You are on page 1of 10

Seismic Design of Lifeline Bridge using Hybrid

Seismic Isolation
Murat Dicleli, P.E., M.ASCE1
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This paper presents the merits of a hybrid seismic isolation system used for the seismic design of a major bridge. The bridge
is analyzed for two different arrangements of seismic isolation systems. The first arrangement consists of friction pendulum bearings at all
substructure locations; the other incorporates a hybrid system where laminated elastomeric bearings are used at the abutments and friction
pendulum bearings at the piers. Analysis results have demonstrated that the hybrid seismic isolation system provided a structure with a
fundamental period long enough to attract smaller seismic forces, while controlling the magnitude of isolation bearings displacements. It
also provided a more uniform distribution of seismic forces among substructure elements. As a result, higher seismic forces on the piers
were reduced, allowing for a more economical design of substructures. The hybrid seismic isolation system helped to control the
wind-induced vibrations and reduced the sizes of the isolation bearings.
DOI: 10.1061/共ASCE兲1084-0702共2002兲7:2共94兲
CE Database keywords: Seismic design; Bridges; Seismic isolation; Friction; Damping.

Introduction The concept of seismic base isolation has been adopted into
practice with the development of natural rubber bearings 共Kelly
Highway 17, in Ontario, Canada, forms part of the Trans-Canada 1997; Naeim and Kelly 1999兲 and became more popular with
Highway and is a strategic link in the interprovincial highway laminated elastomeric 共LE兲 bearings 共Derham et al. 1975; Naeim
system. Along this route, a two-lane highway is currently serving and Kelly 1999兲. Other base isolation systems have also been
provincial, regional, and local traffic. In response to identified developed, including low damping natural rubber bearings 共Kelly
capacity deficiencies and growing public concern, the construc- and Quiroz 1992兲; high damping natural rubber bearings 共Derham
tion of a four-lane highway system was initiated. This new high- et al. 1985兲, lead-plug bearings 共LEP兲, which were invented in
way system contains a number of structures, including the twin New Zealand 共Robinson and Tucker 1977, 1983兲 and have been
used extensively for the seismic base isolation of buildings 共Char-
structures of the Mississippi River crossing.
leson et al. 1987兲 and bridges 共Built 1982兲; and friction pendulum
The design of the Mississippi River twin bridge presented a
共FP兲 bearings 共Zayas et al. 1987, 1996兲, which have been used in
number of unique challenges. The bridge is located in a valley
the seismic design of buildings 共Al-Hussaini et al. 1994兲 and
that, from a geological point of view, is famous for its sensitive
bridges 共Constantinou et al. 1993; Zayas et al. 1996兲. Both the FP
Leda clay and possesses the highest-risk seismic zone classifica-
and LE bearings with or without a lead plug have been incorpo-
tion in Ontario with a zonal acceleration of 0.2g 共Canadian 2000兲. rated into design codes 共Blakeley 1982; Ministry 1983; Interna-
The bridge crosses a 78 m wide river and a Class A provincial tional 1997; AASHTO 1999兲.
wetland. The substructures would not be allowed within the riv- A number of different seismic isolation bearings mentioned
er’s watercourse, and their size and locations were restricted to above were investigated, and a hybrid solution was deemed the
minimize the disturbance to the wetlands. most appropriate for the seismic design of the bridges. The hybrid
The relatively large size of the structures and the higher-risk system consists of FP bearings at the piers and LE bearings at the
seismic zone classification of the area resulted in seismic forces abutments. This is the first bridge to be built with a seismic iso-
that could not be accommodated with a conventional structural lation design of this kind in Canada and possibly in the United
system and satisfy the imposed environmental constraints at the States 共Michael Constantinou, personal communication, June 14,
same time 共the limitation on the footing plan area was 64 m2兲. 2000兲.
Therefore, a structure with a seismic isolation system was consid- This paper describes the step-by-step procedure for the seismic
ered to reduce and evenly distribute the horizontal forces to the design of the bridges and outlines the merits of hybrid seismic
substructures. isolation. In the subsequent sections, first the site and bridge prop-
erties will be defined, and then the preliminary design criteria
1 employed in the selection of the structural system will be ex-
Assistant Professor, Dept. of Civil Engineering and Construction,
Bradley Univ., Peoria, IL 61625. E-mail: mdicleli@bradley.edu plained. The seismic isolation system, seismic loading, and struc-
Note. Discussion open until August 1, 2002. Separate discussions tural model used in the analysis and design of the structure will
must be submitted for individual papers. To extend the closing date by also be defined. These will be followed by the presentation of the
one month, a written request must be filed with the ASCE Managing analysis results for the bridge with the FP bearings alone and the
Editor. The manuscript for this paper was submitted for review and pos- hybrid system.
sible publication on December 8, 2000; approved on May 24, 2001. This
paper is part of the Journal of Bridge Engineering, Vol. 7, No. 2, March
1, 2002. ©ASCE, ISSN 1084-0702/2002/2-94 –103/$8.00⫹$.50 per Site Geotechnical Properties
page.
The results of geotechnical investigation at the locations of sub-
94 / JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 structures have indicated that the site soil profile consists of three

