You are on page 1of 17

Structures 46 (2022) 1154–1170

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Seismic design and performance evaluation of typical lightweight and


heavyweight RC girder bridges with friction pendulum bearing in Indonesia
Muslinang Moestopo a, Iswandi Imran a, c, Dionysius M. Siringoringo b, *, James Michael a,
Erwin Lim a, Shinsuke Yamazaki d
a
Civil Engineering Department, Institut Teknologi Bandung, West Java, Indonesia
b
Institute of Advanced Sciences, Yokohama National University, Yokohama, Kanagawa, Japan
c
Research Center for Disaster Mitigation, Institut Teknologi Bandung, Bandung, West Java, Indonesia
d
Nippon Steel Engineering Co. Ltd, Tokyo, Japan

A R T I C L E I N F O A B S T R A C T

Keywords: Seismically isolated bridges have shown better seismic performance than bridges with conventional bearings
Seismic isolation during large earthquakes. While elastomeric type of isolation has been introduced in various new highway
Friction pendulum bearing bridges in the country, the use of friction pendulum system for bridge bearing is not yet common in Indonesia, a
Seismic performance evaluation
highly seismically active country. In this paper, design and seismic performance evaluation of two typical
Bridge seismic design
Indonesian seismic code
reinforced concrete bridges seismically isolated by single concave friction pendulum (SCFP) bearings system and
designed according to Indonesian seismic code is described. The reference bridges are selected from typical light
rapid transit (LRT) bridge and highway reinforced concrete (RC) bridge in Indonesia. Seismic performances of
SCFP-isolated bridges with different bearing configurations and the conventional non-isolated bridges under
1000-year design earthquake were investigated and compared using nonlinear finite element analyses. The re­
sults show that SCFP-isolated bridges performed better than non-isolated bridges under design earthquake as
shown by reductions in the base shear force, absolute deck acceleration, and deck acceleration amplification
ratio. Effects of SCFP on the seismic performance of isolated bridge was evaluated by comparing the moment and
rotation values of each pier and the possible formation of plastic hinges. Performance comparisons of SCFP-
isolated bridges with various friction types and arrangements revealed that the LRT bridge with all higher
friction and the highway bridge with lower friction have the best seismic performance among all isolated bridge
models analyzed.

other one is by controling distribution of the reduced lateral forces


among the supporting structures and foundations. In the past large
1. Introduction
earthquakes, reports have shown that seismic performances of bridges
with various types of seismic isolations and vibration control systems
Observed damages on the bridges designed by conventional ductility
improved considerably compared to conventional non-isolated ones of
design concept reveal that significant structural damages occurred pri­
similar types [5–7].
marily on the piers. This type of damage may eventually result in bridge
Various seismic isolation bearings for bridges and buildings are now
collapse [1–4]. Bridge structure has limited structural redundancy that
available and they generally divided to four types: elastomeric type,
may work against the benefit provided by the ductility design concept.
spring type, frictional sliding type, and rocking type. For lateral isola­
Therefore, efforts to protect bridges against damaging earthquake have
tion, two of the most common systems applied on bridge structure are
been focused on minimizing the forces, particularly shear force, carried
the elastomeric bearings and frictional sliding bearings [8,9]. Elasto­
by the piers. For this purpose, seismic isolation has been seen as one of
meric bearings have been utilized to accommodate thermal expansion
the most promising alternatives. The isolation system provides two
and rotation at bridge supports. These bearings have high tolerance for
significant design features for a bridge. One is by lengthening natural
movement, overload, while require minimum maintenance [10]. Natu­
period of the structure and shifting the main natural period from the
ral rubber bearing (NRB) and lead rubber bearing (LRB) is the most
seismic resonance range it reduces seismic forces substantially. The

* Corresponding author.
E-mail address: dion@ynu.ac.jp (D.M. Siringoringo).

https://doi.org/10.1016/j.istruc.2022.11.003
Received 25 June 2022; Received in revised form 11 October 2022; Accepted 1 November 2022
Available online 9 November 2022
2352-0124/© 2022 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
M. Moestopo et al. Structures 46 (2022) 1154–1170

Nomenclature Q Lateral Force


Qd Characteristic strength
a Rate parameter Rs Radius of curvature of the concave plate
Ceff Effective damping T0 Natural period of isolator
d Total displacement Teff Effective natural period of the bridge
diso Isolator displacement W Structural weight
dy Isolator yield displacement Wsup Weight of superstructure
Ff Friction force u Slider relative horizontal displacement
g Gravity constant v Velocity of slider
h Pier height μ Friction coefficient
Keff Effective stiffness of the bridge μ0 Static friction coefficient
Kh1 Elastic initial stiffness μfast Friction coefficient at fast velocity
Ksub Stiffness of substructure μslow Friction coefficient at slow velocity
Kp Post-elastic stiffness ξ Effective damping ratio
N Normal force σ Bearing pressure
Pv Axial force on the bearing

common base-isolation device of this type [11]. It normally consists of high seismicity [31], bridge with tall pier [32], and in combination with
dual steel plate cover, multiple elastomeric layers, and in the case of viscous damper [33]. In the developing countries, like Indonesia, the use
LRB, steel shims in between with a lead core. Elastomeric layers provide of seismic isolation system for bridges are not yet popular. FPS with its
restoring force and lateral flexibility, lead core acts as energy dissipator, improved performance such as higher durability under severe environ­
and steel shims provide vertical bearing capacity. LRB is notable for mental conditions, reduced height, and insensitivity to the frequency
having low maintenance requirement and good life-time cost [12]. As a content of the ground motions, is a viable option for bridge seismic
result, it has been a primary choice for seismic isolator in bridges during isolation system in Indonesia. The country is known for having severe
the last two decades in Japan, the United States, New Zealand, and Italy tropical environmental conditions, complex tectonic settings with
[10,11,13,14,15]. However, elastomeric bearing size must be increased various types of seismic sources, and unfavorable soil conditions.
to maintain good stability when subjected to large displacement. Un­ Implementations of seismic isolation or any vibration control devices on
fortunately, this would increase lateral stiffness of the isolator and make highway bridges are still rare. Lack of experience with design procedure,
it ineffective. Furthermore, under low temperatures, crystallization of unawareness of benefits of such systems, and general perception of high
the rubber layers may occur and was found to increase lateral stiffness of cost of the devices are among the factors contributing to unfavorable
a LRB [16]. A study found that in the winter, LRB-isolated bridge pier perception on seismic isolation system in the country [34].
ductility was reduced, and base shear was higher than in the other This paper presents an exploratory study on the use of single concave
seasons [17]. friction pendulum (SCFP), one type of FPS, for seismic isolation device
Development of the frictional sliding bearing began in the late 1980 s in two typical bridges in Indonesia. The paper describes design pro­
[9]. There are two kinds of sliding bearing: flat sliding bearing applied as cedure of the base isolation system using the recent Indonesian seismic
a complement to elastomeric bearing, and friction pendulum bearing. code. Two typical reinforced concrete bridges with and without SCFP
The latter was introduced in 1990 as Friction pendulum system (FPS) isolation bearing are investigated and compared. The study also in­
[18]. FPS is made of stainless-steel plate with polytetrafluoroethylene vestigates important aspect of design that is the effect of variability in
(PTFE) coating and comprises of an articulated slider supporting a friction coefficient on the seismic performance. The main objectives of
spherical concave surface. The device applies engineering principles of the study are: 1) to present and implement the concept of SCFP-isolated
pendulum motion to shift the natural period away from the dominant bridges using the recent Indonesian seismic design code, 2) to compare
period, hence reducing impact of the seismic forces. Period shift can be seismic performance of typical conventional non-isolated bridge and the
achieved easily as since it depends only on the plate curvature radius. ones isolated with SCFP under the same design ground motions, 3) to
Excessive displacement is prevented by applying adequate amount of investigate effect of different type of SCFP with different friction co­
friction damping on the sliding surface [18]. Application of FPS on efficients and determine the most optimum design of SCFP by selecting
bridge has some advantages such as eliminating torsional rotation, the friction type that gives most favorable responses. In the analyses, we
distributing seismic forces uniformly, reducing three to four times present case study involving typical structures of light rapid transfer
seismically induced forces, and reducing equivalent linear stiffness of (LRT) bridge and highway bridge, which were designed by Indonesian
the whole structure [19]. In the past, some issues have been raised in the seismic code dan design practice. The paper is organized in the following
implementation of FPS such as sliding surface abrasion and wear, vari­ order. In the next section, basic information on reference structures and
ation of friction coefficient, accuracy of modeling assumption, effect of isolation systems is provided. Next, design procedure of the isolation
the response under vertical excitation, and heat problem on the sliding system is explained. Finite element simulations of both isolated and non-
surface, to name a few [20–24]. Recent research and developments, isolated bridges are then presented with discussion on the results of
however, have provided more comprehensive and better understanding analyses. The paper ends with discussion on the findings and
of the problems and the solutions. Therefore, improved models to conclusions.
describe actual behavior of FPS have been proposed and verified, and
better materials were utilized in design of the devices [25–30]. 2. Description single concave friction pendulum (SCFP)
Despite comparative advantages of FPS, the number of applications isolation system
on bridges design is still lower than elastomeric type of bearings. In the
United States, FPS in bridges is fewer than 5% of all isolated bridges Single concave friction pendulum (SCFP) named NS-Spherical
constructed [9]. So far, the use of FPS has been limited to long-span Sliding Bearing (NS-SSB) is used in this study (Yamazaki et. al., 2021)
bridges with span length varied between 60 and 100 m. In China, [35]. An isolator unit consists of articulated slider, upper concave plate
recently FPS has been used more widely for bridges located in area with with spherical radius and sliding material, spherical joint, and base

