Conceptual Seismic Design of Cable-Stayed Bridge

This article was downloaded by: [University of Arizona]
On: 12 January 2013, At: 20:34

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of Earthquake Engineering

Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/ueqe20
Conceptual Seismic Design of Cable-

Stayed Bridges
a a b
G. M. Calvi , T. J. Sullivan & A. Villani
a
Department of Structural Mechanics, University of Pavia, Pavia,
Italy
b
EQCO s.r.l., Pavia, Italy
Version of record first published: 30 Sep 2010.
To cite this article: G. M. Calvi , T. J. Sullivan & A. Villani (2010): Conceptual Seismic Design of
Cable-Stayed Bridges, Journal of Earthquake Engineering, 14:8, 1139-1171
To link to this article: http://dx.doi.org/10.1080/13632469.2010.505275
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Journal of Earthquake Engineering, 14:1139–1171, 2010
Copyright © A.S. Elnashai & N.N. Ambraseys
ISSN: 1363-2469 print / 1559-808X online
DOI: 10.1080/13632469.2010.505275
Conceptual Seismic Design of Cable-Stayed Bridges
G. M. CALVI1 , T. J. SULLIVAN1 , and A. VILLANI2

1
Department of Structural Mechanics, University of Pavia, Pavia, Italy
2
EQCO s.r.l., Pavia, Italy
Cable-stayed bridges can be a very effective means of bridging large distances in both seismic and
non-seismic regions. Their design, analysis, and construction can be very challenging, and fortu-
nately there is a considerable amount of literature that can assist engineers with both the analysis and
Downloaded by [University of Arizona] at 20:34 12 January 2013
detailed design of cable-stayed bridges. It is less common, however, to find simple recommendations
for the conceptual design of cable-stayed bridges, in particular for seismic loading. As such, this arti-
cle reviews and discusses some of the important conceptual design considerations for cable-stayed
bridges, first for gravity loads and then for seismic excitation. The advantages and disadvantages
of different cable-stayed bridge solutions are highlighted, with review of deck sections, tower con-
figurations in both the longitudinal and transverse direction, deck-to-pier connections, and cable
arrangements, amongst other things. Reference is made to a number of real cable-stayed bridge
solutions. After reviewing the important conceptual design considerations for cable-stayed bridges,
a simple preliminary sizing procedure is proposed. The preliminary sizing procedure is intended to
offer designers a quick but rational means of identifying reasonable member sizes for cable-stayed
bridges that should then be verified through advanced analyses in the developed and detailed design
stages of the project. A case study application of the preliminary sizing procedure is made for a three-
tower cable-stayed bridge in Ecuador, and by comparing preliminary and final design member sizes
it is concluded that the preliminary sizing procedure may be a useful tool for design.
Keywords Cable-Stayed Bridge; Conceptual Design; Seismic Design; Direct Displacement Based
Design
1. Introduction
As demands for improved infrastructure increase around the world, civil engineers con-
tinue to be challenged to develop large bridges that must perform well even under extreme
loading. An effective means of bridging large distances in both seismic and non-seismic
regions is through the use of cable-stayed bridges. The typical elevation of a cable-stayed
bridge is shown in Fig. 1. The decks of a cable-stayed bridge are supported using cables
that climb diagonally to strong stiff towers, which act as the main load-bearing elements for
the bridge. The orientation and construction methodology adopted for the bridges is such
that under uniform loading the static horizontal forces imposed by the cables on the decks
are typically balanced. Consequently, the towers will be designed to resist the vertical com-
ponent of the gravity load and additional lateral loads associated with live loads, wind and
seismic actions, impact from colliding objects, drag from water flow, and possibly others.
The structural form and detailed behavior of cable-stayed bridges is well understood
and there is a considerable amount of literature on cable-stayed bridges, including the books
by Troitsky [1988], Gimsing [1997], and Walther et al. [1999]. There are fewer publications
Received 27 February 2010; accepted 26 June 2010.

Address correspondence to G. M. Calvi, Department of Structural Mechanics, University of Pavia, Via
Ferrata 1, 27100 Pavia, Italy; E-mail: gm.calvi@unipv.it
1139
1140 G. M. Calvi, T. J. Sullivan, and A. Villani
FIGURE 1 Typical elevation of a cable-stayed bridge.
that consider conceptual design approaches, but a useful discussion and review of differ-
ent cable-stayed bridge configurations is provided by Virlogeux [2001]. From a seismic
engineering perspective, there are many interesting publications that focus on the seismic
control of cable-stayed bridges [Ali and Abdel-Ghaffar, 1995; He et al., 2001; Turan, 2001;
Park et al., 2003, 2005; Jung et al., 2003; Ruangrassamee and Kawashima, 2006; Chang
and Loh, 2006] and also on aspects of modelling and analysis [Kawashima et al. 1988,
1990; Wilson and Gravelle, 1991; Schemmann and Smith, 1998a,b]. However, there are
a number of conceptual design decisions that can be made for cable-stayed bridges early
in the design process that will significantly influence their behavior in earthquake events.
In this article, various conceptual design issues are examined and discussed in order to
identify optimal seismic design strategies for cable-stayed bridges.
2. Performance Objectives and Design Loads

Large bridge structures involve the investment of a great deal of time and money and are
usually very important for the infrastructure of a community and depending on the seis-
micity of the site in question, seismic loads can also be a very important load case for such
bridges.
Performance-based seismic design (PBD) is a framework for seismic design that
has continued to increase in popularity since the early nineties (for example, see Fajfar
and Krawinkler 1997; FEMA 445, 2006]. A number of bridges have been designed
using performance-based design concepts, such as the Tacoma Narrows Parallel Crossing
[Arzoumanidis et al. 2005]. The aim of PBD is essentially to offer engineers an effective
means of controlling the risk posed by earthquakes. As part of a PBD process, engineers
ensure that certain performance levels (or damage states) are satisfied for different design
ground motion intensities, such as those proposed in the PBD matrix of Table 1. With
knowledge of the probability of exceedence of different ground motion intensities, the
TABLE 1 Performance-based design matrix (adapted from Priestley et al., 2007) defining
the earthquake design intensity for different structural categories and different performance
levels
Earthquake design intensity
Importance class∗ Level 1 Level 2 Level 3
I Not Required 50% in 50 years 10% in 50 years
II 50% in 50 years 10% in 50 years 2% in 50 years
III (includes Normal Bridges) 20% in 50 years 4% in 50 years 1% in 50 years
IV (includes Vital Bridges) 10% in 50 years 2% in 50 years 1% in 50 years
∗
refer EC8.1 T4.3 for detailed definitions of importance class.
Conceptual Seismic Design of Cable-Stayed Bridges 1141
engineer can control the risk posed by an earthquake, if they can successfully limit the
structural response under a given ground motion to prescribed performance limits.
Important performance states typically considered for the performance-based seismic
design of bridge structures include:
● Serviceability limit state (Performance Level 1) in which no elements suffer notable

damage and the bridge continues to function normally following the earthquake
without any need for repairs.
● Repairable damage (Performance Level 2) in which inelastic response may develop
in pre-selected and adequately detailed plastic hinge zones. The plastic mechanism
should be carefully selected such that eventual repairs do not require closure of the
bridge. The cover concrete in the plastic hinge zones may spall and the longitudinal
reinforcement may yield but strains are limited to moderate values (without risk of
re-bar buckling). Joints in the deck may be damaged but must remain traversable by
emergency services and should be easily repaired.

