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

net/publication/307545844

Advanced calculation methods for box girder bridges

Article · November 1992

CITATIONS READS
0 1,803

1 author:

Jacques Berthellemy
Ministère de l'Écologie, du Développement durable, des Transports et du Logement
104 PUBLICATIONS   173 CITATIONS   

SEE PROFILE

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

Innovation for steel- concrete composite bridges View project

Tied-arch bridges / Les ponts en arcs autoancrés View project

All content following this page was uploaded by Jacques Berthellemy on 01 September 2016.

The user has requested enhancement of the downloaded file.

 European Steel Design Education Program
1992

Advanced Calculation Methods for Box Girder Bridges

8.5.2.

Jacques Berthellemy
Ingénieur des Travaux Publics de l’État.
SETRA - 46, Avenue A. Briand BP100 92223 BAGNEUX CEDEX - FRANCE

Steel or steel-concrete composite box girders are usually more expensive than plate
girders because they are often used for more complex cases of geometry and because they require
more fabrication time. They have, however several advantages over plate girders which make
their use attractive :

- very high torsional rigidity : In closed box girders, torque is resisted mainly by de
Saint Venant shear stresses because the de Saint Venant torsional stiffness is normally much
greater than the torsional warping stiffness. For highly curved spans, this stiffness of the box
girders are almost essential during their construction , as well as under service loads. Metallic
closed-top boxes even allow torsional stiffness during their erection, without needing expensive
temporary bracing, which would disturb the realisation of the concrete slab.

- very wide flanges allow large span to depth ratio.

- a neater appearance since the stiffening can remain invisible in the box.

- a very good aerodynamic shape , which is equally important for large suspension
or cable-stayed bridges as is the torsional stiffness.

- a very good adaptability to the most difficult conditions. The box girders are able
to cross greater torsional spans than flexional spans using piers with a single bearing as shown in
the following figures :

1
 The plate girders solution.

The box girder solution,

avoiding the unmechanic skew alignments of bearings,
is here cheaper, and more esthetic.

GLOBAL ANALYSIS METHODS

The principal methods that are in use for global analysis, and their applicability to the
various types of steel or steel-concrete composite bridges, are discussed in the following pages.

Most composite bridge deck fall into one of two groups:

Beam theory can be used for the bridges of the first group which can be idealised as a grillage or
an orthotropic plate.
Those in this first group include beam-and-slab and multi-separate-box bridges, and are the
simplest to analyse. The box girders are here assumed to be stiffened enough to stay
undeformable. For uniform loading, elementary beam theory gives useful results, but distribution
analysis by a grillage or an orthotropic plate method is needed for more complex loadings.

Typical cross section for the first group.

2
 In the other group are those in which the longitudinal members consist of a closed
box with eventually a few number of internal webs, which effectively resist vertical shear only. In
general, the effects of loading on such structures are best evaluated by dividing the load into
bending, uniform torsional, warping torsional, distortional and local components. The Bredt and
Leduc theory for thin walled hollow sections gives the primary shear stresses due to torsion.
Timoshenko and Vlassov have shown the validity of this hypothesis with the theory of the
undeformable girder sections : the de Saint Venant torsional stiffness is much greater than the
torsional warping stiffness.

But the shear stresses are accompanied by longitudinal torsional and distortional
warping stresses. The distortional warping stresses are the consequence of the bending
deformation of the section walls in the transverse plane. They usually are several times larger than
the torsional warping stresses. They reach their maximum values at bearings, at cross sections
subjected to eccentric point loads, and at the location of the diaphragms which prevent the
deformation of the girder section.

1 - Grillage

In the grillage analysis, the structure is represented by a plane grillage of discrete but
interconnected beams. Almost any arrangement in plan is possible, so skew, curved, tapering or
irregular decks can be analysed. But the usual layout is sets of parallel beams in two directions.
This will now be discussed, assuming the plane of the grillage to be horizontal.

The method is not adapted to study the warping stresses and shear lag. Local effects
in decks and slabs can only be studied with a grillage by the use of a dense network of beams.

In a simple form of grillage analysis, each beam is alloted a torsional stiffness and a
flexural stiffness in the vertical plane. Vertical loads are applied only at the intersections of the
beams. The matrix stiffness method analysis is used by the existing softwares, to find the
rotations about two horizontal axes and the vertical displacement at these nodes, and hence the
bending and torsional moments and vertical shear forces in the beams at each intersection.

