You are on page 1of 19

Engineering Structures 21 (1999) 737–755

Recent evolution of cable-stayed bridges


*
Michel Virlogeux
24 Rue de la Division Leclerc, Bonnelles 78830, France

Abstract

This paper aims to provide evidence of the rapid progress in cable-stayed bridges. The span record progressed rapidly during
the last 10 years, passing from 465 to almost 900 m; it is now expected that 1200 m will be reached soon, if care is taken with
regard to aerodynamic stability and dynamic response to turbulent wind, which can be now mastered. Cable vibrations, which have
been extensively analysed in the last few years, can also be controlled by different types of countermeasures. However, cable-
stayed bridges have also developed in directions other than very long spans: flexible decks, extradossed cables and multispan cable-
stayed bridges which will certainly receive a wider development for large projects.  1999 Elsevier Science Ltd. All rights reserved.

Keywords: Cable-stayed bridges; Very long spans; Aerodynamic stability; Dynamic response to turbulent wind; Cable vibration; Flexible decks;
Extradossed bridges; Multispan cable-stayed bridges

1. Introduction

Engineers invented the concept of cable-stayed


bridges very early on, at the same time as they began
developing suspension bridges; however, with the col-
lapse of the bridges built over the rivers Tweed and
Saale, at the beginning of the 19th century, the idea was
abandoned. Roebling and others later introduced some
cable-stays in suspension bridges to reduce deform-
ability, such as for the Brooklyn bridge, and some innov-
ative suspension systems very close to those of cable-
stayed bridges were invented, such as those by Gisclard.
Surprisingly, the first “modern” cable-stayed bridges
were built in concrete by Eduardo Torroja in the 1920s
(Tampul aqueduct) and by Albert Caquot in 1952 Fig. 1. The bridge over the Donzère canal in France (1952). Photo:
J. Kerisel.
(Donzère canal bridge; Fig. 1); but the real development
came from Germany with papers published by Franz
Dischinger and with the famous series of steel bridges for cable-stayed bridges, since it provides keys to under-
crossing the river Rhine. stand the evolution of their design.
The international development of this bridge type However, other aspects of the design of cable-stayed
began in the 1970s, but a very big step forward took bridges are worth discussion: the development of slender
place in the 1990s, when cable-stayed bridges entered decks for medium and large spans; the concept of
the domain of very long spans which was previously “extradossed” bridges, considered as an intermediate sol-
reserved for suspension bridges [1–12]. It is extremely ution between concrete box-girders prestressed by exter-
interesting to analyse the progress in the world record nal tendons and cable-stayed bridges; and the design of
multispan cable-stayed bridges.
Finally, we shall return to the very long spans and
* Tel. and fax: ⫹ 33-1-30-88-43-44. their limits, predominantly controlled by the dynamic

0141-0296/99/$ - see front matter  1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 9 8 ) 0 0 0 2 8 - 5
738 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 2. Evolution of record spans for cable-stayed bridges.

response to turbulent wind; not forgetting the cable-


vibrations, the control of which now appears to be one
of the major problems in cable-stayed bridges.

2. New record spans

The world record for cable-stayed bridges progressed


very slowly in the 1970s and 1980s, but since the begin-
ning of the 1990s all records have been broken in a
gigantic step forward (Fig. 2):
the Saint-Nazaire Bridge in France (404 m in 1975),
with an orthotropic box-girder for the deck (Fig. 3); Fig. 4. The Barrios de Luna Bridge, in Spain. Photo: Fernandez Casa-
the Barrios de Luna Bridge-also called the Fernandez do.
Casado Bridge-in Spain (430 m in 1983), in pre-
stressed concrete (Fig. 4);
the Alex Frazer Bridge-also called the Anacis Island
Bridge-in Vancouver, Canada (465 m in 1986), with
a composite deck made of two steel I-shaped beams
supporting a reinforced concrete slab (Fig. 5);

Fig. 5. The Alex Frazer Bridge, in Canada. Photo: Buckland and


Taylor.

the Ikuchi Bridge in Japan (490 m in 1991), a com-


posite structure with prestressed concrete access
spans and with an orthotropic slab supported by two
steel box-girders in the central part of the main span
(Fig. 6);
Fig. 3. The Saint-Nazaire Bridge over the river Loire. Photo: Eiffel the Skarnsund Bridge in Norway (530 m, in 1991),
Constructions Métalliques. in prestressed concrete (Fig. 7);
M. Virlogeux / Engineering Structures 21 (1999) 737–755 739

Fig. 6. The Ikuchi Bridge, Japan. Photo: M. Virlogeux.

Fig. 8. The Yang Pu Bridge, China. Photo: M. Virlogeux.