J. Bridge Eng., 2002, 7(2): 94-103


Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Bridge geometry

separate layers of cohesive soil deposit overlying limestone bed-


rock. The first layer is a desiccated upper crust of silty clay ap-
proximately 4 m thick. The next layer of soil is roughly 30 m
thick and consists of firm to stiff silty clay to clay. The third layer
of soil extends for about 24 m down to the limestone bedrock and Fig. 2. Pier geometry
consists of stiff to very stiff silty clay. The site soil condition is
assumed as stiff for analysis purposes.
The geotechnical analyses have demonstrated that, in the event
gap between the pile and the upper crust of foundation soil are
of an earthquake, the foundation soil may suffer a partial loss of
anticipated in the case of an earthquake. A similar event was
undrained shear strength and decrease in stiffness under cyclic
observed in the 1989 Loma Prieta earthquake, which resulted in
loading. For design purposes, it was estimated that the desiccated
damage to the foundations of Struve Slough Bridge on Highway 1
upper crust of soil and underlying clay might suffer a loss of
共Mitchell et al. 1991兲. In the preliminary design stage, it was
respectively 10 and 25% in the undrained shear strength under
planned to batter as many piles as possible to reduce permanent
seismic conditions. This may produce permanent displacements
lateral displacement of the foundations. The steel piles were de-
of foundation soil in the order of 50 to 100 mm. Consequently, a
signed as laterally unsupported along the potentially unstable
possible formation of a gap between the piles and upper crust of
upper crust of foundation soil. The piles are also embedded suf-
soil is anticipated.
ficiently into the pile cap to prevent a possible pullout failure.
Additional measures were taken at the abutment foundations.
Bridge Properties Dead-man anchors were used to tie back the abutment footings to
reduce the force and displacement demand on the piles. Light-
The plan view and elevation of the twin bridges is illustrated in weight backfill material with a unit weight of 14 kN/m3 was used
Fig. 1. The total length of the bridge is 374 m and the width is behind the abutments to reduce the effect of dynamic soil pressure
11.46 m. Each bridge has six spans carrying two traffic lanes. The in the event of an earthquake.
largest span is 85 m long and is located over the river. The spans
at the west and east ends of the bridge are respectively 54 and 40 Pier Geometry
m long. The rest of the spans are 65 m long. The bridge has a
slab-on-steel-girder deck. Four steel girders are supporting a 225 In the planning stage, a hammerhead pier was recommended for
mm thick concrete slab and are spaced at 3.25 m. The girders are aesthetic reasons. Hammerhead piers are simple cantilever struc-
haunched on the piers on both sides of the river. The depths of the tures. Rocking of such piers in the event of an earthquake is
girders are 4.2 m at the haunches and 2.2 m elsewhere. The bridge usually more severe than that of piers with multiple columns due
deck is continuous from one abutment to the other and supported to their narrower foundations and the presence of only a single
by five piers in between. The geometry and dimensions of the support. Although a rocking motion may favor the seismic per-
piers is shown in Fig. 2. All piers have identical geometry. How- formance of a bridge if it can be controlled 共Priestley et al. 1996兲,
ever, the height of the piers varies between 7.2 and 9.3 m along excessive rocking may result in large displacements at the super-
the bridge. The piers and abutments are supported on piles ex- structure level and uplift of piles. This may lead to a potential
tending down to the limestone bedrock. A total of 24 and 28 pullout failure of the piles, as observed in past earthquakes. Ad-
HP310⫻110 steel piles are used respectively under each abutment ditionally, relying on the skin friction between the piles and the
and pier. soil to resist such uplift forces is generally not recommended in
cohesive soils due to the possible formation of a gap between the
pile and the soil upon cyclic movements. The flexural demand at
Preliminary Design Considerations the base of a single column is also quite large, as any lateral force
applied at the pier top will create a moment equal to its product
with the column height.
Abutments and Foundations
To overcome such performance deficiencies and to have an
As mentioned earlier, a permanent displacement of foundation aesthetic appearance similar to that of a hammerhead pier, two
soil of the order of 50 to 100 mm and a possible formation of a smaller hammerhead piers were joined using a tie beam to obtain

JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 / 95

J. Bridge Eng., 2002, 7(2): 94-103


a moment-resisting frame, as shown in Fig. 2. This new pier may produced a transverse direction fundamental period of 0.43 s, a
not experience as much hysterical rocking movement due to the maximum seismic shear force of 7,690 kN at pier #2, and very
presence of multiple supports and a wider foundation. Moreover, large pier sizes. In the longitudinal direction, several alternatives
the axial force couple generated at the pier supports reduces the were tried by fixing the bearings at only one or two piers to allow
flexural demand on the columns as they offset some of the base the rest of the structure to expand or contract due to temperature
moment produced by lateral loads in the transverse direction. The variations. This produced an uneven distribution of seismic forces
reduction in flexural moments also leads to an improvement in among the substructures and even larger pier sizes at the location
shear resistance of the columns. Although, substructures of of longitudinally fixed bearings. Such large pier sizes would cause
bridges with seismic isolation systems are designed to remain significant disturbance to the wetlands and violate the imposed
within the elastic range 共Priestley et al. 1996; Canadian 2000兲, the environmental constraints. Furthermore, a poor structural behav-
energy dissipation due to possible cyclic deformation of column ior was anticipated in the event of an earthquake due to the pos-
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