Choice of grillage :

Care must be taken to select an appropriate idealisation for a continuous structure,

and also to deduce the stress resultant in it from the results of the grillage analysis. The principal
conclusions relevant to the design of composite structures are as follows :

Grillage analysis is not the most suitable method for one simple beam-and-slab deck.

Each longitudinal web should be represented by a grillage beam ( so that a simple box
becomes two beams ) . The alternative for this would be to use one single beam in the axis of the
box to represent it, and to dispose on each side of the longitudinal beam, very rigid extensions at
the location of every transverse beam with a total length equal to the box's breadth : the global
effects in the structure are easier to catch with this second method.

3
 Each transverse diaphragm or cross-frame should be represented by a transverse
beam. If there are none within the span, or if their spacing exceeds the mean spacing of the
longitudinal beams by more than 50 % , additional transverse beams representing the slab alone
should be added, and there should be at least nine in all, within every span.

The transverse grillage members should extend to the edge of the real slab and their
ends should be attached to longitudinal grillage beams, even if the real slab has no significant
edge stiffening.

The case of skew bridges :

Skew bridges should be avoided, as soon as possible in the design scenario of the
road, especially when high torsional stiffness is caused by the choice of a box girder section.

For skew decks, the grillage cross beams should be parallel to the real cross-members,
if any. The stiffness and position of the real supports should be accurately modelled in the
grillage. Real cross members can with advantage be skewed for skews up to about 20° but for
high skews, they should be orthogonal. For decks without cross members, skew grillage are
preferred for skews up to about 30° because the preparation of input data for an orthogonal
grillage is more complex. For highly skewed decks, the cross members , if any , and the grillage
beams should be orthogonal, to avoid the high torsional moments inherent in skews layouts.
Transverse reinforcement in the slab should be parallel to the transverse grillage beams.

Local effects in decks :

If the purpose is to catch the local bending and shear effects in deck slabs too, the
concrete slab of a composite bridge should be represented by a dense network of members. The
distance between two parallel members of the grillage should be of 0.50 m. In the case of a plane
grid analysis, the torsional stiffness of such a concrete strip should not for instance be b.t 3/3 but
should be reduced to only b.t3/6 , where b is the breadth of 0.50 m and t the thickness of the deck.
This reduction is due to the fact that no de Saint Venant shear stress flux goes around the
perimeter of the strip's cross section. Vertical local loads should be only applied at the
intersections nodes of the beams.

How to model the equivalent loads :

Approximately one half of the local load can be distributed over the eight nodes of the
vicinity to get correct results, even near the loaded point.

4
Torsional and flexional rigidities of grillage members :

In the case of steel-concrete composite box girders, the areas of structural concrete are
transformed to steel on a modular basis. The Young modular ratio is used for flexion, and the
ratio G steel / G concrete = 5 is used for torsion due to temporary traffic loads.

In the case of a composite box girder bridge, with a concrete upper deck, the torsional
stiffness K is for example:

Shear lag ( traînée de cisaillements ) can usually be neglected in global analysis.

For regions where concrete slab is in compression, the flexural stiffness is strictly that
of the uncracked reinforced section. Even where the concrete has cracked under a traffic load, its
stiffness remains. For these reasons, it is common to assume, in global analysis, that for the
computation of the beam rigidities, all concrete is uncracked and unreinforced.

Longitudinal grillage members :

Flexural rigidities of longitudinal members depend on the breadth of concrete slab

that is assumed to be associated with each steel web. For the deck shown in the figure below,
arbitrary division of the deck midway between webs would result in the neutral axis of the beam
web EH being higher than that of the beam with web FG. More rigourous analysis of concrete
decks shows the neutral axes of all the longitudinal members to be almost at the same level across
the width of the deck. This also suggests that the breadth of deck allocated to the web EH

5
and FG should be more nearly equal, as shown by the lines A,B and C. This alternately suggests
that the transverse position of the steel boxes under the slab can in this case be improved if
possible.

Interpretation of the output of a grillage analysis :

A typical output gives the external reactions at each support : these results should
always be controlled first. The softwares usually also give values of the vertical shear, bending
moments and torsional moment for each grillage member on both side of each joint in the
grillage. Care should be taken in deducing design values for the real members from these results.
The values midway between two adjacent nodes are usually more representative of the real
values.