1155
M. Moestopo et al. Structures 46 (2022) 1154–1170

plate. The sliding plate is attached to concave plate. All plate compo­ expressed in the Equation (6).
nents are made of stainless steel. The upper concave that functions as ( )
μ(v) = μfast − μfast − μslow × e− av (6)
housing plate is placed on the top of the articulated slider, whereas the
spherical joint is placed on the bottom and connected to the bridge pier The friction coefficients at fast and slow velocities and rate param­
by base plate (Fig. 1). The sliding material is made of polytetrafluoro­ eter (a) are given in Table 2 along with the corresponding bearing
ethylene (PTFE) and high strength fiber. The material is selected to pressure conditions for both friction type A and type B.
ensure frictional force stability under high bearing pressure. Along perimeter of the sliding plate, restraining rim is provided as
In case of earthquake, the lateral force will induce displacement of shown in Fig. 1. The restraining rim prevents the slider from falling off
the girder and causing the slider to move on the sliding surface. The the sliding surface when displacement exceeds the capacity during
device starts to slide when Q = Ff, where Ff denotes the resultant of extreme large earthquake. A related study describes detailed experi­
friction force on sliding surface. Fig. 2(b) shows the concave section’s mental and numerical investigation on the limit state of SCFP’s
free-body diagram, from which the following horizontal and vertical restraining rim under extreme earthquake [36].
force equilibrium relationships are described:
Horizontal direction : Q − Ff cosθ − Nsinθ = 0 (1) 3. Design code and reference structures

Vertical direction : W + Ff sinθ − Ncosθ = 0 (2) 3.1. Seismicity and seismic design code

The slider’s relative horizontal on the concave surface u can be Located in the area where three tectonic plates meet, namely, Pacific,
written as: Australian, and Sunda Block plates, Indonesia is known as one of most
u = Rs sinθ (3) tectonically active regions in the world. Intricate interactions among the
major plates and at least four minor ones result in very active plate
Note that compared to the effective radius, the relative displacement boundaries regions that extend to more than 18,000 km. Within the past
in horizontal direction is negligibly small so that cosθ is approaching 20 years, there have been numerous cases of earthquake exceeding
one. Next, considering Ff = μW, one can obtain the force and Mw7.5 in Indonesia. Some of the most devastating ones are the 2004
displacement relationship as: Great Sumatra (Mw 9.2), the 2005 Nias (Mw 8.6), the 2009 Padang (Mw
W W 7.6), the 2009 Papua (Mw 7.6), and the recent 2018 Palu-Koro (Mw 7.5)
Q= u + F f = u + μW (4) earthquakes. Fatalities exceeded 200,000 people in the 2004 Great
Rs Rs
Sumatra earthquake and the economic loss was enormous due to the
It should be mentioned that the term of W
Rs
can be regarded as the widespread destructions of infrastructure. After the 2004 Great Sumatra
post-elastic lateral stiffness, therefore the natural period of isolator can earthquake, a new and more comprehensive seismic hazard map along
be expressed as the following equation: with new seismic code were released. They are updated regularly by
√̅̅̅̅̅ considering recent sources of earthquakes, developments of probabi­
Rs listic seismic hazard analysis method, and recent development of in­
T0 = 2π (5)
g ternational building and bridge code [37]. The most recent Indonesian
seismic design code for bridge is the SNI 2833:2016 [38] which is used
Equation (5) implies that the natural period of the SCFP is influenced
as the basis for seismic design and performance evaluation method
only by the radius of curvature of the concave plate. Characteristic
presented in this study.
strength is defined as the friction force, Ff of the isolator at zero
The reference bridges used as case study were located in Jakarta city,
displacement, Qd = μ0 W. Meanwhile, the post-elastic stiffness Kp = RWs is
Indonesia. Fig. 3(a) illustrates the information on seismicity including
the actual stiffness of the isolator once the characteristic force is
geological and tectonic setting within radius 250–500 km of Jakarta
exceeded. The static friction coefficient, μ0 value is assumed to be equal
city. Two main sources of earthquake are shown in the map, namely
to μfast value listed in Table 2. The idealized hysteretic loop of the various historical earthquakes on major fault lines with maximum
isolator follows a bilinear model of characteristic force, Qd = μ0 W and Mw6.5–7.6 and megathrust earthquakes with maximum Mw 7.8–9.0
post-elastic lateral stiffness, Kp = RWs illustrated by Fig. 2b. deep on intra-slab subzone southwest of Jakarta city (Fig. 3(a)). The
Three types of SCFP bearings, namely NSSSB-40-200-400, NSSSB-40- seismic hazard map and design spectra were built based on information
200-525, and NSSSB-60-200-525 were used in this study. All bearings on the historical and actives earthquake sources. Fig. 3(b) shows the
have the same radius of curvature of 4500 mm, and thus also have the Jakarta city design acceleration response spectrum with 7% probability
same natural period of 4.26 s. There were differences of bearing capacity of exceedance in 75 years (i.e., approximately 1000-year return period)
and stiffness as can be seen on Table 1. Two types of friction of SCFP and 5% damping assumption for soft soil type. The spectra are provided
bearing were used in this study: type A and type B. In general, SCFP by Directorate General of Highway, Ministry of Public Works and
friction coefficient depends on velocity of the slider and can be Human Settlement [39]. Selected spectral parameters according to the

Fig. 1. Description of NS-SSB SCFP bearing: (a) at stationary condition, (b) in moving condition.

1156
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 2. Schematic figure of SCFP bearing, (a) Free body diagram of SCFP of force equilibrium, (b) force–deformation hysteresis curve.

Table 1
Models and specifications of SCFP bearings.
Model Bearing Natural Effective Diameter Diameter of Vertical Post-elastic Nominal Ultimate Horizontal
Pressure (σ) Period (sec) of Concave Plate Slider (mm) Stiffness (kN/ Horizontal Stiffness Vertical Load Displacement (mm)
[MPa] (mm) mm) (kN/mm) (kN)

NS-SSB- 40 4.26 1070 200 2500 0.279 1256 410


40–200-
410
N-SSSB- 40 4.26 1320 200 2500 0.279 1256 535
40–200-
535
NS-SSB- 60 4.26 1320 200 3341 0.419 1884 535
60–200-
535

