Вы находитесь на странице: 1из 8

Paper: Hambly/Nicholson

Paper

Prestressed beam integral bridges


E.c. Hmbly, MA(Cantab), PhD, FEng, FIStructE, FICE
Edmund Hambly Ltd.

B. A. Nicholson, MA(Cantab), CEng, MIStructE


Edmund Hambly Ltd

Tim Hambly has had his own practice since


1974 as a consulting engineer in structural,
offshore and civil engineering. Previously he
worked for Giffrds, Kiers, and Arups, and in
research at Cambridge. He is now a Visiting
Professor at Oxford. His books Bridge deck
behaviour and Bridge foundatlons and
substructures are very widely used. He is a
former member of the Institutions Council and
recipient of Murray Buxton and Guthrie Brown
awards.

1920 1930 1940 1950 1960 1980


1970
Bruce Nicholson graduatedfrom Cambridge Year
University in 1982. He then worked for 3 years
for Rendel Palmer & Tritton on several bridge Fig I . Increasing use of integral bridges in the USA
and highway projects. In the 5 years since joining
Edmund Hambly Ltd. he has worked on projects
involving bridges, buildings, pipelines, and
offshore structures. Spans skew
between bearings
9 15.000 -1-
l-
20000 _I
r-
20.000 _I_ 15.000
1-
4
1 :fj@

Elevation
Synopsis
Expansion joints on bridges have provided severe maintenance
problems for bridge authorities throughout the world. Bridge
engineers are increasingly endeavouring to design bridges as
continuous structures, and not as simply supported structures,
in order to minimise these maintenance problems. This paper
explains the concept of integral bridges in which the deck is
made continuous and also connected to the abutments, so that
all expansion joints are eliminated. A design method is \ i. \ \ \\ \

explained for a deck of composite construction using 15.000 h\\20.000 \L\


20000
prestressed concrete bridge beams, and a demonstration design /-
be$zt%:g?- \
is presented. This design uses Y-beams, which are being
Plan
introduced by the Prestressed Concrete Association to replace
M-beams. The design method is straightforward and avoids 12.600
the complication of calculating the effects of creep. y3Prestressed
beams 900 deep
Introduction
What are integral bridges? 65
Integral bridges are bridges which are constructed without any movement
joints between spans or between spans and abutments. The road surfaces
are continuous from one approach embankment to the other. Integral
bridges are becoming more widespread as engineers seek ways of avoiding
the very expensive maintenance problems due to thepenetration of water
and deicing salts throughmovement joints. TheDepartment of Transport
is understood to be considering recommending to designers that bridges
should be constructed as continuous structures, unless there are good reasons
for not doing so, for this reason.
Integral bridges have been used widely in the USA and Canada, andthis
report draws on the experience of bridge engineers from those countries m
as well as from the UK. Two important references are Reports NCHRP Fig 2. General arrangement of demonstration design
141 and 322 of the American National Cooperative Highway Research
Program. NCHRP 141 is about Bridge deck joints including integral
construction of bridges without deck joints. Fig 1, taken from NCHRP Objectives of thb paper
141, illustrates the developing popularity of integralabutments from1930 In the past, many of the bridges in the UK constructed using prestressed
to 1989. NCHRP 322 is concerned with Design of precast prestressed beams were designed as simply supported. Continuous bridges using
concrete girders made continuous and reviews the design methods of the prestressed beams have not been popular because of the supposed design
various States and makes recommendations. Burke also discusses these problem caused bycreep and the constructional problem involved in joining
subjects. the beams end to end. This paper offers simple solutions to both these

474 The Structural Engineer/Volume 68/No.23/4 December 1990


Paper: Hambly/Nicholson

I /

A-A
LB
B- B
o 17 No. fully bonded 158 strand
Q 6 NO.debonded a t 2500
0 6 No. debonded at 4500
0 6 No. T20 bottom flange reinforcement
each end (epoxy coated 1