Fig. 7. The Skarnsund Bridge, Norway. Photo: Highway Department


of Norway.

the Yangpu Bridge in Shangai, China (602 m, in


1993), with a composite deck made of two twin I-
shaped beams and with a reinforced and prestressed
concrete slab (Fig. 8);
the Normandie Bridge, in France (856 m, completed
in 1994), a composite structure again with concrete
access spans and with an orthotropic box-girder in
the central part of the main span (Figs. 9 and 10);
the Tatara Bridge, in Japan, with a similar design,
will extend the record to 890 m in 1998 (Fig. 11). Fig. 9. The Normandie Bridge. Photo: G. Forquet (SETRA).
The Normandie and Tatara bridges are the first to
enter the domain of very long spans, which was before tropic (Saint-Nazaire, Ikuchi, Normandie), pre-
exclusively reserved for suspension bridges. However, stressed concrete (Barrios de Luna, Skarnsund) and
this very large step forward is not a real surprise: composite decks (Anacis, Yangpu) was a clear indi-
cation that we were very far from the limits: to
We must not forget the major German cable-stayed
reduce the cost of cables, it is necessary to reduce
bridges built in Köln and Düsseldorf with a unique
the weight of the deck for very long spans, and to
pylon. Their long spans—302 m for the Severin
give preference to orthotropic box-girders over
Bridge in Köln, in 1959; 320 m for the Düsseldorf
800 m; as stated by René Walther, prestressed con-
Kniebrücke in 1969; 368 m for the Düsseldorf Flehe
crete decks can be economically built up to 500 or
Bridge in 1979—evidenced that cable-stayed bridges
600 m, depending on site conditions (including wind
could be built with two pylons for spans ranging
effects), and composite decks up to about 800 m.
between 600 and 700 m.
The competition, for 20 years, between steel ortho- As for the future, it is clear that much longer spans
740 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Composite decks, with two steel I-shaped beams of


limited depth, even for rather long spans, up to
602 m for the Yangpu Bridge. Limits only arise with
aeroelastic stability, due to the rather unfavourable
shape of the deck which calls for some aerodynamic
amendments (fairings on both sides to give some
streamlining; baffles between the main beams to div-
ide the open void below the deck and limit torsional
wind effects). In these bridges the main problem is
the possible development of cracks in the concrete
slab; this can be prevented by prestressing the slab
where cables do not introduce high compressive
forces.
Prestressed concrete slabs stiffened by two rectangu-
lar edge beams. Very long spans can be built with
such designs, such as the Penang Bridge in Malaysia
or the Dames Point Bridge in Jacksonville, Florida
(400 m, in 1988).
However, the greatest flexibility is reached with the
rectangular slabs designed by René Walther; the
pioneer bridge is at Dieppoldsau over the Rhine
(97 m in 1985; Fig. 12). The idea was re-used by
Fig. 10. The Normandie Bridge. Photo: M. Virlogeux.
Jorg Schlaich for the Evripos Bridge, in Greece
(215 m, in 1993), a rectangular slab 45 cm thick
(Figs. 13–16).
Since the French Code introduces extremely severe
requirements for Service Limit States in prestressed con-
crete structures, we could not design such slabs. How-
ever, for the Châlon-sur-Saône bridge—also called the
Bourgogne Bridge—we designed a cross-section with
two main ribs connected by a top slab at the upper level
for road traffic, and with a cantilevered slab on each side
at the lower level for pedestrians. We could thus have
both an open section, with an easy and economical erec-
tion, and a rather important flexural inertia for a very
limited depth of less than 1 m (Figs. 17–20).
Finally, we tried to design ribbed slabs with stream-
Fig. 11. Architect’s impression of the Tatara Bridge. Photo: Honshu lined shapes to avoid fairings and baffles, which we con-
Shikoku Bridge Authority.
sider to be the result of imperfect designs: the depth of
the ribs is limited and their width enlarged, and the trans-
than those of the Normandie and Tatara bridges will be
built in the coming decades, over 1000 or 1200 m. Two
projects have already been designed for such long spans:
for the Messina Straights long ago, by Fritz Leonhardt,
and for the Eastern Bridge of the Storebaelt by Cowi-
Consult, with a main span 1200 m long. We have been
invited to give an opinion on this last design which
looked perfectly feasible; only navigation requirements,
which called for a much longer span, prevented its con-
struction.

3. Slender decks

Recent years also show the evolution towards slender


and flexible decks for medium spans: Fig. 12. The Dieppoldsau Bridge, Switzerland. Photo: René Walther.
M. Virlogeux / Engineering Structures 21 (1999) 737–755 741

Fig. 13. The Evripos Bridge: longitudinal view.

Fig. 14. The Evripos Bridge: cross-section.

Fig. 15. The Evripos Bridge: view of the completed bridge. Photo:
M. Stathopoulos.

verse cross-beams which connect them and support the


top slab—and which are preferably steel I-shaped beams
for lighter decks—are of variable depth. We designed in
such a way—without success—a cable-stayed solution
for the Ting Kau bridge in Hong Kong. It proved
extremely efficient for aerodynamic stability (Fig. 21).
Nevertheless we must note the opinion of some
designers who consider that such flexible decks are stati-
cally unstable due to the second-order effects and to the
reduction of rigidity produced by cracks and plast- Fig. 16. The Evripos Bridge: view of the completed bridge. Photo:
ification in concrete. It should be noted that we could not M. Stathopoulos.
742 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 17. The Burgundy Bridge at Chalon sur Saône: longitudinal view.