3.2. Description of the reference bridges and finite element models


Table 2
Friction coefficients of each type and bearing pressures.
There were two reference bridges investigated in this study: 4-span
Friction Bearing Friction Friction Rate light rapid transit (LRT) railway bridge with total length of 100 m and
Type Pressure (σ) Coefficient at Coefficient at Parameter
5-span highway bridge with total length 225 m. Oblique view, girder
[MPa] maximum minimum (a) [s/mm]
velocity (µ-fast) velocity (µ-slow) and piers cross sections, and piers arrangement are illustrated in Fig. 4
and Fig. 5 for LRT bridge and highway bridge, respectively. The piers of
Type A 40 0.050 0.030 0.008
60 0.037 0.027 0.009
LRT bridge have varying height between 7.3 m and 8.2 m, whereas the
Type B 40 0.046 0.025 0.017 height of highway bridge piers varies between 12.8 and 14.7 m. Both
60 0.043 0.023 0.017 bridges were constructed using reinforced concrete with the following
material specifications: 33 MPa compressive strength, Grade 60 rein­
forcement bar (ASTM A615M) with 413 MPa yield strength and 620
Indonesian code are listed in Table 3. It is a common engineering
MPa ultimate strength. The weights of superstructures sustained at each
practice in Indonesia to also refer to AASHTO guide and specifications
pier are listed in Table 4.
from the United States along with the national bridge design codes. The
Structural analysis of the LRT and highway bridges was carried out
most widely referred to bridge design codes are the AASHTO LRFD
using three-dimensional finite element model. The bridge was modelled
Bridge Design Specifications 2012 [40], the NCHRP 949 Guidelines for
in CSI SAP2000 [43] and the fishbone model was used where the bridge
Performance-Based Seismic Bridge Design [41], and the ASHTO Guide
girder, piers, and pier caps were modeled using frame elements. This
Specifications for Seismic Isolation Design 4th Edition 2014 [42].
element includes the effects of biaxial bending, torsion, axial deforma­
Note that the analysis described here did not include the effects of
tion, and biaxial shear deformations. Meanwhile, bearings are modelled
near-fault ground motion because there were no known active faults
as non-linear link elements. Detailed configuration of girder, piers and
near location of the bridges. Therefore, specific analysis associated with
other structural elements are illustrated in Fig. 4 (b) and Fig. 5(b), for
influences of near-field ground motions such as forward-directivity ef­
LRT and highway bridges, respectively. The LRT bridge consists of four
fect, high intensity velocity pulse, and effect of vertical ground motion
separated spans and five piers, all piers have the same dimension cross-
were not included in this study. However, if information on active faults
sectional dimensions while differ in height. The girder of LRT bridge
in the vicinity of bridge location is known a priori, then the above­
consists of two U-shaped RC box girder as shown in Fig. 4(c). They are
mentioned effects of near-field earthquake and its consequences on
modeled as two sets of beams which were discontinued at the pier caps
seismic performance of the bridge should be considered at design stage.

1157
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 3. Information on seismicity of Jakarta city: (a) seismotectonic and seismic activities around the city, (b) Design acceleration response spectra for the city for 7%
probability of exceedance in 75 years (approximately a 1000- year return period) for soft soil type and 5% damping ratio. (After [34,39]).

as illustrated by Fig. 5(b). Similar to the LRT bridge, pier caps supporting
Table 3
the girders of highway bridge have the same dimensions and they are
Parameters used in the design spectra of the reference bridges.
tapered towards both ends.
Design spectra for 7% probability of exceedance in 75 years The number of SCFP bearings on one pier was determined by
(1000-year return period)
considering the type of the bearing which gives the area and device
(SNI 2833:2016)
capacity, and the maximum structural weight imposed on each device at
Site class SE each pier. The latter quantities were known from the linear and modal
Seismic acceleration for 0.2 s period (SS) 0.572 g analysis performed beforehand; and by imposing the maximum bearing
Design peak ground acceleration (PGAD) 0.363 g capacity larger than the maximum structural weight on each pier, the
Seismic acceleration for period 1 s (S1) 0.234 g
suitable number and arrangement of device type for each pier was
Peak ground acceleration (PGA) 0.288 g
Amplification factor for PGA (FPGA) 1.26 determined. Details of the bearing capacity and maximum structural
Amplification factor for period 0.2 s (FA) 1.556 weight per pier are described in Table 4. Based on design calculations, it
Amplification factor for period 1 s (FV) 3.064 was decided that each box girder of LRT bridge was supported by two
Starting period of constant acceleration (T0) 0.161 s units of SCFP bearings on the girder end. Therefore, eight units of SCFP
Ending period of constant acceleration (TS) 0.806 s
Design acceleration for period 0.2 s (SDS) 0.89 g
bearings were placed on each LRT bridge pier. Meanwhile each box
Design acceleration for period 1 s (SD1) 0.717 g girder of highway bridge was supported by three units of SCFP bearings
on the girder end. Therefore, twelve units of SCFP bearings were placed
on each of highway bridge pier. The SCFP bearings were modelled as the
and supported by the SCFP bearings. Arrangement and properties of the nonlinear link element of friction isolator available in SAP2000. Design
right and left girders are shown in in Fig. 4(b). The pier caps supporting parameters of the nonlinear links were determined using design pro­
the girders have the same dimensions for all LRT bridge and are tapered cedure explained in the following section.
towards both ends. Meanwhile, the highway bridge consists of five In both LRT and highway bridges, gaps exist between the adjacent
separated spans and six piers. The girder consists of two RC box girder girders to allow for girder longitudinal movement. In real application,
with the cross-section properties shown in Fig. 5(c). In the finite element the gaps would be closed by expansion joints at the pavement level. The
model, one unit of box girder is modeled as beam separated at the piers size of gaps is determined such that it can accommodate the maximum

Fig. 4. (a) LRT Bridge Oblique View, (b) plan view of the LRT bridge used in FE model with dimension, (c) U-shaped cross-section of one section of bridge girder (d)
cross-section of the pier.

1158
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 5. (a) Highway bridge oblique view, (b) plan view of the highway bridge used in the FE model with dimension, (c) Typical box-girder cross-section of one girder
section (d) cross-section of the pier.

Table 4 Table 6
Detailed information on pier height and superstructure weight sustained at in­ Detailed information on structural components used in the Highway Bridge
dividual piers. finite element model.
LRT bridge Structural Frames of FE model for Highway Bridge*
Pier Height Superstructure Device Bearing Bearing Capacity Components
SV LS Pier Pier Head (PH)
(m) Weight (kN) Type per unit per unit (kN)
P1 8.073 2261.72 40 MPa 1256.64 PH1 PH2 PH3
P2 8.148 3577.52 40 MPa 1256.64
Cross section A 6.742 16.549 10.500 7.000 7.963 12.250
P3 7.336 3276.27 40 MPa 1256.64
[m2]
P4 8.026 3555.40 40 MPa 1256.64
Moment of inertia 6.288 11.273 10.719 2.333 3.434 12.505
P5 7.445 1974.43 40 MPa 1256.64
about 3 axis [I33]
(m4)
Highway Bridge Moment of inertia 69.812 134.600 7.875 7.146 8.128 12.505
Pier Height Superstructure Device Bearing Bearing Capacity about 2 axis [I22]
(m) Weight (kN) Type per unit per unit (kN) (m4)
P1 12.811 9124.87 40 MPa 1256.64 Shear Area on axis 2 2.054 10.350 8.750 5.833 6.635 10.208
P2 13.262 16446.0.5 60 MPa 1884.96 [m2]
P3 13.606 16675.8 60 MPa 1884.96 Shear Area on axis 3 4.928 14.334 8.750 5.833 6.635 10.208
P4 13.821 16700.0 60 MPa 1884.96 [m2]
P5 14.262 14270.3 60 MPa 1884.96 Section modulus 3.648 74.049 6.125 2.333 3.019 7.146
P6 14.739 9315.44 40 MPa 1256.64 about 2 axis [m3]
Section modulus 10.658 20.550 5.250 4.083 4.645 7.146
about 3 axis [m3]
Torsional constant 13.326 23.710 15.255 6.003 8.195 21.134
[N-m/rad]
displacement demand calculated using the design earthquake. This will
*Note: location and notation of frame elements are given in Fig. 5.
also determine the size of expansion joints used in design. With such
arrangement, the longitudinal girder pounding can be avoided because
conventional non-isolated models (i.e., model EL and model SS) are on
the maximum demand of girder longitudinal displacement under design
the assumption on seismic behavior of the piers. Piers in the model EL
earthquake will still below size of the gap between two adjacent girders.
are assumed to behave elastically when subjected to 1000-year earth­
Details on basic dimensions of piers, girder and pier caps cross-sectional
quake, while piers in Model SS are expected to behave inelastically
properties used in the finite element models are described in Table 5 and
following elastoplastic relationship when subjected to similar earth­
Table 6 for LRT bridge and highway bridge, respectively.
quake. Meanwhile, the isolated models (i.e., models A, B, C, and D)
Six models were constructed in the FE analysis for both LRT and
represent the SCFP-isolated bridges with different friction bearing types
highway bridges. The six models consist of two conventional non-
and arrangements. In the simulation, linear and nonlinear time-history
isolated bridge models and four SCFP-isolated bridge models. Details
analyses using finite element models were conducted. The linear
on each model are presented in Table 7. The differences in the

Table 5
Detailed information on structural components used in the LRT Bridge finite element model.
Structural Components Frames of FE model for LRT Bridge*

GR53 GR49 GL53 GL49 Pier Pier Head (PH)

PHM PHI PHJ

Cross section A [m2] 2.082 1.988 2.082 1.988 4.000 3.268 2.697 1.763
Moment of inertia about 3 axis [I33] (m4) 0.740 0.718 0.740 0.718 1.333 0.453 0.235 0.056
Moment of inertia about 2 axis [I22] (m4) 7.353 6.008 7.353 6.008 1.333 1.789 1.544 1.199
Shear Area on axis 2 [m2] 0.843 0.841 0.843 0.841 3.333 2.781 2.503 1.430
Shear Area on axis 3 [m2] 1.162 1.076 1.170 1.084 3.333 2.888 2.347 1.474
Section modulus about 2 axis [m3] 0.637 0.630 0.637 0.630 1.333 0.632 0.354 0.178
Section modulus about 3 axis [m3] 2.709 2.389 2.709 2.389 1.333 1.193 1.029 0.800
Torsional constant [N-m/rad] 0.046 0.044 0.046 0.044 2.253 1.174 0.735 0.194

*Note: location and notation of frame elements are given in Fig. 4.

