Seismic Protection on Cable-Stayed Bridges

Simão M. Freire
Departamento de Engenharia Civil of Instituto Superior Técnico, October 2014
____________________________________________________________
Abstract I Seismic Protection is a fundamental issue when it comes to high seismic risk areas design.
This protection becomes even more important when we talk about crucial structures such as bridges.
Cable-stayed bridges are obviously included in this crucial group of structures. There are several seismic
protection systems available and it´s important to make the right choice when one of them is needed. The
right approach to this subject is to make an exhaustive comparison between the systems considered
relevant in a particular case. When dealing with a cable-stayed bridge the designer should first of all
understand correctly the dynamic behavior of the structure. Ground motions force the deck, the pylons
and the stays to oscillate. By this time the points of the structure with potential to install a device should be
identified. This study aims at identifying these points and also to install the most suitable devices in the
most suitable places. The tests will include a viscous damper, a hysteretic damper and also a Tuned Mass
Damper (TMD). The three systems will be installed in a cable-stayed bridge with a span of 420m. The
goal of this study is the minimization of deck displacement. This displacement is important because it
affects the design of structural joints, connections with overpasses and also abutments. In Seismic
Protection the forces the devices introduce in the structure should be evaluated. The final choice of a
device of this type should always be a balanced commitment between the minimizing of the longitudinal
deck displacement and the increase of the stresses transmitted to the original structure.

Keywords :Seismic Protective Systems, Cable Stayed Bridges, Longitudinal Displacements, Viscous
Damper, Hysteretic Damper, Tuned Mass Damper
____________________________________________________________
1.Introduction
displacement. Any action taken to reduce the
longitudinal displacement, should be analyzed in
The design of bridges in seismic areas, such as comparison with the situation in which the deck
Portugal, requires quite important concerns has a joint to accommodate the movement . The
since the conception phase. This includes challenge is the reduction of the deck
dynamic analysis and design (Reis, 2006). So it displacement and the efficiency of the protection
is not possible to separate the structural solution system tested. A compromise between them
from the seismic resistance system of the bridge must be found. In this paper was only
whatever it is. A cable-stayed bridge is no considered the longitudinal component of the
exception and this should, since its conception, seismic action. The vertical and transversal
contemplate earthquake resistance through its components were neglected because the design
own capacity of deformation or through of long bridges in seismic areas frequently leads
protection systems. It is important to realize to solutions with very low horizontal stiffness in
what an earthquake is to such a structure and the longitudinal direction (Guerreiro et al, 1998)
how much damage can it cause to it. After the
designer can decide which is the best strategy Many seismic isolation construction designs and
to protect the bridge. technologies have been developed over the
years in attempts to mitigate the effects of
The goal of this study is the minimization of deck earthquakes on buildings, bridges and

1
vulnerable contents (Skinner et al, 1993). Where F is the force in the damper, C and α are
Developments on this area have been made damper`s coefficients and is the velocity
mainly using the deformability of the structure between two points of the damper.
itself, changing its dynamic properties or its The initial design of this damper was made
ability to dissipate energy. considering α=0,2 and C as 10% of the deck
weight. It was also defined that the viscous
-5
It is intended to study two widely known damper has a maximum deformation of 10 m.
dampers: the viscous damper and the hysteretic
damper. These two types of damper exhibit a Hysteretic Damper
versatility that favors their use, because they
are easy to introduce into the structural system This device is usually a steel cantilever that uses
and also because they give the designer metal plasticity to dissipate energy (Figure 2)
freedom to define their characteristics This device is defined by its stiffness (K1) and by
(Guerreiro, 2006). its yielding force (Fy). When the damper reaches
it’s yielding force the stiffness beyond that
This paper also aims to study the applicability of moment is a percentage of K1. This stiffness is
a Tuned Mass Damper (TMD) to a cable-stayed called post-yielding stiffness (K2).The initial
bridge. The final choice of a device of this type sizing of this device was made considering
should always be a balanced commitment and K1 corresponding to an elastic
between the minimizing of the longitudinal deck
displacement of the damper equal to 6% of the
displacement and the increase of stresses
deck displacement in seismic case. The yielding
transmitted to the original structure.
force (Fy) was considered equal to the maximum
force registered in the viscous damper.
The devices are designed and applied to a case
study – alternative Vasco da Gama Bridge –
and the main results and conclusions are a
summary of the research project on “Seismic
Protection on Cable-Stayed Bridges” (Freire,
2014). Nonlinear analysis follows Eurocode 8
requirements applicable to Lisbon area.

