You are on page 1of 13

Bull Earthquake Eng (2009) 7:363–375

DOI 10.1007/s10518-008-9071-8

ORIGINAL RESEARCH PAPER

Effect of torsional stiffness of prestressed concrete box


girders and uplift of abutment bearings on seismic
performance of bridges

Christos P. Katsaras · Telemachos B. Panagiotakos ·


Basil Kolias

Received: 28 December 2007 / Accepted: 8 May 2008 / Published online: 18 July 2008
© Springer Science+Business Media B.V. 2008

Abstract Uplift of certain abutment bearings during earthquake may be utilized as a safety
mechanism for the protection of the bridge against excessive stress. For this reason design
codes such as Eurocode 8 permit the uplift of such bearings under certain conditions. Uplift of
an abutment bearing occurs when the torque at the deck end exceeds a critical value. Therefore
the torsional stiffness of the deck, before or after concrete cracking, is an important factor. In
this work realistic values for the cracked torsional stiffness of the bridge deck are estimated
from a parametric study of typical thin-walled prestressed box girders, based on non-linear
analysis of softened space truss models proposed in the literature. Moreover the interaction
between bearing uplift and pier flexural response is investigated and recommendations for
the seismic design of bridges are proposed, taking into account the possibility of uplift of
abutment bearings before or after yielding of the piers.

Keywords Bearings · Box-girder · Bridge design · Prestressed concrete · Seismic design ·


Structural analysis · Torsional stiffness · Uplift

1 Introduction

The uplift of abutment bearings during earthquake induces a modification in the seismic
response of the bridge that may lead to damage of the bearings and of the structure in their
vicinity. On the other hand bearing uplift acts as a safety mechanism that protects the bridge
deck from excessive torque. For these reasons certain codes, such as Eurocode 8—Part 2
(EN1998-2), permit uplift of bearings under certain conditions for ductile bridges when the
design earthquake is exceeded 1/q times, where q is the Eurocode 8 behaviour factor.
An important factor for the correct estimation of bearing uplift is the torque T at the ends
of the deck that is developed due to compatibility between bending deformation of piers,
corresponding rigidity of the abutments and torsional deformation of deck. As a consequence

C. P. Katsaras · T. B. Panagiotakos (B) · B. Kolias


DENCO – Development & Engineering Consultants S.A., 16 Kifissias Ave, 151 25 Maroussi, Greece
e-mail: tpanagiotakos@denco.gr

123
364 Bull Earthquake Eng (2009) 7:363–375

bearing uplift depends strongly on the torsional stiffness of the deck. Test results of prestressed
concrete elements under combined torsion, bending and shear have been reported in the liter-
ature (McGee and Zia 1976; Batchelor and McEwen 1977). Based on the aforementioned test
results it is concluded that before concrete cracking the torsional stiffness remains constant
and it is not significantly affected by the reinforcement and the relative values of torque,
bending moment, and shear force. Moreover, the torsional stiffness before concrete cracking
can be estimated with adequate accuracy from the theory of elasticity.
After concrete cracking the torsional behaviour of the element is non-linear with signif-
icant reduction of the corresponding stiffness that depends strongly on the reinforcement.
According to EN1998-2, the recommended value for the torsional stiffness of prestressed
box-girder decks for seismic analysis is equal to 50% of the torsional stiffness of the uncracked
cross-section.
The softened space truss model has been utilized by many researchers for the estimation
of torsional behaviour of closed thin-walled reinforced concrete cross-sections after con-
crete cracking (Vecchio and Collins 1986; Hsu and Mo 1985a; Hsu 1991; Rahal and Collins
1996; Fu and Tang 2001). A significant aspect of the aforementioned theory is the reduction
of the concrete strut strength due to the elongation of transverse reinforcement bars. This
phenomenon is called concrete softening and it is particularly important for the estimation of
torsional behaviour (Vecchio and Collins 1981; Hsu and Mo 1985a). The softened space truss
model theory allows for the unified consideration of reinforced and prestressed elements by
introducing additional initial strain in the prestress steel which is equal to the prestress steel
strain at concrete decompression (Hsu and Mo 1985c).