Fig. 18. The Burgundy Bridge at Chalon sur Saône: cross-sections, typical and upon supports.

Fig. 19. The Burgundy Bridge at Chalon sur Saône: view of the com-
pleted bridge. Photo: G. Forquet (SETRA).

analyse many load cases when we controlled the Evripos


Bridge for the Greek Administration, due to our limited
computational capacities; but we had the impression that
the bridge was far from unstable. Some analyses must
be developed, with more concentrated loads than already
applied; but we have the impression that those who con-
sider stability a problem are not aware of the detailed
design of these slender bridges, aimed at perfectly bal-
ancing permanent loads and thus reducing nonlinear
effects in prestressed concrete. In addition, it is clear
that:
Fig. 20. The Burgundy Bridge at Chalon sur Saône: view of the com-
an insufficient understanding of code requirements pleted bridge. Photo: G. Forquet (SETRA).
M. Virlogeux / Engineering Structures 21 (1999) 737–755 743

Fig. 21. Cross-section of a project for the Ting Kau Bridge.

can lead to a large underestimation of the real struc- viaduct in South West France. The concept consists of
tural safety; designing very short pylons, rigidly connected to the
the computational process can be unstable when the bridge superstructure, a prestressed concrete box-girder
structure is perfectly safe. of constant depth for spans 100 m long; cables are pass-
ing on saddles on these short pylons to act as external
Great modesty is necessary in this area. The most
tendons much more than as cable-stays, since tension
positive aspect is that none of the slender bridges already
variations are presumed to be very small as compared
built shows any sign of an unexpected second-order
with those in cable-stayed bridges. These extradossed
effect.
tendons mainly act through their initial tension, produc-
ing a vertical, lifting action on the box-girder to reduce
its effective weight, as well as a compressive force as
4. Extradossed bridges
classical tendons do (Fig. 22).
It is clear that a part of the concept was based on
Since the allowable stress in cable-stays is rather low
a distorsion of code specifications, which differ greatly
(generally between 40 and 45% of the Guaranteed Ulti-
between cable-stays and tendons, but the project
mate Tensile Stress, GUTS, for the Service Limit
developed with the architect Charles Lavigne for the
States), as compared with tendons (between 60 and 72%
Arrêt Darré viaduct was extremely elegant; it would
of GUTS, to give an idea, depending on code specifi-
have been an excellent pendant to the Ganter Bridge if
cations and prestressing losses), some engineers
it had been built.
imagined solutions where cable-stays could be replaced
Logically this design inspired several other bridges,
by tendons.
mainly in Japan where extradossed bridges are now a
The pioneer bridge is the famous Ganter Bridge
specific bridge type. Among five or six other bridges,
designed by Christian Menn. The concrete box-girder is
“stayed” by prestressed concrete walls which “suspend”
it from very short pylons. However, this bridge has been
much more admired for its structural elegance and its
perfect integration in the landscape of the Swiss moun-
tains than as a technical innovation.
In fact, the installation of the “tendon stays” in con-
crete walls has some drawbacks, such as the fact that
tendons cannot be replaced as traditional stays; not for-
getting the cost of the erection of these concrete walls.
This is why very few bridges have been built in that way
after the Ganterbrücke: the Papagayo Bridge in Mexico,
the Barton Creek Bridge in Texas, USA, and the Socor-
ridos Bridge in Portugal by Reiss.
However, before these three bridges were erected, Jac-
ques Mathivat developed the concept of “extradossed” Fig. 22. Model of the extradossed bridge proposed for the Arrêt
bridges with his alternative design for the Arrêt Darré Darré viaduct. Photo for SECOA and Charles Lavigne.
744 M. Virlogeux / Engineering Structures 21 (1999) 737–755

