutment
non-linear analysis, design of diaphragms and spanning and pier 1 are founded on pad foundations on grade A/B chalk.
anchor beams, constraints on placement of external The remaining foundations are founded on bored piles,
cables and simplification of the details for deviator blocks 1200 mm in diameter with the exception of those for the main
while maintaining adequate tolerances for construction. river piers which were 2400 mm in diameter due to the need to
resist ship impact. Typical pile lengths vary from 21 to 29 m.
1. INTRODUCTION Load tests were conducted on trial piles in the chalk and the
The New Medway Bridge comprises 12 continuous spans of results were used to refine the detailed design of subsequent
externally post-tensioned twin cell box girder with an overall piles to be less conservative than an initial design using CIRIA
length of 953 m. It carries four London-bound lanes and a hard Report 11.1
shoulder of the M2 and was constructed as part of the A2/M2
widening project in Kent for the Highways Agency. The bridge 2. GLOBAL DECK DESIGN
was constructed between the existing Medway Road Bridge,
built in the 1960s, and the new Channel Tunnel Rail Link 2.1. Loading and global analysis
bridge, which was under construction simultaneously for most The bridge is designed for the worst of 45 units of HB or AIL
of this Design and Build contract. The detailed design was loading together with associated HA loading. Since the bridge
carried out by Atkins for the joint venture contractor Costain– is externally post-tensioned, BD582 was used in conjunction
Skanska–Mowlem. The completed bridge is illustrated in Fig. 1 with BS5400:Part 43 in the design. The bridge was designed to
and general arrangement illustrated in Fig. 2. satisfy all the specified code serviceability and ultimate limit
state (SLS and ULS) criteria with one tendon having been
The bridge deck width is typically 21 m. The main span of removed for a replacement operation. Additionally, the bridge
152·4 m, the longest span in the UK for an externally post- was designed to survive at ULS under dead load with 25% of
tensioned bridge, was built by balanced cantilever construction the cables at any section ignored or any two cables removed.
(see Fig. 3) with a parabolic profile to the deck soffit to match
the appearance of the existing bridge. A single twin-cell box Analysis of the bridge was performed using a spaceframe global
girder was selected in preference to the existing bridge model with separate shear flexible grillage analysis to deter-
Bridge Engineering 157 Issue BE1 Design of the new Medway Bridge Smith Hendy 27
Downloaded by [ UNIVERSITY OF SHEFFIELD] on [16/09/16]. Copyright © ICE Publishing, all rights reserved.
rigorously since all increases
had a significant impact on
reducing reinforcement quan-
tities in such a large struc-
ture.
Strain increases were calcu-