2 Equilibrium during bearing uplift

The equilibrium of the deck end is examined in Fig. 1. The typical case of two bearings
carrying only vertical force at distance d/2 from the deck axis and one central bearing car-
rying the horizontal transverse reaction is assumed. Following forces come into play: deck
internal forces acting on the shear centre of deck cross-section (torque T , vertical shear force
VZ , transverse shear force Vϒ ), weight of the transverse beam W0 , vertical bearing reactions
(F1 , F2 ) and transverse support reaction (F3 at distance h below the shear centre of the deck
cross-section). For all time instances when one of the bearings has been uplifted, i.e. when
either F1 = 0 or F2 = 0, the following relation is true:
T + VY · h = (VZ + W0 ) · d/2 (1)
This relation is derived from equilibrium of forces and moments at the deck end. The right
part of the equation is affected mainly by the longitudinal and vertical component of the
earthquake. If only the transverse response is considered during uplift, Vz is essentially due
to permanent actions and not affected by the seismic response, i.e. constant. Moreover the
moments T and VY · h have the same sign when the first transverse mode is dominant. As a
consequence the following cases are distinguished: (a) when there is no transverse support
(i.e. Vϒ = 0) or when the transverse support is located at the level of the shear centre (i.e.
h = 0) then the torque T is fairly constant during uplift, (b) when the transverse support is
located below the level of the shear centre (i.e. h > 0) then the torque T is reduced during
uplift, and c) when the transverse support is located above the level of the shear centre (i.e.
h < 0) then the torque T is increased during uplift. Case c is neither practical nor expedient,
therefore it can be concluded that torque T does not increase during bearing uplift. Because
the end spans of the bridge have the largest torque it is concluded that bearing uplift intro-

123
Bull Earthquake Eng (2009) 7:363–375 365

Fig. 1 Equilibrium of deck end

duces an upper limit to the deck torque T which is determined approximately by applying
relation (1) at the time of uplift.

3 Torsional stiffness of thin-walled prestressed cross-sections

3.1 Torsional behaviour before concrete cracking

According to the theory of elasticity for the case of closed thin-walled closed cross-sections
(Bredt theory) the relation between the torque, T, and the twist θ , is (Timoshenko and Goodier
1951):

T = (G IT ) θ, IT = (2 Am )2 (bi /ti ) (2)
i

where θ is the twist (torsional rotation per unit of length), G is the shear modulus of the
cross-section material, IT is the torsional constant of the cross-section, the product GIT is
the torsional stiffness of the cross-section, Am is the area enclosed by the centreline of the
walls of the cross-section and bi , ti are the width and the thickness of wall i respectively. The
cracking torque Tcr is determined by setting the principal tensile stress equal to the tensile
strength of the concrete f ct :

Tcr = 2 Am tmin f ct (3)

where tmin is the minimum thickness of the cross-section walls.

123
366 Bull Earthquake Eng (2009) 7:363–375

Fig. 2 Truss model for torsional member: (a) general view, (b) element in shear flow zone

For the case of a prestressed cross-section under pure torsion, the increase of the cracking
torque due to the influence of the mean compressive stress σ p due to prestress can be taken into
account by means of a multiplication factor γ , which is determined as (Hsu and Mo 1985c):
  
 σp 
γ = 1 +   (4)
f ct

For the case of typical prestressed concrete cross-sections that are used in bridge design the
value of the multiplication factor γ is in the order of 2.

