1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 1/6

DESIGN OF JOINTS IN SEGMENTAL HOLLOW BOX GIRDER

BRIDGES
G. Rombach
A. Specker
Department of Concrete Structures
Technical University Hamburg-Harburg, D 20171 Hamburg, Germany

Keywords: segmental bridges, joints, external prestressing

INTRODUCTION

Segmental box girder bridges externally post-tensioned are one of the major new developments in
bridge engineering in the last years. In contrast to classical monolithic constructions a segmental
bridge consists of small precast elements stressed together by external tendons (Fig. 1). The many
advantages of this type of structure like fast and versatile construction, no disruption at ground level,
high controlled quality and cost savings have made them the preferred solution for many long elevated
highways, especially in South East Asia (see [1], [2]), and bridges.
Although many segmental bridges had been built in the last years the design of the unreinforced
joints between the segments, which is of critical importance regarding the safety of the structure, is
still under discussion (Fig.9). There is a big discrepancies between the various design models. The
known models are either too conservative and thus too uneconomic (German Specification [3]) or not
valid for high compressive stresses (AASHTO [4]). Therefore numerical calculations had been
conducted and verified by full-scale tests. The results, which will be presented in this paper, lead to a
better understanding of the behaviour of segmental constructions and a more realistic design of the
joints.

NUMERICAL SIMULATION OF A SINGLE SPAN SEGMENTAL BRIDGE

The load deformation characteristic of a segmental
construction is different from a monolithic one due to the dry
unreinforced joints between the precast elements.
Examinations of the behaviour of a segmental bridge and the
forces in the joints finite element calculations had been
conducted taking into account the non-linear behaviour due to
the opening of the dry joints under tension. In contrast to
known numerical investigations, the fine indentation of the
joints had been modelled which is of great importance
regarding torsion effects (Fig. 2).

Fig. 1

Segmental bridge under

construction

Fig. 2 Finite element mesh of a segment

1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 2/6

(12 segments 340 cm + 2 segments 172,5 cm )

[cm]

10,2 m

2,4 m

Cross-section

A real existing single span

segmental bridge with external posttensioning, a standard span of the
elevated highway Second Stage
Expressway System in Bangkok [2]
been modelled (Fig1, Fig 3). This
structure is used as data from a fullscale test [5] is available to verify the
results of the complex numerical
simulations. The opening of the dry
joints is modelled by interface
elements.

Fig. 3 Standard Span of SES, Bangkok

Fig. 4 shows the calculated moment-deflection curve which is typical for a single span segmental
bridge with dry joints. At the beginning of loading the whole structure is under compression due to the
high post-tension normal forces. Thus the structure behave like a monolithic one. The deflection
increases linear with the load. At a midspan moment of M 37 MNm due to live load the first joint near
midspan starts to open rapidly resulting in a great decrease of stiffness. The lever arm of the inner
forces keeps nearly constant. Thus the moment deflection curve is again nearly linear. The structure
fails due to crushing of the concrete in the top slab. Nevertheless a ductile behaviour of the segmental
bridge can be seen.
Only 3 of 13 joints are open under failure load. Thus a great part of the bridge keeps under full
compression.

deflection in midspan due to live loads [m]

0,40

q
g
h

0,30
point

0,20

Finite Element
experiment
opening of the joint
up to 1/
3h

0,10

20

38 40

2/3 h

46,9

Stresses in mid-span before failure

60

bending moment in midspan due to live load [MNm]

Fig. 4 Comparison between full-scale test and numerical results

Further shown in Figure 4 are the results from a full-scale test carried out in Bangkok. A good
agreement between the numerical results and the test data can be seen. This demonstrates that the
finite element model is capable to model the real behaviour of a segmental bridge.
Several load combinations corresponding to bending, shear and torsion are examined to determine
the stresses resp. the forces in the joint [6]. In a single span bridge the joints near the support are
always closed due to the small bending moment. As the behaviour of an open joint is main of interest
also a single span bridge restraint on one side with a modified tendon profile has been modelled as
insetted in Fig. 5.
Figure 5 shows the resulting shear forces in the first joint close to the support in the webs and the
slabs due to torsion with increasing load. The results from three different numerical models are
presented. The first one is a monolithic girder which behaves always linear. Further the shear forces
for a segmental bridge with smooth and keyed joints are shown.
There are no differences between the models as long as all joints are closed. When the joint starts
to open, the force in the top slab (tensile region) decreases. A great difference in the behaviour of a
bridge with plain and keyed joints can be noticed. Smooth joints can only transfer forces when they