● Collapse prevention (Performance Level 3) in which significant damage may
develop in pre-selected and adequately detailed plastic hinge zones, and minor
localised damage may occur in other parts of the bridge. The damage may require
closure of the bridge to repair, but collapse of the bridge is avoided. Deformations
in plastic hinge regions could result in significant concrete spalling and longitudinal
re-bar strains may be close to critical values for the onset of re-bar buckling. Joints
in the deck may be damaged but should remain traversable by emergency services.
One may also consider other intermediate limit states, depending on the needs of the client
and local authorities. Selection of an appropriate nonlinear mechanism (typically referred
to as an inelastic mechanism, but may also involve nonlinear elastic response) is key for
the adequate performance of the bridge. In order to ensure functionality, the bridge deck,
cables, and connections are usually always designed to respond elastically or with minor
amounts of inelasticity (e.g., partial yielding of a steel beam section or small inelastic
demands on a limited number of reinforcing bars at the perimeter of a large concrete sec-
tion). For the collapse prevention limit state, some design strategies will permit individual
cables to exceed the yield strength, provided that there is still adequate protection against
collapse through redistribution of loads to adjacent cables. This approach is similar to an
impact loading scenario in which it might be assumed that due to vehicle collision, one
or more cables fail and the bridge must be verified to sustain the likely gravity loads in
its damaged state. In light of these considerations, a desirable mechanism for the lateral
response of a cable-stayed bridge under extreme loads will typically involve formation of a
flexural-plastic hinge in the main piers and fuses at connections between the deck and piers.
Another alternative, that could also be incorporated into the pier-yielding mechanism, is to
design and detail foundation systems at the base of the towers that limit seismic forces; this
option will be discussed further in later sections. Joints at the deck-abutment connections
will usually be detailed to sustain large displacements without offering resistance.
A design seismic intensity is assigned for each of the important performance levels.
As illustrated in the PBD matrix of Table 1, the acceptable probability of exceeding a given
performance state for important structures (Importance Class III & IV) is lower than for
normal structures. Note that the probability of exceedence for a certain period of time can
be directly correlated to an equivalent earthquake return period using Eq. (1).
p = 1 − e−LN , (1)
where N is the inverse of the return period (Tr = 1/N), p is the probability of exceedence,
and L is the design life considered. As such, a design intensity event having 10% probabil-
ity of exceedence in 50 years is equivalent to what is commonly referred to as a 475-year
return period event, and a design intensity having 2% probability of exceedence in 50 years
is equivalent to a 2,475-year return period event. Cable-stayed bridges are typically clas-
sified as important structures, with design intensities in the range of those indicated for
Performance Class III and IV of Table 1. However, minimum design intensities for cable-
stayed bridges do not yet appear to have been set with designers typically consulting with
clients in order to set suitable probabilities of exceedence for design.
Care should be taken not to lose sight of the potential that performance-based design
offers for the control of risk. PBD should not simply be considered a framework that
requires performance criteria to be checked for a number of different earthquake inten-
sities. If it were, one could argue that performance-based design has been going on since
the 1970s, from the time that either gravity, wind, or earthquake loads were checked for
both serviceability and ultimate limit states. Seismic risk can be defined [Fournier d’Albe,
1982] as the convolution of seismic hazard and the vulnerability:
Seismic Risk = Hazard x Vulnerability (2)
There are also alternative, expanded definitions of seismic risk in the literature, such as that
proposed by Grant et al. [2006] which introduces Exposure and Cost, but this basic defini-
tion is sufficient to introduce the concepts of risk here. The hazard considers the probability
of different levels of seismic intensity at a site, whereas the vulnerability indicates the sus-
ceptibility of a building to sustain losses due to ground shaking. If the risk due to natural
hazards is to be insured, then it would appear that the acceptable seismic risk in a region
of low wind demands (for example) should be greater than that of a region of high wind
demands. Nonetheless, it appears that current design tools do not yet permit practicing
engineers to rationally control the risk posed by natural hazards in such a refined manner.
This may be due to the difficulty in quantifying losses for a given performance level. As a
result, current practice is to undertake the performance-based design of bridges without the
explicit calculation of the seismic risk.
Design and verification of a bridge for a range of performance levels requires dif-
ferent representations of the seismic loads and this is typically done through reference to
response spectra. For cable-stayed bridge structures it is important that the seismic hazard
assessment used to set the design spectrum is applicable out to long periods, as noted in
Fig. 2, since the flexible structures are often characterised by fundamental periods in excess
of 7.0 s.
Of particular significance to the seismic demands on large cable-stayed bridges will be
the peak spectral displacement demand, D , and the spectral displacement corner period,
TD , both of which are indicated in Fig. 2. Traditionally, code design spectra did not define
these parameters and instead displacement demands were effectively assumed to increase
linearly beyond TD at a slope equal to that in the TC -TD range. More recently, however,
the EC8 [CEN EC8, 2004] has indicated values for TD around 2s, whereas the IBC2006
[ICC, 2006] indicates values from 6–12 s. As argued by Faccioli et al. [2004], it appears
that the corner period and spectral displacement demand are principally dependent on
the earthquake magnitude and distance from the site. As the fundamental mode response
will be significantly affected by these characteristics of the design spectra, it is important
that seismic hazard assessments assess and define these long-period response-spectrum
characteristics carefully.
Spectral displacement
demands become large at
Spectral Displacement
Spectral Acceleration
Care required to long periods
ΔD
accurately define long-
period spectral demands
for bridge design
TC TD TC TD TE TF
Period Period
(a) (b)
FIGURE 2 Typical shape of (a) design acceleration spectrum, (b) design displacement
spectrum (note that the long period portion of the displacement spectrum has been con-
densed to fit within the page as the TD to TE period range is often several times greater than
the TC to TD period range) .
Finally, in addition to the performance requirements specified for the working life of
the bridge, one may also need to consider the risk posed by earthquakes during the construc-
tion stage. Due to the low probability of experiencing an earthquake during the relatively
short construction period, the design seismic intensity for these checks can be rather low.
If one wishes to limit the risk to levels similar to those accepted during the working-life of
the bridge, then a suitable return period event can be established using Eqs, (1) and (2). It
should be noted that as the bridge structure will evolve in different stages, it will not usu-
ally be necessary to consider the whole construction period in identifying the design return
period, but rather the maximum likely period for each construction phase. Consequently,
seismic loads are reduced and the seismic design intensity for a 6-month construction stage
may be around 20% of the collapse-prevention seismic intensity of the finished bridge.
While the seismic demands considered for each of the different construction phases may
be rather low, the seismic capacity of the unfinished bridge can also be rather low, depend-
ing on the construction methodology. As the construction will typically involve cantilever
decks from either side of a central tower pier, either lateral propping of the deck or a rigid
connection to the tower may be required to avoid in-plan torsion problems until the adja-
cent spans are connected. Temporary propping of the pier legs may also be necessary and
some reflection should be given as to the appropriate design loads for temporary works
structures. Clearly, the options for propping could be significantly affected by the site
characteristics.
3. Important Conceptual Design Considerations for Non-Seismic Loads

There are different ways in which the requirements of non seismic load cases can affect the
seismic design solution. Gravity loading on the bridge will have a strong influence on the
peak compression forces that develop in the longitudinal axis of the deck. Therefore, one
might consider a strong heavy deck to resist the compression but of course the gravity loads
themselves are a function of the weight of the deck. In addition, while a heavy deck may
not be problematic for wind loading, it would certainly be an issue for seismic loading. As
such, the ideal deck section is a strong but lightweight deck.
One might also consider the benefits of a flexurally stiff deck for non-seismic loads.
An extensive parametric study undertaken by Walther et al. [1999] showed that for static
loads the use of a stiff deck section is not ideal for cable-stayed bridges, since it attracts sig-
nificant bending moments at three critical zones: deck-to-pier, abutments, and mid spans
(also see Virlogeux, 2001). However, this observation comes from static considerations,
and a stiff deck can instead be quite beneficial when dynamic wind and earthquake loads
are considered. Also note that vertical displacements of the deck may be most significantly
affected by the stiffness of the cable-pier system. This can be appreciated simply consid-
ering the typical span-to-depth ratios of decks in cable-stayed bridges, which tend to be
in the range of 100–200, well above normal ratios used to control deformations associated
with beam flexure. As such, the deck stiffness will be more relevant for local deformations
between cable support points and for dynamic vibrations associated with wind and seismic
response.
As a result of the above considerations, composite decks, of the form shown in Fig. 3,
are a relatively common solution for cable-stayed stayed bridges. The use of composite
construction has been shown to provide an economical solution for cable-stayed bridges
with spans running from 350 to around 600–700 m [Schlaich, 2001]. Composite deck sec-
tions will typically consist of two longitudinal beams, aligned with the cable connection
points. The spacing of cables should be set with due regard to construction lifting and
transport requirements for the main beams, in addition to limiting static flexural demands
on the main beams. The transverse beams are required to transmit gravity loads to the
main beams, but can also be an effective means of stabilizing the main beams against
lateral torsional buckling. Longitudinal and transverse stiffeners may often be necessary
for the longitudinal beams but do add cost to the construction and therefore one might
consider the use of stockier sections in order to avoid the need for a large number of
stiffeners.
The gravity loads are also likely to impose the greatest axial loads on the piers and
foundations. However, as both the piers and foundations of cable-stayed bridges tend to
be massive structures in order to provide adequate lateral stiffness for wind, earthquake,
and eccentric gravity loading, the vertical loading for these elements is not usually criti-
cal. Eccentricity in vertical loads along the length of the deck can require the following
considerations to be made:
Cables Aligned with

Main Beams
Composite Concrete Deck

(either pre-cast planks or cast-in-situ)
Secondary steel beams to transfer

gravity loads and brace main beams
Note that longitudinal and transverse

Main Longitudinal stiffeners are often required but are
Steel beams not shown for clarity
FIGURE 3 Typical composite deck section of a cable-stayed bridge.

● Deck continuity: The deck should be continuous in order to permit redistribution of

loads to different bridge segments and piers, and to aid overall stability.
● Decompression of cables: The sagging of one bridge deck segment could cause the
adjacent segment to go into hogging, with a resulting reduction in cable tension
force.
● Horizontal deck restraint: For single pier bridges the unbalanced compression forces
in the deck imply that an external reaction is required. This could either be provided
through a horizontal restraint at the deck-pier interface, or with a connection at the
abutments. However, the deck-pier connection is preferred for temperature effects
and differential vertical movements of nearby cables relative to the abutments.
● Flexural response of towers: The eccentric load on the deck will typically imply
that the piers are subject to bending and shear. For this reason, gravity load
considerations imply that the piers should possess high lateral stiffness.
● Deck vertical displacements: The eccentric load on the deck can produce large deck
displacements and is another motivation for a stiff tower and a limited cable spacing.
Note that for relatively common three span bridges the “back stay effect” can help
control the displacements without needing very high tower stiffness (see Virlogeux,
2001).
Wind loading can prove to be a critical non-seismic load case in the transverse direction.
Unfortunately, due to the aeroelastic response of the bridge structures, it is typically nec-
essary to undertake wind-tunnel testing to properly gauge the wind loads and dynamic
effects due the wind. Important characteristics of the bridge will be the deck stiffness (lat-
eral, vertical, and torsional) as well as the shape of the deck section itself, with sharp angled
sections (such as box-girders) being more likely to evoke instability effects such as vortex
shedding. For a general text on wind effects on structures see Simiu and Scanlan [1996].
Practical relationships for preliminary control of the bridge deck aerodynamics are pro-
vided in Walther et al. [1999]. Conti et al. [1996], Ge and Tanaka [2000], and Jones and
Scanlan [2001] discussed specific aspects of wind response for cable-stayed bridges.
The ideal pier configuration deserves some consideration. Ideally, the pier head will
be positioned centrally above the deck so that under gravity-only loads the cables do not
induce bending in the piers. As most bridges have single decks, then common pylon shapes
are as shown in Fig. 4a–d. Centrally located piers are also possible, as in the case of the
Millau Viaduct, France, which has a configuration similar to that in Fig. 4e. Note that for
decks supported by a single central line of cables such as that of Fig. 4e, increased torsional
stiffness of the deck section may be necessary and pre-cast box sections are sometimes used
for this purpose. When two decks are desired, or a very wide deck solution is desired, there
is also the possibility of using a centrally located pier with angled arms tied at their top, as
shown in Fig. 4f.
Longitudinally, there are also different options available for the tower configuration,
although the variation in typical configurations is reduced. Figure 5 shows some typical
configurations in the longitudinal direction.
From a gravity load point of view, stiffer pier configurations are advantageous since
vertical displacements of the deck due to eccentric gravity loads will be greatly affected by
the lateral stiffness of the piers. Such considerations may motivate designers to use inclined
piers, and by coupling piers designers can also maximise the lateral stiffness. With this in
mind, one could argue that the configurations shown in Figs. 4a, c, and d, as well as those
in Figs. 5b and c, are desirable solutions for gravity load requirements. However, as will
be explained and discussed in the next section, such configurations may not always be an
optimum solution for seismic design.
(a) (b) (c) (d) (e) (f)
FIGURE 4 Transverse view of tower pier configurations for cable-stayed bridges.