3.2 Torsional behaviour after concrete cracking

The non-linear behaviour of a deck element after concrete cracking can be estimated using
softened space truss models. Every wall of the element is considered as a truss with lon-
gitudinal ties corresponding to the longitudinal reinforcement, transverse ties correspond-
ing to the transverse reinforcement, and inclined struts corresponding to cracked concrete.
The inclination of the concrete struts with respect to the longitudinal axis is about 45◦
for the case of reinforced concrete elements or less for the case of prestressed sections due to
the normal compressive stress on concrete. Failure of the space truss model occurs in one of
the following modes: (1) yielding of longitudinal and transverse reinforcement without fail-
ure of concrete struts, (2) yielding of reinforcement in the weakest direction (longitudinal or
transverse) followed by failure of concrete struts in compression, and (3) failure of concrete
struts in compression before yielding of reinforcement.
The analysis that follows is based on the softened space truss model proposed by Hsu (Hsu
and Mo 1985a,b,c; Hsu 1991). A typical truss model is illustrated in Fig. 2. A prestressed
concrete member with hollow rectangular cross-section is subjected to a torque T . Shear flow
q will develop along the shear flow zone with thickness t. Based on the truss model concept,
the concrete struts between the cracks are subjected to a compressive stress σd , where α is the
inclination angle of the struts. The equilibrium, compatibility, and stress–strain relationships
of the membrane element shown in Fig. 2b have been discussed by Hsu (1991). A total of
11 equilibrium and compatibility nonlinear equations are formulated: 3 stress equilibrium
equations, 1 total torque equilibrium equation, 3 strain compatibility equations and 4 com-
patibility equations for the bending of the concrete struts due to warping of the membrane
element in the shear flow zone. These equations are coupled with the non-linear stress–strain
constitutive laws of the materials. The behaviour of the softened concrete is described by the

123
Bull Earthquake Eng (2009) 7:363–375 367

modified compression field theory which gives concrete strength as a function of strain in
the transverse steel (Vecchio and Collins 1981, 1986). The stress–strain relation of reinforce-
ment steel is modelled by an elastic–perfectly plastic law. The stress–strain relationship of
high-strength prestressing steel when the proportionality limit is exceeded is more accurately
described by a Ramberg-Osgood curve rather than a bilinear idealization (Hsu and Mo 1985c):
  m  m1
E p εps
fps = E p ε ps 1 + (5)
fpu

where fps is the stress of prestressing steel when the proportionality limit is exceeded, Ep
is the modulus of elasticity of prestressing steel, εps is the total strain of prestressing steel,
fpu is the failure stress of prestressing steel and m is a coefficient determined from tests. The
value m = 4 that was used by Hsu (1991) is assumed in this work.
The equilibrium, compatibility and constitutive law equations that govern the torsional
member are solved for the unknown variables which include the stresses, strains and the
inclination angle α of the concrete struts. For a given twist θ the corresponding torque T is
calculated from the solution of the set of nonlinear equations. In this manner the nonlinear
torque–twist relationship of the prestressed member after concrete cracking is produced. The
governing equations and the solution procedure of the space truss model used in this work
have been presented in detail by Hsu (1991).

4 Parametric study for the estimation of torsional stiffness of prestressed box


cross-sections

The reduction of the torsional stiffness of the cross-section after concrete cracking is exam-
ined through a series of parametric analyses that cover a range of typical prestressed box
girder cross-sections. The torsional stiffness before cracking (GIT ) as well as the crack-
ing torque Tcr of the prestressed cross-section are determined from the theory of elasticity
according to relations (2), (3), and (4). The secant torsional stiffness after concrete cracking
(GIT )cr is determined from the T − θ curve based on the non-linear analysis of the space
truss model that is presented in 3.2.
For all examined cross sections the following quantities were kept constant: box width
5.8 m, web thickness 0.50 m, top slab thickness 0.26 m, mean concrete compressive strength
f cm = 43 MPa, mean concrete tensile strength f ctm = 3.2 MPa, concrete elastic modulus
34 GPa, reinforcement steel yield strength 500 MPa, prestress steel ultimate strength
1,770 MPa, prestress tendon stress 1,250 MPa. The values of the variable parameters that
have been examined in the parametric study are presented in Table 1. Four different cross-
section sizes have been examined with height varying from 2.5 to 5.5 m and bottom slab
thickness varying from 0.25 to 0.70 m, which correspond to four different cross-sections of a
bridge deck constructed by the method of balanced cantilever construction. Moreover various
values of the volumetric ratio of reinforcement and prestress tendons have been examined. A
series of non-linear analyses of the softened space truss model is carried out for each of the
examined variable parameters while the rest of the variable parameters take constant values
equal to the values indicated with bold characters in Table 1.
The results of the study are presented in Fig. 3 for each of the examined parameters. The
ratio of the cross-section torsional stiffnesses after and before concrete cracking, (G IT )cr /
(G IT ), is presented as a function of torque, T , normalized to the value of cracking torque Tcr .
For all examined cases an abrupt change in the stiffness ratio is observed, which corresponds