Fig 3. Beam details

problems. The secondary objective is to encourage bridge designers,whether The slab is 200mm thick withf,, = 40MN/m2, giving a modular ratio
or not they use prestressed beams, to consider the benefits of integral of 0.91 relative to the beam. The composite section has properties of:
bridges. height = 1.065 m . I = 0.729m4
The demonstration design is for a four-span bridge carrying a single area = 0-548m2(equivalent) Zslabt= 0 . 154m3
carriageway over a motorway as shown in Fig 2. The dimensions have been -
y = 0-545m Z,,,,, = 0.205 m3
based on the DTp standard bridges. The width of the bridge allows for Zbeamb = 0 - 134m3
a 7.3 m carriageway, two 2-Om footpaths, plus 0.4m overwideningto allow
for a slight curvature. The spans cross a D3M motorway of width 36.5 m. Further details are described in later sections.
A skew of 20"has been chosen, so that skews of between 0 and about40"
can be built using similar details. The demonstration design uses Y3 beams Grillage analysis
at 1 m spacing. The design calculations are in accordance with BS 5400: The load distribution was calculated using a grillage analysis, illustrated
Part 4 4 , individual clauses of which are referred to in this paper using the in Fig 4, following the methods of Hambly'. One longitudinal member in
notation 'Pt4:4.2.2'. the grillage represents one prestressed beam, acting compositely with 1 m
width of slab with cantileverat edge. The mesh was relativelyfine to simplify
Prestressed concrete beams load description. The inertia of each memberwas based on the gross concrete
Design principles section. The torsion constant was set to zero in the analysis, as permitted
For the purposes of design of the beams, it is assumed that each span is by BS5400 Pt4: 5.3.4.2, meaning that the beams were assumed to carry
simply supported for live load as well as dead load (as is the case for a no torsional moment. This resulted in beam movements being increased
simply supported bridge). In the long term, creep under prestress may cause by about 12 Yo. A check was made at the end of the design with a grillage
the beams to assume a slight hogging curvature, and it is presumed that using full torsion stiffnesses, and it was found that thebeams had adequate
this would open partially the construction joints at the beam ends. The strength for coexisting torsion.
live load would have to close up these joints before any hogging moments The length of the beams is 19.8m. The span assumed in the analysis
could be generated over the supports. No account is therefore taken of was 19*6m,allowing for the centre of thebearing area at 0.1 m fromeach
the benefits of continuity over piers for live load at the serviceability limit end.
state, even though there are substantial reserves of strength at the ultimate
limit state. Conversely, no account is taken of the sagging moment which Loadings
can develop along the beams because of the restraint of creep deflections The dead load of the beams and the slab is carried by the beams acting
by bottom flange reinforcement. NCHRP 322 has shown that the effects alone, and the stresses were calculated by hand. Residual stress patterns
of restraint moments and continuity moments virtually cancel out in the were also calculated for temperature difference loading and differential
span, so that stresses are similar to simply supported conditions. The effect shrinkage, on the assumption that the beams were simply supported. The
of ignoring continuity is relatively small when the critical Code provisions grillage model was usedfor SDL (2.4kN/m2 for the carriageway, 6kN/m2
for pretensioned beams relate to stresses at the serviceability limit state. for the footpath), footpath live load (5 kN/m2), HA load (45 1kN/m),
If, at alater date, the Codeenables beams to be designed for ultimate limit and HB load (45 units). The loads and load combinations were taken from
state conditions, economy may be achieved by taking advantage of the BS5400: Part 2, as amended in BD37/886.
continuity. Five loadcombinations were defined in Pt2. However, only load
combinations 1 and 3 were considered relevant for thedesign of this bridge
Beam and slab details deck. Wind load, collision loads and frictional restraint on this deck were
The details of the beams are shown in Fig 3. The beams have Y3 section negligible, although collision loads have to be considered in the design of
with the following-~properties:
~
the piers and parapets.
Table 1 summarises the load combinationsconsidered for the design of
height 0-90m
ztI =
= 0.0265m4
= o. 0479 m 3 prestressed beams and lists the applicable partial load factorsfrom Pt2:
area 0-373m2
=
- Table 1. Pt 4:4.2.2 requires a modified version of load combination 1, in
Y 0-347m
= Z, = 0.0763 m3
which the HB load is limited to 25 units, as well as the full version.
feu = 50MN/m2
Thetotal weight of a25 unitHB vehicle = 25 x lOkN X 4
fci = 40 MN/m2 (transfer)
axles = 1000kN.Thetotal weightof HAloadononelane
weight = 9.OkN/m
= 45.1 kN/m x 20m + 120kN = 1020kN.Thus it was not obvious
The prestress comprises 29 strands of 15 -2mm diameter, for which the whether HA load alone, or HA+ 25 HB, would be critical for themodified
characteristic strength is 29 X 0-232 = 6-73MN, andthe initial prestress load combination 1, so both cases were considered. However, for the full
force Pi = 0.75 x 6.73 = 5.05MN. load combination 1,45 units HB was clearly more critical than HAalone.