Possibility of coupling
at pier tops
(a) (b) (c)
FIGURE 5 Longitudinal view of tower pier configurations for cable-stayed bridges.
4. Concept Design Considerations for Seismic Loads

In order to develop an optimal seismic design solution the engineer must be aware of the
ideal structural characteristics for non gravity load cases identified in the previous section.
There are several important issues to be dealt with in developing the seismic design. It may
appear that compromises are required at times, since the ideal characteristics for seismic
load may be completely different to those for non-seismic loads. However, through good
engineering and with the help of special structural elements, optimal characteristics can
be identified for both seismic and non-seismic load cases, as explained in the following
paragraphs.
4.1. Deck to Pier Connection

An important decision in the concept design of a cable-stayed bridge will be whether to
connect the bridge deck(s) to the pier(s), as illustrated in Fig. 6. There are essentially
three options available to the engineer: (i) provide no connection; (ii) provide a strong
stiff (essentially rigid) connection; or (iii) provide an intermediate solution with limited
transfer force and intermediate stiffness.
If no connection is provided between the deck and the piers, the deck will exhibit a
pendulum-type response with little lateral stiffness. Given that the majority of the bridge
mass is distributed along the deck, this response mode will be characterized by a very long
period of vibration. Note that the period of a pure pendulum can be obtained from:

H
Tp = 2π , (3)
g
Detail A Connect Deck

to Piers?
TOWER
ELEVATION
DETAIL A
FIGURE 6 Deck-to-pier connection consideration.
where H is the pendulum length which could be considered equivalent to the pier height,
and g is the acceleration due to gravity. As a reasonable pier height could be assumed to be
the deck span, Lspan , divided by four, the pendulum period could be estimated to be:

Tp = Lspan (4)
In reality however, considerable longitudinal stiffness is also provided by the cables and
piers, and this will reduce the longitudinal period from that given by Eq. (4). The exact
period will therefore depend on the pier configuration adopted, but from experience, one
should expect a longitudinal period in the order of Tp /2. Some more refined considerations
of the modal characteristics of a cable-stayed bridge are provided in Walther et al. [1999].
With reference to the shape of typical response spectra shown Fig. 2, it is clear that as
the period of a structure increases, the acceleration demands decrease and the displacement
demands increase. The benefits of avoiding a deck-to-pier connection are therefore that the
seismic design forces are low and the drawback is that the bridge has to be detailed to be
able to sustain very large relative displacements between the deck and piers and the deck
and abutments. In addition, without a deck-to-pier connection, the low lateral stiffness may
imply that the response under eccentric live loading and wind loading is unacceptable.
If a stiff, strong deck-to-pier connection is provided, the period of the system reduces,
the acceleration demands increase and therefore the force demands also increase rather
dramatically. While the connection itself can be sized to transmit such forces, the large
demands will be particularly problematic for the piers and possibly for the composite deck
in the region of the connection.
The third alternative is to provide an intermediate solution for the deck-to-pier connec-
tion which could be achieved through the use of fuses, springs or dampers. The connection
should be designed to ensure adequate stiffness and strength under non-seismic loading, but
with good flexibility, displacement capacity and energy dissipation characteristics under
seismic loading. As seen in Fig. 7, the Rion-Antirion bridge adopts such a connection strat-
egy for the transverse response direction, using a stiff metallic fuse device sized to transfer
wind and static loads to the pier, in parallel to a number of viscous dampers that dissipate
large amounts of energy after the fuses yield during an intense earthquake event (see Infanti
et al., 2004).
FIGURE 7 Deck-to-pier connection utilized in the Rion-Antirion Bridge.
FIGURE 8 View of computer model of South Crossing bridge proposed for Guayaquil.
It is likely that an intermediate solution will offer the best solution for cable-stayed
bridges in seismic regions. To illustrate this, consider the three-pier cable-stayed South-
Crossing bridge proposed for Guayaquil, shown in Fig. 8. The finite element model (FEM)
of the bridge shown in Fig, 8 was developed using frame, shell, and cable elements in
SAP2000 v11, and for details of the modelling and analysis readers should refer to Calvi
et al. [2011]. An intermediate type of deck-to-pier connection is proposed for the bridge,
as indicated in Fig. 9. The connection consists of two nonlinear viscous dampers placed
horizontally at 45◦ to the longitudinal axis of each deck together with a central damper
perpendicular to the deck and inclined at 15◦ . In addition, metallic fuse devices also
connect the pier to each deck, but for clarity, these are not shown in the sketch of the
connection.
In order to illustrate the benefit of the intermediate connection solution for this bridge,
nonlinear time-history analyses were run for three different cases: (i) with actual inter-
mediate connection details; (ii) with a rigid pier-to-deck connection; and (iii) with no
connection between the deck and pier. Figure 10 provides a simplified representation of
the pier and deck moment distributions in a cable-stayed bridge, illustrating the manner
in which seismic actions can cause large tower shears and transverse deck moments when
stiff, strong deck-to-pier connections are adopted. Clearly, if the deck-to-pier connection
is freed, then the deck transverse moments would be expected to reduce and the deck dis-
placements increase at the connection region. Tables 2 and 3 compare the tower base forces
DECK 1
Central damper TOWER Two dampers at 45° to

inclined at 15° deck positioned
horizontally
DECK 2
FIGURE 9 Plan view of deck-to-pier connection of the proposed South Crossing bridge
in Guayaquil.
TABLE 2 Effect of pier-to-deck connection typology on the forces of the Guayaquil South
Crossing bridge central tower base, as predicted through nonlinear time-history analyses
Central tower base actions from nonlinear time-history analyses
Connection Longitudinal Transverse Bending about Bending about
type shear (MN) shear (MN) trans. axis (MNm) long. axis (MNm)
Intermediate 244 223 468 377
Rigid 313 277 580 450
Free 251 258 476 427
TABLE 3 Effect of pier-to-deck connection typology on the displacements of the

Guayaquil South Crossing bridge deck at mid-span and pier locations, as predicted through
nonlinear time-history analyses
Minimum displacements at deck level as predicted through nonlinear
time-history analyses
Longitudinal (m) Transverse (m) Vertical (m)
Connection Mid- Mid- Mid-
type Pier P1 span Pier P2 Pier P1 span Pier P2 Pier P1 span Pier P2
Intermediate 0.48 0.53 0.50 0.49 1.04 0.68 0.26 −1.01 0.24
Rigid 0.08 0.15 0.12 0.05 2.20 0.06 0.05 −0.95 0.05
Free 1.11 1.19 1.17 0.64 0.65 0.78 0.29 1.29 0.28
Elevation of Cable Forces & Pier Moments under

Longitudinal Excitation
Plan View of Deck Moments under

Transverse Excitation
FIGURE 10 Simplified representation of force distributions for longitudinal response

(top) and transverse response (bottom) of a cable-stayed bridge when stiff strong deck-
to-pier connections are adopted.
and deck displacements, respectively, for the three different connection typologies as pre-
dicted through the non-linear time-history analyses. Observe that the intermediate solution
records the lowest design forces in the towers (lower than free solution because of addi-
tional energy dissipation of viscous damping devices) and is the most effective means of
limiting the deck displacements. As such, it is clear that the intermediate solution offers
significant advantages for the South Crossing bridge.
The design of a pier-to-deck connection should consider the control of both longitu-
dinal and transverse response. The critical response direction will depend on the relative
magnitude of seismic and eccentric gravity loads, as well as displacement limits for the dif-
ferent response directions. For standard cable-stayed bridge configurations, the connection
in the longitudinal direction will need good stiffness to limit deck displacements due to
eccentric gravity loads. However, if eccentric gravity loads in the longitudinal direction do
not cause excessive displacements even when the deck-to-pier connection is omitted, then
provided that the earthquake displacements can be sustained, the deck-to-pier connection in
the longitudinal direction can be freed. This was the solution adopted in the Rion-Antirion
bridge that is characterized by very stiff towers and a slightly arched deck that meant that
eccentric gravity loads are not problematic.
In the transverse direction, the displacements should be limited to avoid pounding
between the deck and piers, and the characteristics of the deck-to-pier connection should
be set with this objective in mind. If the displacements at the deck-to-pier connection are
relatively limited in the transverse direction, it is clear that the longitudinal response may
engage more of the deck mass. However, identification of a critical response direction is
not necessarily straightforward since longitudinal resistance is provided by both the cable-
pier system in addition to the deck-to-pier connection. In contrast, the cables provide little
restraint to the deck movements in the transverse direction and therefore practically the full
proportion of the deck mass excited in the transverse direction must be transferred at the
deck-to-pier connection.
In line with the above considerations, assuming that either a rigid or an intermedi-
ate deck-to-pier connection is adopted, large longitudinal and transverse shears should be
expected in the pier. The magnitude of these forces suggests that large tower bases will be
necessary, but this is typically acceptable given the large stiffness that the towers should
possess in order to limit deformations of the deck under eccentric gravity loads and the
benefits that large tower bases have on distributing loads onto the foundations. Note that
in addition to large shears, the deck-to-pier connection may also induce potentially signif-
icant torsional moments, depending on the pier configuration and the response direction
being considered. If single pier towers are adopted, with deck segments either side of each
tower, then torsional moments due to the deck-to-pier connection should be relatively small.
However, for two-pier tower options, such as those shown in Figs. 4a–d, the longitudinal
response will require that the piers resist torsion and this should be considered together
with the shear requirements in verifying the pier sections. Note also that in addition to pro-
viding lateral stiffness and strength, the transverse beams that link typical two-pier tower
configurations immediately below the deck level (shown in Figs. 4a–d) can be an effective
means of reducing the torsion component below the deck level.
The deck-to-pier connections may involve construction of inclined structural members,
providing some vertical restraint to the deck. This has the practical advantages of offering
good space for the installation of the devices up under the deck, and also has aesthetic
advantages in that the connection devices are likely to be relatively hidden. The structural
benefits of the vertical restraint offered by the deck-to-pier connection can also be consid-
ered as by inclining the connection devices, some vertical restraint and energy dissipation
can be expected. However, for the case of the South Crossing in Guayaquil, it was observed
that the vertical displacements of the deck varied by less than 5% when the inclination of
the central dampers (see Fig. 9) was changed from 15◦ to 0◦ . This suggests that the vertical
response of the deck is principally associated with longitudinal seismic excitation (with
flexure of piers resulting in vertical displacements of deck) and not the vertical excitation
component of the ground motion.
4.2. Tower Pier Design