1159
M. Moestopo et al. Structures 46 (2022) 1154–1170

Table 7 Next, the first assumed total displacement was estimated by


Details on SCFP bearings configurations of each bridge model. assuming damping ratio as 30%. Using this assumption, the effective
Isolated LRT bridge Highway bridge stiffness of each isolator group on each pier was estimated as:
models 8 Devices per pier (total 40) 12 devices per pier (total 72) ( )
αj
Model A • P1, P2, P3, P4, P5: Type A, 40 • P1, P6: Type A, 40 MPa Keff ,j = Ksub,j (11)
1 + αj
MPa P2, P3, P4, P5: Type A, 60
MPa
Model B • P1, P2, P4, P5: Type A, 40 • P1, P6: Type A, 40 MPa where αj is a coefficient of each pier j estimated by the following
MPa P2, P5: Type A, 60 MPa equation:
P3: Type B, 40 MPa P3, P4: Type B, 60 MPa
Model C • P1: Type A, 40 MPa • P1, P6: Type A, 40 MPa Kp2,j d + Qd,j
αj = (12)
P2, P3, P4: Type B, 40 MPa P2, P3, P4, P5: Type B, 60 Ksub,j d − Qd,j
P5: Type A, 40 MPa MPa
Model D • P1, P2, P3, P4, P5: Type B, 40 • P1, P6: Type B, 40 MPa In Equation (12) above, Kp2,j denotes the post-elastic stiffness defined
MPa P2, P3, P4, P5: Type B, 60 as Kp2,j = nj × Kp where nj is the total number of SCFP on pier j. Qd,j
MPa
denotes the total design activation force on pier j, defined as:
Conventional Non-isolated models
Model EL Piers are assumed to have elastic behavior
Qd,j = nj × Qd , and d denotes the total displacement of each pier
Model SS Piers are assumed to have elastoplastic behavior comprising the displacement of isolator and pier head.
After obtaining the effective stiffness, the effective natural period
was estimated using the effective stiffness of each isolator as:
analysis was performed to confirm that the selected structural di­ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
mensions have satisfied the code minimum requirements. The purposes W
of having six models are twofold, one is to compare performance of Teff = 2π (13)
gΣKeff ,j
SCFP-isolated bridge with conventional non-isolated bridge and the
other one is to obtain the model with SCFP bearing configuration that Using the results from Equation (10), the displacements at the pier
gives the best seismic performance subjected to the design ground head (dsub,j ) and isolator (diso,j ) were computed by the following
motion. equations:
d
4. Design procedure of the isolation system diso,j = (14.a)
1 + αj

The preliminary design of isolation bearings started with determi­ dsub,j = d − diso,j (14.b)
nation of vertical loads applied on each bridge pier. The maximum
vertical loads on each pier were determined as a basis for selecting the Furthermore, the effective damping ratio was estimated as:
suitable bearing, which in this case, was the bearing pressure of 40 MPa ΣEd,j Σ2Qd,j (diso,j − dy )
or 60 MPa. In addition, to find the most optimum performance, different ξ= = ( ) (15)
ΣEs,j ΣπKeff ,j diso,j + dsub 2
friction coefficients namely type A and B were also considered for each
bearing type. A simplified analysis method following the AASHTO GSID The above procedure involved iteration, where calculations were
2014 [42] was used in the preliminary phase to predict the effective repeated until the difference of isolator displacement in two consecutive
parameters of the isolation system. iterations was less than 5%. The final isolators displacements were then
The procedure started with determination of stiffness value of the compared with the deformation limit of the bearing. In the case when
substructure (Ksub) from known period (T) and weight of the super­ the required deformation was not met, the bearing size must be adjusted.
structure (Wsup). In this case, the weight of the superstructure was Flowchart in Fig. 6(b) illustrates the step-by-step analysis procedure for
assumed to be a lumped mass: isolated bridge. In addition, the flowchart of analysis procedure for
( )2 conventional non-isolated bridge is also included (Fig. 6a). Following
Wsup 2π the abovementioned procedure, the total design displacement and
Ksub = (7)
g T effective damping were obtained as displayed in Table 8.
The stiffness of individual pier (Ksub,j ) was further determined by After obtaining the final design displacement, restorability of the
assuming stiffness value was proportional to the third power of pier isolation system was confirmed to prevent excessive residual deforma­
height (hj): tion under significant seismic forces. As stipulated in the AASHTO GSID
2014 [42], the difference between the lateral restoring force of an
isolation system at the total design displacement (TDD) and at the half of
1
h3
Ksub,j = ∑j 1 Ksub (8) total design displacement should be greater than 1.25% of the super­
h3j
structural weight. This requirement was satisfied for each bridge model
Next, the design activation force for individual SCFP bearing (Qd) as shown by Table 9 and Table 10. Note that in both tables, the forces
was determined using formula below: and stiffness values are described as a ratio of the superstructure weight.
After confirming that all requirements were satisfied, the effective
Q d = μ0 P v (9)
isolator parameters, namely the effective stiffness (Keff) and effective
In this equation, μ0 value was assumed to be equal to μfast value listed damping coefficient (Ceff) of all isolated models were the utilized as the
in Table 2, and Pv was the axial force acting on the bearing. Afterwards, input parameters to each SCFP bearing. They were modelled as the link
the elastic and inelastic horizontal stiffnesses of individual SCFP bearing element of friction isolator in the finite element model developed in
were determined, in which the elastic initial stiffness (Kh1) was assumed SAP2000 [43].
to be 100 times that of the inelastic stiffness (Kp). Based on the initial
stiffness, post-elastic stiffness, and the design activation force, the yield 5. Seismic performance evaluation criteria
displacement (dy) of individual SCFPB was estimated as:
Following the requirements in Indonesian seismic code SNI
dy =
Qd
(10) 2833:2016 [38] and AASHTO GSID 2014, nonlinear time history ana­
Kh1 − Kp lyzes were conducted for all conventional non-isolated and the SCFP

1160
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 6. Flowcharts of the analysis procedure for: (a) conventional non-isolated bridge, (b) SCFP-bearing isolated bridge.

Table 8 Table 9
Design Displacement and Damping of Isolated LRT and Highway Bridges. Lateral restoring force capacity in isolated LRT bridge models.
LRT Bridge Parameter Model Model Model Model
Parameter Model Model Model Model A B C D
A B C D
Characteristic Strength (% of 5.03 4.95 4.78 4.62
Transversal Displacement (mm) 141.92 143.08 145.45 147.85
Weight)
Longitudinal Displacement (mm) 142.76 143.90 146.26 148.64
TDD (Total Design Displacement) 201.30 202.93 206.27 209.65
Total Design Displacement (mm) 201.30 202.93 206.27 209.65
(mm)
Effective Longitudinal Natural 1.701 1.724 1.751 1.758
Radius (mm) 4500 4500 4500 4500
Period (s)
Post-sliding stiffness (% of Weight/ 0.022 0.022 0.022 0.022
Effective Transversal Natural 1.639 1.654 1.684 1.692
mm)
Period (s)
Restoring force at 0.5 TDD (% of 7.26 7.20 7.08 6.95
Effective Damping Ratio 34.7% 34.4% 33.9% 33.3%
Weight)
Restoring force at TDD (% of 9.50 9.46 9.37 9.28
Highway Bridge Weight)
Parameter Model Model Model Model Restoring force difference (% of 2.24 2.25 2.29 2.33
A B C D Weight)
Transversal Displacement (mm) 290.59 281.36 272.60 277.10 Minimum demand (% of Weight) 1.25 1.25 1.25 1.25
Longitudinal Displacement (mm) 308.79 299.42 290.61 294.82 Status OK OK OK OK
Total Design Displacement (mm) 424.02 410.87 398.45 404.60
Effective Longitudinal Natural 2.948 2.908 2.868 2.888
Period (s) Furthermore, as mentioned in the AASHTO GSID 2014, the analysis can
Effective Transversal Natural 3.257 3.210 3.170 3.177 be performed using average response if there are seven pairs of record
Period (s)
Effective Damping Ratio 19.0% 19.8% 20.5% 20.1%
used. In this study, seismic responses of the bridge models were simu­
lated under seven pairs of ground motion records. Unfortunately, the
number of ground motion data recorded in Indonesia, especially in
bearing isolated models. In the seismic code and design practice in Jakarta city is not adequate to be used in the simulation. Therefore,
Indonesia, seismic analyses were normally conducted using ground ground motion records with characteristics as similar as possible to the
motions from Benioff zone, megathrust, shallow crustal, and shallow most frequent earthquake mechanisms in Indonesia, namely the mega­
background earthquake mechanisms. This characterizes the three main thrust, shallow crustal, Benioff, and shallow background were selected
earthquake sources in Indonesia: megathrust subduction earthquake, for the analysis. The list of selected earthquakes and their characteristics
strike-slip earthquake along the defined faults, and diffused seismic zone are given in Table 11.
that is the area not connected with a specific fault or fault type. Each ground motion pair was scaled to match the target response