The bending and torsional moments will, in general, show a discontinuity at each
joint. For an orthogonal grillage, each change in bending moment is equal to the change in
torsional moment at that joint in the member at right angles to the one considered. Similarly, the
change in torsional moment equals the change in bending moment in the perpendicular member.

typical bending or torsional moments given by a

grillage analysis

If the values midway between two following nodes are not directly available, the
design bending and torsional moments at each joint can be taken as the mean of the values on
each side of the joint, for the member considered.

6
 2 - Orthotropic Plate Analysis

In orthotropic plate analysis, the deck structure is smoothed across its length and
breadth and treated as a continuum.

The elastic properties of an orthotropic plate are defined by the two flexural rigidities
Dx and Dy and a plate torsional rigidity H. The governing equation relating deflection w to load P
acting normal to the plane of the plate is :

Design charts for decks that can be idealised as orthotropic plates have been derived
from series solutions. They give deflections and longitudinal and transverse moments due to a
point load, and so provide a rapid method for distribution analysis. Their applicability is limited
to simply supported decks of skew not exceeding 20° whose elastic properties can be represented
solely by length, breadth, and the three quantities Dx, Dy and H.

In composite structures, they can be used for beam-and-slab decks with not less than
five equally spaced longitudinal members of uniform diaphragms over the supports.

In practice, the method has often been superseded by grillage analysis.

3 - Folded plate analysis

The folded -plate method is normally limited to assemblages of rectangular plates,

and is not applicable to skew decks due to coupling between the harmonics. These orthotropic
plates may extend over several spans but must be simply-supported at the extreme ends, with
rigid diaphragms over the end supports. When folded plates diaphragms are used to represent the
transverse frames, its advantages are that it can give a complete and accurate solution in much
less computer time than is needed for finite element method, and can accept a wide variety of
types of loading and both displacement and force boundary conditions.

Folded plate analysis : beam on elastic foundation analogy

If we assume that the deformations of the cross section of one box girder bridges are
concentrated in the corners of the box girder, we find an analogy between this folded plate
analysis and the beam theory. The cross frames or the diaphragms transform the effect of

7
eccentric loads into uniform of de Saint Venant torsional shear stresses by resisting to the
resulting distortion, therefore they become the elastic supports for the box girder representing
beam. This simplification offers an economic way to compute the internal forces in the cross
frames of the box girder, as well as the distortion warping stresses.

To apply the same method to a double cellular box-girder bridge with one single
internal web, the distortion must be decomposed into symmetric and antisymmetic deformations.
For boxes with more internal webs, it is possible to decompose the deformations of the cross
section into eigenvalue functions of deformation.

4 - Finite element analysis

The finite element method has more and more application in civil engineering. This is
the most versatile among the matrix stiffness methods of elastic analysis, and can in principle
approach the solution of almost any problem of global analysis of a bridge deck. The finite
elements method is not only used for the study of statical problem, but for dynamic problems
involving elasticity and/or plasticity. The finite elements method allows the study of shear lag and
the computation of effective flange breadths. It is also useful to study local effects in slabs.

Its least disadvantage is its cost, especially because of the high level expert time
required for the idealisation of the structure. The expert's know-how is needed, in selecting an
appropriate pattern of elements, determining the right limit conditions for boundary nodes along
the supports or the symmetry axes, and interpreting results. The choice of wrong elements can be
misleading in regions of steep stress gradient, because the conditions of static equilibrium are not
necessary satisfied. The selection of the discretisation density level, or of the material behaviour
may have serious repercussions on the accuracy of the results.

Other problems are that the method is generally not well adapted to compute right
deformations and fabrication cambers.

Manual preparation of computer input and interpretation of output takes so much time
that procedures for mesh generation and the plotting of trajectories of principal moment or stress
are now incorporated into the programs. The result plotting procedures may dangerously lead to
underestimate the maximum values of some wanted effects, by computing non representative
average values.

Because high level expert's judgement and experience are needed for a quality
insuring use of the method, it should have little application in routine design work and will not be
further summarised here.

8
 CROSS SECTIONAL DISTORTION

The distortional loading causes the cross section of the box to change shape by
bending of its walls in the transverse plane. The associated transverse bending and shear stresses
are usually acceptable in the relatively thick walls of concrete boxes, but in steel and steel-
concrete composite boxes, diaphragms or cross-frames may be required to control the distortion.
There are also longitudinal distortional warping stresses. In steel and steel-concrete composite
beams, their relative size depends on the spacing of the cross frames.