The 68/No.23/4 December 1990 475


Paper: Hambly/Nicholson

Fig 4. Grillage model

Load combination 3includes the effects of temperature difference. The as 9 070 at transfer ( l Vo relaxation, 8 070 elastic deformation of concrete),
residual stresses will not affect the ultimate strengthof the beams, so this increasing to 28 Vo in the long term(additional1 070 relaxation, 5 070
load combination was considered only at the serviceability limit state. shrinkage, 13070 creep). Calculation of stresses for the two criticalconditions
resulted in the requirement for the stress due to prestress (after allowing
Y-beam design in bending for all losses) to lie within the range 16.8N/mm2 to 20.3N/mm2. A
As indicated by Pt4:6.1.2.1, the design is controlled by the serviceability pattern of 29 prestressing strands, 15.2mm diameter, was selectedas shown
limit state. The various cracking criteria and stress limits which apply to in Fig 3. After losses, the prestressing force is 3 -64MN at aneccentricity
this beam are as follows. of 0.186m, giving a stress at the soffit of 18.6N/mm2. This level of
prestress ensures that the beam complies with all the requirements of the
Cracking:
serviceability limit state, as can be seen from Table 2.
(a) Pt4:4.2.2 states thatthe beam is categorised as class 1 for load
The ultimate capacity of the composite section was determined by trial
combination 1, with HB load reduced to 25 units. For class 1, Pt4:4.1 .l .l@)
and error. The neutral axis (2nd guess) was 0.35m from the top of the
states that no tensile stress is permitted.
slab with maximum tensile strain in tendons of 0.011 occurring when
(b) Similarly, the beam is categorised as class 2 for load combinations 2
concrete strain reached 0.0035 compression. The tensile force in the tendons
to 5 . In this case, tensile stress is permitted and Pt4: Table 24 gives the
was then 4-3 MN, providing a moment of 3 6MNm. This was downgraded
maximum tensile stress as 3 - 2N/mm2.
by factor of 1 15, because tendons were not at yield, to 3-l MNm. This
(c) Pt4:6.3.2.4(b) limits the tensile stress at transfer, solely because of the
compares with a maximum ULS moment of 2.8MNm fora simply
prestress and dead loads, to 1N/mm2.
supportedspan(madeup of DL = 0-75MNm, SDL = O-2MNm,
Stress limits: LL = 1-6MNm, yf3= 1 1). At the ultimate limit state, there is an
(d) Pt4: Table 22 gives the allowable compressive stress, for members in additional reserve of strength, not used here, due to the beneficial effects
bending, as 0.4f,,. Thus the stress in the top of the beam must be limited of continuity.
to 0-4 x 50 = 20N/mmz.
(e)
Similarly,
the
stress
the
inslab
must
be
limited
to
0.4 x 40 = 16N/mm2.
(f) Attransfer, Pt4:Table 23gives allowable compressive stress of 1.0 x Footpath
O-Sf,,= 0.5 x 40 = 20N/mmZ. This is the same as 0.4fcu.
It was found that (b) and (f) were the two critical conditions governing
the prestress design; (b) gave rise to a lower limit on the stress in the beam
soffit due tothe prestress, while (f) gave an upper limit on this stress. The
maximum bending moment for combination 3 was found by the grillage
analysis to be 1 -31MNm (including SDL and partial factors). The pattern
of loading producing this moment is shown in Fig 5 . To this must be added
\\
1.0 X 45 HB
-

I
9

- --I --= \
\
the effects of a midspanbending moment due todead loadsof 0.65 MNm
and the residual stresses due to temperature difference and shrinkage which \ 1.0 x Footpath

are illustrated in Fig 6. Prestress losses, based on Pt4:6.7.2, were estimated +1.2 X SDL
Fig 5. Load pattern f o r maximum midspan moment at SLS
TABLE I - Load factors forthe load combinations consideredfor the
design of the beams (shrinkage was ignored when its effect would have
-02 O

-"l
been beneficial)

SLS ULS 1.0


Load
Comb.1 Comb. l Comb.1
Comb.3 Comb.1
Prestress 1- 0 1.0 1.0 1.0
Dead 1-0 1 s o 1.0 1.0 1-15
SDL 1.2 1-75 1.2 1.2 1.2
HA alone 1- 2
HA + 25HB 1.1
'1.7 - 0.4
HA + 45HB 1-3 1.0 1.1(a) Positive (b) Negative (c) Differential
Footpath 1- 0 1.0 1* o 1.0 1.5 temperature temperature shrinkage
Temp. dif. 0.8 difference difference of slab
Shrinkage (1 -0) (1 *O) (1 -0) (1 -0) Fig 6. Residual stresses

476 The Structural Engineer/Volume 68/No.23/4 December 1990


Paper: Hambly/Nicholson

TABLE 2 - Serviceability limit states stresschecksfor midspan of beams Y-beam design for shear
(all stresses are in units of N/mm2) Pt4:6.3.4.1 states that calculations for shear should be carried out at the
ultimate limit state. Combination 1 will clearly be critical, as the temperature
SLS stresses and cracking difference loading in combination 3 does not give riseto vertical shear forces.
Pt4:6.3.4.1 also requires that the ultimate shear resistance be calculated

fi Grillage Temp. Shrnk. Total for both a cracked and an uncracked section. Both maximum shear and
coexistent moment, and maximum moment and coexistent shear, must be
Comb.1 considered in the case of the cracked section (moment does not come into
(reduced) M=0.65 M = 1.04 Class l the calculation for the uncracked section).
Beam The highest shear forces occur at the ends of the beams and have been
bottom 18.6 -8.5 -7.7 - 0.4 +2.0 > 0 calculated for asection 0.5 m fromcentre of bearing. The maximum shear
at ULS, due to all combination 1 loads, was found to be 0.61 MN (made
Comb. 3 M=0.65 M=1.31 Class 2
up of DL = 0.14MN, SDL = O.04MN, LL = 0-37MN, yo = 1 1) with
Beam coexisting M = 0.31 MNm. The shear strength calculations, following
bottom 18.6 -8.5 -9.8 - 1.3 - 0.4 -1.4 >- 3.2 Pt4:6.3.4, was based on the Y3 beam alone, except that composite section
Beam top -4.4 13.5 6.4 0.2 1.0 16.7 < 20 properties were used for the overall depth and section modulus for cracking
~~

Comb.1 M = 0.65 M = 1-43 moment. Near the end the critical strength was for section uncracked in
Beam top -4.4 13.5 +7.0 1*o 17.1 < 20 flexure, for which V,, = 0.43MN. Shearreinforcement of about
Slab top + 9.3 ( - 0.2) 9.3 < 20 650mm2/m was therefore required according to Pt4:6.3.4.4. T12 links at
a spacing of 250mm have been detailed, providing 905mm2/m.
Comb. 3 M = 1.31 The Code formula for the shear capacity of an uncracked section, V,,,
Slab top +8.5 +2.1 (- 0.2) 10.6 < 20 was derived by setting the maximum principal tensile stress to the tensile
strength of the concrete. Theformula is theoretically correct for a
SLS transfer rectangular beam, where the maximum shear stress, and so maximum
principal tensile stress, occurs at the centroid. The factor of 0.67 in the
Code formula relates to the average shear stress of an elastic distribution.
Total The Y-beam is not rectangular, and the maximum shear stress will always
~~

Beam
bottom 123.5
Beam top - 5 6 -
1 1 17.9
3.4
< 20
> -1
occur near the neck at the bottomof the web, although for thebeams used
in this design this is close to the centroid. An elastic analysis of this Y-
beam section indicated a shear stress in the neck 20% less than for a
rectangular beam of the same width as theneck. Thus this Y3 beam should
have 20 070 greater uncracked strength V,, than was calculated. This reserve
was ignored.