lated by an iterative pro-
cedure considering the non-
linear material behaviour of
reinforcement, concrete and
tendons, with the increase in
length of the cable obtained
by strain integration along
the axis of the cable. Both
hand and computer methods
were used to check behaviour.
Generally all the strain
increase took place on the
elastic part of the tendon
stress–strain diagram due to
Fig. 1. Completed bridge the losses from creep, shrink-
age, draw-in and friction. The
magnitude of increases in the
mine the transverse distribution. Finite element models were cable strain possible led to the conclusion that, for Medway
used in the design of local elements. The construction sequence Bridge, it was generally not possible to increase the tendon
had a significant effect on the load effects to be built up and force significantly past the stress at which the tendon stress–
this is discussed further later. strain relationship first becomes non-linear, except for the
shortest cantilevering tendons.
2.2. Efficient design of the prestressing-strain increases
Externally prestressed bridges are less efficient than internally In the course of the calculations, it became clear that the
prestressed ones in flexure at ULS because the eccentricity of majority of the deformation, and thus strain increase, occurred
the tendons is likely to be less. Furthermore, the strain in the in relatively small regions where the reinforcement was
tendons does not increase at the same rate as the strain in the yielding. Therefore a considerable time-saving could be
adjacent concrete at the section being checked, as the tendons achieved by only integrating the strain in these plastic regions.
are not bonded to the concrete. On the new Medway Bridge, the
box was sufficiently deep that the loss of eccentricity from Normally in design it is safe to ignore tension stiffening.
placing tendons next to the flanges was not significant. The However, when calculating strain increases in external tendons
lack of strain increase however was very significant and this led this is unsafe as a tension-stiffened section will not undergo as
to ULS flexure governing the design as expected. much deformation as a fully cracked section. Where the tensile
stresses did not exceed the tensile strength of the concrete, the
Typically, it might be assumed that the tendon force does not section was deemed not to be cracked and the gross elastic
increase at ULS as any strain increase results only from the cross-section stiffness was used. Where stresses exceeded this
overall deflections of the structure and not from local high value, section stiffness was calculated from a cracked analysis
strains. BD582 allows a small strain increase without proof for using the stress–strain curves for reinforcement and concrete
mid-span regions of cables which do not extend beyond the given in BS5400:Part 4.3 However, account was taken of
supports, but most of the Medway Bridge tendons did not tension stiffening by adjusting the stiffness. For the typical
comply. The design therefore considered strain increases more reinforcement content on the Medway Bridge, it was expected
WA P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 EA
West viaduct Mainspans East viaduct
50.0 m 67.1 m 82.3 m 82.3 m 82.3 m 95.3 m 152.4 m 95.3 m 78.5 m 67.5 m 55.5 m 45.0 m
West East
abutment abutment
S S S F F F F S S S S F = Fixed bearing
S = Sliding bearings
Spread
foundation
Elevation
Fig. 2. Bridge general arrangement
28 Bridge Engineering 157 Issue BE1 Design of the new Medway Bridge Smith Hendy
required links around the
larger bars even if they had
been lapped straight.
3. PIER DESIGN
The most slender bridge piers
(see Fig. 4) were found to
develop very large additional
moments when designed in
accordance with the simple
formulae and effective
lengths in BS5400:Part 4,3 so
non-linear analysis (both
material and geometric) was
used to reduce the quantity of
reinforcement required.
The stiffness of the pier sec-

tions was determined from
the stress–strain relationships
in Figs 1 and 2 of
Fig. 3. Balanced cantilever main spans during construction BS5400:Part 4.3 A gf3 value
of 1·15 rather than 1·1 was
also used at the Technical
that tension stiffening would have little effect on sections Approval Authority’s instruction. Creep f factors from Appen-
where the steel strain from a fully cracked analysis exceeded dix C of BS5400:Part 4 were derived to account for the effects
1 millistrain. From a study of Clark,4 an average increase in of creep. To allow for this creep, a modification factor of 2·0
stiffness of 2·0 was applied to the cracked stiffness of sections was then used to stretch the concrete stress–strain diagram
where the steel strain from the non-linear cracked analysis was along the strain axis. It is noted that Eurocode 2 now gives a
less than 1 millistrain. method very similar to this.
2.3. Efficient design of the prestressing force at lock-off Initial imperfections were applied to the modelled piers. The
BS5400 Part 43 requires the force in a tendon at lock-off piers with free bearings were modelled individually with a
nowhere to exceed 70% of its ultimate tensile strength. This simple lean imperfection. Elastic critical buckling analysis was
effectively limits the maximum jacking force that can be used used to investigate the buckling mode shapes for the system
which must in no event exceed 80% of ultimate tensile comprising monolithic and pinned piers and similar imperfec-
strength. A departure from standards was sought to use tions were applied to this complete system. Eurocode 25was
Eurocode 2.5 This permitted a higher stress of 73·1% of tensile used for guidance on the initial imperfection, which for the free
characteristic breaking stress to be locked in. piers was taken as height/200. On site, the construction was
limited to a tolerance on verticality of half this design value.
2.4. Other design aspects^curved soffit The lateral deflections under load found by the non-linear
The depth of the main bridge box was varied parabolically so analysis were considerably less than those found to BS5400:
that the soffit was theoretically on an approximately constant Part 43 and a large saving on reinforcement was produced.
radius. The soffit was however formed from a series of chords.
The continuity tendons have to follow the bottom flange profile 4. CABLE LAYOUTS
via a series of deviators which gives rise to a series of
concentrated downward forces. These forces must be resisted by 4.1. Cable routing
the flange and deviator spanning transversely between webs. There appears to be far greater freedom in placing tendons in
an externally prestressed bridge as the whole space within the
A similar additional downward force arises from the long- box can be used (as long as access is still provided) and the
itudinal tension reinforcement in the flange required for tendons do not have to be confined within the concrete outline.
ultimate bending capacity. This was a significant force as the The disadvantage however is that the tendons cannot be
largest soffit bars were T40. As the bars were also placed as a gradually deflected throughout the span by the encasing
series of straight bars and lapped at segments at an angle of concrete, as can be done with internally prestressed bridges,
1·08 (including allowance for tolerance), there was concern over and physical deviators have to be provided.
the integrity of the lapping regions and the possibility that the
lapped bars could pull out through the cover. After looking at Tendons in a typical viaduct span were anchored at an anchor
test results for straight bars lapping and by considering the beam ahead of the pier (see Fig. 5). The tendons then deviate
possibility of bars shearing out, it was considered that small T6 downwards from the top of the pier diaphragm to a span
links should be placed around the lapping bars where their size deviator (see Fig. 6), pass horizontally to another span deviator
exceeded T25. It was noted that Eurocode 25 would have and then deviate back up to the previous pier diaphragm and
Fig. 4. Slender viaduct piers
WA P1
A B C D WB1WB5 WB7 WB11 E F
Tendons WA1, WA3, WA5, WA7,