1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 3/6

are under compression whereas keyed joints can still transfer forces until a certain gap is reached.
Even bigger differences can be seen in the webs. The plain joint reach the limit condition lim Fz =
0,7n just after the joint opens whereas the force in the keyed joint still increases.

q DA

Horizontal force Fy [MN]

3,0
g+q

joint no. 1

q
g
y

2,0

top slab
1,0

-1,0

Segmental bridge
with shear keys
monolithic girder
Segmental bridge with plain joint

-2,0
-3,0

bottom slab

0 2

10

15

1 increment = 0,1xlive load

right web

-1,0
-2,0
-3,0

lim F =
z 0,7
n

-4,0
-5,0

-1,0
Horizontal force Fz [MN]

Horizontal force Fz [MN]

loading increased

0 2

10

15

20

26

30

35

15

-3,0

30

20

35

39

joint opening
up to 2/3 h
26

30

35

39

left web

-2,0
lim F

-4,0
-5,0
-6,0
-7,0

0 2

10

26

joint opening
up to 1/3 h

1,0x(dead + live load) loading

increased
0

20

= 0,7

Segmental bridge
with shear keys
monolithic girder
Segmental bridge with plain joint

39

Fig. 5 Forces in the webs and the slabs due to torsion

The results emphazise that the shear keys have a significant influence on the behaviour of a
segmental bridge under torsion loads. Calculations with plain joints are insufficient when torsion
effects become significant.

Fig. 6 Joint opening due to positive resp. negative bending moments

DESIGN OF SEGMENTAL JOINTS

There is a great uncertainty regarding the design of the joints between the segments (see Fig. 9).
This is surprising as the behaviour of the joint is of critical importance for the safety of a segmental
structure.
The shear capacity of a keyed joint is a combination of the friction between the plain surfaces and
the shear capacity of the keys. The latter one is neglected in the German regulations.

1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 4/6

3.1 Existing design models

The joints of many segmental bridges had been designed according to the AASHTO
Recommendations [4]. Equation 1 mainly is based on tests with small specimens having usually one
shear key only [7] similar to that shown in Figure 7 [8].
V j = Akey 6, 792 103 f ck (12 + 2, 466 n ) + 0, 6 Asm n

(1)

[m, MN, MPa ]

where:

Asm
fck
Akey

average compressive stress across the joint

area of contact between smooth surfaces in
the failure plane
characteristic concrete compressive strength
min. area of the base of all keys in the failure plane

A k A sm

A sm

Ak

According to the German recommendations for design of segmental bridges [3] only the frictional
forces should be considered in the design. The load bearing of the shear keys is neglected as only
epoxy joints can be used. Please note the difference between eq. (1) and (2) regarding the frictional
area Asm resp. AT.
L L
3 L+H

V j = n AT

(2)

b(z)

where: AT effective shear area

zi

AT
L/2

The results of both models will be discussed together with the

proposed design concept in section 3.3.

3.2 Tests and numerical verification

80 20 50

100

150

To develop a design concept for the joints tests with specimens, similar to that described in [7]
having one or multiple shear keys (Fig. 7) were conducted to calibrate the finite element model. The
study includes dry and glued joints. The dimensions of the shear keys are representative for
segmental bridges. The non-linear material behaviour of the concrete like e.g. crushing and cracking
and the interaction between the indented surfaces (bond, slippage, friction) has been considered in
the numerical model.

115

35

150

[mm]

Fig. 7 Test specimen

The test specimens are first stressed normal to the joint and than loaded with a vertical force up to
failure. Fig. 8 shows the experimental and calculated load-deformation curve. The behaviour of the
joint and the ultimate load are well predicted. The highly complex concrete behaviour near the failure
load has not been modelled as this region is not relevant for the load bearing capacity of a joint.

1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 5/6

250

250
dry joint

200
shear force in kN

shear force in kN

glued joint

205,5 kN

198,5 kN

200

150

100

150

100

50

50

experiment
Finite Element calc.

experiment
Finite Element calc.
0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

vertical deflections at top of the specimen in mm

0,0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

vertical deflections at top of the specimen in mm

Fig. 8 Test results versus numerical results for a dry and epoxy joint
3.3 New design model
After the verification of the finite element model, a numerical parametric study had been conducted
with various number and shapes of shear keys, concrete qualities etc. [6]. The results lead to a design
model that differs from the existing concepts. The shear capacity of a keyed dry joint Vd,j is a
combination of a frictional and a shear part. For the first one the total area of the joint Ajoint is used and
not only the smooth parts (ASm) like in AASHTO recommendations. The load bearing capacity of the
keys depends on the concrete tensile resp. compressive strength and the area of the failure plane
Akey.
Akey = min hne.bn
left
right
min hne = h ne,1-3 < h ne,1-4