1161
M. Moestopo et al. Structures 46 (2022) 1154–1170

Table 10 6.1. Performance evaluation based on bridge responses


Lateral restoring force capacity in isolated highway bridge models.
Parameter Model Model Model Model 6.1.1. Girder acceleration and amplification ratio
A B C D One of the benefits of using seismic isolation bridge is the high-
Characteristic Strength (% of 4.17 4.35 4.54 4.41 frequency filtering effect. Isolation system prevents transmission of
Weight) high frequency components of ground motion from pier to girder which
TDD (Total Design Displacement) 424.02 410.87 398.45 404.6 will result in significant reduction of girder vibration. Reduction of
(mm) girder acceleration is desirable because it minimizes the inertia force
Radius (mm) 4500 4500 4500 4500
Post-sliding stiffness (% of Weight/ 0.022 0.022 0.022 0.022
and seismic demand on the piers, thus preventing damage to bridge piers
mm) [6]. Typical seismic responses of the bridge deck and pier P3 (i.e., pier in
Restoring force at 0.5 TDD (% of 8.88 8.92 8.97 8.91 the middle) in horizontal direction subjected to ground motion ILA051
Weight) by 1999 Chi-Chi Earthquake are shown by Fig. 8 for LRT and Fig. 9 for
Restoring force at TDD (% of 13.59 13.48 13.39 13.40
highway bridges. The figures shown here are the results of non-linear
Weight)
Restoring force difference (% of 4.71 4.57 4.43 4.50 time-history analysis for Model A of LRT and highway bridge. Two
Weight) characteristics can be observed from the responses, one is that girder
Minimum demand (% of Weight) 1.25 1.25 1.25 1.25 responses in time domain are smaller than that of the piers in respective
Status OK OK OK OK direction during the large excitation. The other one is the frequency
contents of the spectra showing that high frequency components of the
spectrum and then applied to at least 3 directions with 45◦ angle to find pier accelerations are not transmitted to the girder spectra. One can see
the maximum response. Fig. 7 shows the seven selected ground motions in the frequency spectra figure that for LRT bridge in longitudinal di­
and their response spectra matched to the design response spectra of rection, the high frequency contents of pier acceleration around 6 Hz do
Jakarta city for 7% probability of exceedance in 75 years. The matching not appear in the girder acceleration spectrum. Meanwhile, for the
period window was selected between 0.2 T and 1.5 T. The figure also highway bridge, the high frequency contents of pier acceleration around
includes mean absolute error (MAE) as well as the standard deviation of 4–5 Hz do not appear at the girder acceleration spectra. This phenom­
mean absolute error (SDMAE) to quantify accuracy of spectra-matching enon, known as high-filtering effect, indicates that the SCFPs have
process. Note that MAE values are less than 3% and SDMAE values are functioned effectively in decoupling the pier and girder, and preventing
less than 0.5 indicating excellent accuracy of spectra-matching for all the high frequency content of piers’ vibration from being transmitted to
ground motions. the girder, hence creating a more flexible structure.
Next, the nonlinear time-history analysis was conducted using the Fig. 10 illustrates the summary of the girder absolute acceleration
scaled ground motions, and structural responses such as absolute deck resulting from nonlinear time history analysis under the selected ground
accelerations, isolator displacements, base shear forces, and number of motions. The accelerations shown here are resultants of the horizontal
plastic hinges were extracted. The linear modal eigenvalue analysis of absolute accelerations. In addition, it should be mentioned that although
non-isolated LRT bridge model provided the natural frequencies of the girders are discontinuous at the piers, the accelerations on each
2.141 Hz and 2.232 Hz for the first transversal and longitudinal mode, girder are equal because they were subjected to same set of ground
respectively. Meanwhile for the non-isolated highway bridge model, the motions. It can be observed that the highway girder accelerations are
finite element model provided the natural frequency of 0.665 Hz for the generally smaller than that of the LRT bridge girder. This is under­
first transversal mode, and 1.434 Hz for the first longitudinal mode. standable considering that highway bridge girder has larger mass and
longer period, which result in smaller acceleration. The SCFP bearings
6. Results of analyses and discussions were successful in reducing girder acceleration as evident by substantial
decrease of acceleration in all isolated models A-D compared to the
Seismic performance was evaluated based on two criteria. 1) Seismic conventional non-isolated bridge Model EL and Model SS. The conven­
responses resulted from non-linear finite element analyses: girder ac­ tional non-isolated bridge SS has smaller accelerations than Model EL
celerations, hysteresis curves, amplification factors, base shear forces, because the elastoplastic behavior of the piers reduced the accelerations
and isolator displacements. 2) Analysis on the possibility of plastic but increased the displacement. On the other hand, the elastic behavior
hinges development on the piers. The first criterion is to confirm the of Model EL renders that pier and girder behave elastically, and the high-
benefit of isolators type and arrangement in reducing structural re­ filtering effect do not appear in the girder accelerations in contrast to the
sponses as expected in the basic principles of seismic isolation design, case of isolated bridges. This means that the high frequency contents of
whereas investigation on the plastic hinges formation is conducted pier accelerations were transferred to the girder because pier and girder
based on procedure described in NCHRP 949 [41]. are coupled.
Results of nonolinear analysis reveal that the average peak acceler­
ation for isolated LRT bridge models was 0.188 g, while the average
peak accelerations for conventional non-isolated bridge model SS and

Table 11
Details on selected ground motion records.
Mechanism Code Source Earthquake Magnitude Epicentral Distance Duration PGA of Each Matched Ground
(MW) (km) (s) Motion (g)

Megathrust ILA051 PEER Chi-Chi Earthquake 20 September 7.62 160.21 43.00 0.610
TAP075 PEER 1999 45.00 0.573
MYG012 K-NET Tohoku Earthquake 11 March 2011 9.00 170.00 300.00 0.697
MYG013 K-NET 300.00 0.737
Benioff PDG USGS Padang Earthquake 30 September 7.60 81.00 129.00 0.567
2009
Shallow ORR PEER Whittier Narrows Earthquake 5.99 77.07 40.00 0.542
Background 10 January 1987
Shallow Crustal SER PEER Landers Earthquake 29 June 1992 7.28 75.20 55.43 0.557

1162
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 7. Time-history of spectra-matched seven historical ground accelerations used in nonlinear time-history analysis in: (a) east–west direction, (b) north–south
direction. Response spectra of ground accelerations matched to the design response spectra in Jakarta (7% probability of exceedance in 75 years) in: (c) east–west
direction (d) in north–south direction. Note: the values inside the bracket of the legend denote the mean absolute error (MAE) and standard deviation of mean
absolute error (SDMAE), respectively.

model EL were 0.545 g and 1.288 g, respectively. This suggests that in were 0.525 g and 1.675 g, respectively. The results suggest that in
average there was 65.5% reduction of peak girder acceleration when the average there was 67.2% reduction of peak girder acceleration when the
bearing system was changed from conventional non-isolated bridge bearing system of highway bridge was changed from the conventional
model SS to the SCFP-isolated model. Of all LRT bridge isolated models, model SS to the SCFP-isolated model. Model B exhibited the smallest
Model D exhibited the smallest deck absolute acceleration for all ground deck acceleration. However, similar to the case of LRT bridge, the dif­
motions. The differences in deck absolute acceleration among the iso­ ferences in deck absolute accelerations among the isolated highway
lated bridge models were quite insignificant with standard deviation bridge models were quite insignificant with the standard deviation of
0.0002 g. For the highway bridge, the average peak acceleration of the 0.001 g.
isolated bridge models was 0.172 g. Whereas the average peak accel­ Another important indicator of seismic isolation efficacy is the
erations for conventional non-isolated bridge model SS and model EL amplification ratio. It is a ratio between girder’s peak acceleration and

1163
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 8. Time histories of LRT bridge accelerations subjected to ILA051 ground motions in: (a) x-direction (along bridge axis) and (b) lateral direction. Fourier spectra
amplitude of bridge accelerations in: (c) x-direction (along bridge axis) and (d) lateral direction.