WA9, WA11 & WA13 omitted from WB3 WB9 WB13
plans for clarity. See sections for
details
Tendons WB2, WB4, WB6, WB8, WB10,

WA14 WA10 WA4 WB12 & WB14 omitted from plans
WA8 WA6 WA2 for clarity. See sections for details
WA12
A B C D E F
Plan
Anchorage beam Anchorage beam
WA tendons WB1-14 P1 tendons WA1-14
300
Deviator Deviator
100
W2 W9
Tendons WA1-14 W01 W02
300
Drainage pipe Drainage pipe

2330
100
100
600
100
Elevation
box box
2x7 anchorages box
WB11
WA11
WB13
WB1
WB5
WB7
WA13
WA1
WA5
WA7
WB10
WB14
WA10
WA14
WB12
WA12
WB9
WB3
WA9
WA3
WB4
WA4
WB8
WB6
WB2
WA8
WA6
WA2
WA13,
2x7 deviators
WA11
WA10,
WA14,
WA1,
WA5,
WA7,
2x14 deviators
WA9,
WA3,
WA4,
WA12,
WA8,
WA6,
WA2
Section C_C
Section A_A Section E_E
box box
box 2x7 anchorages 2x7 anchorages
WB11
WA11
WB13
WB1
WB5
WB7
WB10
WB14
WA1
WA5
WA7
WA10
WA14
WB12
WA12
WB9
WB3
WB4
WA4
WB8
WB6
WB2
WA8
WA6
WA2
2x7 deviators
Section B_B
Section D_D Section F_F
Fig. 5. Typical prestressing arrangement in viaduct spans (dimensions in mm)
tendon at the blister in front.
Each newly anchored tendon
has to pass through the blister
in the previous segment,
because it is too close to the
previous tendon to pass out-
side the blister and allow
sufficient cover to its reinfor-
cement. The tendons must
then also be deviated at the
segment previous to this to
avoid clashes as they come
together towards the pier. The
longer tendons, from the
ninth segment onwards, were
all in the upper layer and
were straight between the
main piers and the segment 8
deviator. As these later seg-
ments are longer and the plan
angle of the tendons is smal-
Fig. 6. Viaduct cable routingin-span deviator and anchor beam (during construction) ler, the tendons only needed
to pass through the previous
one blister, without deviation,
onto the remote end anchorage beam at the far side of the pier. until they reached the main deviator beam at segment 8 (see
Tendons are thus lapped over the piers when the subsequent Fig. 10). The setting out of these deviator pipes was crucial both
span is completed so one main constraint on tendon position- to ensure that tendons did not clash and also to guarantee a
ing was to ensure that these two sets of cables missed each smooth profile without potentially damaging kinks at pipe
other at diaphragms (see Fig. 7). The other main constraint was entries and exits.
to ensure that a clear walkway was still provided throughout
the box so tendons move back towards the webs from their 4.2. Design of deviator pipes
spaced positions at the piers. The tendons must also keep clear The deviator pipes comprise 5·4 mm thick MDPE 140 mm outer
of the drainage. diameter pipes cast into the concrete. The HDPE duct around
each tendon, itself 5·3 mm thick with 110 mm outer diameter,
There were greater difficulties, however, with the balanced was then taken continuously through this larger pipe without
cantilever section because of the necessarily large number of any connection between the two. This was essential for
tendons. There is effectively less width available for routing the replaceability of the tendon. In all, there were approximately
cantilevering tendons than for an internally prestressed design 1600 deviator pipes required for the viaducts and main span.
since they cannot be run out
into the box cantilever
flanges. Consequently the
cantilevering cables would
not all fit within one layer
and a double layer was
necessary between the main
piers (Fig. 8) and the eighth
segment of the balanced can-
tilever, where a full width
deviator was provided (Fig. 9).
Between segment 8 and the