cally, it is not logical to accept the same—and rather


high—allowable stress in all extradossed bridges when
some suffer very low stress variations under traffic loads
(since the deck is rigidly connected to piers and has a
variable depth) and when some other ones suffer high
stress variations (with a deck of constant depth and with
simple supports between piers and deck). Akio Kasuga
published a diagram which demonstrates the very large
differences in stress variation for the Japanese
extradossed bridges.
We have designed an extradossed bridge crossing the
Les Usses valley, in association with some colleagues
who later developed the project for themselves, with a
Fig. 23. The Odawara Blue Bridge. By courtesy of Mr. Akio Kasuga. constant depth and simple supports; in such conditions,
the stress variation range corresponding to traffic loads
we can cite two, the more elegant in our opinion, the is almost the same as for a prestressed concrete cable-
Odawara Blue Bridge (Figs. 23 and 24) and the Tsuku- stayed bridge and specifications must be much closer to
hara Bridge under construction in 1997. those of cable-stays than those of tendons.
Like the Arrêt Darré project, these Japanese bridges
have medium spans, between 100 and 150 m (180 m for
the longer); in our opinion this is the optimum span 5. Multiple cable-stayed spans
range for this type of bridge.
All the Japanese bridges have two lateral planes of Finally, a very recent trend must be mentioned: the
extradossed cables, one on each side of the box-girder. design of multiple cable-stayed spans.
This is not as light and elegant as with a unique plane The pioneer construction is evidently the Maracaibo
of extradossed cables in the bridge axis, but the shapes Bridge designed by Ricardo Morandi. The structural
selected for some of these bridges are almost perfect and concept is clear: a series of extremely rigid pylons—
they are a real architectural success. prestressed concrete trusses—support cable-stayed canti-
The final differences concern two important structural levers with a simple span suspended between two suc-
details: for most of these bridges, the box-girder is of cessive pylons (Figs. 25–27).
variable depth and there is a rigid connection between This static system was reproduced by Morandi with a
pier and deck. We have a preference generally for box- unique main span for two bridges in Italy and for the
girders of constant depth, but a variable depth, especially Wadi Kuf bridge in Libya, which has once been the
when associated with a rigid connection between pier world record; and also by French engineers for the
and deck, reduces stress variations produced by traffic Chaco Corrientes bridge. However, it is well adapted to
loads in extradossed cables. a succession of cable-stayed spans, since longitudinal
Clearly, specifications concerning internal tendons, length variations can develop freely at the joints between
external tendons, extradossed cables and cable-stays cantilevers and simple spans, and since the very rigid
must be more coherent, to avoid distorsions such as pylons can easily balance the effects of traffic loads
those observed with extradossed bridges. More specifi- when distributed on one complete span on one side and
later on the other.
A very recent project follows this line to resist
extremely severe earthquakes and adapt to the possible
movements of active faults: the bridge which has been
designed by GTM-Dumez to cross the Patras bay
between Rion and Antirion. Four large off-shore struc-
tures will constitute in the same time the foundation cais-
sons and piers; each will support a cable-stayed cantil-
ever, with simple spans between the cantilevers (Figs.
28 and 29). The conceptual design, which could be
improved later, exactly corresponds to the basic ideas
developed long ago by Ricardo Morandi.
Cable-stayed bridges developed with a very different
concept of a continuous deck suspended from “flexible”
pylons, following the design of the first German bridges.
Fig. 24. The Odawara Blue Bridge. By courtesy of Mr. Akio Kasuga. However, this design is not at all adapted to the erec-
Fig. 25. The Maracaibo Bridge, by Ricardo Morandi: longitudinal views, general (from The Bridge spanning Lake Maracaibo in Venezuela, Bauverlag, Wiesbaden, 1963).
M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 26. The Maracaibo Bridge, by Ricardo Morandi: longitudinal view, local (from The Bridge spanning Lake Maracaibo in Venezuela, Bauverlag, Wiesbaden, 1963).
745
746 M. Virlogeux / Engineering Structures 21 (1999) 737–755