2. Seismic Protection Devices

Viscous Damper

This device operates according to the equation

[1] (Figure 1). Figure 2 – Force-Displacement relation for
Hysteretic Damper (Guerreiro, 2006)
( )

Tuned Mass Damper

This device is a mass attached to the structure

in order to reduce its dynamic response (Figure
3). The TMD is tuned to the structure
fundamental frequency to produce inertia forces
that counterbalance the original motion.

The initial design of this device was made by

fixing its mass between 0,5% and 10% of the
deck´s mass.

Figure 1 – Force-Velocity relation for Viscous

Damper ( Guerreiro, 2009)

2
 The bridge structure is similar to the one
constructed for Vasco da Gama Bridge. The
main difference is the material adopted for the
deck (Figures 5 and 6). The original structure
has a pre-stressed concrete deck and the case
study a composite steel-concrete one.

The bridge has two towers of 150m high (Figure

7), one central span and two side spans. The
Figure 3 – Tuned Mass Damper model (N1) solution in this case consists of a central span of
420 m and two side spans of 194.7m, reaching a
3. Seismic Data total length of 829 m. The deck is completely
independent from the towers. Regarding the
The Eurocode 8 requirements were applied but Cable-Stayed system case study is similar to
the maximum value of soil acceleration was Vasco da Gama Bridge. The system is a vertical
2
defined as 4 m/s . Structural damping was group of stays located in each one of the main
considered as 5%. deck beams. As said deck is a composite steel-
concrete element. The concrete is C45/55 class
and the reinforcements made of S500 NR. The
4. Case Study structural steel class is S355 NL Towers and
intermediate piers are made of reinforced
4. 1 Bridge Description concrete also (C40/50 and S500 NR for
reinforcements).The modulus of elasticity of the
The case study is a Cable-Stayed bridge stays was considered 195 GPa (constant). For
developed by José Oliveira Pedro as an the stays was used high-resistance stainless
alternative study for the conception of the Vasco steel.
da Gama Bridge in Lisbon (Figure 4).

4 7

Figure 4 – General view of Vasco da Gama Bridge (N2) ; Figure 5 - Transversal section of the Deck
(Oliveira Pedro, 2007) ; Figure 6 - Modeling detail of the composite section (Oliveira Pedro, 2007)
Figure 7 – Tower (Gattel, 1999)

3
The next step was the dynamic analysis of the This is the confirmation that the structure is in
fixed base structure. The mass of the structure fact very flexible due to longitudinal horizontal
is divided in the following way (half bridge): displacement. The seismic analysis was made
for 10 artificial earthquakes generated according
Table 1 – Structure´s Mass by element to the mentioned conditions. As expected the
Mass [ton] % Structure´s Mass bridge without any seismic protection device
leads to important longitudinal displacements
Beam 2661 9% (Table 2). The displacements were monitored in
Deck 11799 42% three different points. These points are the top of
the tower (Top), the point of the tower at deck
Stays 577 2% height (Tower) and a point of the deck (Deck).
Towers 13152 47% The deck oscillates as a rigid body so all his
points have the same displacements.
Total 28190 100%
The displacements presented are the linear
Modal dynamic analysis demonstrated that the combination of the maximum value obtained for
bridge is characterized by a fundamental each earthquake.
frequency of 0,18 Hz. This fundamental
frequency corresponds to the longitudinal mode. Table 2 – Displacements for free structure
Point Disp. [m]
Top 0,59
Deck 0,52
Tower 0,05
st
1 Mode f=0,18 Hz Relative Deck-Tower 0,47

The deck presents considerable displacements

of average value of 0,52 m.
Unexpectedly the top of the tower presents
displacements even bigger than the deck. It is
nd
2 Mode f=0,32 Hz important to understand if the deck and the top
oscillate always together or if they are moving
out of phase. The tower remains almost at the
same position and so there is a big relative
motion between it and the deck. Here is the
opportunity to install some seismic protection
rd
3 Mode f=0,41 Hz devices such as viscous and hysteretic
Figure 8 – Fundamental vibration modes of the dampers. In figure 9 is possible to compare
bridge different point displacements during an
earthquake.