123
368 Bull Earthquake Eng (2009) 7:363–375

Table 1 Examined cases in the parametric study

Parameter Values Number of


examined cases

Cross-section size (height h/bottom (5.5 m/0.70 m), (4.4 m/0.55 m), 4
flange thickness t) (3.2 m, 0.35 m), (2.5 m/0.25 m)
Longitudinal reinf. volumetric ratio ρ l 0.5%, 1.0%, 1.5%, 2.0% 4
Transverse reinf. volumetric ratio ρ t 0.5%, 1.0%, 1.5%, 2.0% 4
Prestress tendons volumetric ratio ρ lp 0.25%, 0.5%, 0.75% 3

Note: The volumetric ratios correspond to the concrete volume of the torsion box; Bolded entries indicate the
assumed value when the corresponding variable is not the main parameter of the parametric analysis

Fig. 3 Results of parametric analyses for the reduction of torsional stiffness of thin-walled prestressed
concrete cross sections after concrete cracking

to yielding of the transverse reinforcement. After that, failure occurs when the concrete struts
fail in compression.
It is concluded that for all examined cases the ratio of cracked-to-uncracked torsional stiff-
ness is between 10% and 30% just after concrete cracking and between 8% and 12% when
the transverse reinforcement yields. This significant reduction of the torsional stiffness after

123
Bull Earthquake Eng (2009) 7:363–375 369

concrete cracking has also been reported by other researchers (Mukhopadhyay and Roy 1979;
Wafa et al. 1995; Rahal and Collins 1996). In particular from the test results of prestressed
concrete elements subjected to in-plane shear forces reported by Marti and Meyboom (1992)
it is calculated from the experimental curves that the ratio of cracked to uncracked stiffness
is between 25% and 40% after concrete cracking, 13–29% at the first yield of reinforcement,
13–23% at the last yield of reinforcement, and 7–8% at failure.
According to Fig. 3, the ratio of cracked-to-uncracked torsional stiffness just after crack-
ing is affected by the cross-section size, the transverse reinforcement volumetric ratio and
the prestress tendons volumetric ratio. On the other hand for the examined cases the longitu-
dinal reinforcement volumetric ratio did not have a significant effect. The ratio of cracked-
to-uncracked torsional stiffness at yield of the transverse reinforcement was only mildly
influenced by the prestress steel volumetric ratio.
Consequently, a secant torsional stiffness between 10% and 30% of the uncracked tor-
sional stiffness is proposed by the authors as an appropriate value for the torsional stiffness
of prestressed concrete box girders after concrete cracking.

5 Influence of bearing uplift

5.1 Elastoplastic pier behaviour without bearing uplift

When no uplift of abutment bearings occurs, the torque at deck ends can be estimated from
compatibility of deformations for elastoplastic pier behaviour. The first two spans of the deck
are investigated (Fig. 4). The deck is connected monolithically to pier BC. δ is the displace-
ment and ϕ the rotation of the pier top, and Mt , Mb are the bending moments at the top and
bottom of the pier respectively. The torque diagram of the deck is illustrated in Fig. 4 for
the case where no bearing uplift occurs. The ratio T2 /T1 of the second-to-first span torque
is denoted as κ (i.e. T2 = κ T1 ). The first span is subjected to a significantly larger torque
than the others, and the direction of T1 and T2 is the same when the first transverse mode is
dominant, therefore κ is generally positive and near zero.
The following relations are derived; the first one expresses the moment equilibrium of
joint C and the other two express the deformation compatibility:

Mt = T1 − T2 = T1 (1 − κ) (6)
GIT
T1 = ϕ (7)
L1
6EI p 4EI p
Mt = δ− ϕ (8)
H2 H
where GIT is the torsional stiffness of deck cross-section, L 1 is the length of the end span,
EI p is the bending stiffness of pier cross-section for bending about the longitudinal bridge
axis, and H is the pier height. Depending on the magnitude of the earthquake induced forces,
the cracked or uncracked condition of the elements during the earthquake should be taken
into account in the values of GIT and EI p .
By solving the system of Eqs. 6–8 the following relations are derived:
6E I p λ
T1 = · ·δ (9)
H 2 4 + λ(1 − κ)
6E I p 2 + λ(1 − κ)
Mb = · ·δ (10)
H 2 4 + λ(1 − κ)

123
370 Bull Earthquake Eng (2009) 7:363–375

Fig. 4 Simplified model of the first two spans of the bridge when no uplift of the bearings occurs

where λ is the ratio of deck torsional stiffness to pier bending stiffness:


GIT /L 1
λ= (11)
EI p /H
As a simplification it is assumed that pier bending behaviour is initially elastic until the
formation of plastic hinges at their base for bending moment equal to M y and displacement
at the pier top equal to δ y , and then perfectly plastic until the plastic deformation capacity
of the plastic hinge is reached for a displacement at pier top equal to δeq . The ratio δeq /δ y
expresses the ductility of the pier in terms of displacements. For typical ductile systems the
equal displacement rule is assumed, i.e. the elastoplastic system is considered to develop
the same maximum displacement δeq as the equivalent elastic system having linear stiffness
equal to the secant stiffness at yield of the elastoplastic system. Then the ratio δeq /δ y corre-
sponds to the seismic behaviour factor q defined in EN1998-2 as the ratio of the force of the
equivalent elastic system to the force of the elastoplastic system.
If a plastic hinge has formed at the bottom of the pier then the increment of bending
moment for elastic–perfectly plastic behaviour is Mb = 0. The deck torque increment T1
is:
3E I p λ
T1 = · ·δ (12)
H 2 3 + λ(1 − κ)
By combining relations (9) and (12) the following deck torque relation is derived for elasto-
plastic pier behaviour, assuming that plastic hinges form only at pier base:

δeq
[2 + λ(1 − κ)] + [4 + λ(1 − κ)]
My λ δy
T1 = · (13)
2 [2 + λ(1 − κ)] · [3 + λ(1 − κ)]
The ratio of the peak torque of the elastoplastic system Te− p to the peak torque of the unlim-
ited elastic system Tunl is always smaller than 1.0. Based on Eq. 13 this ratio is given by the
following relation as a function of δeq /δ y and λ(1 − κ):

123
Bull Earthquake Eng (2009) 7:363–375 371

Fig. 5 Ratio of maximum deck torque for elastic–perfectly plastic pier behaviour Te− p to the maximum deck
torque for unlimited elastic pier behaviour Tunl

δeq
[2 + λ(1 − κ)] + [4 + λ(1 − κ)]
Te− p δy
γT = = (14)
Tunl δeq
[6 + 2λ(1 − κ)]
δy

Relation (14) is plotted in Fig. 5. The ratio γT = Te− p /Tunl is equal to 1.0 for elastic systems
(i.e. δeq /δ y = 1.0) while for elastoplastic systems it is between 0.5 and 1.0. The values of
Te− p and Tunl depend strongly on the torsional stiffness of the deck. However the effect of
the factor λ(1 − κ) on the ratio γT = Te− p /Tunl is not significant (less than 10%). It is also
observed that for typical ductile bridges with behaviour factor q = δeq /δ y ≤ 3.5 and no
bearing uplift, the torque for the design of the deck should be Te− p = γT · Tunl , that may
be substantially larger than the capacity design torque according to EN1998-2, amounting
to approximately γ0 · Tunl /q ≤ Tunl , (γ0 = 1.40 = overstrength factor of the plastic hinge).
This is due to the fact that following the formation of the plastic hinges at the base of piers,
no complete plastic mechanism is formed i.e. the system can resist additional seismic forces.