It is clear that we have to introduce rigidity some-


where: either in the piers and pylons, with a continuous
transmission of bending moments between them, or in
the deck. Two recent projects evidence these solutions.
The first one is for the Millau viaduct: between 1989
and 1993, with Emmanuel Bouchon and Daniel Lecoin-
tre, we developed the preliminary design of a cable-
stayed bridge, 2500 m long, with seven pylons and six
main cable-stayed spans 320 m long. After the rec-
ommendation from René Walther to eliminate an inter-
mediate expansion joint, we decided to attach the deck
to all piers, which are very high, the pylons also being
rigidly connected in the deck. To allow for longitudinal
expansions and contractions, the piers at both ends were
made of two columns, longitudinally, with a single line
Fig. 27. The Maracaibo Bridge, by Ricardo Morandi: a view of the of fixed bearings on top of each for a greater flexibility.
bridge (from The Bridge spanning Lake Maracaibo in Venezuela, Bau- The central main piers, 240 m high, are rigidly connec-
verlag, Wiesbaden, 1963).
ted to the deck; they evidently have to be much stiffer
for their stability. The deck was 5.50 m deep to limit
tion of a series of cable-stayed spans, and only very lim- deflections and bending moments (Figs. 40–43).
ited applications were made: the Kwang Fu Bridge, in Between 1993 and 1994, Jean-François Klein, Pierre
Taiwan, has two successive cable-stayed spans (134 m, Moı̈a and their colleagues developed a slightly different
in 1978); the Colindres Bridge, in Spain, also has two solution for a bridge crossing the lake of Geneva. The
successive spans (125 m, in 1993). The single existing box-girder depth is much smaller: 3.50 m; however, they
bridge with multiple cable-stayed spans is the Arena designed much stronger pylons, in perfect continuity
Viaduct, designed in Spain by Juan José Arenas, which with the piers below, also made of two independent col-
has five successive cable-stayed spans, but of very lim- umns for the same reasons (Figs. 44 and 45).
ited length (105 m, in 1993; Figs. 30 and 31). There is
one example with longer spans: the Mezcala Bridge, in
Mexico (312 m, in 1993); but there are two cable-stayed 6. Dynamic response in turbulent wind
spans only, which are rather well stiffened by the back-
stays in the side-span on each side (Figs. 32 and 33). Very long cable-stayed spans are controlled by their
We must also evoke two recent bridges with a specific aerodynamic behaviour which classically has three dif-
design: the Macau Bridge, designed by José Luis Cancio ferent aspects: the effects of vortex shedding; aerody-
Martins, has two successive cable-stayed spans, but with namic stability; and the dynamic response to turbulent
a double pylon at the center; it works like two successive wind.
classical cable-stayed bridges (Figs. 34 and 35); the Ting The orientation towards streamlined box-girder sec-
Kau bridge, designed by Jorg Schlaich, has three pylons. tions inspired from the English suspension bridges, and
The stability of the central one is improved by very long with two lateral planes of cable-stays, practically solves
cables, anchored at its top on one side, and attached at the problems related to vortex shedding and aerody-
the deck level to a “side-pylon” on the other side (Figs. namic stability.
36–39). For the Normandie Bridge, for example, vortex shed-
This last solution is not extremely elegant in our opi- ding has been practically eliminated; no vortex shedding
nion, and none of the others answers the real problem could be evidenced through the measure of the Küssner
which arose with the development of very large projects coefficients in laminar wind, and only very limited ones
and of very long bridges: how to efficiently build a series could be measured, for the second vertical mode, during
of long cable-stayed spans? a taut-strip model test.
When a single span is loaded, the corresponding The aerodynamic stability is not a question either for
cables receive an important tension variation which streamlined box-girders, provided that the width to span
bends the two adjacent pylons towards the load; if their ratio is high enough, at least about 1/40.
deflection is important, the corresponding backstays— The design of very long spans is practically governed
the cables anchored in the same pylons and in the adjac- by the dynamic response to turbulent wind, which can
ent spans—lift upwards the adjacent spans. This results be now predicted with a good accuracy.
in important bending moments, alternatively in one Vertical vibrations are mainly controlled by the rigid-
direction and the other according to the loaded span, ity of cables and are very similar for the bridge in oper-
with corresponding deflections. ation and during construction. The bridge box-girder
Fig. 28. The Rion-Antirion project in Greece: longitudinal view, general. By courtesy of Dumez-GTM.
M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 29. The Rion-Antirion project in Greece: longitudinal view, local. By courtesy of Dumez-GTM.
747
748 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 30. The Arena viaduct, Spain. By courtesy of Freyssinet and Fig. 33. The Mezcala Bridge, Mexico. Completed bridge. By cour-
Juan José Arenas. tesy of Alain Chauvin.

Fig. 34. The Macau Bridge. Longitudinal view.

sional rigidity of the box-girder; the lateral planes of


cable-stays provide additional rigidity, between 10 and
20%, for example, in the Normandie Bridge. However,
it appears extremely hazardous to think of a unique axial
plane of cables, as has been done for more limited spans
Fig. 31. The Arena viaduct, Spain. By courtesy of Freyssinet and
Juan José Arenas. (450 m in the Chao Phraya Bridge in Bangkok, now the
longest in the world).
Transverse vibrations (sway, also named
must resist the induced bending moments; a structural balancement) are only controlled by the deck inertia and
depth between 3.00 and 4.00 m—depending on wind also by the connection conditions between deck and
conditions and on specific design provisions—appears pylons, and deck and intermediate piers in the side-
perfectly adapted. spans, if any. When the deck can be rigidly connected
Torsional vibrations are mainly controlled by the tor- to the pylons, or when the deck transverse displacements

Fig. 32. The Mezcala Bridge, Mexico. Longitudinal view.


M. Virlogeux / Engineering Structures 21 (1999) 737–755 749

Fig. 35. The Macau Bridge. Completed bridge. By courtesy of Freys- Fig. 38. The Ting Kau Bridge. Construction situation of the bridge.
sinet and José Luis Cancio Martins. By courtesy of Freyssinet and Bureau Schlaich and Partners.

Fig. 36. The Ting Kau Bridge. Longitudinal view.

Fig. 37. The Ting Kau Bridge. Construction situation of the bridge. By courtesy of Freyssinet and Bureau Schlaich and Partners.

are prevented at the pylons as well as on intermediate Of course, construction situations are much more sev-
supports in the side-spans (and provided that these sup- ere for transverse vibrations than in the completed
ports are rigid enough), supports help limit sway bridge. This is the main drawback of cable-stayed
vibrations, as for, example, in the Normandie Bridge and bridges as compared with suspension bridges. Tune mass
the Skarnsund Bridge. The width to span ratio can then dampers, or other stabilization systems, can be used to
reach 1/40. With more classical conditions, the deck damp or reduce sway movements during erection, as was
must be slightly wider to resist high wind forces. done for the Normandie Bridge; but the tune mass dam-
750 M. Virlogeux / Engineering Structures 21 (1999) 737–755

with a very low width to span ratio, either due to the


span or to a rather limited width (Fig. 46).
The Ting Kau bridge is not exactly in this vein, since
the division of the superstructure into two separate decks
aimed at erecting a series of unique axial pylons. How-
ever, this division certainly helped the aerodynamic stab-
ility in torsion, despite the unfavourable shape of classi-
cal composite decks with two steel I-girders.
Finally, we must mention the decrease in rigidity of
cable-stays for increasing spans, due to sag effects, as
considered by Leonhardt long ago. For his cable-stayed
project in the Messina Straights, he proposed stiffening
main cables by the installation of cross-cables (also
called aiguilles).
Aiguilles have been installed in some cable-stayed
bridges, in Denmark, Japan and France, but for a very
different reason as we shall see: to limit cable-vibration.