Similar calculations at SLS and ULS were carried out for the stresses in Interface shear
the deck 5 m and 3m fromthe ends of the beams. Different grillage load Calculations for the longitudinalshear at the interfacebetween the top of
combinations were required to find the largest bending moments at these the beam and the slab were carried out using Pt4:7.4.2.3 for the ultimate
positions. The check at 5 m indicated that the same prestress could be used limit state. It was assumed that the topsurface of the precast beam would
as at midspan. The check at 3m indicated that six strands needed to be be 'type 2' (rough as cast). The longitudinal shear was derived from the
debonded to satisfy the stress limits at transfer. vertical shear using the elastic distribution expression: V, = VAy/I.At the
Calculations were also carried out for a position 0.5m from the ends ends of the beams, with V = 0.61 MN, the amount of steel required was
of the beams (at end of transmission length). Six morestrands were 1590mm2/m, which was over double that in the links needed for vertical
debonded to keep the minimum stress above zero for long-term conditions. shear only. However, the reinforcement for interface shear is needed only
Edge beams are subject to different loading to the beams under the in the immediate vicinity of the interface. The links have therefore been
carriageway. Although the live load is smaller, the dead load is substantially detailed with an extra loop at the top, to double the areaof steel crossing
greater because of the extra weight of the parapet beam. It was assumed the interface, as shown in Fig 3.
that the parapet beam would be cast after the slab, and would therefore Calculations for thecentral portion of the span showed that only nominal
be supported by the composite section rather than the beam acting alone. links were required for vertical shear, but over twice as much steel was
It was found that thesame prestress design could be used as for theinternal required for interface shear. The spacing of the T12 links increased from
beams. -
250mm to 500mm at a distance of 4 5 m from the ends of the beams.

(a) Plan view showing staggered treatment of skew (b) Section I

Fig 7. Diaphragm details

The Structural Engineer/Volume 68/No.23 / 4 December 1990 477


Paper: Hambly/Nicholson

Fig 8. Construction joint atdiaphragm


(a) Bridge in Toronto built in 1967 (bottom flange reinforcement is two
bars 5/8in diameter) (Notealso the presence of the pigeon, which is very
common in Toronto)
(b) Bridge in Tennessee built in 1981 (bottom flangereinforcement is six
bars %in diameter)

Continuity at supports will occur over the piers because of live load and must therefore be
Diaphragms considered inthe design. Any beneficial continuity thatmay existat adjacent
At each pier the connecting beamsare placed onto a sharedbearing with supports is ignored, as the construction joints
may not be fully closedthere.
a gap of 200mm between end faces of beams, as shown in Fig 7. Beam The live load hogging moment is therefore calculated as for the central
ends are rightto beams, even on skew decks since compression forces may support of a two-span continuous .deck, and the slab reinforcement is
be transferredacrossthem.Diaphragmsareconstructed by placing designed assuming that the slab forms the tension flange of the composite
reinforcement and concrete in the gaps between beams; however, the beam over the piers. The design of this reinforcement was found to be
diaphragms are notdesigned to provide any primary structural function. governed by the crack width criterion at the serviceability limit state.
The diaphragmshold the beams in placeover the piers and provide surfaces The single span grillage used for the design of the beams was extended
for jacking during maintenance. to two spans. Cracked section properties were used for the beams for a
Six T20 reinforcing bars(epoxy coated) protrude from the bottom flange length of 2m to 3m each side of the central pier. It is expected that the
of each beam into the diaphragms over the piers (the bars are straightwhen slab will crack right through when acting as the tension flange, and there
cast and bent to L later). Thisis referred to in this paper as bottom flange is a construction jointat theends of the precast beams,so any tensile force
reinforcement. Some designers call it positive moment reinforcement; must be carried by the steel in the slab.The section propertieswere based
however, it is not intended to develop full continuity against sagging on T12 longitudinal bars at l00mm spacingin both the top and bottom
moments. Over a period of time the beams may hog upwards slightly of the slab. The neutralaxis wasfound tobe about 150mm above the soffit
because of the effects of creep and shrinkage of the concrete. This deflection of the precast beam for this condition.
is partially resisted by the bottom flange reinforcement, but even so the Pt4:4.2.2 states that only load combinationneed 1 be consideredat SLS
construction joints between beam end faces and diaphragm concrete are for thecrack width check, with HB loadinglimited to 25 units. It was found
likely to open slightly. NCHRP 322 states that no serviceability problems that the HA loading alone was the mostonerous loadcase. The knife edge
have been reported relating to these construction joints, even in bridges loads were placed close to the centreof one of the spans. At the ultimate
that do nothave any bottomflange reinforcement.The construction joints limit state combination 1with 45 units HBwas the most onerous case. The
are sheltered from deicing salts. In the demonstration design, the diaphragms maximum hogging moments obtained from the grillage analysis (including
have been recessed about lOmm at the edge beams in order to mask any superimposed dead load and partial load factors) were:
opening of the joints. Fig 8 illustrates the exposed faces of diaphragms
on bridges in Ontario and Tennessee. SLS moment = 0-28MNm
The diaphragms will flex and twist slightly whenindividual beams deflect
ULS moment = 0-51MNm
under live load. At the ends of the beams the surfaces of the bottomflange
adjacent to the diaphragm are coated with a slip layerto prevent spalling The strain at the top of the slab due to the live load hogging moment
due to relative movement between beam and diaphragm. of 0.28MNmis 0.00078. Crack widths were calculated to Pt45.8.8.2 (b),
During construction attention shouldbe paid to thepossibility of a rapid for flanges in overall tension:
drop in temperature while the concrete in the diaphragm is setting. The
beam ends should notbe moving apart when the concreteis setting. This Crack width = 3acrem
also appliesto the connectionof the end spans to the abutment. I f a large
temperature change is expected over a long length of the deck, special Pt4:eqn. (25) allowsforthetensionstiffening of theconcrete.
precautions may be necessary, such as control of the time at which the Unfortunately, in this case the moment is due entirely to live load, and
concrete is placed or control of deck temperature by spraying with water. not at all due to permanent loads,so the, equationdoes not provide any
benefit from the stiffening effect of the concrete. The strain quoted above
Hogging moment reinforcement must be used directly in the crack width formula.
The bridgedeck may on occasions act ascontinuous over the piers for live The slab topwas designed with a cover of 35mm toT12 transverse bars.
load if the jointsbetween the beams arenot yet opened by creep orif they Hence cover to longitudinal barsis 47 mm to their surface and53 mm to
are closed by differential settlement. In these situations,
hogging moments centreline. With bars at l00mm spacing