ends in this new rigid frame pier is an added safety if the isolation sibility of the bridge deck hammering the abutments. Based on
system malfunctions. Note that if the substructures of a seismi- the above observations, a decision was made to use a seismic
cally isolated bridge is designed to have lateral resistance lower isolation system.
than that of the isolator units and allowed to undergo inelastic
deformation, the seismic force level will be controlled by the
substructures, and the isolator units above will be rendered inef- Seismic Isolation Bearings
fective.
Inadequate ductility due to poor confinement has been respon- It has been proven that increasing the fundamental period of vi-
sible for damage to bridge columns 共Mitchell et al. 1995兲. As bration and energy dissipation capacity of a structure is an effi-
properly reinforced spiral or circular columns provide a superior cient way of improving its seismic behavior 共Priestley et al. 1996;
ductility at plastic hinging regions 共Priestley et al. 1996兲, a circu- Naeim and Kelly 1999兲. This can be achieved by using a seismic
lar cross section was selected for the columns. isolation system that possesses a built-in energy dissipation
mechanism and is capable of providing a longer fundamental pe-
riod. As mentioned earlier, the fundamental period of the structure
Expansion Joints
without seismic isolation bearings is 0.43 s, and the bridge site
In the preliminary design stage, two alternative configurations has stiff soil conditions. For design response spectra developed
were considered for the expansion joints along the bridge deck. for such soil conditions, an elongation of period beyond 0.6 s
The first alternative was to provide several expansion joints be- generally results in a considerable reduction in spectral accelera-
tween the abutments to effectively accommodate the movements tions 共International 2000; AASHTO 1998; Canadian 2000兲.
due to temperature variations. The other was to build a continuous Therefore, using a seismic isolation system is appropriate.
superstructure with expansion joints only at the abutments. This A number of different seismic isolation bearings were investi-
second alternative requires bearings and modular expansion joints gated: FP and LEP bearings were found to be the most commonly
that can accommodate large thermal movements 共130 mm兲 due to used seismic isolation systems in North America 共Naeim and
the substantial length of the structure. Kelly 1999兲, and both of them have been incorporated into design
It was anticipated that the first alternative might result in a codes 共AASHTO 1999; International 1997兲. A decision was made
higher initial cost and expensive maintenance due to the presence to use one of these bearings.
of a number of intermediate expansion joints 共Burke 1988兲. Such
bridges are also more vulnerable to seismic forces due to the lack
of superstructure continuity 共Dicleli and Bruneau 1995a; Priestley LEP Bearings
et al. 1996兲 and may require larger support widths and restrainer LEP bearings are made of vulcanized rubber laminated between
ties to accommodate the seismically induced relative displace- steel plates, confining a cylindrical lead core at the center of the
ments at expansion joints 共Tseng and Penzien 1973; Penzien and bearing. The rubber provides the lateral flexibility to elongate the
Chen 1975兲. Failures of this kind of bridges have been recurrently structure period. The steel plates prevent bulging of the rubber,
observed in past earthquakes 共EERI 1991; Bruneau et al. 1996兲. provide a high vertical stiffness, and confine the lead core. The
The elimination of the intermediate expansion joints may re- lead core dissipates energy upon cyclic movement and provides
sult in less-expensive maintenance and a lower initial cost as resistance to wind. The plan area and the total height of the rubber
smaller girder sizes may be obtained due to the effect of continu- layers determine the flexibility of the bearing, and the diameter of
ity. Additionally, continuity provides a better distribution of seis- the lead core determines the desired level of energy dissipation or
mic forces to substructure elements 共FHWA 1987; Dicleli and equivalent hysteretic damping 共EHD兲.
Bruneau 1995b兲 and reduces the vulnerability of the structure to
large ground displacements and liquefaction. Thus, a final deci-
sion was made to use a continuous superstructure based on the FP Bearings
above considerations. FP bearings are composed of an articulated slider coated with a
self-lubricated composite liner and a concave spherical surface as
Bearings shown in Fig. 3. A pendulum motion for the supported structure is
achieved as the articulated slider rides on the spherical surface
For the range of loads and thermal displacements, spherical and and at the same time dissipates energy due to friction. The weight
pot bearings 共Watson 1992兲 were one of the few bearing types of the bridge also acts as a restoring force as the bridge rides on
available in the market at the time the structure was designed and the concave surface. The period and stiffness of the isolation bear-
were initially considered. In the preliminary design stage, all the ing are selected by choosing the appropriate radius of curvature.
bearings were fixed in the transverse direction of the bridge to The level of EHD is chosen by selecting an appropriate dynamic
restrain the lateral movements due to wind-induced forces. This friction for the bearing.

96 / JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002

J. Bridge Eng., 2002, 7(2): 94-103


Fig. 4. Hybrid design response spectrum
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

role in the selection of the isolation bearing. Table 1 clearly indi-


cates that the properties of the FP bearings are not affected by
temperature variation as much as those of LEP bearings. LEP
bearings are not recommended for temperatures below ⫺40°C as
glassification of rubber material may occur at such temperatures.
Additionally, the stiffness of the LEP bearings varies as a function
of temperature. A 60°C temperature variation may result in a 30%
change in bearing stiffness. Based on this information, FP bear-
ings were found to be more suitable for the bridge.
Fig. 3. Friction pendulum bearings

Design Response Spectrum


Selection of Isolation Bearings
The Guide Specifications for Seismic Isolation Design 共AASHTO The bridge falls in seismic performance zone 3 where the multi-
1999兲 contains information for the designer to consider the effects mode response spectrum method is required for the seismic analy-
of aging, temperature, wear, contamination, and scragging on the sis of the bridge by the code 共AASHTO 1998; Canadian 2000兲.
performance of the isolation system. These effects are reflected on Therefore, an appropriate design response spectrum was required
the stiffness and characteristic strength of the isolator using prop- for the analysis of the bridge. Preliminary analyses indicated that
erty adjustment factors. These factors are provided for both elas- the structure possesses different levels of damping for the struc-
tomeric bearings and sliding isolation systems with PTFE and tural and isolated modes of vibration. Structural modes primarily
bimetallic interfaces. The composite liner used for the FP bear- represent the vibration of the substructure elements where a 5%
ings is different than those mentioned above and meets stringent damping ratio is assumed. For the modes governed by the seismic
specifications for use in military applications. Therefore, the isolation system, the structure has an EHD ratio of 30%. Accord-
guide specification was not used in the selection process. The ingly, a hybrid response spectrum with different levels of damp-
selection of the isolation bearings was based on the information ing ratio for structural and isolated modes of vibration is used in
listed in Table 1, which was obtained from the manufacturers. The the analysis 共AASHTO 1999兲. The response spectra that consti-
size, cost, and energy dissipation properties of both LEP and FP tute the hybrid response spectrum used in the analysis are shown
bearings are comparable 共Priestley et al. 1996兲. However, the un- in Fig. 4. They are defined by the following expression 共Canadian
usually cold climatic conditions of the bridge site played a major 2000兲:

Table 1. Comparison of Seismic Isolaton Bearings


Characteristic LEP bearing FP bearing
Cost Comparable Comparable
Size Comparable Comparable
Energy Comparable Comparable
dissipation
Stability Stability problem under vertical Stability under vertical load is
load may exist at large not as much of a concern
displacements
Sensitivity to Not recommended for May be used at such
temperature temperatures lower than ⫺40°C as temperatures
glassification may occur
Stiffness Lateral stiffness is a function of Equivalent lateral stiffness
temperature 共30% change in slightly changes due to
stiffness for a temperature temperature variation
variation of 60°C兲

JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 / 97

J. Bridge Eng., 2002, 7(2): 94-103


Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Structural model of bridge deck