7. Cable vibration

Many cable-stayed bridges had to suffer important


cable vibrations, beginning with some of the early Ger-
man bridges and with the Brotonne Bridge in 1977.
Fig. 39. The Ting Kau Bridge. Construction situation of the bridge.
By courtesy of Freyssinet and Bureau Schlaich and Partners. Unfortunately, owners, designers and contractors gener-
ally hide these problems and engineers had to wait years
before they could begin to understand and solve them.
pers installed on the cantilevers never or hardly ever It should be stated that many factors can produce cable
moved, since the wind velocity never reached high vibrations. We cannot detail them here, but only give a
values during erection. general view of the possible phenomena.
A new solution has been proposed—basically for very The first one is classical: vortex shedding. The fre-
long span suspension bridges, but it is well adapted to quency of the exciting force is given by:
cable-stayed bridges—which consists of designing two
parallel, streamlined box-girders connected by a series of US
cross-beams. The distance between the two box-girders N⫽
produces the necessary inertia and rigidity to resist high D
winds with very long spans; in the same time, the tor-
sional aerodynamic damping becomes very high, since where U is the wind velocity, D the cable diameter and
a torsional vibration of the whole deck produces vertical S the Strouhal number, equal to 0.18 for a circular cylin-
displacements of the two individual box-girders (in der. Vortex shedding can thus excite a cable on its mode
opposite directions) and correspondingly high damping. k (the period of which is Tk) when:
However, this solution—proposed for crossing the Mes-
sina Straights—has never received applications; it can D
U⫽
be useful only for bridges (suspended or cable-stayed) 0.18•Tk

Fig. 40. The Millau viaduct; longitudinal view of the SETRA preliminary project.
M. Virlogeux / Engineering Structures 21 (1999) 737–755 751

but they do not appear dangerous except if buffeting can


produce shocks in cross-cables, if any, when they have
received very low tensions.
However, buffeting can produce a specific type of
aerodynamic instability in bridges with two parallel
planes of stay-cables: gusts striking the upwind plane of
stays will strike the downwind stays at a time B/U later,
where B is the distance between the two planes of stays.
If this difference corresponds to half a cycle of torsional
movement of the deck (the period of which is Tt), an
instability can take place. The critical velocity is then
Fig. 41. The Millau viaduct; cross-section of the SETRA prelimi-
nary project. given by:

2B
U⫽
Tt

Wake effects are of different types. Cables can be in


the wake of other structural elements, such as a tower
leg or construction equipment during erection, and thus
in the vortices produced by this other structural element.
The critical velocity is then given by vortex shedding on
this element:

H
U⫽
S•Tk

where H is the transverse dimension of the waking


element, S the corresponding Strouhal number and Tk the
Fig. 42. The Millau viaduct. An architect’s impression of the
awarded project. By courtesy of Sir Norman Foster and Partners.
period of the cable on mode k.
Another very classical situation comes when cables
are installed in pairs. The second cable is in the wind
that is only applicable for very low wind velocities. of the first one and then excited by the vortices produced
Since energy is then very low, vortex shedding is not a on the first one; but the movements of the second cable
real danger for cables. then disturb the flow around the first one and both cables
Buffeting has direct effects on cables, as on any flex- vibrate. This phenomenon decreases and disappears
ible structure. These effects grow with the wind velocity, when the distance between cables increases.

Fig. 43. The Millau viaduct. An architect’s impression of the awarded project. By courtesy of Sir Norman Foster and Partners.
752 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 44. The project across the Lake of Geneva: longitudinal view. By courtesy of Bureau Tremblet.

Fig. 45. The project across the Lake of Geneva: cross-section. By courtesy of Bureau Tremblet.

Fig. 46. Twin stream-lined box-girders for long spans.

Of course, the same phenomenon develops when sev- cable tension (in Newtons). This formula does not apply
eral cables are grouped, for example to constitute a for the first vertical mode due to sag effects; it becomes:
strong back-stay.