478 The Structural Engineer/Volume 68/No.23/4December 1990


Paper: Hambly/Nicholson

a,, = (5d+532)0'5-6= 67mm Integral abutments


Crack width = 3 X 67mm X 0.00078 = 0.16mm Abutment design
The integral abutments, shown in Fig 9, are connected to the deck in order
This crack width is less than thecrack width limit of 0.25 mm in Pt 4: Table to avoid any movement joints from one end of the bridge to the other.
1. The integral abutments are small, in order to limit the weight which must
The ultimatemoment capacity was calculated as for a reinforced beam, move with the deck and to avoid excessive passivereactions during thermal
with effective depth measured to centroid of the slab reinforcement = expansion of the deck. However, the fill behind still has sufficient passive
0-965. For reinforcement of 20 T12 bars (10 each in the top and bottom resistance to react with longitudinal braking and traction forces.
of the slab), the ultimate moment capacity was calculated to be 0-82MNm, Each abutment has a run-on slab which is designed to span over the fill
which exceeds the maximum ULS moment of 0.51 MNm. immediately behind the abutment in order to prevent water inflow and
Hogging moments reduce to zero at 2m to 3meach side of the piers, traffic compaction of material where the fill is partially disturbed by
so it is suggestedthat any slab reinforcement provided over the piers, which abutment movement. Relative movement between the structure and the
is additional to the longitudinal slab reinforcement elsewhere could be highway pavement must be absorbed by localdeformation of the pavement
curtailed with a staggered layout 3m from the piers. near the end of therun-on slabs, whichwill require regular but
straightforward maintenance. If the pavement is of concrete construction
Restraint moments due to creep and shrinkage a compression joint must be placed between the run-on slab and the
Calculations in Appendix A illustrate how the bottom flange reinforcement pavement. Loveal17and Wasserman' indicate that the State of Tennessee
6T20 (1900mm2)should be sufficient to resist the restraint moments which has used integral concrete bridges of over 250m overall length, with
could develop at supports due tocreep and shrinkage of the beams. Creep abutment movements of up to 50mm. The run-on slab must be attached
and shrinkage effects are secondaryin comparison to weight and prestress to the abutmentwith reinforcement (epoxy coated) in order to pull it back
and cannot be predicted with the same accuracy at the design stage. The when the deck shrinks in cold weather. The joint may be filled withbitumen
authors believe separate calculations for each bridge to be unwarranted. to prevent water ingress. The reinforcement tie should be robust enough
The demonstration design has bottom flange reinforcement of 6T20. It to resist friction forces (and braking), The run-on slab should also be
is suggested that designers use standard bottom flange reinforcement, designed as abridge span for its full length without support from the ballast
varying only with beam size. Ontario have standard details for different below.
beam sizes - e.g. 0.9m-deep I-beams have U-bars providing 1400mm2, The baseslab has been placedas high as possible while keepingthe bottom
1 -2m beams have 1900mm2 and l .4m beams have 2500mm2. (bearing) face 1 m from the ground surface (for frost protection). It has
been assumed that the footing will be constructed on stiff ground, or
Piers and bearings properly compacted selected granular fill, which can provide an allowable
Pier design bearing pressure of about 0.2MN/m2 at topof embankment. Under HA
The deck and piers are articulated so that they form a combined structure loading with 45 units of HB on deck and run-on slab, the total live load
in accommodating longitudinal forces and movements. Elastomeric bearing plus dead load reaction is calculated to be about 5MN,and footing
pads are designed to support vertical load coexisting with rotation of the dimensions of 14m length and 2.5m width are adequate.
deck, while relative lateral movement is restricted by dowel bars. During
thermal expansion/contraction of the deck, the pier is expected to flex and Longitudinal forces and movements
rock on its foundation. (On high skew decks the columns and footings may The maximum longitudinal forces are likely to occur during thermal