Important decisions for the design of the tower piers should be whether to utilize one, two,
or four legs and whether to connect the cables up the height of the piers or only at the pier
tops (see Fig. 13).
Regarding the number of legs in the tower, one should consider the different lateral load
resisting mechanisms that could be expected to develop as well as the stiffness provided.
Coupled tower piers are relatively stiff and therefore offer an effective means of limiting
deformations under eccentric gravity loads. This consideration supports the use of four-leg
towers, such as the towers realised in the Rion Antirion bridge (see Combault, 2005).
As shown in Fig. 11a, a multiple leg configuration in which the piers are connected
by transverse beams will be expected to develop significant coupling forces. The coupling
action subjects the piers to varying axial loads and this implies that changes in the pier
flexural resistance and stiffness should be expected during the seismic response. The piers
will be subject to very high axial forces due to gravity loads alone, and the additional axial
(a) (b)
FIGURE 11 Effect of number of tower legs on likely load resisting mechanisms in towers:
(a) coupled multiple leg tower elevation and bending moment diagram; (b) single leg tower
elevation and bending moment diagram.
load due to seismic response should be carefully evaluated as it may lead to an earlier
onset of concrete crushing in critical piers. This is considered particularly relevant for the
capacity design of four-leg towers in which, under diagonal attack, three of the four legs
could go into tension, imposing very large axial loads on a single pier. As such, doubts
should be raised about the reliability of a coupled mechanism for the bridge piers, and some
account for the uncertainties in capacity should be made in the design. One possibility for
controlling the axial force due to coupling is to design the coupling beams to fuse, thereby
limiting the axial loads that can be transferred to the piers as part of a capacity design
approach.
The uncoupled pier configuration, shown in Fig. 11b, appears to be advantageous from
a seismic viewpoint, therefore, since the lateral loading does not alter the axial loads in the
piers and this implies that the flexural resistance, stiffness and ductility capacity of the pier
can be more reliably estimated. However, despite the simple cantilever form, the axial load
in the piers may still vary significantly due to vertical earthquake excitation components.
As argued by Papazoglou and Elnashai [1996] and more recently by Elnashai et al. [2006],
the performance of a concrete pier and in particular its shear resistance, can be significantly
affected by the vertical component of excitation. Given the fundamental importance of the
piers to the overall performance of the bridge structures, the seismic design should therefore
account for axial load variation on the seismic response. In practice, this can be done by
first sizing the elements without account for vertical excitation effects, and then verifying
the response of the structure when subject to three ground motion excitation components
through nonlinear time-history analyses. However, additional research is also required to
ensure that shear resistance models adequately account for axial load variation effects.
In the longitudinal direction of the bridge, the designer could choose to connect cables
to a single pier or to multiple coupled and or inclined piers, as indicated in Fig. 12. Coupled
piers may again be prone to some of the uncertainties in resistance due to axial load vari-
ations. However, there are some practical advantages to a coupled pier solution in the
longitudinal direction as, in contrast to the transverse direction, piers can be positioned
relatively close to each other and therefore vertical piers can be coupled by relatively short
devices up the full height of the piers, as indicated in Fig. 12c.
As pointed out earlier, another important decision to make in developing the structural
scheme is whether to connect the cables in a distributed fashion up the height of the piers
or only at the pier tops in a fan-type arrangement, as indicated in Fig. 13. The option of
fixing the cables at different levels up the height of the pier may at first seem desirable
Single Piers in Would coupled piers …pier coupling up

Longitudinal be better? full height or only
Direction? at top?
(a) (b) (c)
FIGURE 12 Tower leg configuration possibilities for longitudinal response: (a) single
tower pier; (b) inclined piers coupled at top; (c) straight piers coupled up height.
Critical cable strains likely

to develop in cables at 45°
Hc to deck.
Δx Δx
Forward of pylon, deck Critical cable strains
rises in fundamental mode likely to develop in
due to pendulum action. shortest diagonal cables.
(a) (b)
FIGURE 13 Two alternative strategies for the connection of cables to piers: (a) cables
connect to top of piers in a fan-type arrangement; (b) cable are connected up the height of
the piers.
since it avoids congestion of cables at a single point on the pier, easing construction needs,
requires a shorter total length of cables and is arguably more aesthetically pleasing than
the single connection alternative. However, in seismic regions the fan arrangement might
be preferred owing to the increased deformation capacity it offers the bridge. This can be
appreciated by considering the strains that develop in the cables for the two configurations
shown in Fig. 13 when the deck is displaced longitudinally an amount x . The cables of
the single connection point solution are all relatively long in comparison to the multiple
connection point solution, in which the cables close to the base of the piers are short. As
such, a uniform longitudinal displacement of the deck imposes much larger strains on the
short cables, which implies that the longitudinal displacement capacity of the bridge is
much lower than an identical bridge with a fan-type cable arrangement.
While the increased displacement capacity offered by the fan-arrangement of Fig. 13a
may be attractive, there are also reasons for which the distributed cable arrangement of Fig.
13b may be preferred, as pointed out by Kawashima et al. [1993]. For example, the shorter
cables will provide a stiffer solution, which may be particularly useful in limiting the lat-
eral displacement of the deck, thereby limiting demands on dampers and expansion joints.
Kawashima et al. [1993] also pointed out that greater damping could be expected in the
longitudinal mode when a distributed cable arrangement is adopted. As such, in deciding
on a cable arrangement, the designer should weigh the benefits of reducing deformation
demands on dampers and joints against the increased cable diameters that are likely to be
required to sustain the larger design forces associated with the stiffer system.
In order to assess the critical cable position for fan-type arrangements, consider that
the change in cable length, δ cable , due to a longitudinal displacement of the deck can be
approximated by Eq. (5).
δcable = (x − p ) cos α, (5)
where x is the longitudinal displacement of the deck (see Fig. 13), p is the lateral dis-
placement of the pier at the cable connection point, and α is the angle of the cable relative
to the deck.
In addition, note that the length of each cable, Lcable , is given by:
Hc
Lcable = , (6)
sin α
where Hc is the height of the cable-to-pier connection point above the deck, as indicated in
Fig. 13a.
Consequently, the cable strain due to the longitudinal displacement of the deck can be
obtained through the division of Eq. (5) by Eq. (6), to give:
(x − p )
εcable = sin 2α. (7)
2Hc
There are of course other aspects of the response that will affect the strains, such as deck
vertical deformations, but Eq. (7) is of interest since it suggests that for bridges with fan-
type cable arrangements (in which Hc and p are approximately the same for all the cables),
the critical cable strains due to longitudinal seismic response are likely to develop in the
cables inclined at an angle of 45◦ .
4.3. Tower Base Design

The tower base sections, below the deck level, are subject to very large forces, with large
forces transferred from the deck-to-pier connection and from the cables at the top of the
towers adding to the seismic excitation of the towers themselves. Continuity in bending
resistance must be ensured from the tower piers to the tower base as indicated in Fig. 14.
This implies that the base section should be at least as large as the upper sections, and
in order to resist the additional bending that develops due to the large deck-level shears,
the pier sections are often combined to form a large single section for the tower base. For
bridges in deep water this has practical advantages for construction in which the tower base
is initially constructed in a dry dock, then floated in a wet dock, and positioned as a single
unit in place for the superstructure above. By having only a single pier base to manoeuvre
into position and connect to typically underwater foundations, the construction challenges
are restricted to a single installation event for each tower. This was the construction method
and pylon base solution adopted for the Rion Antirion bridge. When towers are not located
in deep water, it is likely that the pier base will be constructed directly in the final position,
after the installation of a working platform at each tower location.
4.4. Foundation Design and Soil-Structure-Interaction Considerations

As with most civil engineering structures, the foundations for cable-stayed bridges require
careful consideration. Suitable foundation solutions will clearly depend on the site in
Tower piers from

above deck level
Require continuity of flexural resistance

from tower piers to tower base
Large tower base section

to transfer loads
elastically to foundations
FIGURE 14 Partial elevation of typical tower base section.
FIGURE 15 Sketch of foundation system adopted for Rion Antirion Bridge. [Combault
et al., 2005.]
question. However, some general observations can be made regarding cable-stayed bridge
foundations and some seismic design issues should be highlighted.
Given the magnitude of the design forces, the foundation area will typically need to be
large and some soil improvement measures may be required. Improvement measures might
include the use of driven piles to reinforce the soil as was done for the Rion Antirion bridge
[Pecker and Salecon, 1999]. Construction possibilities may also influence the choice of
foundation system and the choice between drilled, driven, or specialty piles will depend on
the water level, soil properties, and site conditions. Predictions of the foundation stiffness
should then consider eventual uncertainties in the site characteristics and pile group effects.
Uncertainties in foundation stiffness will make it difficult to analytically predict the differ-
ential displacements of adjacent towers and as such, a construction methodology should be
developed that permits suitable adjustment of the adjoining deck sections.
Soil-structure interaction effects (also now referred to as soil-foundation-structure
interaction effects) can considerably affect the seismic response of structures (refer to Wolf,
1985, and Mylonakis and Gazetas, 2000, for further information). While this is true, it is
considered that proper evaluation of soil structure interaction effects need not be undertaken
until a developed or detailed design stage. Typically, initial conceptual design considera-
tions should instead aim to mitigate the influence of the foundations on the seismic response
by providing strong, rigid foundations. An interesting alternative concept, however, as pro-
posed by Pecker [1998], is to develop foundation systems that fuse as part of a capacity
design approach. Such a concept was incorporated into the design of the Rion-Antirion
bridge, where circular tower footings sit atop a 2.8 m thick gravel layer, as indicated in
Fig. 15, in order to enable sliding in the event of extreme shear forces at the footing-gravel
interface. Not only does this mechanism help to limit the forces imposed on the founda-
tions, it also dissipates an important amount of energy, thereby helping to reduce the bridge
response above.
5. Preliminary Sizing of the Main Structural Elements of Cable-Stayed

Bridges in Seismic Regions

In light of the various conceptual design considerations made in the previous sections, one
can identify a strategy for the preliminary sizing of the main structural elements of cable-
stayed bridges. The intention of the expressions presented in the following sub-sections is
not to accurately establish member sizes, but instead, to illustrate how reasonable member
sizes can be quickly set for the purposes of initial feasibility studies, costing exercises,
and initial structural modeling. The expressions that will be presented in Eqs. (8)–(15) of
this article are considered a new contribution of the authors for the simplified conceptual
design of cable-stayed bridges. In addition, Sec. 5.2 includes a novel formulation of the
Direct DBD approach of Priestley et al. [2007], put forward for the identification of suitable
deck-to-pier connection characteristics. The member sizes obtained from the procedure
should clearly be refined as the design develops, following the application of accurate,
albeit time-consuming, analysis techniques.
Preliminary sizing should start by setting the spacing of the towers. This may depend
on the ground conditions and general characteristics of the site in question. Typically, cable-
stayed bridges should be competitive for spans of between 50–500 m, although longer
cable-stayed bridges have also been constructed, suggesting they are also effective for
longer spans. For bridges with multiple towers the spacing should be regular so as to ensure
balanced gravity loads. Once the distance between the towers, Lspan , has been established,
the height of the towers, Htower , can be found as:

Htower = Hdeck + Lspan 4, (8)
where Hdeck is the height from the tower foundations to the level of the deck, as indicated
in Fig. 16, and is usually dictated by a given navigation clearance. The basis of Eq. (8)
is simply to provide an average cable inclination of 45◦ . An average cable angle of 45◦
implies that the deck compression due to gravity loading only is equal to the total gravity
load acting on the deck.
5.1. Preliminary Member Sizing for Gravity Loads
Having established the bridge span and tower heights, an initial sizing of the deck can be
made with knowledge of the intended deck width and proposed use. For a traffic bridge one
might utilise preliminary gravity load estimates of 10 kN/m2 for dead load and 10 kN/m2
for live load over the full deck width. However, the dead load will very much depend on
the type of deck solution adopted and more accurate estimates of the live load could be
Htower
Hdeck
Lspan
FIGURE 16 Elevation of a typical cable-stayed bridge with key dimensions indicated.

obtained directly from the design brief. The expected axial force in the deck under gravity
loading only is then given by:

Ndeck = wdeck .Lspan 2, (9)
where wdeck is the factored deck gravity loading (units of force per m length of deck) and
Lspan is again the distance between bridge towers.
The deck axial load obtained from Eq. (9) can then be used to estimate the required
section area of the longitudinal steel beams (Fig. 3) from:
Ndeck − tslab .beff .0.85f c

Abeams = , (10)
Fy,d
where Abeams , is the total area of longitudinal beams (to be divided by number of beams to
obtain required section area of a single beam), Fy,d is the design resistance of the steel to be
specified for the beams, tslab is the thickness of the reinforced concrete slab deck possessing
effective width beff , f’c is the concrete compressive strength, and Ndeck is taken from Eq. (9).
Note that the approach proposed here assumes that transverse beams will be used to brace
the main longitudinal beams at sufficiently short centres to ensure that reductions in axial
resistance due to instability are not required. A reasonable spacing for transverse beams
might be one-third to one-half the cable spacing and the concrete deck thickness can be
initially sized to provide sufficient strength and stiffness under gravity loadings to span
between the transverse beams. A typical cable spacing will be between 5 and 10 m, and in
setting the cable spacing, one should consider the construction methodology as this may
require main beams to cantilever unsupported prior to the cable connection and may also
dictate transport and lifting sizes for main beams.
An average size for the cables can be set as a function of gravity loads using the following
expression:
wdeck .s
Acable = √ , (11)
Fy,d 2
where Acable is the average area required of the cables, Fy,d is the design strength (stress)
of the cables, s is the spacing between cables (assumed constant), and wdeck is again the
factored ultimate gravity load of the deck (force per unit length). Note that the value of
Acable given by Eq. (11) assumes that two lines of cable would be provided to each deck.
The area required by Eq. (11) is the average area necessary to resist standard grav-
ity loads. Seismic loading could also influence the necessary cable sizes, and accurate
checks should be undertaken following nonlinear time-history analyses. A capacity-design
estimate of the required cable area for seismic loads can be made considering that the maxi-
mum longitudinal force that be applied to the deck is that which exceeds the restoring force
caused by the self weight, as indicated in Fig. 17. As such, the maximum longitudinal force
that will be transferred through the cables if the deck is not held down, can be given by:
√ wdeck,EQ .s
Acable = 2 . (12)
Fy,d
wdeck,EQ is the gravity load of the deck expected to be present at the time of the maximum
design intensity earthquake event and the other symbols are as for Eq. (11). While Eq.
(12) might be useful as a capacity design estimate, if limited inelastic deformations are
permitted in the cables this will add considerable deformation capacity and the preliminary
cable design may be based only on the requirements of Eq. (11).
After only a few quick calculations, therefore, the overall geometry is set and preliminary
sizes for the deck elements and cables are established. The remaining preliminary design
tasks are to size the main tower section, the deck-to-pier connection and the foundations.
As discussed earlier in the article, an important function that the tower piers have is to
resist out-of-balance forces due to eccentric gravity loading. As such, preliminary sizing of
the piers should first be based on the eccentric gravity loads expected. In the longitudinal
direction, one assumes that live loads could act on one side of the towers only, thereby
providing the out-of-balance moment acting on the system. For multiple tower bridges and
for cable-stayed bridges with vertical restraint at abutments, this out-of-balance moment is
resisted not only by the tower piers, but also by the deck in bending. However, it has been
found that by making the simplifying analogy shown in Fig. 18, one can quickly gauge the
approximate section size required for the piers.
The approach illustrated in Fig. 18 is to assume that the eccentric gravity loading
imposes a moment on the tower base, Mbase , given by:
2
Lspam
Mbase = ζ .w , (13)
8
where w is the eccentric gravity loading on the deck (units of kN/m), Lspan is the deck span
between towers, and ζ is a continuity coefficient intended to take account of the flexural
resistance offered by the continuous deck. The continuity coefficient will clearly depend on
Maximum longitudinal force

resisted by cables may be
h approximated as that which tends to
lift the deck to rotate it around the
pier top.
F x x
i.e. For equilibrium, F.h does
W W not exceed ΣW.x
FIGURE 17 Capacity design force considerations for longitudinal seismic response of a

cable-stayed bridge.
Mbase = ζ.w.Lspan2/ 8
P = w.Lspan / 2
FIGURE 18 Idealization of eccentric gravity loading for preliminary sizing of tower piers.
the relative stiffness of the deck versus the tower-cable system and designers could select
a value considering the bridge in question. For preliminary sizing purposes a value of 0.5
may be reasonable.
In the transverse direction the importance of eccentric gravity load will depend on the
selected pier configuration. For pier configurations in which the piers are positioned either
side of the deck, eccentric loading in the transverse direction will not cause significant
bending in the piers and instead will result in different pier axial loads. For a single pier
solution, however, the full moment due to eccentric gravity loading in the transverse direc-
tion will need to be resisted by the pier sections. This consideration also provides a simple
means of setting initial pier section dimensions as a design moment can be taken as:
Mpier = wdeck .Lspan .x, (14)
where x is the eccentricity of the gravity loads with respect to the pier centreline, and the
other symbols are as defined earlier.
Another important consideration for the tower pier design is the lateral stiffness
required to control vertical displacements of the deck. The internal forces and displaced
shape of a cable-strayed bridge subject to eccentric gravity loading are indicated in Fig. 19.
It is clear that if the tower piers do not possess high stiffness in the longitudinal direc-
tion, the vertical displacements of the deck can be rather large. With this in mind, one
can select a vertical displacement limit (such as deck span on 500) and, accounting for
cable elongation under the eccentric gravity loading, identify the necessary pier lateral
stiffness. It is also noted, however, that for spans close to abutments, a back-stay effect
offered by the vertical restrain of the abutment can provide an effective means of control-
ling the deck displacements without relying on the stiffness of the pylons. Considering
this, one may choose to do some initial elastic analyses under static loads to check the
displacements and modify the tower pier sizes required by Eq. (14). In addition, the pier
stiffness could also be important in order to control the vertical dynamic response of the
deck when subject to wind and traffic loadings. While this does introduce additional uncer-
tainty as to the required lateral stiffness, it can be expected that if vertical displacements
are relatively limited under eccentric gravity loading then the vertical frequencies should be
relatively high.
5.2. Preliminary Seismic Design in the Longitudinal Direction to Check Pier Sections
and Establish Connection Characteristics
After using gravity load considerations to set the bridge geometry and main member
sizes, suitable characteristics for the deck-to-pier connections should be established with
(a) Eccentric Gravity Loading

Back-stay effect at spans

Maximum Deck Displacement adjacent to abutments can
Under Eccentric Live-Loading reduce vertical
of say L/500 ? displacements significantly
(b) Deformed Shape
FIGURE 19 Eccentric gravity loading affects on (a) internal forces and (b) deformed
shape.
consideration of the seismic loads. In addition, for highly seismic regions, the tower pier
sizing procedure outlined in the previous section may be not provide sufficient stiffness and
strength to control the vertical response of the deck under seismic actions.
The vertical response of the deck becomes relevant for the longitudinal seismic
response because, as illustrated earlier in Fig. 13 and also in Fig. 20, as the deck displaces
longitudinally, the pendulum nature of the response causes portions of the deck to either
sag or lift. This in turn excites the vertical modes of vibration of the deck. The fundamental
longitudinal mode will typically see the deck segments forward of the pier rise due to the
pendulum nature of the response. However, an important higher mode in the longitudinal
response is that in which the deck displaces the tops of the towers, with the deck segments
located forward accelerating downwards, and deck segments behind the towers accelerat-
ing upwards. Such a longitudinal seismic response can impose significant eccentric forces
on the towers.
The actual eccentric load that must be carried by the tower piers will depend on the
vertical acceleration of the deck. As this will depend on the deck frequencies and damping,
a quick means of calculating the vertical accelerations is not available. However, in order
to conservatively size the tower pier sections at the conceptual design stage with allowance
for vertical deck response, one can consider the extreme scenario (illustrated in Fig. 20)
in which the rising side of the deck slackens the cables completely, such that the cable
loads associated with the sagging portion of the deck must be carried by the tower piers.
F F
x x
Deck vertical excitation can
be significant for tower
W W
bending forces.
FIGURE 20 Deformed shape of bridge under longitudinal seismic excitation.
Accordingly, the conceptual design force considered to act at the top of the tower piers is
given by:
wdeck,EQ .Lspan
Fpier,long = ζ , (15)
2
where wdeck,EQ is the gravity load of the deck expected to be present at the time of the
maximum design intensity earthquake event, Lspan is the deck span between towers, and
ζ is a continuity coefficient intended to take account of the flexural resistance offered
by the continuous deck. The continuity coefficient will clearly depend on the relative
stiffness of the deck vs. the tower-cable system and designers could select a value con-
sidering the bridge in question. For preliminary sizing purposes a value of 0.5 may be
reasonable.
For the deck-to-pier connections, reasonable characteristics can be set in order to suit-
ably limit the seismic response of the deck. Important criteria for the seismic design in the
longitudinal direction may include requirements to:
● Limit the longitudinal displacements of the deck in order to ease joint require-
ments. The displacement capacity of movement joints is dependent on manufacturer
capabilities. However, a typical movement joint may have a limit of up to 0.5 m,
whereas specialist movement joints can be manufactured to sustain over a metre of
movement.
● Limit the vertical displacements of the deck by controlling the lateral displacements
at the top of the tower piers.
● Limit the curvature demands in the tower piers.
As satisfaction of the last two of these criteria will depend on the excitation of the tower
itself in addition to the longitudinal movements of the deck, they are difficult to control via
the deck-to-pier connection alone. However, it is argued that the first design requirement
can be effectively controlled through good selection of the deck-to-pier connections. In
order to do this, one can use a direct displacement-based design procedure, such as that
shown in the flowchart of Fig. 21.
The procedure outlined in Fig. 21 is based on the Direct DBD approach of Priestley
et al. [2007] with the required viscous damping force and constant obtained in line with
the arguments made by Sullivan [2009]. For details of Direct DBD, readers should refer to
the text by Priestley et al. [2007]. The procedure shown in Fig. 21 is intended to be appli-
cable for systems with intermediate type deck-to-pier connections that incorporate viscous
Step 1: Identify design displacement, Δd, based on

deck and joint deformation capacity.
Step 2: Approximate effective mass, me, as being

total seismic weight of deck.
Step 3: Select a system damping level, ξsys

(typically around 20%–30%).
Step 4: Enter displacement spectrum at ξsys damping

level to find effective period, Te.
Step 5: Determine required effective stiffness, Ke