冪冉
A specific problem of the same type exists when
cables are made of a bunch of parallel, individually pro- m


T1,␽ ⫽ 2ᐉ
tected strands, a technique developed by Freyssinet. Due ES ␲2f 2
to interaction between strands, the external ones can F 1⫹
F 2ᐉ2
move outwards and then inwards, shocking the central
ones; and finally producing some limited global where S is the cable area, E the physical modulus of
vibration of the cable. These movements—which we call elasticity, and f the sag counted perpendicularly to the
breathing of strands—and the shocks produce an segment joining anchorages, which can be estimated by:
important and disagreeable rattling noise.
Resonance can amplify in cables the vibrations of 4pᐉ2 cos␣
f ⫽
deck or towers when the natural frequencies of cables ␲3F
are close to some of the structural ones.
The period of natural vibration for mode k is given where p ( ⫽ mg) is the lineic weight of the cable and
by the classical formula: ␣ its inclination angle.
The amplification in a cable, for its mode k (the pul-
sation of which is ␻k), of a structural vibration corre-
冪F
2ᐉ m sponding to a pulsation ␻ is given by:
Tk ⫽
k
2␻ 2
␳⫽ H(␻)
where ᐉ is the cable length, m the lineic mass and F the k␲␻2k
M. Virlogeux / Engineering Structures 21 (1999) 737–755 753

with Severn, Burlington, Baytown, Glebe Island and Erasmus


bridges, as well as in several Japanese bridges and
1 many others.
H(␻) ⫽
冪冉1 ⫺ 冉␻ 冊 冊 冉 冉 冊冊
Countermeasures are of different types, some being
␻ 2 2
␻ 2
⫹ 2␰ specifically adapted to a special type of cable vibrations,
k ␻k others being more widely efficient.
To avoid breathing of a group of strands, strands can
where ␰ is the cable damping (ratio to critical). be attached together at small intervals; but the best idea
If the peak can be very high (reaching 1/k ␲ ␰ when is to install them in an external duct which, at the same
␻k is equal to ␻), the amplification ratio can be high time, decreases the drag coefficient and aerodynamic
only when ␻k is very close to ␻ (by less than 1%). It forces.
must also be noted that, except at peak, the amplification Wake effects in twin cables—or in a group of several
ratio is not influenced by the value of the damping coef- cables—can be eliminated by attaching cables. This was
ficient. efficiently done, for example, for the backstays of the
Another family of cable vibrations corresponds to Erasmus Bridge. For the cable-stayed bridges of the
aerodynamic stability. The shape of cables can produce Kojima-Sakaide route of the Honshu Shikoku link, twin
cable galloping. For example, to reduce drag forces on cables were attached by aiguilles with such a design that
cables—which become very high for long spans—some damping was introduced into the system. It could not
engineers proposed to group the individually protected totally avoid cable vibrations, however, and due to a
strands of the Normandie Bridge in a flat hexagonal rather low tension one of the cross-cables broke.
arrangement. However, this would have produced lift Dampers of different types can help to reduce many
forces with oblique winds and potential galloping effects. It is generally considered that rain/wind-induced
effects; we preferred installing the strands in a circular vibrations—the most frequent problem—can be elimin-
duct to avoid this possibility. ated when the cable damping is over 0.5% (ratio to
Cable-strand effects, when there is no external duct critical). In our opinion, an important damping in cables
around a large strand, can produce specific vibrations. is a decisive element of the design; damping must be
Ice on cables may be dangerous also, as on electric lines, considered as necessary in any case.
since ice may change the cable shapes and produce We must remember that damping is not efficient at
unstable forces. Even the classical circular shape appears reducing the amplification of structural vibrations except
elliptical in oblique winds, and some consider that this at peak. In this situation the designer must avoid having
could explain some specific vibrations. vibration periods of cables too close to those of the struc-
Finally, the influence of the Reynolds number on the ture, analysed with simple linear models.
drag coefficient can produce what is called “drag crisis It is necessary, for large bridges, to compute fre-
vibration”. The drag coefficient drops down from 1.20 quencies considering cables as flexible structural
to about 0.50–0.60 when the Reynolds number increases elements, with their distributed mass and their sag effect;
from 2·105 to 5·105 (depending on rugosity); a limited a clear analysis of possible interaction between cables
increase in the wind velocity—in the range of those pro- and structure must be developed to avoid problems.
duced by the wind turbulence—can produce a rapid The only solution to radically change the vibration
reduction of the drag force acting on the cable and create periods of cables consists of tying them by cross-cables,
a backwards movement. The movement back and forth as we did for the Normandie Bridge. However, these
produces a change in the relative wind velocity and aiguilles must receive rather high tension to avoid shocks
maintains the vibration. in high turbulent winds, and damping must be introduced
Nevertheless, many of cable-stayed bridges had to in the cross-cables to improve the system and limit
suffer from rain- and wind-induced vibrations. As evi- dynamic effects.
denced by Japanese researchers, two water rivulets flow However, cross-cables were also used—or pro-
down a cable, a stable one at the lower edge of the cable, posed—for another purpose: to increase the stiffness of
and an unstable one on the upper edge, the dynamic the cable-staying system as proposed by Leonhardt; to
equilibrium of which is governed by capillarity forces, avoid wake effects in twin cables, as already mentioned
weight and wind forces. The existence and the move- for the Honshu-Shikoku project, but also for the Yobuko
ments of these water rivulets flowing down cables Bridge; and finally to avoid rain/wind-induced vibration
change their shape and produce aerodynamic forces and in the Faro Bridge, Denmark. For the latter, very small
vibrations under specific conditions. aiguilles were installed, with low tension; due to the
This problem has been extensively analysed and the dynamic effects of turbulent wind, one of them broke
vibration conditions are well known. They are deemed and they were replaced by stronger ones with higher ten-
responsible (at least partly) for vibrations which sion.
occurred in the Brotonne, Ben Ahin, Wandre, Second The last countermeasure is specifically adapted to the
754 M. Virlogeux / Engineering Structures 21 (1999) 737–755