need to be more flexible than here along the plane of the pier.) If the piers expansion and contraction of the deck. The range of effective bridge
stand on piles, the pile groups should not be stiff in resisting moments. temperatures is from about - 12C to +36"C, i.e. a range of 48C.
The footing is designed to be as small as practicable (while being wide The overall movement on half of the bridge length of 35m is
enough for stability during construction.) It has been assumed that the 48 x O.ooOo12 x 35 = 0-020m, i.e. -+0-010m relative to the mean
footing will be constructed on stiff ground which has an allowable bearing position.
pressure q, of about 0.3MN/m2 (at formation level). Under HA with 45 The abutment is attached to the deck and is too short to flex significantly.
units of HB loading on the deck of the total live load plus dead location Consequently, it is assumed to slide on the bed of rounded gravel. This
is calculated to be about 8MN. Footing dimensions of 14m length and gravel is here assumed to have a peak angle of friction 9 of 35" withpartial
3m width lead to an average net bearing pressure of about 0.2MN/m2. factors yrnof 0.67 and 1.5 for upper and lower bound estimates. The
The abutmentsof this bridge provide adequate resistance to longitudinal friction resistance F is given by F = Vtan(9) /vm where V is the vertical
loads, so the piers needto carry only the vertical load. However, longitudinal reaction on the abutment.
forces and thermal effects cause deflections of up to 7mm which must be Vertical reaction V has a value of 3 * 1MN (with partialfactors
accommodated in the design of the piers and their foundations. Yn = 1-0) and 3 7 MN with partial factors of 1 15 for dead load and
1-75 for superimposed load. Hence upper and lower bound estimates of
Bearings sliding resistance are.
The deck issupported at the piers on elastomericpad bearings of dimensions
700mm x 500mm x 15mm. Each beam end rests on anarea of upper bound F = 3 - 7 tan (35")/0.67 = 3.9MN
250mm x 500mm. The area under the diaphragm is ignored in the bearing
lower b o u n d F = 3.1 tan (35")/1.5 = 1.4MN
design. The bearing under one beam end is subjected to design loading of
0.16MN due to dead loads and 0.31 MNdue tolive loads (45HB). It also
When sliding is towards the embankment the sliding resistance will be
experiences rotations across the 200mm length of about 0.004 due tolive
accompanied by passive resistance from the backfill, which is 2m high.
load and 0.002 due to dead load, and rotation across the 500mm width
The horizontal movement of lOmm represents only 0.5 070 of the height
of about 0.001 due to live load. By following Pt9.1:10, it is found that
and, since the soil strain willbe of the same order, the earth pressure
these loads and rotations can be carried by an elastomeric pad 15mm-thick.
mobilised is likely to be only about half full passive pressure (see Lambe
The bearing pads are shown in Fig 3 as700 x 500 x 15 with two dowels
& Whitman?. Hence the loose granular fill has K of about 2, density
passing through. The bearings will need to be placed in two halves if they
0.016MN/m3, and provides a resistance on 12m width of
are later to be replaced without interference from the dowels.
22
Bearing shelf and dowels P = 2 X 0.016 X -X 12 = 0.77MN
The bearings rest on the bearing shelf which has transverse slots between 2
plinths to facilitate jacking (if later necessary). Reinforcement is located upper bound P = 0.77/0.67 = 1 -2MN
below the plinths to contain bearing forces, and additional links are placed
at the side plinths to enable the pier width to be within the width of the lower bound P = 0.77/1.5 = 0.5MN
deck soffit (for aesthetic reasons). Two dowel bars pass up through each
bearing to hold the deck in position. (Only one dowel bar is used at the The lower bound combined friction forces on two abutments with passive
side beams to avoid local stresses near exposed faces.) The dowel bars can resistance on one is 3 - 3MN. This greatly exceeds the maximum braking
be fixed after construction of the piers by grouting them into drilled holes, force of 1-2MN, and hence braking requires no further attention.
but care has to be taken not to drill the holes through reinforcement. The The upper bound friction force and passive resistance could overload
upper ends are fitted with a plastic sleeve with a 12mmclearance at the top. the abutment wall in bending or shear, and the reinforcement has to be