Fig. 9. Time histories of highway bridge accelerations subjected to ILA051 ground motions in: (a) x-direction (along bridge axis) and (b) lateral direction. Fourier
spectra amplitude of bridge accelerations in: (c) x-direction (along bridge axis) and (d) lateral direction.

Fig. 10. Girder absolute acceleration obtained from nonlinear time-history analysis of selected ground motion scenarios: (a) piers in LRT bridge, (b) piers in
highway bridge.

1164
M. Moestopo et al. Structures 46 (2022) 1154–1170

peak ground acceleration. Amplification ratio higher than one means the 50% were observed on the isolated models of LRT bridge and highway
structure amplifies ground acceleration, while a ratio less than one in­ bridge. The isolation system reduced earthquake forces significantly
dicates that structure system successfully de-amplifies the ground ac­ because SCFP bearings had smaller effective horizontal stiffness
celeration. Fig. 11 demonstrates the amplification ratios for all models compared to conventional non-isolated bridge. This essentially length­
caused by seven selected ground motions. It is evident from the figure ened natural period of the bridge and decreased the effect of inertia force
that conventional non-isolated highway bridge Model EL has amplifi­ on the piers. It should be pointed out that the peak of base shear forces of
cation ratios higher than one for all earthquakes in both LRT bridge and all SCFP-isolated models was very similar indicating that there was little
highway bridge. This amplification is understandable because of fixed influence of friction coefficient difference on the base shear force.
connection between pier and girder, and the fact that girder acceleration Comparisons of hysteresis curves resulting from nonlinear analysis of
increases as a consequence of the linear behavior assumption. The figure all SCFP-isolated LRT bridge models are illustrated by Fig. 13. Note that
shows that amplification ratios of girder for conventional non-isolated all hysteresis curves are still inside the design limit of SCFP bearings
bridge Model SS, which is a nonlinear (elastoplastic) model are less suggesting that base shear forces and displacements of SCFP in all
than that of model EL but still somewhat slightly higher than one for LRT models are within the design limit defined in the preliminary design. All
bridge and nearly close to one for highway bridge. As for the isolated hysteresis curves have the same force–deformation characteristics as the
bridge models, all amplification ratios were less than 0.5, with the one described for the SCFP bearing (Fig. 2(b)). Off all selected ground
average of 0.310 and 0.287 for LRT and highway bridge, respectively. motions, the largest force–displacement relationship is the one resulted
Note that the amplification ratios of all SCFP-isolated bridge models from the 2009 Padang (PDG) ground motion. Also note that the differ­
were substantially smaller compared to the conventional non-isolated ence among the four hysteresis curves is not so significant. This is un­
model including the model with nonlinear assumption. The reductions derstandable because the main source of dissipative energy which is
of amplification ratio were 65.7% and 67% for LRT and highway bridge, frictional damping of SCFP bearing is also very similar as denoted by
respectively. effective damping values in Table 8. Nevertheless, albeit small differ­
The results reveal that the SCFP isolation bearing was effective in ence, Model A of LRT bridge has slightly larger area under the hysteresis
decreasing accelerations as shown by significant reduction of girder curve indicating the largest dissipative energy.
accelerations. Among all isolated models of LRT bridge, amplification Hysteretic curves of all isolated highway bridge models are shown by
ratio of Model D was the smallest with average ratio 0.309. Meanwhile, Fig. 14. Again, it can be observed that all hysteresis curves are inside the
with the average ratio 0.285, Model B showed the smallest amplification design limit of SCFP bearings. Although, there is a slight deformation
ratio among the highway isolated bridge models. However, as in the case limit exceedance occurred in PDG (Padang Earthquake) ground motion
of absolute deck acceleration, the differences in amplification ratios scenario. Similar to the case of LRT bridge, the difference among the four
among the isolated bridge models were quite insignificant with standard hysteresis curves of highway bridge models is not so significant. This is
deviations 0.0002 and 0.0021 for LRT and highway bridges, understandable because the selected friction type of the bearings results
respectively. in the very similar effective damping values as shown in Table 8.
However, in the case of highway bridge, albeit small difference, Model C
6.1.2. Base shear forces, hysteresis curves, and isolators displacements slightly has larger area under the curve indicating the largest dissipative
Fig. 12 displays comparisons of the peaks of base shear forces under energy.
the selected ground motions. Note that base shear forces shown in the Peak isolator displacements in all isolated models are listed in
figure are the total of peak base shear forces experienced by all piers for Table 12. Ultimate lateral deformation limit of the bearing in isolated
all bridge models. The figure clearly demonstrates that Model EL of both LRT bridge models is 410 mm and in highway bridge models is 535 mm.
bridges have the greatest base shear forces under all ground motions, By applying safety reduction factor (α = 0.75), allowable isolator
which is reasonable because Model EL is assumed to have elastic deformation in LRT bridge models is 307.5 mm, and in highway bridge
behavior. On the other hand, Model SS of both bridges show significant models is 401.25 mm. Safety factor of the isolation system is determined
reduction in base shear, because of inelastic behavior of the piers. As by dividing the peak isolator displacement value with the allowable
explained previously, significant reduction of base shear force is antic­ deformation. The system is assumed to be safe if it has safety factor of
ipated in the conventional non-isolated Model SS because of the inelastic greater than 1.0. Table 12 lists the peak isolator displacements in iso­
behavior. However, it should be noted that inelastic behavior would lated LRT and highway bridge models, respectively. The table clearly
lead to a substantial consequence on the structural performance because demonstrates that the peak isolator displacements of all isolated models
it may cause formation of plastic hinges on the structure. are still within the suggested design limit. Note that highway bridge
The isolated models (A – D) generally had smaller peak base shear models that have heavier girder than LRT bridge display relatively
forces than the conventional ones for all earthquakes. Compared to the smaller lateral deformation with SCFP bearing type B (partial majority
inelastic conventional bridge system, reduction of shear force more than application combined with type A), as can be observed in highway

Fig. 11. Amplification factor obtained from NLTHA of selected ground motion scenarios: (a) piers in LRT bridge, (b) piers in highway bridge.

1165
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 12. Base shear reactions from NLTHA of selected ground motion scenarios: (a) piers in LRT bridge, (b) piers in highway bridge.

Fig. 13. Comparison of hysteresis curves of the LRT bridge models isolated with SCFP system obtained from nonlinear analysis of finite element model under selected
ground motions: (a) Model A, (b) Model B, (c) Model C and (d) Model D in longitudinal direction (i.e., bridge axis).

bridge Model C. On the other hand, LRT bridge models, which have the plastic hinge length (Lp) was determined based on the formula
lighter girder display relatively smaller lateral deformation of SCFP proposed by Priestley, et al. [47] shown schematically by Fig. 15.
friction type A (total application), as can be observed in LRT bridge Using information of the cross-sectional dimension, material, size of
model A. Table 8 shows that with the arrangement type of friction on the reinforcing bar, and the vertical load, the monotonic moment–curvature
piers, both LRT bridge Model A and highway bridge Model C have the curves were obtained from calculation in XTRACT software [46]. In
largest effective damping. Large effective damping yields smaller re­ order to construct a cyclic backbone curve according to NCHRP 949,
sponses. As the consequence of small lateral deformation, LRT bridge several adjustments have been made to the monotonic curves. PEER/
Model A and highway bridge Model C have the largest safety factor ATC 72–1 [48] recommends reducing the curve parameters by the
among their respective bridge type. following procedures: 1) capping strength: the strength is taken as 0.9
times of the monotonic curve, but not less than the yield strength, 2) pre-
6.2. Performance evaluation on the basis of formation of plastic hinges capping deformation: deformation is taken as 0.7 times of the monotonic
curve, 3) residual strength is taken as 0.7 times of the monotonic curve.
In the nonlinear analysis, plastic hinges on the bridge piers were The backbone curves were assigned as nonlinear hinges in the finite
modeled using moment-rotation backbone curves. First, the strength element models. Strength degradation of the reinforced concrete was
characteristics of confined concrete were manually calculated following assumed to follow the Takeda hysteresis model because of its charac­
the Mander’s model [44] and unconfined concrete following the Hog­ teristics, namely, the change of stiffness due to flexural crack in concrete
nestad’s model [45]. Characteristics of the unconfined and confined and tensile yielding of longitudinal reinforcements and repetitive
concrete for both bridges were then inputted into computer program deflection reversal that results in degradation in stiffness and deforma­
XTRACT [46] for further analysis. Weight of the upper structure on each tion capacity. Considering all factors above, the final backbone curve for
pier in Table 4 were applied as the constant axial load on each pier, and each pier of both bridges can be constructed as shown by Fig. 16 for LRT

1166
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 14. Comparison of hysteresis curves of the highway bridge models isolated with SCFP system obtained from nonlinear analysis of finite element model under
selected ground motions: (a) Model A, (b) Model B, (c) Model C and (d) Model D in longitudinal direction (i.e., bridge axis).