Fig. 47. A typical section of a cable for the Higachi-Kobe Bridge. Fig. 49. A stay-cable with its aerodynamically shaped duct, and a
Photo: Vu Bui. cross-cable. Photo: G. Forquet (SETRA).

elimination of rain/wind-induced vibrations. It consists


of shaping the ducts in such a way that the water rivulets
on the cable are either eliminated, or channelled in such distribute “dimpels” on the cable ducts. Japanese Engin-
a way that they cannot produce vibration, or disor- eers are confident in the results of this countermeasure,
ganized so that forces cannot be coherent and produce in conjunction with the installation of dampers, since the
vibrations. design is without any cross-cable (Fig. 50).
The first application was for the Higachi Kobe Bridge Finally, we noticed that many of the bridges which
in Japan. Deep longitudinal channels were created in the experienced large cable vibrations have a deck which is
High Density Polyethylene ducts, but with an important not streamlined, which is more or less sensible to the
drawback: the drag coefficient increased up to about 1.35 development of vortices for wind velocities between 10
(Fig. 47). and 20 m/s. A systematic analysis of cable-stayed
For the Normandie Bridge the problem was high- bridges, based on the results of their wind-tunnel tests
lighted by Cowi Consult which worked as the design and the evaluation of the vibration periods of cables
office for the steel contractor, Monberg and Thorsen. related to the dynamic behaviour of the structure, could
The solution—suggested by an engineer of the Danish help to decide whether structure aerodynamic behaviour
Maritime Institute—was tested by the CSTB at Nantes influences cable vibrations. We consider it likely, and
and is now systematically used by Freyssinet. It consists this is why we prefer designing well streamlined cross-
of creating two imbricated helical fillets, slightly more sections. Unfortunately, they are generally slightly more
than 1 mm deep and each with a pitch length of 60 cm; expensive and not always successful in economic com-
the drag coefficient is only 0.63 in the range of natural petition.
wind velocities (Figs. 48 and 49).
For the Tatara bridge, Japanese Engineers decided to

Fig. 48. The test for the ducts of the Normandie Bridge. Photo: Fig. 50. A typical view of a cable for the Tatara Bridge, with its
C.S.T.B., Nantes. dimples. By courtesy of S.E. Corporation.
M. Virlogeux / Engineering Structures 21 (1999) 737–755 755

8. Conclusion [3] Troitsky MS. Cable-stayed bridges. Oxford: BSP Professional


Books, 1988.
[4] Gimsing NJ. Cable-supported bridges. Chichester: John Wiley,
The evolution of cable-stayed bridges is proceeding 1983.
rapidly in many different directions: the development of [5] Cable-stayed bridges. Experience and practice. In: Kanok N, edi-
slender and flexible decks which open many possibilities tor. Proceedings of the International Conference on cable-stayed
bridges. Bangkok, November, 1987.
for medium spans, including competition with other [6] Seminar on cable-stayed bridges. Proceedings of the Seminar
bridge types; the application of cable-stayed bridges to organized by the Indian group of IABSE. Bangalore, October,
multiple spans, which will certainly have some impor- 1988.
tance for large projects; and, of course, the rapid increase [7] Innovation in cable-stayed bridges. In: Otsuka H, editor. Proceed-
in span length, in competition with suspension bridges. ings of the International Symposium for innovation in cable-
stayed bridges. Fukuoka, April, 1991. Maruzen.
However, in this last domain, the condition is a complete [8] Cable-stayed bridges. Recent developments and their future. In:
mastering of wind forces, for the bridge as well as for Ito, Fujino, Miyata and Narita, editors. Proceedings of the Sem-
cables. inar. Yokohama: Elsevier, December, 1991.
[9] Aerodynamics of large bridges. In: Larsen A, editor. Proceedings
of the first International Symposium of Aerodynamics of Large
Bridges. Copenhagen: Balkema, February, 1992.
References [10] 1994 International Symposium on cable-stayed bridges. Proceed-
ings. Shanghai.
[11] Ponts suspendus et à haubans. Proceedings of the International
[1] Walther R, Houriet B, Isler W, Moı̈a P. Ponts haubanés. Laus- Conference on Cable-stayed and Suspension Bridges. Deauville,
anne: Presses Polytechniques Romandes, 1985. October, 1994.
[2] Podolny W, Scalzi JB. Construction and design of cable-stayed [12] International Symposium on Cable Dynamics. Proceedings.
bridges. New York: John Wiley, 1976. Liège, October, 1995.

You might also like