5.2 Elastoplastic pier behaviour with bearing uplift

When uplift of the bearings occurs, the maximum torque T induced at the deck ends is limited
to the value that satisfies Eq. 1 during bearing uplift. This torque value is generally smaller
than the torque for capacity design of the deck that was estimated in the previous section.
Therefore bearing uplift may be utilized as a safety mechanism for the protection of the deck
from extensive torque.
The post-elastic behaviour of the bridge taking into account the possibility of bearing uplift,
before or after plastic hinges at the ends of the piers are formed, is examined through nonlinear
static analyses of a simple three span bridge (60 m + 100 m + 60 m) with 29 m tall piers (Fig. 6).
The examined case is an actual bridge constructed by the method of balanced cantilever con-
struction. For this type of bridge the abutment bearings are likely to uplift during earthquake
due to the small vertical bearing reaction for permanent loads and the large torsional stiffness
of the deck box. The height H and the bottom slab thickness t of the box cross-section are
variable along the deck length (H = 5.8 m, t = 0.75 m at the piers, H = 2.6 m, t = 0.26 m at
the midpoint of the central span and at the abutments). The deck is supported at the abutments

123
372 Bull Earthquake Eng (2009) 7:363–375

Fig. 6 Bridge model for nonlinear static analysis

in the transverse direction 1.2 m below the shear centre. The distance d between the bearings
at the abutment is a variable parameter. The torsional stiffness of the deck is assumed equal
to the uncracked value. The validity of this assumption for the examined case is verified later
by the analysis results.
The space model of the bridge consists of linear elastic beam elements. The nonlinear
behaviour of uplift-capable bearings is modeled by compression-only gap elements. The
plastic hinges that form at the base and top of piers are modeled by nonlinear rotational
springs with rigid–perfectly plastic law, concentrated at pier ends. For the nonlinear static
analyses the computer program SAP2000 is used. A gradually increasing load pattern that is
proportional to the first mode shape in the transverse direction is applied (Fig. 6a).

5.2.1 Elastoplastic pier behaviour with bearing uplift before the yielding of piers

It should be noted that this case is not permitted by EN1998-2. The dependence of deck torque
and pier bending moment on the normalized base shear is presented for this case in Fig. 7a,
based on the results of nonlinear static analysis for the bridge presented in Sect. 5.2, with
distance between the bearings equal to 5 m. The assumption of uncracked torsional stiffness
for the deck is valid because the maximum deck torque does not exceed the cracking torque
of the smallest cross-section Tcr = 29 MNm.
It is observed that after bearing uplift the deck torque is reduced, and even more after
pier yielding. Because bearing uplift occurs before the formation of pier plastic hinges, the

123
Bull Earthquake Eng (2009) 7:363–375 373

Fig. 7 Results of nonlinear static analysis for the pier base bending moment and the deck torque for
(a) distance between bearings equal to 5 m and (b) distance between bearings equal to 11 m

behaviour of the system is trilinear. However the deviation from the bilinear behaviour is
not significant for the examined case. The deviation from the bilinear behaviour introduces
uncertainties in the estimation of the seismic response of the system.
If bearing uplift occurs before pier yielding then in addition to the ductile behaviour of
the piers, part of the nonlinear system response results from the uplift of bearings. Bearing
uplift is not a reliable mechanism of energy dissipation compared to plastic response of piers.
Moreover if uplift occurs early, uplift displacements and the corresponding impact loads on
the bearings are larger. This results in increased possibility of damage of the bearings and the
structure in their vicinity. These facts are in agreement with EN1998-2 provision that does
not permit uplift of bearings before the formation of plastic hinges (i.e. before the design
earthquake is exceeded 1/q times).

5.2.2 Elastoplastic pier behaviour with bearing uplift after the yielding of piers

This case is permitted by EN1998-2. The dependence of deck torque and pier bending moment
on the normalized base shear is presented for this case in Fig. 7b, based on the results of non-
linear static analysis for the bridge presented in Sect. 5.2, with distance between the bearings
equal to 11 m. The assumption of uncracked torsional stiffness for the deck is valid because
the maximum deck torque does not exceed the cracking torque of the smallest cross-section
Tcr = 29 MNm.