Force computation in the diaphragms

The cross frames or the diaphragms transform the effect of eccentric loads into
uniform of de Saint Venant torsional shear stresses by resisting to the resulting distortion. By this
way, their rigidity avoids the girder section deformations in the transverse plane.

The distortion effects have to be cumulated with the local bending and shear in deck
slabs in the vicinity of point loads. The calculation of local bending stresses in plates is
simplified by the use of influence surfaces : those for simply supported plates tend to give high
values in box girders as they do not consider the true boundary conditions of the plate panel.
- simple method :
The method often used in practice consists of isolating by the thought a segment of
the box girder bridge which length is equal to the distance between two adjacent diaphragms or
cross-frames. One diaphragm or cross-frame is located in the centre of this segment. All the
external loads applied at the segment are concentrated on the cross-frame. We also assume that
the segment exchanges uniform of de Saint Venant torsional shear stresses, as well as bending
shearing forces with the rest of the bridge at both segment's ends. All these shear stresses are also
concentrated on the cross-frame.

The diaphragm or cross-frame, is then calculated under the equilibred set of actions
that has been described. The resulting dimensions remain usually economic and acceptable.
- refined method :
The distortion forces due to the loads applied to the above described segment are
resisted in fact by more than a single diaphragm or cross-frame. The method of the beam on
elastic foundation is one way to compute the distribution of the distortion loads over the different
diaphragms or cross-frames considered as elastic supports.

9
 The following diagram shows the distortion force which will be reduced in the refined
method. For the simplicity of the example, the one box girder section is only loaded by two
opposite forces applied to the top corners.

The PLATE 1 page 12 is a parametric study showing the influence of diaphragm

separation for a simple box.

The PLATE 2 page 13 shows the distribution of the distortion forces in the different
cross-frames of a four spans bridge in one of the two main spans of 80 m. The figures show the
results of a folded plate analysis using diaphragms to represent the frames.

The PLATE 3 page 14 shows the warping stresses along the Chevire's Bridge. This
isostatic bridge is a 162m length single span across the River Loire. The influence of two
additional bracings has been studied here in order to demonstrate that additional transverses
bracings were useless.

Advices for the diaphragms on piers

The local effects of reactions at bearings cause complex states of stress in the support
diaphragms and stress concentrations in the adjacent webs. The cross frame is usually a
diaphragm on piers to obtain a great rigidity to resist distortion. If the bearings are under the
diaphragm on piers, care must be taken to avoid dilatation problems due to longitudinal
eccentricity occurring at bearings. This is a particular problem with steel diaphragms which
suggest that concrete or composite members could be used more often in this location, where
extra weight is easily carried. But this mechanic point of view is not the only one to take into
account for the conception of these elements, that also must resist to corrosion during their life,
and allow the regular inspection of the welded connections.

10
 FLANGE BUCKLING AND SHEAR LAG

In very wide flanges, the shear lag effects, which are neglected in the global analysis,
have to be taken into account for the verification of stresses, especially for short spans.

Shear lag must be considered when verifying sections and calculating stresses, as it
frequently causes the longitudinal stress at a flange/web intersection to exceed the mean stress in
the flange by 20%.

It is allowed in calculations based on the elementary theory of bending to use an

effective flange breadth less than the real breadth. This effective flange breadth depends on the
ratio of width to span.

For a simply supported beam, for example, the effective breadth Fe.b of the portion
between webs is given in the table below, where :

11
12
13
14
 BIBLIOGRAPHY

(1) P. Dubas and E. Gehri : Behaviour and Design of Steel Plated Structures. Technical
Commitee 8 Group 8.3 ECCS - CECM - EKS (n°44 1986).
(2) RP. Johnson and RJ. Buckby : Composite structures of steel and concrete. Volume 2 :
Bridges. COLLINS (1986 London).
(3) British Standard 5400 : Part 3 : Steel, concrete and composite bridges. Part 3 : Code of
practice for design of steel bridges. BS Institution (1982).
(4) CF. Kollbrunner et K. Basler : TORSION SPES / BORDAS (1955 Lausanne / Paris).
(5) Stahlbau Handbuch : Stahlbau Handbuch für Studium und Praxis. Band 1. STAHLBAU
VERLAG (1982 Köln).
(6) DC. Dalton and B. Richmond : Twisting of thin walled box girders. Proceedings of the
institution of civil engineers. (January 1968).

15

View publication stats