Table 12
Peak isolator displacement in isolated LRT and highway bridge models.
LRT Bridge
Ground motions Model A Model B Model C Model D
ILA051 199.92 210.82 207.15 206.72
MYG012 230.90 239.07 248.94 248.76
MYG013 227.31 232.01 231.92 231.59
ORR 181.14 191.81 192.66 191.31
PDG 289.07 297.17 298.91 298.78
SER 244.94 256.65 262.95 262.33
TAP075 250.30 262.80 259.19 258.67
Average 231.94 241.48 243.10 242.59
Safety Factor 1.326 1.273 1.265 1.268

Highway Bridge
Ground motions Model A Model B Model C Model D
ILA051 304.96 294.35 292.07 292.59
MYG012 381.08 355.96 351.88 353.45
MYG013 326.74 309.87 294.08 296.86
ORR 319.27 310.63 310.40 310.42 Fig. 15. Plastic Hinge Method from Priestley et.al. [47] used to model the
PDG 487.13 464.72 465.97 464.78 plastic hinges formed in the columns of LRT and highway bridges.
SER 463.39 423.57 415.36 417.48
TAP075 399.09 377.71 378.52 378.65
Model SS of the conventional LRT bridge are classified as operational
Average 383.09 362.40 358.32 359.18
Safety Factor 1.047 1.107 1.120 1.117
level (PL2), which means the performance level of the entire LRT bridge
Model SS is classified as operational (PL2). Note that Fig. 18(a) reveals
that plastic hinges are not formed in any piers of all isolated LRT bridge
bridge and Fig. 17 for highway bridge. models (A – D) suggesting that classification of performance level of all
Effects of SCFP bearings on the seismic performance of isolated isolated LRT bridge models was fully operational (PL3). It is evident
bridge were evaluated using the performance criteria explained above. from this comparison that seismic performances of all SCFP-isolated LRT
By comparing the moment and rotation values of each pier, structural bridge models were better than the non-isolated conventional one.
performance of the bridge was determined as fully operational (PL3), Fig. 18(b) reveals that no piers in all SCFP-isolated highway bridge
operational (PL2), or life safety (PL1) as defined in NCHRP 949 [41] models were classified as the life safety level (PL1). On the other hand,
using formulas listed in Table 13. In total, there were five piers evaluated several piers (from 1 to 5) were classified in the life safety level (PL1) of
in each finite element model of the LRT and six piers of the highway non-isolated conventional highway bridge Model SS. Thus, performance
bridge. of the entire highway bridge Model SS can be classified as the life safety
Results of nonlinear time-history analysis and investigation of the (PL1). The absence of piers in all SCFP-isolated LRT and highway bridge
formation of plastic hinges on the piers are shown in Fig. 18. Model EL of models, with operational level (PL2) (Fig. 18 (a)) and life safety level
both bridges were excluded from the evaluation because they were (PL1) (Fig. 18 (b)), respectively, can be seen as the consequence of
assumed to have elastic behavior. Fig. 18(a) shows that all piers in smaller accelerations of the isolated bridge girder compared to that of

1167
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 16. Cyclic backbone curves used in the analysis of the LRT Bridge in longitudinal(X) and lateral (Y) directions: (a) Pier P1, (b) Pier P2, (c) Pier P3, (d) Pier P4,
and (d) Pier P5.

Fig. 17. Cyclic backbone curves used in the analysis of the Highway Bridge in longitudinal(X) and lateral (Y) directions: (a) Pier P1, (b) Pier P2, (c) Pier P3, (d) Pier
P4, (d) Pier P5, and (e) Pier P6.

all isolated models of highway bridge can thus be classified as opera­


Table 13
tional (PL2). In general, the results demonstrated that for both LRT and
Performance level of hinge defined in NCHRP 949 [41].
highway bridges, seismic performances of the bridge models with SCFP
Performance Level (PL) Description Compressive Strain Limit (εc ) bearings have improved considerably compared to the conventional
PL1 Life Safety ( ρ fyh εsu
)
non-isolated models.
εc = 1.4 0.004 +1.4 v ′
fcc
PL2 Operational ρ fyh εsu
εc = 0.004 + 1.4 v ′ 7. Conclusions
fcc
PL3 Fully Operational εc ≤ 0.004
This paper presents a study on the implementation of single-concave
friction pendulum (SCFP) bearings for seismically isolated lightweight
the conventional non-isolated bridge. Reductions of acceleration resul­ LRT and heavyweight highway reinforced concrete bridges according to
ted in the reduced seismic forces and smaller moments on the piers and the recent Indonesian seismic design code. The paper describes the
consequently improved performance level. design procedure and seismic performance evaluation of the bridges
Fig. 18 (c) reveals that all piers with operational level (PL2) exists in under 1000-year design earthquake using nonlinear finite element
all models including the isolated bridge model. The results suggest that analysis. The study also provides comparisons of seismic performance of

1168
M. Moestopo et al. Structures 46 (2022) 1154–1170

Fig. 18. Number of piers and their respective acceptance criteria from nonlinear time-history analysis of selected ground motions: (a) piers in LRT bridge with
operational (PL2) criteria, (b) piers in highway bridge with life safety (PL1) criteria, (c) piers in highway bridge with operational (PL2) criteria.

the SCFP-isolated bridges under different friction types and arrange­ Declaration of Competing Interest
ments to the typical conventional non-isolated bridges, and obtains the
model with most favorable SCFP bearing arrangement. Based on the The authors declare that they have no known competing financial
finite element analyses and observations on the structural responses, the interests or personal relationships that could have appeared to influence
following conclusions can be drawn from the study. the work reported in this paper.