(3)

minimal
failure
surface
r

hne,2

hne,3

h ne,2

h ne,3

hne
bn

bn

Ajoint
fck
b
f = 0,14
Akey

coefficient of friction
safety coefficient
average compressive stress across the joint
area of the compression zone
characteristic concrete compressive strength
width of the web
factor for the indentation of the joint
min. area of the base of all keys in the failure
plane
height of keys, with hne 6bn
width of the keys

hne,4

= 0,65
F = 2,0
n

A joint = h.b

where:

b
bn

h ne,1

hne,1

for dry joints: V = 1 ( A + f f A )

d, j
n
jo int
ck
key
F

The failure plane Akey will have the least area of key breakage. A relatively high safety coefficient of

F = 2,0 should be used as the failure of the joint is brittle.

For glued joints only the frictional part can be used (eq. 4). Experiments showed a relatively small
increase in strength of appr. 20% between a glued and a dry joint. Furthermore a sufficient quality of
the glue can not be guaranteed on site.
for glued joints:

Vd , j =

n Ajo int

(4)

To compare the results of both models, the shear stress = Vd,j / Ajoint is calculated for a standard
segment of the segmental bridge in Bangkok [2]. The relevant joints are fully closed. The concrete
compressive strength is fck = 40 MPa.
Fig. 9 shows the load bearing capacity of a keyed joint according to various design models. The
great differences between AASHTO and the German regulations can be seen. The first model can not
be used for high compressive stresses, which may occur near the ultimate design load of a multispan
segmental bridge. Furthermore it seems to overestimate the load bearing capacity of a joint.

1st FIB Kongress, Osaka, Japan,2002

Session 2 Paper E-73 6/6

510cm

SH
TO
AA

Shear Stress [MPa]

15
10

an
rm
e
G

n
co
gn
i
s
de
w
Ne
on
ati
c
i
f
ci
sp e

Detail A

35

t
cep
50

200

75

240 cm

22.5

20

185cm

Shear Keys
fck = 40 MPa

10

15

20

25

30

35

Compressive Stress [MPa]

AASHTO [4]:
DBV [3]:

= 4,17 + 1, 06 n
= 0, 7 n

[MPa]

Fig. 9 Comparison between different design models (standard segment [2])

SUMMARY

A non-linear finite element model of a single span segmental bridge is presented. The results show
that the behaviour is dominated by the dry joints. The indentation of the joint is of great significance
when torsion effects have to be considered. Based on experimental and numerical studies a new
concept for the design of dry and glued joints is proposed.
REFERENCES
[1]

Brockmann, Ch., Shafer, G.: Design and Construction of the Bang Na-Bang Pli-Bang Pakong
Expressway. in: Stoelhorst, D. et al: Challenges for Concrete in the Next Millenium, Vol. 1,
pp. 275-280, Rotterdam 1998

[2]

Rombach, G.: Bangkok Expressway - Segmentbrckenbau contra Verkehrschaos, aus: Aus

dem Massivbau und seinem Umfeld (Hilsdorf, Kobler ed.), Schriftenreihe des Institutes fr
Massivbau und Baustofftechnologie, University of Karlsruhe 1995, pp. 645-656

[3]

Deutscher Beton-Verein: Empfehlungen fr Segmentfertigteilbrcken mit externen

Spanngliedern, 1999

[4]

AASHTO 89 (American Association of State Highway and Transportation Officials): Guide

Specifications for Design and Construction of Segmental Concrete Bridges, 1989, Interim
Specifications 1990 1999

[5]

Takebayashi, T., Deeprasertwong, K., Leung, Y.: A Full-Scale Destructive Test of a Precast
Segmental Box Girder Bridge with Dry Joints and External Tendons, Proceedings of the
Institution of Civil Engineers, August 1994, pp. 297-315

[6]

Specker, A.: Der Einfluss der Fugen auf die Querkraft- und Torsionstragfhigkeit extern
vorgespannter Segmentbrcken. Thesis, Technical University of Hamburg-Harburg, 2001

[7]

Buyukozturk, O., Bakhoum, M., Beattie, S.: Shear Behaviour of Joints in Precast Concrete
Segmental Bridges, Journal of Structural Engineering, No. 12, December 1990, pp. 33803401

[8]

Roberts, C.L., Breen, J.E., Kreger, M.E.: Measurements Based Revisions for Segmental
Bridge Design and Construction Criteria. Research Report 1234-3F, The University of Texas
at Austin, Austin 1993