pier, each newly anchored
tendon has to pass through
the anchorage blister in the
previous segment and the one
prior to that. From their end
anchorages, the tendons set
off at a small horizontal
angle into the box such that
the tendon misses the burst-
ing reinforcement and Fig. 7. Viaduct cable routingtypical diaphragm at pier (during construction)
anchorage for the previous
misaligned pipe. Tolerance is
needed in any case even with
‘perfect’ setting out as the
deck changes shape during
construction from the pre-
cambered geometry and
deflects under load. A
number of possibilities were
considered to provide toler-
ance. These included:
. a large bell-mouth at pipe

entry
. compressible material
added around the pipe ends
to allow flexing
. over-bent pipe in the plane
of curvature.
The bell-mouth was rejected

because of the added con-
Fig. 8. Double layer of balanced cantilever tendons at main pier diaphragm struction expense and setting
out difficulty together with
the difficulty in detailing
There was some concern initially over the use of plastic reinforcement around it. The compressible material option was
deviator pipes because it was felt that they might be prone also ruled out because it would only transfer any kink in the
to floating during concreting and might not hold their shape tendon to the point at the end of the compressible material.
once bent to the specified radii. Neither of these concerns was This might prevent the concrete from spalling but could allow
actually found to be a problem on site during construction. the tendon to cut through the duct on stressing.
The setting out information for each pipe comprised a bend The over-bent pipe solution was selected. The theoretical
radius, an offset and level at each end of the pipe and a tangent point was kept some way inside the pipe. Typically an
rotation angle for the pipe from an axis through the plane of angular error in pipe setting out was provided so that the pipe
curvature of the pipe. The pipe was marked with a stripe at this could comfortably be 10 mm out of position in any direction at
angle from the plane of curvature and these marks were the each end. An overall translation of a pipe was less problematic.
setting out points on the pipe. It was appreciated that providing Allowance for these unintentional deviations was also made in
tolerance in the placement of these pipes was essential to avoid the design of the supporting elements. The over-bent pipe
problems with spalling of the concrete at the entry to a solution only really appears to provide tolerance in the plane of
curvature and not perpendi-
cular to it. However, since the
tendon should not exit
touching a side of the pipe, it
was considered that some
tolerance would be provided
in that plane also. A final
amount of tolerance was pro-
vided by the double thickness
of plastic in the two pipes,
which could allow an amount
of reshaping under com-
pression without damaging
the duct. Where the deviator
pipes had very large radii or
were, in some cases, straight,
the pipes also had a small
radiused flare machined into
the end of the plastic to give
a more gradual transition for
a misaligned tendon.
Fig. 9. Deviation and anchorage of the balanced cantilever tendons There was only one incident
of any concrete spalling
5. LOCAL ELEMENTS^
ANCHORAGES AND
DIAPHRAGMS
5.1. Stressing blisters and

anchor beams
For externally post-tensioned
bridges, the stressing
anchorages and surrounding
areas must be designed for
the characteristic strength of
the tendon as it is possible for
cables, particularly short
ones, to undergo sufficient
strain increase to reach yield.
Furthermore, anchor blocks
in internal post-tensioning
usually transmit the load
directly into the concrete
ahead of the anchorage. For
externally post-tensioned
structures, the anchor blocks
have to span as a slab or
beam back to the webs and
flanges in addition to resist-
ing bursting, spalling and
equilibrium effects within the
anchor block. Consequently,
the reinforcement for the
most heavily loaded anchor
beams became particularly
congested and therefore the
cages were detailed to be pre-
assembled and craned into
position wherever possible.
Extensive use of strut-and-tie
models was used in the
design of the anchor blocks.
Cantilevering stressing blis-