0,6
Long. Displacement [m]

0,4

0,2

-1E-15
0 5 10 15 20 25 30 35 40
-0,2

-0,4

-0,6
Time [s] Deck Top Torre
Tower
Figure 9 – Free structure response to an earthquake

4
Deck and top move in fact together. It will be In spite of this the relation between the two
interesting to discover if it is possible to things is not linear. In fact for small forces the
decrease deck displacements by decreasing the decrease of deck displacement is considerable
top ones. Although is more likely the top follows but for bigger forces is quite small. The
the deck instead of the opposite. optimized damper should be the one where the
decrease of displacement compensates
4.2 Viscous Damper increase of forces. To make this analysis was
built a graphic where is possible to relate deck
It was decided that the bridge would have four displacement and force transmitted to the bridge
dampers per tower. So the Viscous Damper for each value of C.
modeled for each tower corresponds to two real 100%
dampers. This happens because the

% (C=8000 𝑘𝑁/(𝑚/𝑠)
computational model was built for half of the 80%
bridge. The optimization of the Viscous Damper
60%
coefficients was made initializing the system
with the following values: 40%

20%

0%
0 2000 4000 6000 8000
C [𝑘𝑁/(𝑚/𝑠)]
Table 3 – Initial Viscous Damper results F Deck disp.
Force [kN] Deck Disp. [m]
5115 0,18 Figure 11 – Influence of coefficient C in the
maximum forces transmitted to the towers and
The results for this initial Viscous Damper were maximum deck displacements
quite interesting. It was achieved a decrease of
deck displacement of 66%. Of course this The optimal Viscous Damper due to the criteria
results lead to a force in the damper that is defined is C=3500 .It was suggested by the
transmitted to the structure. Values of the intersection of the curves in figure 11.
coefficient C between 0 and 8000 were and α=0,2 were values already
considered in the optimization process. The fixed.
stiffness remained the same. The results
obtained are in figure 10. The main goal of this paper is to reduce seismic
response by energy dissipation. So it is
7000 important to analyze the hysteresis loop of the
6000 Damper designed (Figure 12)
5000 4000
4000 3000
F [kN]

3000 2000
2000
F [kN]

1000
1000
0
0
-0,3 -0,2 -0,1-1000 0 0,1 0,2
0,100 0,200 0,300 0,400 0,500 0,600
Deck Displacement [m] -2000
Figure 10 – Maximum forces transmitted to the -3000
towers and maximum deck displacements for
different C coefficients -4000
Displacement [m]
The deck displacements decrease when the Figure 12 – Optimized Viscous Damper
forces transmitted to the towers increase. hysteresis loop

5
Deck displacement obtained for the optimized Of course knowing the results from the Viscous
Viscous Damper was 0,23m. This is a reduction damper was a big help. Just as said before this
of 56% in deck displacement. The force study aims to achieve the optimization of the
transmitted to the tower was 3048 kN. The hysteretic damper. The results obtained for this
hysteresis loop correspondent to this solution initial solution were the ones presented in table
(Figure 12) indicates the expected energy 3.
dissipation.
Table 3 – Initial Hysteretic Damper results
4.3 Hysteretic Damper Force [kN] Deck Disp. [m]
3553 0,25
The optimization of the Hysteretic Damper was
made knowing already the results for the The initial results show a reduction of deck
optimized Viscous Damper. The rules defined to displacement of 52%. These results were
initialize the design were used but starting considered good. In spite of that the optimized
already from the previous results. hysteretic damper was still the goal to achieve.

To understand the influence of the damper

stiffness and yielding force in deck displacement
their values were changed. The results were
analyzed always taking into account the force
transmitted to the towers. The ratio was
considered as a fixed value.

The first analysis was made considering a fixed

As mentioned this study pretends to obtain stiffness of 250000 and yielding forces
equivalent hysteresis loops for both viscous and
hysteretic dampers. At this stage the between 0 and 4000 kN. The maximum forces
comparison between the two loops was made transmitted to the towers were assumed as
(Figure 13). being very close to Fy values in each case. So
figure 14 is the influence of yielding force in
4000 maximum deck displacement. Of course yielding
force is representing the maximum force
3000 transmitted to the towers.
2000
0,6
Deck diplacement [m]

1000
F [kN]

0 0,5
-0,25 -0,15 -0,05
-1000 0,05 0,15
0,4
-2000
0,3
-3000
0,2
-4000
Displacement [m] 0 1000 2000 3000 4000
Fy [kN]
Viscous Series2
Hysteretic
Figure 14 – Influence of yielding force in the
Figure 13 – Comparison between Optimized maximum deck displacements
Viscous Damper and Initial Hysteretic Damper
hysteresis loops Once more the reduction of deck displacement
is achieved by increasing structure´s forces. The
The two loops obtained in figure 13 show that relation between these two parameters is not
energy dissipation is equivalent in the two linear. At first the increase of yielding force is
Dampers. So it is possible to conclude that the very attractive but then it is causing less and
initial rules defined for the design of the less deck displacement reduction.
Hysteretic Damper were correct.