(= 4π2me/Te2) and base shear Vb = Ke.Δd
Step 6: Determine fraction of force that can be Alternatively, if total Vb

transferred by cables, Vtop, and take difference with Vb can be taken by cables, can
to find deck-to-pier design force, Vdeck = Vb – Vtop. use viscous dampers only.
Step 7: Estimate displacement of tower/pier at deck level Δpier,deck. Subtract

displacement from deck displacement, Δd, to find displacement demands on
deck-to-pier connections (dampers): Δdamp = Δd - Δpier,deck
Step 8: If deck-to-pier connection has a displacement-dependent force component

(i.e., if Vdeck > 0) calculate required stiffness of connection: Kd-t-p = Vdeck/Δdamp and
identify damping characteristics of connection device, ξd-t-p.
Step 9: Find required damping force contribution, Fdamp, of

deck-to-pier connections:
2Vtopξpier + 2Vdeck.ξd−t−p + Fdamp
ξsys =
2Vb
Step 10: Calculate required damper constants from:

C = Fdamp.Te/(2πΔdamp)
FIGURE 21 Displacement-based seismic design procedure for preliminary sizing of deck-

to-pier connection characteristics in longitudinal response direction.
dampers, possibly in parallel to elastic displacement-dependent devices (such as rubber

bearings). Details of the above procedure will be illustrated in a subsequent application
later in this article.
5.3. Seismic Design Considerations for the Transverse Direction

While the transverse seismic response of a cable-stayed bridge incorporating a full-isolation
concept may at first appear rather straightforward, there are a number of factors that make
aspects of this response difficult to predict, as will be discussed subsequently. However, it
will also be argued that for what concerns the preliminary conceptual design, the transverse
response direction will not typically be critical for member sizing.
In the transverse direction it may be assumed that the cables and therefore piers provide
little lateral restraint to the deck. As such, if the deck is isolated from the piers at the deck
level, the period of vibration in the transverse direction should correspond to that predicted
by the period expression for a classic pendulum, presented earlier in Eq. (3). This transverse
period will of course be significantly reduced if a connection is provided that has an elastic,
displacement-dependent force component. However, in the case that the bridge does not
have elastic restrainers, and instead incorporates linear or nonlinear viscous dampers, as
in the case of the South Crossing of Guayaquil, then a long period of vibration could be
expected. Considering the spectral displacement demands in the long period range, one
would estimate that the peak displacement in the transverse direction would be equal to the
peak ground displacement, or at most the peak spectral displacement (from periods TD to
TE in Fig. 2). Consequently, if such a displacement demand can be accommodated through
provision of a sufficient gap between the deck and the piers, then one could argue that the
transverse response should be fine.
However, the pendulum-type dynamic response of the deck is likely to be influenced
by the response of the tower piers with considerable interaction with the higher modes of
vibration of the tower itself. Such interaction could be due to both the transverse and verti-
cal excitation of the tower piers. Furthermore, the higher modes of vibration of the deck are
likely to have long periods in the transverse direction, with significant mass participation.
Such higher modes may also be excited through interaction of the transverse and longitudi-
nal response of the deck-to-pier connections. Consider the damper orientation utilized for
the Guayaquil bridge shown earlier in Fig. 9. The longitudinal response is limited through
incorporation of nonlinear viscous dampers between the deck and the piers. In order to
allow for transverse movements of the deck, and to provide some damping in the trans-
verse direction, the dampers are oriented at 45◦ to the line of the deck. As such, as the
longitudinal response develops, the dampers develop forces which will also act to resist the
transverse response of the deck. The balanced configuration of the dampers implies that
the longitudinal excitation is not expected to excite the transverse response, but it may still
permit transverse excitation of the pier to be imposed onto the deck when the dampers are
moving at a high velocity due to the longitudinal response. Note that the dampers may well
even tend to filter the imposed action in the transverse direction to that associated with the
frequencies of vibration in the longitudinal direction.
Evidently, therefore, the seismic response in the transverse direction is rather diffi-
cult to accurately predict without nonlinear dynamic analyses. However, this should not
impact too greatly on the preliminary conceptual design requirements. As outlined in
Sec. 5.1, preliminary sizing for the tower piers and tower bases in the transverse direc-
tion can be made through consideration of eccentric gravity loads. The cables should
not be affected by transverse seismic response and nor should the deck section. Finally,
while the dampers will assist in limiting the response in the transverse direction, it is
suggested that their preliminary sizing be based on the longitudinal seismic response
direction.
In regions of very high seismicity the pier sections could also be sized considering that
the vertical modes of vibration of the deck could also impose large moments on the tower
piers in the transverse direction in a similar manner to that described for the longitudinal
direction. For the case of two-deck bridges, such as the South Crossing bridge in Guayaquil,
one deck could move vertically out of phase with respect to the other (see Fig. 22). Again,
for sizing the piers one could consider that the rising deck slackens the cables completely,
such that the weight of the sagging portion of the deck must be carried by the tower piers.
W W/2
FIGURE 22 Eccentric loads to be carried by tower piers due to vertical excitation of the
deck(s).
In line with this, the design overturning for the tower piers in the transverse direction is
given by:
Mpiers,trans = ζ .wdeck,EQ .Lspan .x, (16)
where wdeck,EQ is the gravity load expected on the deck at the time of the design earthquake,
Lspan is the tower-to-tower span of the deck, and x is the eccentricity, as indicated in Fig. 22a
below. Note that for single-deck bridges one can also use a similar analogy considering that
the vertical excitation will also excite torsional modes of vibration in the deck, potentially
lifting one edge of the deck and sagging the other, as indicated in Fig. 22b.
Some consideration should also be directed towards the ends of the bridge where the
transverse response may be more critical for what regards the joint design. The dampers
at the ends of the deck should be sized to control the displacement demands on the joints,
and while this could be achieved using a DBD approach such as that proposed for the
longitudinal direction, the participating mass and equivalent SDOF characteristics are more
difficult to establish, suggesting that one might resort to a more trial-and-error approach
using advanced analyses.
5.4. Example Application of Preliminary Sizing Strategy

In this section, the preliminary sizing approach is illustrated for a bridge possessing mul-
tiple towers, spaced 415 m apart, to be formed with a composite deck of the type shown
in Fig. 3. The deck is 23 m wide and must carry an unfactored deadload of 10 kPa and
an imposed live load of 5 kPa. The average height of the deck above foundation level is
60 m. These characteristics approximately match those of the South Crossing cable-stayed
bridge of Guayaquil and therefore the results of the application can be compared to the
detailed design solution that were obtained after detailed nonlinear time-history analyses
of the bridge.
Step 1. Set height of piers
In order to achieve an average cable angle of 45◦ , the tower height is set as:

Htower = Hdeck + Lspan 4 = 60 + 415/4 = 164 m.
Step 2. Obtain gravity axial force in deck to size main beams of deck
The factored gravity loading on the deck, wdeck , is taken as:
wdeck = 1.5 × (10 + 5) × 23 = 517.5 kN/m.
The expected axial force in the deck under gravity loading only is then given by:

Ndeck = wdeck .Lspan 2 = 517.5 × 415/2 = 107400 kN.
The deck axial load is then be used to estimate the sectional area of the longitudinal steel
beams:
Ndeck − tslab .beff .0.85f c 107.400 − 0.25 × 5.0 × 0.85 × 40

Abeams = = = 0.156 m2 ,
Fy,d 414
where a slab thickness of 250 mm and a total effective width of 5,000 mm has been
assumed, with material properties of f’c of 40 MPa and a design yield strength of
414 MN/m2 .
Step 3. Estimate average area of cables
Assuming a cable spacing of 12.5 m and a cable design resistance of 900 MPa, the average
area of cables can also then be obtained, as per Eq. (11), as:
wdeck .s 517.5 × 12.5

Acable = √ = √ = 0.007 m2 .
Fy,d 2 900, 000 × 2
For this example it will be assumed that small inelastic deformations will be allowed to
develop in the cables and that therefore preliminary sizing of the cables does not need to
consider seismic loading.
Step 4. Identify design loads for tower pier sizing
In order to identify a reasonable design force for the preliminary sizing of the tower piers,
the analogy presented in Fig. 18 is first applied, and as such, a design moment in the
longitudinal direction is taken as:
2
Lspam 4152
Mbase,l = α.w = 0.5 × 172.5 × = 1857000 kNm,
8 8
where the eccentric load w is taken as 172.5 kN/m, approximated using the live load of
5 kPa, factored and assumed to act over the full width of the deck.
In addition to bending, the pier is subject to a large axial force that can be taken as
the dead load plus the eccentric live load, which gives an axial load of approximately
250,000 kN plus self weight of the tower itself. Using these forces, and assuming a square
rectangular hollow concrete section, it can be shown that a 7.5 m wide × 7.5 m deep RC
pier with 2 m thick walls and with 2.5% longitudinal re-bar can provide a resistance of
around 2000 MNm.
The design moment for the transverse direction will depend on the desired pier con-
figuration, as discussed earlier. Assuming the pier configuration shown in Fig.4f, eccentric
gravity loading will subject the piers to bending in the transverse direction. Continuing
with the pier size estimated for the longitudinal direction, one can estimate the transverse
bending moment as:
Mbase,t = wLspam x = 172.5 × 415 × (23/2 + 1 + 7.5/2) = 1163000 kNm.
As this moment is less that the longitudinal direction, this simple estimate suggests that
a 7.5 m square hollow section with 2 m thick walls would be a reasonable preliminary
section size. However, given that the bridge under examination is located in a region of high
seismicity, the eccentric moments due to the vertical excitation of the decks, as described
in Secs. 5.2 and 5.3, should be estimated.
Applying Eq. (15) to the South Crossing bridge, the design lateral force on the tower
piers in the longitudinal direction is:
374 × 415
wdeck,EQ .Lspan
Fpier,long = ς = 0.5 = 38800 kN.
2 2
The piers in the longitudinal direction are coupled at their top. The degree of coupling will
affect the design moments up the height of the piers. As the pier sections are typically
tapered, one might consider an effective amount of coupling to be that which gives a pier
base design moment two to four times the magnitude of the pier top design moment. If the
coupling is such that the base moment is four times the top moment, the design moment in
the longitudinal direction at the base of coupled piers becomes:
2Fpier,long .Hpier 2 × 77500 × 104