ters were initially sized to
give adequate shear connec-
Fig. 10. Main span cantilevering tendon layout
tion to the main box, based
on the maximum allowable
shear stress. The length was
around pipe exits and entries during tendon stressing and this however limited to ensure sufficient longitudinal clearance
resulted from a duct floating during concreting. between blisters for positioning jacks during possible future
tendon replacement.
4.3. Design of deviator beams
The reinforced concrete deviators themselves were designed A simplified strut-and-tie model formulated for a typical outer
using strut-and-tie idealisations in the same way as the web blister is shown in Fig. 11 where solid lines are ties and
diaphragms and anchorages described later. The large ‘gussets’ dashed lines are compression struts. It can be seen that the top
so derived were then included in the global shear flexible flange and webs require reinforcement to carry the overall
grillage analysis as it was expected that they would stiffen the bending effect induced by the eccentricity of the anchor from
box and attract distortional moments from live load. The the web–flange intersection. It was essential to ensure that the
prestress loads were also applied to this model to check the strut-and-tie model was a realistic idealisation of the true
distortional effects of these also. Significant moments were behaviour as although adequate ULS reinforcement is guaran-
indeed found to be attracted and the bottom flange was teed, performance at SLS is not. The behaviour was verified
therefore locally thickened to give an upstand beam. This by a three-dimensional finite element model using brick
additional transverse distribution was however also included in elements, that gave good agreement with the strut-and-tie
the global shear flexible grillage analysis, giving benefit to the predictions. A lower limiting reinforcement stress was used for
design of the webs and flanges. SLS design.
A general arrangement of a
typical viaduct diaphragm
region is shown in Fig. 12
and was well suited to a
strut-and-tie analysis as indi-
cated (where again, solid lines
are ties and dashed lines are
compression struts). Load
from the central web was
suspended over the access
holes by inclined bars with
large radii at the bends to
prevent overstress of the con-
crete. Suspension steel was
also required to carry load
from the bottom of the outer
webs back to the top of the
section. A critical aspect of
the detailing was ensuring
that the bend position of the
inclined bars did not allow
force to be transmitted on a
path through the top of the
access hole on its way back to
Fig. 11. Simplified strut-and-tie model for typical outer web blister the bearings.
Similar, but more compli-

The viaduct anchor beams are supported by both the webs and cated, models were devised to design the balanced cantilever
the top flange, but with some additional shear connection into web–diaphragm region.
the webs at the interface with the underside of the beams. A
typical anchorage beam can be seen behind the deviator beam 6. DESIGN FOR TEMPORARY CONDITIONS DURING
in Fig. 6. They were designed as a beam in bending and torsion CONSTRUCTION
with the reactions from the top flange treated as applied forces
and torques. The most heavily loaded anchor beam anchored 20 6.1. Propping to balanced cantilevers
cables, giving an applied design load of 10 000 t. At some For the permanent situation, the river piers were made slender
anchor beam locations, the webs were locally thickened to take so they could flex under temperature, creep, shrinkage and
the high compression at the interface with the web. The beam prestress without attracting excessive moments. Since the
was heavily reinforced for shear and torsion and the reinforce- overturning moments during construction of such structures
ment was therefore carefully detailed so the cage could be can be considerably in excess of the permanent design
prefabricated on the ground and craned into position in one moments, it was suspected that the piers might not possess
piece. adequate strength during construction without either propping
or an increase in column reinforcement.
A three-dimensional finite element model with brick elements
was also set up to check general behaviour and the assumed Natural frequency calculation was performed on all lengths of
distribution of pressures on the concrete ahead of the anchor the cantilever during construction. The lowest frequency
beam. This gave good agreement with the beam approach. occurred with the bridge ‘nodding’ in the longitudinal direction,
with twisting of the pier shaft next lowest at a similar
5.2. Diaphragms frequency. The nodding motion gives rise to large overturning
The viaduct diaphragms at
supports have to transmit the
bearing reactions into the
webs as shear and transmit
torsion (without distortion)
into the box girder from
uneven bearing loads. The
balanced cantilever dia-
phragm region also has to
transmit out-of-balance
moments from the fixity into
the pier. Strut-and-tie models
were developed for both Fig. 12. Simplified strut-and-tie model for typical diaphragm region (tendons not shown)
situations.
moments in the pier. At full
extension, the natural fre-
quency was only 0·2 Hz
which meant that turbulence
excitation could be a prob-
lem. BD496 requires this to be
checked where frequencies
are less than 1 Hz so the
0·2 Hz frequency was a con-
cern. A further concern was
that proximity of the existing
Medway Bridge and the new
Channel Tunnel Rail Link
bridge would increase the
turbulence experienced.
A dynamic analysis was per-