123
374 Bull Earthquake Eng (2009) 7:363–375

It is observed that after pier yielding the deck torque is increased, and then reduced when
bearing uplift occurs. Uplift occurs after the formation of plastic hinges and does not affect
the bilinear behaviour of the piers. Therefore, the post elastic behaviour of the whole system
is not affected significantly as it is mainly determined by the post-elastic behaviour of the
piers.
Bearing uplift after pier yielding does not introduce additional uncertainties in the estima-
tion of seismic response and also acts as an additional safety mechanism preventing brittle
failure of the deck in torsion before the capacity of pier plastic hinges is exceeded. These facts
are in agreement with EN1998-2 provision that permits the uplift of bearings after formation
of pier plastic hinges (i.e. after the design earthquake is exceeded 1/q times). It should be
verified however that the consequences of bearing uplift to the bearings, the deck, and the
substructure are not so severe as to hinder the performance requirements of the bridge during
earthquake.

6 Conclusions

Based on the results presented in this work it is concluded that uplift of a bearing over the
abutment may act as an additional safety mechanism preventing torsional deck failure before
the plastic capacity of the ductile elements of the bridge is exceeded. If it is ensured that
bearing uplift occurs after the formation of pier plastic hinges then there are no additional
uncertainties in the estimation of the seismic response of the bridge and the consequences
of bearing uplift are less likely to induce damage to the bearings and the structure in their
vicinity. These results are in agreement with EN1998-2 provision which allows bearing uplift
when the the design earthquake is exceeded 1/q times.
The equilibrium conditions that apply when bearing uplift occurs establish an upper limit
to the deck torque that prevents brittle torsional failure of the deck. In order to estimate the
deck toque during bearing uplift it is important to estimate the torsional stiffness of the deck
in a realistic manner. The presented parametric study for the case of prestressed deck box
girders indicates that the torsional stiffness after concrete cracking is reduced significantly
and it is typically in the order of 10–30% of the uncracked torsional stiffness.

Acknowledgements The support of the Hellenic General Secretariat for Research and Technology through
research project ASPROGE (Antiseismic Protection of Bridges) is gratefully acknowledged.

References

Batchelor B, McEwen D (1977) Eccentrically prestressed box girders under combined bending and torsion.
ACI J Proc 74(4):156–162
Fu CC, Tang Y (2001) Torsional analysis for prestressed concrete multiple cell box. J Eng Mech 127(1):45–51.
doi:10.1061/(ASCE)0733-9399
Hsu TTC (1991) Nonlinear analysis of concrete torsional members. ACI Struct J 88(6):674–682
Hsu TTC, Mo YL (1985a) Softening of concrete in torsional members—theory and tests. ACI J
Proc 82(3):290–303
Hsu TTC, Mo YL (1985b) Softening of concrete in torsional members—design recommendations. ACI J Proc
82(4):290–303
Hsu TTC, Mo YL (1985c) Softening of concrete in torsional members—prestressed concrete. ACI J Proc
82(5):290–303
Marti P, Meyboom J (1992) Response of prestressed concrete elements to in-plane shear forces. ACI Struct J
89(5):503–514
McGee D, Zia P (1976) Prestressed concrete under torsion shear and bending. ACI J Proc 73(1):26–32

123
Bull Earthquake Eng (2009) 7:363–375 375

Mukhopadhyay M, Roy SK (1979) Pure torsion of prestressed flanged sections. Build Environ 14:135–146.
doi:10.1016/0360-1323(79)90019-2
Rahal KN, Collins MP (1996) Simple model for predicting torsional strength of reinforced and prestressed
concrete sections. ACI Struct J 93(6):658–666
Timoshenko S, Goodier JN (1951) Theory of elasticity (2nd edn). McGraw Hill
Vecchio F, Collins MP (1981) Stress–strain characteristics of reinforced concrete in pure shear. Final report
of International Association of Bridge and Structural Engineering (IABSE) colloquium on advanced
mechanics of reinforced Concrete, pp 211–225
Vecchio F, Collins MP (1986) The modified compression-field theory for reinforced concrete elements
subjected to shear. ACI J Proc 83:219–231
Wafa FF, Shihata SA, Ashour SA, Akhtaruzzaman AA (1995) Prestressed high-strength concrete beams under
torsion. J Struct Eng 121(9):1280–1286. doi:10.1061/(ASCE)0733-9445

123

You might also like