6
Understanding the most effective value of Fy is Deck displacement is very little influenced by the
needed. For that was produced the graphic stiffness of the damper. In fact deck
(Figure 15), where force transmitted to the displacement is almost constant over the entire
structure and deck displacement were plotted range of stiffness tested. The adequate stiffness
for every Fy value. remains the initial one.

Due to the criteria defined for yielding force and

100% stiffness the optimized hysteretic damper is
90% characterized by Fy=2000 kN and K1=250000 .
% [Fy=4000 kN]

80% was an equation already fixed.

70%
60% The hysteretic loop obtained for this Damper is
50% presented in figure 17.
40%
3000
30%
0 1000 2000 3000 4000 2000
Fy [kN]

F [kN]
1000
Deck disp. F
0
Figure 15 - Influence of yielding force Fy in the -0,25 -0,15 -0,05 0,05 0,15
maximum forces transmitted to the towers and -1000
maximum deck displacements
-2000
The optimized Hysteretic Damper due to the -3000
criteria defined is Fy=2000 kN. It was suggested Displacement [m]
by the intersection of the curves in figure 15.
Figure 17 – Optimized Hysteretic Damper
Beyond this moment Fy value was fixed. As
hysteresis loop
mentioned the next step was to considered
different values of stiffness K1. Stiffness was
The deck displacement obtained for the
considered between 50000 and 1000000 . optimized Hysteretic Damper was 0,28m. This is
a reduction of 47% in deck displacement. The
The forces transmitted to the structure were force transmitted to the tower was 2634 kN. The
very increased by the increase of stiffness. On hysteresis loop correspondent to this solution
the other hand deck displacement didn´t suffer a (Figure 17) indicates the expected energy
lot with this modification. In figure 16 is possible dissipation.
to conclude that changes in stiffness value are
not important. 4.4 Tuned Mass Damper
100% The research project on “Seismic Protection on
% (K=1000000 kN/m)

90% Cable-Stayed Bridges” (Freire, 2014) includes,

80% as mentioned, the case study response for
viscous and hysteretic dampers but also the
70% response for a Tuned Mass Damper. In this
60% research project is proved that TMD`s are not
50% effective for seismic protection on cable-stayed
bridges. The tests include a TMD attached to the
40%
deck and also one positioned in the top of each
0 K [kN/m] 500000 1000000 tower. None of them is effective. In the study is
Deck disp. F also proved that the TMD is not effective in this
Figure 16 - Influence of stiffness K1 in the case because the seismic action is not
maximum forces transmitted to the towers and harmonic. In case the action considered was
maximum deck displacements harmonic the TMD attached to the bridge´s deck

7
would have been as much effective as any of 5. Comparative analysis
the dampers tested. This paper does not include
the complete TMD analysis study. For that is In this section is made a brief comparison
suggested that the research project is between the viscous damper and the hysteretic
consulted. Anyway is interesting to make the damper. Both solutions are analyzed as possible
comparison between the seismic and the protection systems for a cable-stayed bridge. In
harmonic action. So it is presented the deck section 4.4 the TMD was excluded from the valid
displacement for a seismic and an harmonic options.
action (sine). The case presented (Figure 18) is
the bridge attached with a TMD in its deck. The optimization process led to two devices with
very good results in the decreasing of deck
0,6 displacements. These two devices also assure
Deck displacement [m]

0,5 that the forces transmitted to the structure are

not exaggerated. The devices this study elects
0,5 are presented in table 4.
0,4
Table 4 – Elected Devices
0,4 Hysteretic Damper Viscous Damper
0,3
n=20
0,3 α=0,2
0,2
0,0% 1,0% 2,0% 3,0% 4,0%
mTMD [%mTOTAL]
Sine Earthquake
Figure 18 – Structure with TMD response to
seismic and harmonic action

0,6

0,4
Deck displacement [m]

0,2

0
0 5 10 15 20 25 30 35 40

-0,2

-0,4

-0,6
Time [s]
Original Hysteretic Viscous

Figure 19 – Deck displacement during an earthquake for original and protected structure (optimized)

8
 3000

2000

1000
F [kN]

0
0 5 10 15 20 25 30 35 40

-1000

-2000

-3000
Time [s]
Hysteretic Viscous
Figure 20 – Force in optimized dampers during an earthquake