Fig. 6. Structural model of piers

AS i A
C sm ⫽ ⭐2.5 (1) ignored deliberately to avoid triggering unwanted modes of vibra-
BT e B
tion that are not useful in the analysis. All six static degrees
where C sm ⫽elastic seismic response coefficient for isolated struc- of freedom were used in the analysis.
tures; A⫽zonal acceleration ratio, S i ⫽site coefficient; T e ⫽period The in-plane translational stiffness of the deck is relatively
of vibration; and B⫽damping coefficient listed for various damp- much higher than that of the other members of the bridge. Ac-
ing ratios in the code 共Canadian 2000兲. cordingly, at the abutment and pier locations, the bridge deck is
The zonal acceleration ratio is 0.2 g for the bridge site. For modeled as a transverse rigid bar of length equal to the deck
stiff soil conditions with depth to rock larger than 60 m, the site width, as shown in Fig. 5. This transverse rigid bar is used to
coefficient is 1.50. The damping coefficient is 1.0 for 5% damp- simulate the interaction between the axial deformation of the col-
ing and 1.7 for 30% damping. umns and torsional rotation of the bridge deck, as well as the
interaction between the in-plane rotation of the deck and displace-
ments of the isolation bearings. The transverse rigid bar was el-
Structure Model evated to the level of the center of gravity of the bridge deck
using a set of vertical rigid elements attached to it. This was done
A 3D model is necessary for a realistic representation of the be- to accurately define the vertical location of the mass of the bridge
havior of the bridge and load distribution among its various com- deck. The rigid vertical elements are then connected to the isola-
ponents when it is subjected to seismic loads. Some of the re- tion bearings, as shown in Fig. 5.
search data on the correlation of computer analysis results with
actual seismic behavior 共Kawashima and Penzien 1976; Douglas
Substructures
1979; Imbsen and Penzien 1979; Wilson 1986兲 were used to
model the structure. The analytical models for each component of The total weight of the piers constitutes 20% of the total weight
the bridge are presented in the following subsections. of the bridge. Therefore, their influence on the structure response
could be significant. Each pier is modeled as a rigid frame, as
shown in Fig. 6. Both the cap-beam and the columns are divided
Deck
into a number of segments, and their tributary mass is lumped at
The bridge superstructure is modeled as a 3D beam element, as the nodes connecting each segment. The parts of the column and
shown in Fig. 5. Full composite action between the slab and steel cap-beam within the joint are modeled as rigid elements.
girders is assumed in the model 共Douglas 1979; Dicleli 2000a,b兲. The columns are connected to a rigid pile cap that is modeled
The moment of inertia of the superstructure about the Y-axis (I y using a horizontal and a vertical rigid bar, as shown in Figs. 5 and
⫽3.55 m4 ) is obtained by first calculating the moment of inertia 6. The length of the vertical rigid bar is set equal to the footing
of each composite girder using an effective slab width 共Canadian depth to accurately estimate the effect of seismic forces trans-
2000兲 and then multiplying the result by the number of girders. ferred to the piles 共Duan 1996兲. Rotational and translational
The moment of inertia of the superstructure about the Z-axis (I z springs are connected at the end of the vertical rigid bar to simu-
⫽59.50 m4 ) is also calculated assuming full composite action late the effect of foundation stiffness on the structure response.
between the slab and the girders. The strong diaphragms provided Two separate plane frame models for the foundations are con-
between the girders are assumed to help achieve this full compos- structed to calculate the stiffness of the springs in two orthogonal
ite action. The superstructure is divided into a number of seg- directions. To obtain a plane frame structure, the piles in each row
ments, and its mass 共14,580 kg/m兲 is lumped at each nodal point perpendicular to the plane of the frame are lumped together and
connecting the segments. Each mass was assigned four dynamic represented by a single equivalent frame element, as shown in
degrees of freedom; translations in the X- and Y-directions of the Fig. 7. Each equivalent frame element is assigned 21 nodes, di-
bridge and rotations about the X- and Z-axes, as shown in Fig. 5. viding it into 20 equal segments. The lateral stiffness of the foun-
The remaining two dynamic degrees of freedom are dation soil is calculated at each node level using the coefficient of

98 / JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002

J. Bridge Eng., 2002, 7(2): 94-103


The above-horizontal force is the resistance of the bearing to
lateral loads and is directly proportional to the friction coefficient,
displacement, and reaction force applied on the bearing.
In the response spectrum analysis of the structure, an equiva-
lent bearing stiffness, k e , and an EHD ratio representing the en-
ergy dissipation of the bearing are used. The equivalent bearing
stiffness shown in Fig. 8 is obtained by simply dividing the maxi-
mum horizontal force by the corresponding maximum bearing
displacement, D max .
␮W W
k e⫽ ⫹ (3)
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

D max R
The EHD ratio, B EH , is expressed as 共Earthquake 1999兲:

冉 冊
Fig. 7. Structural model to calculate spring constants at pier base 2 ␮
B EH ⫽ (4)
␲ D
␮⫹
R
subgrade reaction for the soil, which is recommended as a con- The bearings are modeled as 3D beam elements and connected
stant, 33 MN/m3, to a depth of 5 m and linearly increasing at a between the vertical rigid bars and the pier cap under each girder
rate of 5.80 MN/m3/m beyond 5 m. Spring elements with stiffness location as shown in Fig. 5. Pin connection is assumed at the
equal to that of the soil are then attached to each node. The joints linking the pier cap to the bearings. The product of the
stiffness of the springs is calculated by multiplying the coefficient elastic modulus, E eb , and moment of inertia, I be , of the beam
of subgrade reaction at each node level by the pile width 共0.30 element is obtained by setting the equivalent stiffness of the bear-
m兲, tributary length between the nodes 共2.25 m兲, and number of ing equal to the stiffness of the 3D beam element:
piles lumped to form a frame element. The equivalent frame ele-
ments are connected to the pile cap modeled as a beam element. k eq h 3b
The same procedure is repeated to obtain a plane frame model in E be I be ⫽ (5)
3
the other direction. The models are then assigned a unit horizontal
displacement and rotation to calculate the stiffness of translational where h b 共216 mm兲 is the actual height of the bearing.
and rotational springs. The stiffness of the translational springs in
the X- and Y-directions are calculated as 111,960 and 118,760 LE Bearings
kN/m, respectively. The stiffness of the rotational springs about
the X- and Y-axes are calculated as 19,445,000 and 5,688,160 For the analyses of the bridge with hybrid seismic isolation sys-
kN•m/rad, respectively. tem, the LE bearings at the abutments are idealized as elastic 3D
beam elements and connected to the rigid bars at girder locations.
Pin connection is assumed at the joints linking the bearings to the
FP Bearings supports. The product of the elastic modulus, E eb , and moment of
The force-displacement hysteresis loop for the FP bearing is inertia, I be , of the beam element is obtained by the following
shown in Fig. 8 共Earthquake 1999兲. There are three parameters expression 共Dicleli 2000a兲:
defining the hysteresis behavior: the radius, R, of the concave G b A b h 2b
surface, the friction coefficient, ␮, and the reaction force, W, act- E be I be ⫽ (6)
3
ing on the bearing, R and ␮ for the bridge bearings are 2,235 mm
and 0.07, respectively. The horizontal force, F, at any displace- where G b and A b are respectively the shear modulus and surface
ment D, is expressed as 共Earthquake 1999兲 area of the bearing. The shear modulus is 1.06 MPa at a tempera-
ture of 20°C. The bearings’ plan dimensions are 785⫻785 mm
W
F⫽␮W⫹ D (2) and the thickness is 285 mm.
R

Analysis of Bridge

The analyses of the bridge were conducted using the program


S-Frame 共Softek 2000兲. The bridge is analyzed for two different
configurations of base isolation systems: the first consists of FP
bearings at all substructure locations, and the second incorporates
a hybrid system where LE bearings are used at the abutments and
FP bearings at the piers.
An iterative multimode response spectrum analysis technique
is used to obtain the isolation bearing displacements and other
structural responses. First, a maximum displacement, D i max , is
assumed for each FP bearing, i. The assumed displacements, the
reactions due to the self weight of the structure, and the friction
Fig. 8. Force-displacement hysteresis loop for friction pendulum
coefficient of the bearings are substituted in Eq. 共3兲 to calculate
bearings
the equivalent stiffness for each FP bearing. The calculated

JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 / 99

J. Bridge Eng., 2002, 7(2): 94-103


Table 2. Modal Periods and Percentage of Mass Participation for
Friction Pendulum 共FP兲 and Hybrid Base Isolation Systems
Percentage of mass
participation
Period Longitudinal Transverse
Mode number FP Hybrid FP HYBRID FP Hybrid
1 2.09 1.96 0 0 23 64
2 1.98 1.75 0 80 43 0
3 1.95 1.56 80 0 0 1 Fig. 9. Modes of vibration for friction pendulum base isolation sys-
4 1.63 1.30 0 0 14 15 tem alone
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Higher modes T⬍1.30 T⬍1.30 20 20 20 20

modes of vibration with a 20% mass contribution, which is the


equivalent stiffness is then substituted in Eq. 共5兲 to obtain the total mass of the substructure elements. Modes 1, 2, and 4 are the
stiffness of the beam elements used in the model. The EHD ratio transverse modes of vibration and mode 3 is the longitudinal. The
for each bearing is also calculated by substituting the bearing total mass contribution for the isolated modes of vibration in each
properties 共radius and friction coefficient兲 and displacements in direction is 80% of the structure mass, which is the total mass of
Eq. 共4兲. The average of the EHD ratios obtained for each bearing the superstructure. The fundamental period of the structure is 2.09
is used in the analysis of the structure. The multimode response s.
spectrum analysis of the structure is conducted using the program
S-Frame 共Softek 2000兲. The new bearing displacements obtained
Maximum Isolation Bearing Displacements
from the analysis are compared with the initially assumed bearing
displacements. If the difference is smaller than an assumed level In the longitudinal direction, all bearings have a relative uniform
of accuracy, the iteration is stopped; otherwise the iteration is displacement of 115 mm. The bearing displacements are identical
continued manually with the new displacements until the desired due to the large relative axial rigidity of the bridge deck and
convergence is achieved. lateral stiffness of substructures. In the transverse direction, how-
As the EHD ratio is a function of bearing displacements, dif- ever, the bearing displacements vary along the bridge. This results
ferent values were obtained at each iteration step, but it eventually from the unusual flexibility of the bridge in the transverse direc-
converged to 30%. The 5% modal damping of the structure and tion since the width of the bridge 共11.46 m兲 is relatively small
the 30% EHD of the isolation system is incorporated in the analy- compared to its total length 共374 m兲. The iterative multimode
ses using a hybrid response spectrum. The final hybrid response response spectrum analysis of the structure in the transverse di-
spectrum used in the design is composed of a mixture of spectral rection converged to 2 separate sets of bearing displacements, as
accelerations for 5 and 30% damping ratios. The damping coeffi- shown in Fig. 10. This may be due to the closely spaced modal
cients for damping ratios of 5 and 30% are respectively obtained frequencies of the structure and the sensitivity of the relatively
as 1.0 and 1.7 from the design code 共Canadian 2000兲 and substi- small bearing stiffness to calculated displacements. As seen in
tuted in Eq. 共1兲 to calculate the spectral accelerations. The spec- Fig. 10, the maximum bearing displacements at the west and east
tral accelerations for the 5% damping ratio are used for periods abutments are respectively 213 and 236 mm and are much larger
(T e ⬍1.30 s) of the structural modes of vibration. Those for 30% than those at the piers. This results from the smaller bearing stiff-
damping ratio are used for the larger periods of the isolated ness at the abutments. The dead load reactions at the abutments
modes of vibration. In the analysis of the structure with hybrid are only 20% of those at the piers due to the smaller tributary
seismic isolation system, the elastomeric bearings were assumed weight of shorter end spans and the uplift reactions created by
to have a 5% damping ratio.
The magnitude of the vertical reaction force on the bearings
undoubtedly varies due to vertical ground accelerations, and con-
sequently so does the friction resistance throughout the earth-
quake. However, seismic vertical accelerations generally have a
much higher frequency content than the corresponding horizontal
ones. Thus, it is assumed that the resulting fluctuations of fric-
tional resistance will average themselves during any sliding ex-
cursion, simultaneously abating the significance of vertical accel-
erations 共Dicleli and Bruneau 1995c兲.