1. Seismically isolated bridges with SCFP bearing have better perfor­ Acknowledgement
mance compared to the conventional non-isolated bridges under the
same 1000-year earthquake simulated in Jakarta city. In comparison This research is partially funded by NSE-YNU 2022 research grant to
to the typical conventional non-isolated LRT and highway bridge, all the corresponding author (PI: Dionysius M. Siringoringo). We gratefully
SCFP-isolated bridge models demonstrate favorable seismic perfor­ acknowledge this support.
mances as indicated by significant reductions of base shear forces,
absolute girder accelerations, and response amplifications. All these References
favorable seismic performances are satisfied under the required
conditions of restoring forces restorability of the isolator system. [1] Fujino Y, Hashimoto S, Abe M. Damage analysis of Hanshin Expressway viaducts
during 1995 Kobe earthquake. I: Residual inclination of reinforced concrete piers.
2. Seismic performances level of the LRT and highway bridge models J. Bridge Eng. 2005;10(1):45–53.
with SCFP bearings have improved considerably compared to the [2] Hashimoto S, Fujino Y, Abe M. Damage analysis of Hanshin Expressway viaducts
conventional non-isolated model. Performance level of conventional during 1995 Kobe earthquake. II: Damage mode of single reinforced concrete piers.
J. Bridge Eng. 2005;10(1):54–60.
non-isolated LRT bridge can be classified as operational (PL2). [3] Han Q, Xiuli Du, Liu J, Li Z, Li L, Zhao J. Seismic damage of highway bridges
Whereas, in all isolated LRT bridge models the performance level can during the 2008 Wenchuan earthquake. Earthquake Eng. Eng. Vib. 2009;8(2):
be classified as fully operational (PL3). For the highway bridge, 263–73.
[4] Wang Z, Lee GC. A comparative study of bridge damage due to the Wenchuan,
performance of conventional non-isolated highway bridge model can Northridge, Loma Prieta and San Fernando earthquakes. Earthquake Eng. Eng. Vib.
be classified as the life safety (PL1); while performance level of all 2009;8(2):251–61.
isolated highway bridge models can be classified as operational [5] Limongelli MP, Çelebi M, editors. Seismic Structural Health Monitoring: From
Theory To Successful Applications. Springer; 2019.
(PL2).
[6] Fujino Y, Siringoringo DM, Kikuchi M, Kasai K, Kashima T. Seismic monitoring of
3. Results show that application of type A on all bearings (µ0 = 0.050 at seismically isolated bridges and buildings in japan—case studies and lessons
40 MPa, and µ0 = 0.037 at 60 MPa) in bridge with lightweight girders learned. In: Seismic Structural Health Monitoring. Cham: Springer; 2019.
(LRT bridge), as in Model A, produce the smallest lateral isolator p. 407–47.
[7] Kawashima K, Buckle I. Structural performance of bridges in the Tohoku-oki
deformation, thus the highest safety factor. However, different result earthquake. Earthquake Spectra 2013;29(1_suppl):315–38.
is observed in bridge with heavier girders (highway bridge). In the [8] Naeim F, Kelly JM. Design of seismic isolated structures: from theory to practice.
highway bridge, combination of majority application of type B John Wiley & Sons; 1999.
[9] Buckle IG, Mayes RL. Seismic isolation: history, application, and performance—a
bearings ((µ0 = 0.046 at 40 MPa, and µ0 = 0.043 at 60 MPa) with world view. Earthquake spectra 1990;6(2):161–201.
several type A bearings, such as in Model C, produce the smallest [10] Buckle I, Constantinou M, Dicleli M, Ghasemi H. Seismic Isolation of Highway
lateral deformation instead. This result is expected because type A Bridges. Buffalo: MCEER, University at Buffalo, The State University of New York;
2006.
bearings has notably higher friction reduction when applied with [11] Kelly JM. Aseismic base isolation: review and bibliography. Soil Dyn Earthquake
higher surface pressure compared to that of type B. Eng 1986;5(4):202–16.
[12] Asadi P, Nikfar D, Hajirasouliha I. Life-cycle cost-based design of bridge lead-
rubber isolators in seismic regions. Structures 2020;27:383–95.
LRT and highway bridges are critical part infrastructure system and [13] Unjoh, S. (2014). Menshin (seismic isolation) bridges in Japan. Technical Note of
their functionality after occurrence of a large earthquake is highly PWRI, (4288).
important in a metropolitan city like Jakarta. The study shows that [14] McKay GR, Chapman HE, Kirkcaldie DK. Seismic isolation: New Zealand
applications. Earthquake spectra 1990;6(2):203–22.
seismic performances of the bridges have improved significantly with
[15] Martelli A, Forni M. Seismic isolation and other antiseismic systems: recent
SCFP bearings compared to conventional non-isolated bridges. The applications in Italy and worldwide. Seismic Isol. Protect. Syst. 2010;1(1):75–123.
excellent seismic performance indicates that LRT and highway bridges [16] Cardone D, Gesualdi G, Nigro D. Effects of air temperature on the cyclic behavior of
equipped with SCFP seismic isolation system have higher possibility to elastomeric seismic isolators. Bull Earthquake Eng 2011;9:1227–55.
[17] Billah AHMM, Todorov B. Effects of subfreezing temperature on the seismic
remain functional as usual or can be restored rapidly after a large response of lead rubber bearing isolated bridge. Soil Dyn Earthquake Eng 2019;
earthquake without substantial downtime for structural repairment. 126:1–13.
[18] Zayas VA, Low SS, Mahin SA. A simple pendulum technique for achieving seismic
isolation. Earthquake Spectra 1990;6(2):317–33.
[19] Dicleli M, Mansour MY. Seismic retrofitting of highway bridges in illinois using
friction pendulum seismic isolation bearings and modelling procedures. Eng Struct
2003;25(9):1139–56.

1169
M. Moestopo et al. Structures 46 (2022) 1154–1170

[20] Kumar M, Whittaker AS, Constantinou MC. Characterizing friction in sliding [35] Yamazaki S, Nao N, Tamura Y. Development of highly durable single type spherical
isolation bearings. Earthquake Eng Struct Dyn 2015;44(9):1409–25. sliding bearings. In: NSENGI Technical Report; 2021. p. 1–7.
[21] Mokha A, Constantinou MC, Reinhorn AM, Zayas VA. Experimental study of [36] Nakamura, H., Yamaguchi, J., Takamine, H., Sakai Y. (2021) Analytical study on
friction-pendulum isolation system. J Struct Eng 1991;117(4):1201–17. the response properties of megathrust earthquakes of spherical sliding bearings
[22] Tsopelas P, Constantinou MC, Kim YS, Okamoto S. Experimental study of FPS (SSB) Part 1-3, Proc. of Architecture Institute of Japan (AIJ).
system in bridge seismic isolation. Earthquake Eng Struct Dyn 1996;25(1):65–78. [37] Nugroho WO, Sagara A, Imran I. The evolution of Indonesian seismic and concrete
[23] Eröz M, DesRoches R. The influence of design parameters on the response of building codes: From the past to the present. Structures 2022;41:1092–108.
bridges seismically isolated with the friction pendulum system (FPS). Eng Struct [38] National Standardization Agency of Indonesia. (2016). SNI 2833:2016, Bridge
2013;56:585–99. Design Subjected to Earthquake Load. Jakarta: BSN (in Indonesian).
[24] Eröz M, DesRoches R. Bridge seismic response as a function of the Friction [39] Ministry of Public Works and Housing (Indonesia). (2017). Earthquake Hazard Map
Pendulum System (FPS) modeling assumptions. Eng Struct 2008;30(11):3204–12. for Bridges 2017. Website: http://lini.binamarga.pu.go.id (Accessed 10 July 2021).
[25] Wang YP, Chung LL, Liao WH. Seismic response analysis of bridges isolated with [40] American Association of State Highway and Transportation Officials. AASHTO
friction pendulum bearings. Earthquake Eng Struct Dyn 1998;27(10):1069–93. LRFD Bridge Design Specifications: Customary U.S. Units. Washington, D.C:
[26] Fenz DM, Constantinou MC. Behaviour of the double concave friction pendulum AASHTO; 2012.
bearing. Earthquake Eng Struct Dyn 2006;35(11):1403–24. [41] National Academies of Sciences, Engineering, and Medicine. NCHRP 949: Proposed
[27] Lampropoulos A, Apostolidi E, Dritsos S, Giarlelis C, Jara J, Sutcu F, Takeuchi T, AASHTO Guidelines for Performance-Based Seismic Bridge Design. Washington,
White J. Seismic isolation and response control. IABSE SED 19. 2021. DC: The National Academies Press; 2020.
[28] Furinghetti M, Pavese A, Quaglini V, Dubini P. Experimental investigation of the [42] AASHTO. AASHTO Guide Specifications for Seismic Isolation Design. 4th ed.
cyclic response of double curved surface sliders subjected to radial and Washington, D.C: AASHTO; 2014.
bidirectional sliding motions. Soil Dyn Earthquake Eng 2019;117:190–202. [43] Berkeley CSI. (2018), Computer Program Structural Analysis Program SAP2000
[29] Lomiento G, Bonessio N, Benzoni G. Friction model for sliding bearings under 2018. Berkeley, California: Computers and Structures Inc.; 2018.
seismic excitation. J Earthquake Eng 2013;17(8):1162–91. [44] Mander JB, Priestley MJ, Park R. Theoretical stress-strain model for confined
[30] De Domenico D, Ricciardi G, Benzoni G. Analytical and finite element investigation concrete. J Struct Eng 1988;114(8):1804–26.
on the thermo-mechanical coupled response of friction isolators under [45] Hognestad E. A Study of Combined Bending and Axial Load in Reinforced Concrete
bidirectional excitation. Soil Dyn Earthquake Eng 2018;106:131–47. Members. Bulletin Series No. 399, Engineering Experiment Station, Urbana, USA:
[31] Xia XS, Cui LB, Chen XC. Seismic isolation of long span and super long unit University of Illinois, 1951.
continuous beam bridge with friction pendulum bearings. Eng Mech 2015;32: [46] Chadwell, C. B., & Imbsen, R. A. (2004). XTRACT: A tool for axial force-ultimate
167–71. curvature interactions. In Structures 2004: Building on the past, securing the future
[32] Chen X, Wu P, Li C. Seismic performance assessment of base-isolated tall pier (pp. 1-9).
bridges using friction pendulum bearings achieving resilient design. Structures [47] Priestley MJN, Calvi GM, Kowalsky MJ. Displacement-Based Seismic Design of
2022;38:618–29. Structures. Pavia: IUSS Press; 2007.
[33] He W, Jiang L, Wei B, Wang Z. The influence of pier height on the seismic isolation [48] Pacific Earthquake Engineering Research Center & Applied Technology Council.
effectiveness of friction pendulum bearing for Double-Track railway bridges. PEER/ATC 72–1: Modeling and Acceptance Criteria for Seismic Design and
Structures 2020;28:1870–84. https://doi.org/10.1016/j.istruc.2020.10.022. Analysis of Tall Buildings. Redwood City: ATC; 2010.
[34] Imran I, Siringoringo DM, Michael J. Seismic performance of reinforced concrete
buildings with double concave friction pendulum base isolation system: case study
of design by Indonesian code. Structures 2021;34:462–78.

1170

You might also like