4000

3000

2000

1000
F [kN]

0
-0,25 -0,2 -0,15 -0,1 -0,05 0 0,05 0,1 0,15 0,2 0,25
-1000

-2000

-3000

-4000
Displacement [m]
Hysteretic Viscous
Figure 21 – Hysteresis loop for optimized dampers

In figure 19 the deck displacement is compared A summary of the results in terms of forces
during an earthquake in the original structure transmitted to the structure and displacements is
and also in the structure protected with the provided in table 5.
optimized dampers. In figure 20 is plotted the
force in the optimized dampers and in figure 21 At first glance it may seem that the viscous
the hysteresis loop during the seismic action. damper is more effective than the hysteretic one.
Displacements and forces are very similar in the In fact it is not true because the optimization
two dampers. Hysteresis loops are almost process doesn´t have the detail to confirm that at
equivalent in area as required. The shape of the this level of differences. Furthermore this study
hysteresis loop is of course different in the two doesn´t take in account the economic factor. The
dampers because they have different behavior consideration of this factor can bring a lot of
equations. interesting questions to this field of study.

9
Tabela 5 – Optimized results References
Viscous Hysteretic
Damper Damper
Deck
0,23 0,28 Freire, S.M. – “ Proteção Sísmica de Pontes de
displacement [m] Tirantes ”, Master´s Thesis, Instituto Superior
Deck Técnico – Universidade Técnica de Lisboa,
displacement 56% 47% October 2014
reduction
Force transmitted Gattel – “ Ponte Vasco da Gama ”, Ministério do
to the structure 3048 2634 Equipamento, do Planeamento e da
[kN] Administração do Território, Secretaria de
Estado e das Obras Públicas,1ª Edição, 1999
6. Conclusions Guerreiro, L.; Branco, F.; Azevedo, J. – “
Seismic Displacement Control of the Vasco da
In this type of bridges the towers deform. It is Gama Bridge ”, Fédération Internationale du
not true that the deck oscillates anchored in the Béton, 1998
tower. The deck is so flexible and long that the
top of the two towers follow its movement. The Guerreiro, L. – “ Sistemas de Dissipação de
deck displacement without any seismic Energia ”, Dinâmica e Engenharia Sísmica,
protection device is very high. It is mandatory to Departamento de Engenharia Civil,2006
reduce this displacement. In other case the
designer will have to think in large joints to Guerreiro, L. – “ Estratégias para a Melhoria do
accommodate these movements. It is Comportamento Sísmico de Edifícios ”,
considered appropriate that in future Seminário – Reabilitação Sísmica de Edifícios,
developments, seismic protection and joint Algarve, Portugal,2009
design for these type of bridges are done
together. Oliveira Pedro, José – “ Pontes Atirantadas
Mistas. Estudo do Comportamento Estrutura l”,
Viscous and Hysteretic Dampers were elected Tese de Doutoramento, Departamento de
as the ones capable of decreasing the deck Engenharia Civil do Instituto Superior Técnico,
displacement for cable-stayed bridges. Tuned Lisboa, Portugal, Julho de 2007
Mass Dampers are not effective in this field of
study. Oliveira Pedro, José – “ Pontes de Tirantes –
Concepção, Dimensionamento e Constituição ”,
The follow up of this research project can be the Elementos de Apoio à Disciplina de Pontes de
analysis of different case studies and also the Tirantes do Diploma de Formação Avançada em
inclusion of several devices on the same bridge. Engenharia de Estruturas do Instituto Superior
Testing new positioning for the devices can also Técnico, Lisboa, Portugal, Junho de 2010
be interesting.
Reis, A. J. – “ Folhas da Disciplina de Pontes ”,
The choice of a seismic protection system Departamento de Engenharia Civil do Instituto
should be guided by several variables. Some of Superior Técnico, Lisboa, Portugal,2006
them were not considered in this paper. In fact
the choice of a device of this type should be an Skinner, R. I.; Robinson, W.H.; McVerry, G. H. –
integrated process that include longitudinal “ An Introduction to Seismic Isolation ”, DSIR
displacement, force transmitted to the structure, Physical Sciences, Wellington, New Zealand,
displacement in other directions, devices cost, John Wiley &Sons,1993
devices installation cost, joints cost and
maintenance cost of everything involved during N1, www.iopscience.iop.org, July 2014
the lifetime cycle of a cable-stayed bridge.
N2, www.trevispa.com, September 2014

10