Analysis Results for FP Bearings Alone

Mode Shapes and Periods of Vibration


A total of 25 modes of vibration are considered in the analysis of
the structure. Table 2 displays the periods of vibration of the
structure for the first four modes, and Fig. 9 displays the corre-
Fig. 10. Isolation bearings’ transverse direction displacements for
sponding mode shapes. The first four modes of vibrations are
friction pendulum base isolation system alone
those involving the isolation system. The rest are the nonisolated

100 / JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002

J. Bridge Eng., 2002, 7(2): 94-103


Table 3. Seismic Shear Forces Transferred to Substructures for Fric-
tion Pendulum 共FP兲 and Hybrid Base Isolation Systems
Seismic shear force 共kN兲
Longitudinal Transverse
Dead load reaction
Substructure 共kN兲 FP HYBRID FP HYBRID
East abutment 2,028 246 744 276 536
Pier 1 12,008 1,456 1,373 1,308 1,094
Pier 2 11,564 1,402 1,318 1,146 1,378
Pier 3 8,236 999 951 852 1,096 Fig. 11. Modes of vibration for hybrid base isolation system
Pier 4 9,680 1,174 1,112 1,086 1,173
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Pier 5 8,340 1,012 960 1,224 836


West abutment 1,712 207 744 299 480 at the piers 共⬃134 mm兲 required bearings with a radius of 195
mm. Such bearings have a very steep riding surface due to their
small radius and require a vertical movement of 55 mm to accom-
much longer adjacent spans. Since the lateral resistance of FP modate a horizontal displacement of 134 mm. Therefore, it was
bearings is directly proportional to the dead load reactions acting unfeasible to use such bearings at the abutments since the super-
on the bearings, a very small equivalent stiffness or lateral resis- structure must be designed to accommodate a relative vertical
tance is obtained at the abutments. Consequently, the end spans displacement of 51 mm.
behaved almost like a cantilever beam in the transverse direction. It was anticipated that the above problems might be addressed
This produced even larger bearing displacements as the seismi- if a seismic isolation system with lateral stiffness properties inde-
cally induced forces acquire the shape of the deflected structure pendent of the magnitude of reaction force is used at the abut-
共Chopra 2001兲. ments. Accordingly, it was decided to use a hybrid seismic isola-
It is noteworthy that a time history analysis of the structure tion system consisting of LE bearings at the abutments and FP
would result in a more accurate estimate of the bearing displace- bearings at the piers.
ments. However, it is anticipated that the obtained displacements
are more conservative since the response spectrum analysis
method combines the modal displacements regardless of their di- Analysis Results for Hybrid Seismic Isolation
rection.
The lateral stiffness of LE bearings varies as a function of tem-
Seismic Shear Forces perature. Therefore, the structure was analyzed for a minimum
and a maximum stiffness corresponding to temperatures of re-
The total dead load reactions and the maximum seismic shear spectively 20 and ⫺40°C. The analysis results for the bearings
forces at each substructure location are displayed in Table 3. The with the maximum stiffness are used to design the abutments and
lateral resistance provided by FP bearings is proportional to the those for the bearings with minimum stiffness are used to design
dead-load reactions acting on the bearings. Thus, the seismic the piers and assess the displacement capacity of FP bearings.
shear forces are proportional to these reactions, as shown in the A third analysis is also conducted using an average stiffness
table. Although the abutments are structurally stronger than the obtained by increasing the minimum stiffness by 35%, as
piers due to their massive size, only 7% of the total seismic force recommended elsewhere 共Ministry 1996兲. The results from this
is transferred to the abutments. For example, the seismic shear analysis are used to study the influence of the hybrid seismic
force in the longitudinal direction at the west abutment is 246 kN, isolation system on the performance of the structure.
while the adjacent pier 1 attracts a seismic shear force of 1,456
kN. This uneven distribution of seismic shear forces among sub-
structures may result in larger pier sizes and needs to be avoided. Mode Shapes and Periods of Vibration
Table 2 displays the periods of vibration of the structure for the
Remarks on Performance of Structure first four modes and Fig. 11 displays the corresponding mode
shapes. The first four modes of vibrations are those involving the
The bearing displacements at the abutments are very large. Larger isolation system. The rest are the structural or nonisolated modes
bearings are needed to accommodate such displacements. This of vibration with 20% mass contribution, as in the case of the
may result in a higher structure cost. Therefore, a reduction in the structure with FP bearings alone. Modes 1, 3, and 4 are the trans-
bearing displacements at the abutments is necessary. Moreover, verse direction modes of vibration. The total mass contribution
the abutments attract only a small fraction of the seismic forces for the isolated modes of vibration in each direction is 80% of the
acting on the structure. A more uniform distribution of seismic
forces among substructure elements is needed for a superior struc-
ture performance and a more economical design.
The transverse shear forces due to the effect of wind were also
calculated at each substructure location. It was found that the
wind-induced forces at the abutments exceed the frictional resis-
tance provided by the FP bearings. This may produce wind-
induced vibrations at the end spans. Wind-locking devices need to
be installed at the abutments to prevent such vibrations.
Fig. 12. Isolation bearing’s transverse direction displacements for
Reducing the radius, R, of the friction pendulum bearings at
hybrid base isolation system
the abutments to obtain displacements in the same order as those

JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 / 101

J. Bridge Eng., 2002, 7(2): 94-103


center. The elastomeric bearings may provide a restoring force
to recenter the structure in such circumstances.