ructural
neer/Volume
The 68 /No.23/4 December 1990 4 79
Paper: Hambly/Nicholson

A-A

B- B

4
I; A _I

Plan D - D

c-c E- Epoxy coated


reinforcement

Fig 9. Abutment details

designed for this purpose. The bottom flange reinforcement in the beams a single system in conjunction with the foundations and embankments.
is also checked for themoment from upper bound friction when the (3) Individual spans have been assumed for the purposes of design as simply
abutment is pulled away from the embankment (active pressure ignored). supported for live load as well as for dead load.
In this case, it is found thatthe abutment wall needs T25-200 bars working (4)Continuity reinforcement over the piers has been designed as for a two-
with lever arm of 0 - 4 m . Hambly discusses the global analysis of an span continuous deck.
integral bridge with its foundations, and extends the analysis to estimate (5) Beam ends have been made square, even on a skew bridge, to enable
the distribution of foundation reactions and deck displacements. compressive forces to cross theconstructionjoints andto simplify
manufacture.
Beam seating and diaphragm wall (6) Bottom flange reinforcement is cast into the ends of the prestressed
The beams are seated at their ends during installation on a permanent beams to connect them together over the piers and to connect them into
neoprene pad of about 6mmthickness, 500mm width and 250mm length. the abutments. This reinforcement does not provide full continuity, but
The bottom flange reinforcement in the beams resists splitting of the beam restrains the opening of the construction joints between the beam ends and
above the seating and interacts with the reinforced concrete of the diaphragm the diaphragm. Standard details can be developed so that calculations are
beam and links from the abutment base. The links in the base and shear not needed for each bridge.
key are substantial in order to transfer longitudinal forces into the base (7) Creep and shrinkage may cause the beams to hog slightly, so partially
(see post). The bottom flanges of the beams are covered with a slip layer opening the bottom of the construction joints at the diaphragms.
where they touch the diaphragm beam in order toprevent spalling due to (8) Integral abutments have been designed to be as small as possible so that
relative movement (as at intermediate supports). they can move with the bridge deck, and so that there is sufficient, but
not excessive, resistance to longitudinal movement.
Wing walls (9) Integral abutments must be provided with run-on slabs to prevent water
The wing walls have been made as small as possible so that they can move ingress and traffic compaction of the backfill immediately behind the
with the diaphragm and edge beams. They have been made independent abutment.
of the abutment base so that relative movements can occur if necessary.
It should be posssible for the contractor to build the wing walls after he Acknowledgements
has finished the rest of the abutment andcompacted thebackfill; the wing This paper is based on a report prepared by Edmund Hambly Ltd. for
walls would then be constructed in trench in the compacted backfill. the Prestressed Concrete Association. EdmundHambly Ltd. has also
reported to the PCA on a fact-finding tour to the USA and Canada. The
Conclusions authors are very grateful for the assistance they received from engineers
This paper describes the design of a prestressed beam integral bridge. in the USA andCanada, in particularat Tennessee Department of
Integral bridges, which have no movement joints from end to end, are Transportation and Ontario Ministry of Transport.
becoming more widespread as designers endeavour to avoid the maintenance
problems that develop at joints between simply supported decks. Concrete References
viaducts of over 250m overall length have been constructed in this manner 1. NationalCooperative Highway Research Program,NCHRP 141:
in the USA. The salient features of the design described here are asfollows: Bridge deck joints by M. P. Burke Jr of Burgess & Niple Ltd.,
(1) The composite prestressed beam deck spans have been connected Columbus, Ohio, Washington DC, Transportation Research Board,
together over the piers with in situ concrete diaphragms and connected to 1989
the abutments, to form anuninterrupted structure withoutany movement 2. NationalCooperative Highway Research Program,NCHRP 322:
joints. Design of precast prestressed bridgegirders made continuous by R. G.
(2) The articulation of spans, piers and abutments has been considered as Oesterle, J . D. Gilkin and S. C. Larson of Construction Technology