Mpier,base,long = = = 1600 MNm
5 5
and the average axial load at the base of the pier, estimating the self-weight of the pier using
the pier sized for gravity loads (75% of the 7.5 m square section considering that piers are
tapered), is given as:
wEQ Lspan
Npier,base,long = + Nselfweight = 160 MN
2
which is based on an average cables angle of 45◦ . In addition, it is noted that the cou-
pling will tend to change the axial force in the piers, increasing the axial force on one
side and decreasing it on the other. Considering the axial-moment interaction diagram of
typical concrete sections, it is likely that the overall resistance of the coupled piers is not
significantly affected by this variation in axial force (see Priestley et al., 2007, for details)
and therefore for conceptual design purposes it is neglected and only gravity loads are
considered.
Sizing a pier section to give a flexural strength of 1,600 MNm under an axial load of
160 MN, it is found that a square hollow section 9.5 m × 9.5 m × 2.0 m thick has sufficient
resistance when 2.0% longitudinal reinforcement is provided.
In the transverse direction, the design moment would be
Mpiers,trans = ζ .wdeck,EQ .Lspan .x = 0.5 × 374 × 415 × 22.5 = 1745 MNm
and as this moment would be carried by two piers in parallel, the longitudinal response
appears to be more critical and therefore the 9.5 m square pier section is maintained.
Step 5. Identify deck-to-pier connection characteristics

The other important element that requires sizing is the deck-to-pier connection. As dis-
cussed in Sec. 5.2, the characteristics of this connection should be set considering seismic
loading. In line with the South Crossing system, a connection of the form shown in Fig. 9 is
adopted, with fuse elements to resist wind loads and viscous dampers used to mitigate the
seismic response. The seismic loading at the site for this example, at the collapse preven-
tion limit state, can be represented by the displacement spectrum of Fig. 2b with a spectral
displacement demand of D = 1.5 m at a corner period TD = 7.5 s.
In line with the Direct DBD procedure shown in Fig. 21, the first step is to select
a design displacement. For this example, a design displacement of d = 1.0 m will be
selected. Joint manufacturers should be able to detail a joint to sustain such a movement
without collapse and the vertical displacements of the deck (due to lateral deflection of pier
tops and elongation of cables) should be limited to a similar magnitude.
The effective mass, me , for the seismic design can then be taken as:
me = wdeck,eq Lspam /g = 374 × 415/9.81 = 15800 T,
where wdeck,eq is the distributed load on the deck, set using the unfactored deadload and the
fraction of live load expected at the time of an earthquake.
As per Step 3 from Fig. 21, the system damping level is then selected. With the use
of viscous damping devices, a relatively high system damping value can be selected since
the deck-to-pier connection forces will be an effective means of limiting the deck move-
ments. As such, for this example a viscous damping value of 30% is selected. This system
damping value will be ensured through final sizing of the devices, as will become evident
in subsequent steps.
The design displacement spectrum is then scaled to the system damping value using
an appropriate damping reduction factor. Using the EC8 damping reduction expression, the
scaling factor is:
0.5 0.5
10 10
η= = = 0.54.
5+ξ 5 + 30
As such, the highly-damped design displacement spectrum can be developed, as shown in

Fig. 23, and used to identify the required effective period of the system, Te .
Spectral
Displacement
1.5 m ξd = 5%
Δd = 1.0 m
0.8 m ξsys = 30%
Period
7.5 s Te
FIGURE 23 Use of highly-damped design displacement spectrum to identify required

effective period.
Note that the design displacement actually exceeds the damped corner period dis-
placement demand, and one could consider estimation of lower design forces as explained
by Priestley et al. [2007]. However, for simplicity the corner displacement limit is
conservatively neglected here and hence an effective period of 9.4 s is obtained.
As per Step 5 of the procedure shown in Fig. 21, the required effective stiffness is then
found:
me 15800
Ke = 4π 2 2
= 4π 2 = 7134 kN/m
T 9.42
and hence the design “base” shear:
Vb = Ke d = 7134 kN × 1.0 m = 7134 kN.

This design base shear is not actually the total shear expected at the base of the tower piers,
but is instead the shear associated with the damped deck response. The actual total base
shear will be greatly affected by the seismic response of the towers themselves, but this is
not relevant to the sizing of the deck-to-pier connections.
Step 6 of Fig. 21 indicates that checks should be made to assess whether the cable-pier
system can transfer this design shear. This can be gauged quickly by multiplying the design
shear by the pier height to consider an equivalent moment on the tower-piers, in this case
equal to 742,000 kNm. As this value is less than half the moment due to eccentric gravity
loads in the longitudinal direction, it is clear that the full design shear can be transferred
via the cable-pier system and the deck-to-pier connection does not require a displacement-
dependent force component (i.e., Vdeck from Fig. 21 is zero).
As per Step 7 of Fig, 21, the displacement of the towers at the deck level should be
estimated, although it will be assumed here that the lateral forces due to the deck response
are limited and therefore the damper displacement in the longitudinal direction is assumed
equal to the deck displacement. Note that the effect of the damper configuration will be
accounted for once the total required damping force has been established.
The total damping force can be found by rearranging the system damping expression
shown in Fig. 21, as:
Fdamp = 2Vb ξsys − 2Vtop ξpier − 2Vdeck .ξd−t−p = 2 × 7134 × 0.30 − 2 × 7134 × 0.03 − 0
= 3850 kN
and consequently the required damper constant is then found:
C = Fdamp .Te /(2πdamp ) = 3850 × 9.4/(2π × 1.0) = 5740 kNs/m.
For the damper configuration under consideration (see Fig. 9) it is apparent that this damper
constant will be provided for each deck in the longitudinal direction by two dampers at 45◦ .
As such, each damper should possess a damper constant of 5470/2 × 1.414 = 4,100 kNs/m
= 4.1, MNs/m.
As discussed earlier, the seismic response in the transverse direction can be influ-
enced by various higher modes, and for preliminary sizing it is suggested that the damper
characteristics obtained for the longitudinal direction be utilised also in the transverse
direction.
TABLE 4 Comparison of preliminary member sizes with those obtained after detailed
analyses
Detailed design soln Preliminary sizing
Main Beam Section Area (m2 ) 0.127 0.122
Average Cable Area (m2 ) 0.008 0.007
Tower Pier Section Area at deck 118 120
level (m2 / per deck)
Damper Constant (MNs/m) 4.5 4.1
At this stage therefore, the preliminary sizing procedure is complete; the overall geom-
etry, average cable size, main deck beam sizes, tower piers and deck-to-pier connection
characteristics have been identified. This information permits the development of a reason-
able structural model for advanced verification analyses of the performance of the bridge.
While some modification to the member sizes should be expected as a result of advanced
verification analyses, the preliminary sizing should ensure that any changes required are
not great.
As explained in the introduction to this section, the design scenario examined above
corresponds to the South Crossing cable-stayed bridge of Guayaquil. As such, the effec-
tiveness of the approach can be gauged by comparing the preliminary sizes with the final
sizes identified after detailed verification analyses of the bridge were undertaken. Table 4
compares the detailed design member sizes for the South Crossing with those obtained
from the preliminary sizing procedure.
The comparison presented in Table 4 suggests that the preliminary sizing procedure
can be an effective tool for design. The main beam sections for the deck, the average cable
area, the tower pier sections and the damper constants all match the preliminary estimates
well. Future research should investigate the effectiveness of the approach when applied to
other cable-stayed bridges.
6. Conclusions
This article has reviewed and discussed some of the important conceptual design consider-
ations for cable-stayed bridges, first for gravity loads and then for seismic excitation. The
advantages and disadvantages of different cable-stayed bridge solutions have been high-
lighted, with review of deck sections, tower configurations in both the longitudinal and
transverse directions, deck-to-pier connections, and cable arrangements. It has been argued
that the use of intermediate-type deck-to-pier connections is likely to provide the best con-
trol of seismic response, as they can limit the actions imposed onto the bridge towers which
should be stiff in order to limit deck displacements under both gravity and seismic loads.
Reference has been made to a number of real cable-stayed bridge solutions.
In addition to identifying important conceptual design issues, a preliminary sizing pro-
cedure has been proposed. In addition to a series of simple design expressions for member
sizes and bridge proportions, a novel formulation of the direct displacement-based design
procedure has been proposed for identification of suitable deck-to-pier connection charac-
teristics. The preliminary sizing procedure is intended to offer designers a quick but rational
means of identifying reasonable member sizes for cable-stayed bridges that should then
be verified through advanced analyses in the developed and detailed design stages of the
project. A case study application of the preliminary sizing procedure has been made for a
three-tower cable-stayed bridge in Ecuador and by comparing preliminary and final design
member sizes it is concluded that the preliminary sizing procedure may be a useful tool
for design. Future research should investigate the effectiveness of the simplified conceptual
design approach when applied to other cable-stayed bridges.
References
Ali, H. M. and Abdel-Ghaffar, A. M. [1995] “Seismic passive control of cable-stayed bridges,” Shock
and Vibration 2(4), 259–272.
Arzoumanidis, S., Shama, A., Ostadan, F. [2005] “Performance-based seismic analysis and design of
suspension bridges,” Earthquake Engineering and Structural Dynamics 34, 349–367.
Calvi, G. M., Moratti, M., Pinho, R., Villani, A., and Pietra, D. (2011) “Progettazione Sismica di
un ponte strallato di grande luce: il South Crossing Bridge in Guayaquil, Ecuador,” Submitted to
Progettazione Sismica, IUSS Press, Pavia, Italy.
CEN EC8 [2004], Eurocode 8 – Design Provisions for Earthquake Resistant Structures, EN-1998-1:
2004: E, Comite Europeen de Normalization, Brussels, Belgium.