References
Al-Hussaini, T. M., Zayas, V. A., and Constantinou, M. C. 共1994兲. ‘‘Seis-
mic isolation of a multi-story frame structure using spherical sliding
isolation systems.’’ Technical Rep. NCEER-94-0007, National Center
Fig. 13. Seismic shear force distribution along bridge
for Earthquake Engineering Research, Buffalo, N.Y.
AASHTO. 共1998兲. LRFD bridge design specifications, Washington, D.C.
AASHTO. 共1999兲. Guide specifications for seismic isolation design,
structure mass. The fundamental period of the structure is 1.96 s, Washington, D.C.
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

which is slightly smaller than 2.09 s of the system with FP bear- Blakeley, R. W. G. 共1982兲. ‘‘Code requirements for base isolated struc-
ings alone. tures.’’ Proc., Int. Conf. on Natural Rubber for Earthquake Protection
of Buildings and Vibration Isolation, Malaysian Rubber Research and
Development Board, Kuala Lumpur, Malaysia, 292–311.
Maximum Isolation-Bearing Displacements Bruneau, M., Wilson, J. W., and Tremblay, R. 共1996兲. ‘‘Performance of
steel bridges during the 1995 Hyogoken Nanbu 共Kobe, Japan兲 earth-
In the longitudinal direction, all bearings have a uniform displace-
quake.’’ Can. J. Civ. Eng., 23共3兲, 678 –713.
ment of 102 mm. This is slightly smaller than 115 mm displace- Built, S. M. 共1982兲. ‘‘Lead rubber dissipators for the base isolation of
ment of the system with FP bearings alone. The bearing displace- bridge structures.’’ Rep. 289, School of Engineering, Dept. of Civil
ments in the transverse direction are shown in Fig. 12. These Engineering, Univ. of Auckland, New Zealand.
displacements are considerably different than those of the system Burke, M. P., Jr. 共1988兲. ‘‘Bridge deck joints.’’ NCHRP Synthesis of High-
with FP bearings alone. The LE bearings restrained the displace- way Pract., No. 141, Transportation Research Board, National
ments at the abutments due to their relatively larger lateral stiff- Research Council, Washington, D.C.
ness compared to that of FP bearings. The bearing displacement at Canadian Standards Association. 共2000兲. Canadian highway bridge
the east abutment is now reduced from 236 to 66 mm. However, design code—Final draft, Toronto.
those at piers 2, 3, and 4 are slightly increased. It is noteworthy Charleson, A. W., Wright, P. D., and Skinner, R. I. 共1987兲. ‘‘Wellington
Central Police Station—Base isolation of an essential facility.’’ Proc.,
that, in this case, the iterative multimode response spectrum
Pacific Conf. on Earthquake Engineering, New Zealand National So-
analysis of the structure in the transverse direction converged to ciety for Earthquake Engineering, Wairekei, New Zealand, 2, 377–
only one set of bearing displacements. 388.
Chopra, A. 共2001兲. Dynamics of structures: Theory and applications to
Seismic Shear Forces earthquake engineering, Prentice-Hall, Upper Saddle River, N.J.
Constantinou, M. C., Tsopelas, P., Kim, Y.,-S., and Okamoto, S. 共1993兲.
The seismic shear forces at each substructure location are tabu- ‘‘NCEER-Taisei Corporation research program on sliding isolation
lated in Table 3. The shear forces are more evenly distributed systems for bridges: Experimental and analytical study of friction
among substructure elements compared to those of the system pendulum system.’’ Rep. No. 93-0020, National Center for Earthquake
with FP bearings alone, as shown in Fig. 13. In the case of the Engineering Research, Buffalo, N.Y.
Derham, C. J., Kelly, J. M., and Thomas, A. G. 共1985兲. ‘‘Nonlinear natu-
hybrid seismic isolation system, 21% of the total seismic force in
ral rubber bearings for seismic isolation.’’ Nucl. Eng. Des., 84共3兲,
the longitudinal direction is transferred to the abutments. The use 417– 428.
of LE bearings at the end supports increased the abutment shear Derham, C. J., Learoyd, S. B. B., and Wootton, L. R. 共1975兲. ‘‘Buildings
force in the longitudinal direction from 246 to 744 kN, while the on springs to resist earthquakes.’’ Proc., 5th European Conf. on Earth-
force in pier 1 decreased from 1,456 to 1,373 kN. In the trans- quake Engineering, International Committee on Earthquake
verse direction, the west abutment shear force increased from 276 Engineering, Istanbul, Turkey.
to 536 kN, while the force in pier 5 decreased from 1,224 to 836 Dicleli, M. 共2000a兲. ‘‘A rational design approach for prestressed-
kN. The increase in the seismic shear force at the abutments did concrete-girder integral bridges.’’ Eng. Struct., 22共3兲, 230–245.
not impact their design since their width is determined by the size Dicleli, M. 共2000b兲. ‘‘Simplified model for computer-aided analysis of
of the bearings and expansion joints. integral bridges.’’ J. Bridge Eng., 5共3兲, 240–248.
Dicleli, M., and Bruneau, M. 共1995a兲. ‘‘Seismic performance of multi-
span simply supported slab-on-girder steel highway bridges.’’ Eng.
Struct., 17共1兲, 4 –14.
Conclusions Dicleli, M., and Bruneau, M. 共1995b兲. ‘‘Seismic performance of single-
• The hybrid seismic isolation system provided a structure with span simply supported and continuous slab-on-girder steel highway
a fundamental period long enough to attract smaller seismic bridges.’’ J. Struct. Eng., 121共10兲, 1497–1506.
forces, while controlling the magnitude of bearing displace- Dicleli, M., and Bruneau, M. 共1995c兲. ‘‘An energy approach to sliding of
ments. single-span simply supported slab-on-girder steel highway bridges
• It also resulted in a more uniform distribution of seismic with damaged bearings.’’ Earthquake Eng. Struct. Dyn., 24共3兲, 395–
forces among substructure elements. Consequently, the higher 409.
seismic forces on the piers were reduced, allowing for a more Douglas, M. B. 共1979兲. ‘‘Experimental dynamic response investigations
economical design of substructures. of existing highway bridges.’’ Proc., Workshop on Earthquake Resis-
tance of Highway Bridges, Applied Technology Council, Palo Alto,
• The hybrid seismic isolation system helped to control the
Calif., 497–523.
wind-induced vibrations at end spans and reduce the bearing Duan, L. 共1996兲. ‘‘Bridge-column footings: An improved design proce-
sizes by 38%. dure.’’ Pract. Period. Struct. Des. Constr., 1共1兲, 20–24.
• At the end of a seismic excitation, if the ratio of the final Earthquake Protection Systems Inc. 共1999兲. ‘‘Friction pendulum seismic
displacement to the radius of the friction pendulum bearing is isolation bearing: Details and specifications for bridges.’’ Richmond,
smaller than the friction coefficient, the system may not re- Calif.