480 The Structural Engineer/Volume 68/No.23 / 4 December 1990


Paper: Harnbly/Nicholson

Laboratories, Skokie, Ill., Washington DC, Transportation Research made continuous grow to (1 - e-+j times the moment that would have
Board, 1989 developed if the deck had been constructed monolithically at thestart, where
3. Burke, M. P. Jr: The integrated construction and conversion of simple 4 is the creep factor.
and multiple span bridges, in Bridge Management ed. Harding, J. E., Prestress force P in a beam at eccentricity e causes a restraint moment
Parke, G . H.R.,and Ryall, M.J., University of Surrey First in a monolithic continuous structure of Pe. In this deck the long-term
International Conference on Bridge Management, Elsevier Applied prestress force at midspan is P = 3.5MN at e = 0.38m relative to the
Science, 1990 composite section. With 40% of strands debonded near the ends of the
4. BS 5400 Steel, concrete and composite bridge: Part 4, London, British beams, the average value of Pe over the length of the beams is about 0.83
Standards Institution of value at midspan. Hence average
5. Hambly, E. C.: Bridge deck behaviour, 2nd ed. London, E. & F. N.
Spon, 1991 Pe = 0.83 x 3.5 x 0-38 = 1.1OMNm
6. Departmental Standard BD37/88: Loadsfor highway bridges, Hence the restraint moment due to creep of the composite structure from
Department of Transport, London, 1988 100 days to long term, with 4 = 1-3, is
7. Loveall, C. L.: Jointless bridge decks, Civil Engineering, New York,
American Society of Civil Engineers, November 1985 M,, = ( l - e - & ) P e = (l-e-13) x 1.10 = 0.80MNm
8. Wasserman, E. P.: Jointless bridge decks, Engineering Journal, Beam self-weight would induce a restraint moment in a continuous beam
Chicago, American Institute of Steel Construction, 1987 of - wL2/12 = -0.30MNm. Hence the restraint moment due to creep of
9. Lambe, T. W., and Whitman, R. V.: Soil mechanics, New York, John the composite structure from 100 days to long term is
Wiley , 1969
10. Hambly, E. C., and Nicholson, B. A.: Prestressed beam integral M,, = (l - e - 1 3 )(-0.30) = -0.22MNm
bridges, Leicester, Prestressed Concrete Association, 1990 Dead load of the slab would also induce a restraint moment of - wL2/12.
11. Mattock, A. H.: Precast-prestressed concrete bridges 5. Creep and With 4 = 1 - 0 for loading of the beams at 100 days, restraint moment of
shrinkage studies, Skokie, Ill, Portland Cement Association, May 1961 the composite structure is
12. Clark, L. A.: Concrete bridgedesign to BS 5400, Construction Press,
1983, with supplement 1985 = (1 - e - ? (- 0 - 16) = - 0 . lOMNm
Mslab
13. Ontario Highway Bridge DesignCode 1983, and commentary, Ontario, Superimposed dead load also induces a restraint moment of - wL2/12,
Ministry of Transportation & Communications but since it is placed after the deck is made continuous, no redistribution
occurs due tocreep.
Appendix A. Creep and shrinkage effects
Creep and shrinkage factors M,, = -0.0025 x 202/12 = -0.08MNm
It is assumed below that the deck is made continuous when the beams are Shrinkage of the beams after the slab is cast is about 90 X while the
100 days old. slabshrinks 120 x so that thedifferential shrinkage of the slab
Creep factor is given in Pt4: Appendix C as relative to the beams is30 x (A low value is disadvantageous here,
4 = kL kmkc k, kj whereas a high value is disadvantageous in the beam design.) The slab has
area 0.20m2 at eccentricity of -0-43m. Hence the restraint moment in
The beams were assumed to have: kL = 2- 3 (curing in normalair); a monolithic deck would be
km = 1.6 (transfer after 3 days at 20C or 1 day at 70C); kc = 0.75
(cement content 400kg/m2, water:cement ratio 0.37); k, = 0-8(effective 30 x x 0.20 x 34 OOO x -0.43 = -0.09MNm
thickness h, = 250mm); kj = 0.4 at 100 days (h, = 250mm) and 1 -0long Mattock and Clark both indicate that, in a composite structure,the
term. Whence for creep due to prestress from transfer: shrinkage moment creeps to a factor (1 - e - + ) / @ of the moment needed
4 = 2.3 x 1 - 6 x 0.75 x 0.8 x 0 - 4 = 0-9 at 100 days to restore unrestrained shrinkage. Hence
4 = 2.3 x 1.6 x 0.75 x 0.8 x 1.0 = 2.2 long term = ((l -e-13)/1-3) x (-0.09)
Mshrink = -0.05MNm
and due to prestress from 100 days to long term: Total restraint moment is the sum of the above
4 = 2.2-0.9 = 1.3 MR = 0-80-0.22-0.10-0.08-0.05 = 0.35MNm
Creep due to the weight of the slab from 100 days, for which km = 0.7, The bottom flange reinforcement of 6T20 bars has area A = 0.0019m2
has and lever arm of 0.8m relative to centre of slab. Hence the stress induced
4 = 2.3 X 0.7 X 0.75 X 0.8 X 1.0 = 1.0 by the moment of 0.35 MNm is
Creep of the slab was assumed to have: kL = 2.3 (normal air); km = 1-0; f = 0.35/(0.8 x 0.0019) = 230MN/m2
kc = 0.8 (cement content 350kg/m2, water:cement ratio 0.42); k, = 0.7
The Ontario Bridge Code13recommends a stress limit of 240MN/m2 for
(for h, = 400mm assuming no evaporation through waterproofing
bottom flange reinforcement.
above); kj = 1.0 (long term). Hence
A check of crack width according to Pt45.8.8.2 does not provide an
@ = 2.3 x 1.0 x 0 - 8 x 0.7 x 1 = 1.3 long term indication of the opening of theconstructionjoints because tension
stiffening of the reinforcement is substantial and eqn.(25) produces a
Shrinkage strains are given in Pt4:C.3 as negative value of strain e,. An approximate estimate of the opening of
shrinkage = kLkck,kj thejointcan be obtained by assuming that the stress of 230MN/m2
stretches the reinforcement over a debonded length of about 200mm, giving
The beams were assumed to have: kL = 275 x (normal
air);
an extension of (200 x 230/20 OOO) = 0.23 mm. Any movement of this
kc = 0.75 (as for creep); k, = 0.75 (for h, = 250mm); kj = 0.4 at 100
type that does occur will relax the restraint moment. However, further
days and 1.0 long term. Whence shrinkage of beams is refinement of calculation is not considered meaningful because of the
shrinkage = 275 x x 0.75 x 0-75 x 0.4 = 60 x at 100 days unknowns at the design stage.
= 275 x x 0.75 x 0.75 x 1.0 = 150 x long term
The shrinkage from 100 days to long term is
shrinkage = (150-60) x = 90 x

Shrinkage of the slab was assumed to have: kL = 275 x (normal air);


kc = 0.8 (as for creep); k, = 0.55 (for h, = 400mm); kj = 1.0 (long
term). Hence
shrinkage = 275 x x 0.8 x 0.55 x 1 = 120 x

Long term restraint moments


Mattock and Clark explain how restraint moments in a composite deck

The Structural Engineer/Volume 68 /No.23 /4 December 1990 48 1