formed to assess turbulence
response. The ‘direct method’
described in Chapter 8 of
Barltrop and Adams7 was
used. The conclusion of the
analysis was that turbulent Fig. 13. Falsework for viaduct deck construction
wind produced by far the
largest overturning load on
the structure and this led to
the introduction of propping until adjacent cantilevers were crete applied during the build-up of the final deck cross-
joined (see Fig. 3). section. The pre-camber was therefore very sensitive to the
order in which lengths of deck were constructed as completed
6.2. Construction sequence for viaducts portions of ‘trough section’ (bottom slab and webs only) could
The viaduct spans were generally constructed on birdcage. stiffen up the system for subsequent pours. This sensitivity
Spanning falsework trusses were used across the railway line made it very difficult to vary the order of construction on site if
between piers 2 and 3 on the west viaduct and also on the east needed.
viaduct. The viaduct falsework systems are illustrated in Fig.
13. The box cross-section was built in stages. Firstly, the base The accuracy of pre-camber was particularly important for the
slabs were constructed with a kicker for subsequent construc- bridge. Had too much sag occurred after pre-cambering, the
tion of each of the webs. The webs were then constructed with tendons at mid-span would have lost eccentricity between the
deviators, anchor beams and diaphragms as appropriate. deviators since there was no mid-span deviator. With too much
Finally, the deck slab was
constructed using the webs to
support parts of the form-
work. Fig. 14 illustrates each
of these stages of deck con-
struction. Care was needed in
detailing reinforcement for
early thermal cracking with
this sequence.
The use of spanning form-

work gave a number of prob-
lems in both the design and
the construction. Despite
being approximately 6 m
deep, the trusses were rela-
tively flexible in comparison
with the deck itself, even
without the top slab. Conse-
quently, the majority of the
pre-camber needed for the
deck stemmed from the
deflections of the truss under Fig. 14. Staged construction of viaduct work
the self weight of wet con-
pre-camber, the cables could have fouled the bottom flange provided at the top of the duct, immediately down-stream of
since they ran close to the top side of the flange. To eliminate the deviator.
the risk of such problems, a mid-span deviator (of much less
robust proportions than others actually provided since devi- The anchorage trumpets were conical and plastic so that they
ations would only be second order) could have been used. could be withdrawn from the concrete (and replacements
reinserted) in the event of a replacement operation.
As the trusses were so flexible, the trough section without deck
slab could pick up significant moments under subsequent The proposed sequence for replacement of tendons was based
concrete pours as discussed above. This made compliance with on a similar procedure undertaken successfully in September
the requirement of BS5400:Part 43 to limit stresses during 2000 on the Mid-Bay Bridge in Florida, USA. This would
construction to 1 MPa of tension impossible on the truss spans involve removing the plastic duct along its full length (except
and difficult on the birdcage spans. It was therefore proposed to within deviators), removing the grout from the tendon and then
limit crack widths to 0·1 mm as an alternative. This led to the simply cutting through the stressed tendon with appropriate
need to put additional longitudinal reinforcement at the tops of protection to operatives. Binders would be installed around the
the webs to control the crack widths. However, benefit of the tendon at regular intervals along its length, with closer centres
load carried by the developing deck cross-section was taken in adjacent to cut positions, to prevent strand ‘whip-lash’ as each
reducing the design loads for the truss and the additional strand is cut. Once all the strands had been cut, the grouted
reinforcement provided was also considered in the design of the anchorage would be removed from its anchor beam and a new
permanent structure. trumpet installed into the bearing plate, filling the void left by
the previous trumpet.
7. TENDON REPLACEMENT USING A TYPE 1
8. ACKNOWLEDGEMENTS
TENDON SYSTEM
This paper is published with the permission of the Highways
The external tendons had to be replaceable. Distinction must
Agency and Costain–Skanska–Mowlem joint venture. The
first be made between Type 1 and Type 2 tendon systems. A
authors would also like to acknowledge the contribution from
Type 1 system comprises a conventional multi-strand tendon
VSL International who supplied and installed all prestressing
with bare strands grouted up within a duct. A Type 2 system
components and monitored all stressing operations on site.
also uses a multi-strand tendon, but individual strands are
Parts of this paper were published in the proceedings of the
contained within greased plastic sheaths and the sheathed
international symposium ‘The Role of Concrete Bridges in
strands are themselves grouted up in an outer duct.
Sustainable Development’ in Dundee in September 2003.8
The contractor requested the use of Type 1 tendons as the REFERENCES