102 / JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002

J. Bridge Eng., 2002, 7(2): 94-103


EERI. 共1999兲. ‘‘Costa Rica earthquake reconnaissance report.’’ Earth- Mitchell, D., Tinawi, R., and Sexsmith, R. 共1991兲. ‘‘Performance of
quake Spectra, Special Supplement to Vol. 7. bridges in the 1989 Loma Prieta earthquake—Lessons for Canadian
Federal Highway Administration 共FHWA兲. 共1987兲. ‘‘Seismic design and designers.’’ Can. J. Civ. Eng., 18共4兲, 711–734.
retrofit manual for highway bridges.’’ FHWA-IP-87-6, U.S. Dept. of Naeim, F., and Kelly, J. M. 共1999兲. Design of seismic isolated structures:
Transportation, Washington, D.C. From theory to practice, Wiley, Chichester, U.K.
Imbsen, R. A., and Penzien, J. 共1979兲. ‘‘Evaluation of analytical proce- Penzien, J., and Chen, M. 共1975兲. ‘‘Seismic response of highway
dures used in bridge seismic design.’’ Proc., Workshop on Earthquake bridges.’’ Proc., U.S. National Conf. on Earthquake Engineering,
Resistance of Highway Bridges, Applied Technology Council, Palo Earthquake Engineering Research Institute, Ann Arbor, Mich.
Alto, Calif., 468 – 496. Priestley, M. J. N., Seible, F., and Calvi, G. M. 共1996兲. Seismic design
International Code Council. 共2000兲. International building code, Falls and retrofit of bridges, Wiley, New York.
Church, Va. Robinson, W. H., and Tucker, A. G. 共1977兲. ‘‘A lead-rubber shear
damper.’’ Bull. New Zealand Natl. Soc. Eng., 10共3兲, 151–153.
International Conference of Building Officials. 共1997兲. ‘‘Earthquake
Downloaded from ascelibrary.org by CRRI - Central Road Research Institute on 01/09/18. Copyright ASCE. For personal use only; all rights reserved.

Robinson, W. H., and Tucker, A. G. 共1983兲. ‘‘Test results for lead-rubber


regulations for seismic-isolated structures.’’ Uniform building code,
bearings for the William M. Clayton Building, Toe Toe Bridge, and
Appendix Chapter 16, Whittier, Calif.
Waiotukupuna Bridge.’’ Bull. New Zealand Natl. Soc. Eng., 14共1兲,
Kawashima, K., and Penzien, J. 共1976兲. ‘‘Correlative investigations on
21–33.
theoretical and experimental dynamic behavior of a model bridge Softek. 共2000兲. S-Frame: 3-D Structural Analysis Software, Richmond,
structure.’’ Rep. No. UCB/EERC 76/26, Earthquake Engineering Re- B.C., Canada.
search Center, Univ. of California, Berkeley, Calif. Tseng, W. S., and Penzien, J. 共1973兲. ‘‘Analytical investigations of the
Kelly, J. M. 共1997兲. Earthquake resistant design with rubber, 2nd Ed., seismic response of long multiple span highway bridges.’’ Rep. No.
Earthquake Engineering Research Center, National Information Ser- UCB/EERC 73/12, Earthquake Engineering Research Center, Univ. of
vice for Earthquake Engineering, Springer, London and Richmond, California, Berkeley, Calif.
Calif. Watson Bowman Acme. 共1992兲. WABO-FYFE Structural Bearing
Kelly, J. M., and Quiroz, E. 共1992兲. Mechanical characteristics of neo- Product Data, Amherst, N.Y.
prene isolation bearings.’’ Rep. No. UCB/EERC-92/11, Earthquake Wilson, J. C. 共1986兲. ‘‘Analysis of the observed seismic response of a
Engineering Research Center, Univ. of California, Berkeley, Calif. highway bridge.’’ Earthquake Eng. Struct. Dyn., 14共3兲, 339–354.
Ministry of Transportation. 共1996兲. Structural manual, Quality and Stan- Zayas, V. A., Constantinou, M. C., Tsopelas, P., and Kartoum, A. 共1996兲.
dards, Transportation Engineering Branch, Bridge Office, ‘‘Testing of friction pendulum seismic isolation bearings for bridges.’’
St. Catharines, Ontario, Canada. Proc., 4th World Congress on Joint Sealing and Bearing Systems for
Ministry of Works and Development. 共1983兲. ‘‘Design of lead rubber Concrete Structures, American Concrete Institute, Farmington Hills,
bearings.’’ Civil Division Publication CDP 818/A, Wellington, Mich.
New Zealand. Zayas, V. A., Low, S. S., and Mahin, S. A. 共1987兲. ‘‘The FPS earthquake
Mitchell, D., Bruneau, M., Williams, M., Anderson, D., Saatcioglu, M., resisting system: Experimental report.’’ Rep. No. UCB/EERC 87/01,
and Sexsmith, R. 共1995兲. ‘‘Performance of bridges in the 1994 Earthquake Engineering Research Center, Univ. of California,
Northridge earthquake.’’ Can. J. Civ. Eng., 22共2兲, 415– 427. Berkeley, Calif.

JOURNAL OF BRIDGE ENGINEERING / MARCH/APRIL 2002 / 103

J. Bridge Eng., 2002, 7(2): 94-103

You might also like