Chang, C. M. and Loh, C. H. [2006] “Seismic response control of cable-stayed bridge using different
control strategies,” Journal of Earthquake Engineering 10(4), 481–508.
Combault, J., Pecker, A., Teyssandier, J. P., and Tourtois J. M. [2005] “Rion-Antirion Bridge,
Greece – concept, design, and construction,” Structural Engineering International 1/2005,
22–27.
Conti, E., Grillaudb, G., Jacobb, J., and Cohen, N. [1996] “Wind effects on the Normandie cable-
stayed bridge: Comparison between full aeroelastic model tests and quasi-steady analytical
approach,” Journal of Wind Engineering and Industrial Aerodynamics 65, 189–201.
Elnashai, A. S., Spencer, B. F., Kuchma, D. A., Yang G., Carrion J., Gan Q., and Kim, S. J. [2006]
“The multi-axial full-scale sub-structured testing and simulation (MUST-SIM) facility at the uni-
versity of Illinois at Urbana-Champaign,” in Advances in Earthquake Engineering for Urban Risk
Reduction, eds. S. T. Wasti and G. Ozcebe (Springer, The Netherlands), 66, 245–260.
Faccioli, E., Paolucci, R., and Rey, J. [2004] “Displacement spectra for long periods,” Earthquake
Spectra 20(2), 347–376.
Fajfar, P. and Krawinkler, H., Eds. [1997]Seismic Design Methodologies for the Next Generation of
Codes, Balkema, Rotterdam, The Netherlands.
Fournier d’Albe, E. M. [1982] “An approach to earthquake risk management,” in Engineering
Structures, Vol. 4) (Elsevier Ltd., London, U.K.), pp.147–152.
Ge, Y. J. and Tanaka, H. [2000] “Aerodynamic flutter analysis of cable-supported bridges by multi-
mode and full-mode approaches,” Journal of Wind Engineering and Industrial Aerodynamics 86,
123–153.
Gimsing, N. J. [1997] Cable Supported Bridges: Concept and Design (2nd ed.), John Wiley & Sons,
New York.
Grant, D., Bommer, J. J., Pinho, R., and Calvi, G. M. [2006] “Defining Priorities and Timescales
for Seismic Intervention in School Buildings in Italy,” Research Report ROSE - 2006/03, IUSS
Press, Pavia.
ICC [2006] International Building Code, International Code Council, USA.
He, W. L., Agrawal, A. K., and Mahmoud, K. [2001] “Control of seismically excited cable-stayed
bridge using resetting semiactive stiffness dampers,” Journal of Bridge Engineering (ASCE) 6,
376–384.
Infanti, S., Papanikolas, P., Benzoni, G., and Castellano, M. G. [2004] “Rion-Antirion bridge: design
and full-scale testing of the seismic protection devices,” 13th World Conference on Earthquake
Engineering, Vancouver, B.C., Canada, Paper No. 2174.
Jones, N. P. and Scanlan, R. H. [2001] “Theory and full bridge modeling of wind response of cable
supported bridges,” Journal of Bridge Engineering, ASCE 6(6), 365–375.
Jung, H-J., Spencer, Jr. B. F., and Lee, I-W. [2003] “Control of a seismically excited cable-
stayed bridge employing MR fluid dampers,” ASCE Journal of Structural Engineering 129,
873–883.
Kawashima, K., Unjoh, S., and Azuta, Y. [1988] “Damping characteristics of cable-stayed bridges,”
Proc. 9th WCEE, Tokyo/Kyoto, Japan, pp. 471–476.
Kawashima, K., Unjoh, S., and Azuta, Y. [1990], “Analysis of damping characteristics of a
cable stayed bridge based on strong motion records,” Proc. JSCE, Structural Engineering and
Earthquake Engineering, 7–1, 169–178.
Kawashima, K., Unjoh, S., and Tunomoto M. [1993] “Estimation of damping ratio of cable-stayed
bridges for seismic design,” Journal of Structural Engineering, ASCE 119(4), 1015–1031.
Mylonakis, G. and Gazetas, G. [2000] “Seismic soil-structure interaction: beneficial or detrimental?,”
Journal of Earthquake Engineering 4(3), 277–301.
Papazoglou, A. J. and Elnashai, A. S. [1996] “Analytical and field evidence of the damaging effect
of vertical earthquake ground motion,” Earthquake Engineering and Structural Dynamics 25,
1109–1137.
Park, K-S., Jung, H-J., Yoon, W-H., and Lee, I-W. [2005] “Robust hybrid isolation system for a
seismically excited cable-stayed bridge,” Journal of Earthquake Engineering 9(4), 497–524.
Pecker, A. [1998] “Capacity design principles for shallow foundations in seismic areas,” Keynote
lecture, 11th European Conf. on Earthquake Engineering, September, Paris.

Pecker, A. and Salençon, J. [1999] “Ground reinforcement in seismic areas,” Proc.11th Panamerican
Conf. on Soil Mechanics and Geotechnical Engineering, 2, 799–808.
Priestley, M. J. N., Calvi, G. M., and Kowalsky, M. J. [2007] Direct Displacement-Based Seismic
Design, IUSS Press, Pavia.
Ruangrassamee, A. and Kawashima, K., [2006] “Seismic response control of a cable-stayed bridge
by variable dampers,” Journal of Earthquake Engineering 10(1), 153–165.
Schemmann, A. G. and Smith, H. A. [1998a] “Vibration control of cable-stayed bridges. Part 1:
Modeling issues,” Earthquake Engineering and Structural Dynamics 27, 811–824.
Schemmann, A. G. and Smith, H. A. [1998b] “Vibration control of cable-stayed bridges. Part 2:
Control analyses,” Earthquake Engineering and Structural Dynamics 27, 825–843.
Schlaich, M. [2001] “Erection of cable-stayed bridges having composite decks with precast concrete
slabs,” Journal of Bridge Engineering 6(5), 333–339.
Simiu E. and Scanlan R. H. [1996] Wind Effects on Structures (3rd ed.), Wiley, New York.
Sullivan, T. J. [2009] “Direct displacement-based design of a RC wall-steel EBF system with added
dampers,” Bulletin of the New Zealand Society for Earthquake Engineering 42(3), 167–178.
Turan, G. [2001] “Active control of a cable-stayed bridge against earthquake excitations,” Ph.D.
Dissertation, University of Illinois at Urbana-Champaign, Department of Civil Engineering.
Virlogeux, M. [2001] “Bridges with multiple cable-stayed spans,” Structural Engineering
International 11(1), 61–82.
Walther, R., Houriet, B., Isler, W., Moia, P., and Klein, J. F. [1999] Cable Stayed Bridges (2nd ed.),
Thomas Telford, London, U.K.
Wilson, J. C. and Gravelle, W. [1991] “Modelling of a cable-stayed bridge for dynamic analysis,”
Earthquake Engineering and Structural Dynamics 20, 707–721.
Wolf, J. P. [1985] Dynamic Soil-Structure Interaction, Prentice-Hall, Englewood Cliffs, New Jersey.

Conceptual Seismic Design of Cable-Stayed Bridge

Uploaded by

Copyright:

Available Formats

Conceptual Seismic Design of Cable-Stayed Bridge

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Conceptual Seismic Design of Cable-Stayed Bridge

Uploaded by

Copyright:

Available Formats

This article was downloaded by: [University of Arizona]

On: 12 January 2013, At: 20:34

Journal of Earthquake Engineering

Conceptual Seismic Design of Cable-

To link to this article: http://dx.doi.org/10.1080/13632469.2010.505275

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

Conceptual Seismic Design of Cable-Stayed Bridges

G. M. CALVI1 , T. J. SULLIVAN1 , and A. VILLANI2

Received 27 February 2010; accepted 26 June 2010.

FIGURE 1 Typical elevation of a cable-stayed bridge.

2. Performance Objectives and Design Loads

● Serviceability limit state (Performance Level 1) in which no elements suffer notable

emergency services and should be easily repaired.

Seismic Risk = Hazard x Vulnerability (2)

3. Important Conceptual Design Considerations for Non-Seismic Loads

Cables Aligned with

Composite Concrete Deck

Secondary steel beams to transfer

Note that longitudinal and transverse

FIGURE 3 Typical composite deck section of a cable-stayed bridge.

● Deck continuity: The deck should be continuous in order to permit redistribution of

(a) (b) (c) (d) (e) (f)

FIGURE 4 Transverse view of tower pier configurations for cable-stayed bridges.

(a) (b) (c)

FIGURE 5 Longitudinal view of tower pier configurations for cable-stayed bridges.

4. Concept Design Considerations for Seismic Loads

4.1. Deck to Pier Connection

Detail A Connect Deck

FIGURE 6 Deck-to-pier connection consideration.

FIGURE 7 Deck-to-pier connection utilized in the Rion-Antirion Bridge.

Central damper TOWER Two dampers at 45° to

TABLE 3 Effect of pier-to-deck connection typology on the displacements of the

Elevation of Cable Forces & Pier Moments under

Plan View of Deck Moments under

FIGURE 10 Simplified representation of force distributions for longitudinal response

4.2. Tower Pier Design

elevation and bending moment diagram.

Single Piers in Would coupled piers …pier coupling up

(a) (b) (c)

Critical cable strains likely

δcable = (x − p ) cos α, (5)

4.3. Tower Base Design

4.4. Foundation Design and Soil-Structure-Interaction Considerations

Tower piers from

Require continuity of flexural resistance

Large tower base section

FIGURE 14 Partial elevation of typical tower base section.

5. Preliminary Sizing of the Main Structural Elements of Cable-Stayed

Bridges in Seismic Regions

FIGURE 16 Elevation of a typical cable-stayed bridge with key dimensions indicated.

Ndeck − tslab .beff .0.85f c

Maximum longitudinal force

FIGURE 17 Capacity design force considerations for longitudinal seismic response of a

Mpier = wdeck .Lspan .x, (14)

(a) Eccentric Gravity Loading

Back-stay effect at spans

(b) Deformed Shape

FIGURE 20 Deformed shape of bridge under longitudinal seismic excitation.