subcontractor was concerned that the installation procedure 1. CONSTRUCTION INDUSTRY RESEARCH AND INFORMATION ASSOCIA-
would damage the sheathing for the strands as there was TION. CIRIA Project Report 11 Foundations in Chalk. CIRIA,
insufficient room on site to lay the strands out before pushing. London, 1994.
The designer preferred this option because the Type 1 ducts are 2. HIGHWAYS AGENCY. Design Manual for Roads and Bridges,
smaller than those required for the corresponding Type 2 BD58 Design of Bridges and Concrete Structures with
system and there was limited room within the box. The relative External and Unbonded Prestressing. Highways Agency,
levels of protection were also carefully considered. The Type 2 London, 1994.
system has an additional layer of protection provided by the 3. BRITISH STANDARDS INSTITUTION. Steel, Concrete and Compo-
sheathing. However, in the Type 1 system, the strand is in direct site Bridges, Part 4—Code of Practice for Design of
contact with the grout and therefore corrosion is inhibited by Concrete Bridges. BSI, Milton Keynes, 1990, BS5400.
its passivating alkalinity. 4. CLARK L. A. Tension Stiffening in Reinforced Concrete
Beams and Slabs under Short Term Loading, Cement and
In all cases, the anchorages were not equipped with a grout Concrete Association Technical Report. Cement and Con-
inlet, as this would have compromised their removability. The crete Association, July 1978.
anchorages were provided with a grout cap that is vented. A 5. EUROPEAN COMMITTEE FOR STANDARISATION. Eurocode 2: Design
single grout inlet was provided at the bottom of the duct, at a of Concrete Structures General Rules and Rules for
low point from where grout was pumped in both directions Buildings. CEN, Brussels, 1991, ENV 1992-1-1.
until it reached the anchorages. At this stage the grout was 6. HIGHWAYS AGENCY. Design Manual for Roads and Bridges,
forced through the wedges and filled the grout cap. The grout BD49 Design Rules for Aerodynamic Effects on Bridges.
cap was then vented until the quality of grout vented was the Highways Agency, London, 2001.
same as the grout at the injection point. Side vents were 7. BARLTROP N. D. P. and ADAMS A. J. Dynamics of Fixed
provided to the ducts immediately in front of the anchorage Marine Structures, 3rd edn. Butterworth-Heinemann,
beams to allow easier venting of the air in front of the grout London, 1991.
column. The vents were closed once it was clear that good 8. DHIR R. K., NEWLANDS M. D. and MCCARTHY M. J. Role of
quality grout was being expelled at this location. At locations Concrete Bridges in Sustainable Development. Thomas
where the duct passes through a high-level deviator, vents were Telford, London, 2003, pp. 53–62.
Please email, fax or post your discussion contributions to the secretary by 1 September 2004: email: [email protected];
fax: +44 (0)20 7665 2294; or post to Daniela Wong, Journals Department, Institution of Civil Engineers, 1^7 Great George Street,
London SW1P 3AA.

Medway Bridge Design

Uploaded by

Copyright:

Available Formats

Medway Bridge Design

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Medway Bridge Design

Uploaded by

Copyright:

Available Formats

Proceedings of the Institution of

Design of the New Medway Bridge, England

Strain increases were calcu-

Fig. 2. Bridge general arrangement

The stiffness of the pier sec-

Tendons WA1, WA3, WA5, WA7,

Tendons WB2, WB4, WB6, WB8, WB10,

Drainage pipe Drainage pipe

Fig. 5. Typical prestressing arrangement in viaduct spans (dimensions in mm)

Between segment 8 and the

. a large bell-mouth at pipe

The bell-mouth was rejected

5.1. Stressing blisters and

Cantilevering stressing blis-

Similar, but more compli-

A dynamic analysis was per-

The use of spanning form-

The contractor requested the use of Type 1 tendons as the REFERENCES

You might also like