Beruflich Dokumente
Kultur Dokumente
This paper describes the application of semiactive devices for controlling the earthquake
response of two highway bridges of different cross sections and pier heights. Each of the
bridges consists of a three-span continuous deck supported on the piers and abutments.
Semiactive devices such as the magnetorheological damper, the variable friction damper
and the variable stiffness device are considered as the control devices. These devices are
inserted between the deck and piers or abutments of the isolated bridge. The semiactive
device changes its properties according to the structural response and adds control forces
to the system. Each pier supporting the bridge is modeled as a linear lumped mass
system. The optimum parametric values of the semiactive dampers are evaluated and
considered in analysis of the bridge. A comparative study is performed for different semi-
active devices installed on the bridges under different seismic loadings in the longitudinal
direction. The behaviors of the bridges with different semiactive isolation devices are
compared with the corresponding nonisolated ones. The semiactive dampers are observed
as an effective protective device in reducing the displacements of the isolation bridges
as well as the base shear of the piers.
Keywords: Base isolation; bridge; earthquake; semiactive friction damper; variable stiff-
ness device; MR damper
1. Introduction
Recent devastating earthquakes have revealed that bridges are the most vulnerable
components of the transportation system. It has been observed that the earthquake
damage is mainly due to the excessive displacement of the bridge deck or bearings.
Hence, there is a need for improved performance of highway bridges subjected to
severe earthquake ground motions. The performance of bridges during earthquakes
can be improved by introducing suitable energy dissipation devices into the bridge
system. This can be achieved by installing suitable control devices in the bridge
∗ Research Scholar.
† Professor.
547
November 19, 2008 15:8 WSPC/165-IJSSD 00279
to dissipate the kinetic energy, thereby reducing the deformations of the piers and
deck. The concept of structural control may be utilized to either reduce the amount
of energy transferred into the bridge structure from the ground motion or absorb
some of the energy already transmitted to the bridge structure.
Structural control systems can be classified as passive, active, semiactive and
hybrid. In a passive control system, the properties of the device cannot be modified
after installation and it does not require external power or a computer process for
operation. Active control systems utilize actuators to exert control forces on the
structure, which are determined by incorporating the actuators within a feedback
control system that employs the measured structural response or ground motions
as the input data. Semiactive systems may be viewed as passive control systems,
which have been modified to allow for adjustment of the mechanical properties
(e.g. damping, stiffness and friction level). Semiactive control systems have been
receiving significant attention in recent years, because they provide a mediator
between the active and passive control systems. The three primary categories of
passive, active and semiactive devices can be placed in various combinations to
result in hybrid control systems.
Researchers are more concerned with the vibration control of buildings, for which
many advances have been made. In contrast, the vibration control of bridges has not
been studied to the same extent. Comparatively few studies have been conducted
in the past regarding the effectiveness of control devices for the seismic design of
bridges. Ghobarah and Ali1 investigated the response of a three-span bridge iso-
lated by the lead–rubber bearings subjected to earthquake motions. Using a smart
base isolation concept, the control of bridges under seismic excitations was studied
by Yang and Yu.2 Some experimental and numerical studies of semiactive control of
highway bridges were reported by Kawashima and Unjoh3 and Symans and Kelly.4
Studies have also been carried out on application of semiactive intelligent stiffener
control devices to the Walnut Creek Bridge in the USA.5 Analytical models for
elastomeric bearings were developed by Nagarajaiah and Ferrell6 and Hwang et al.7
Application of the semiactive variable stiffness damper (SAVSD) to the cable-stayed
bridge was studied by He et al.8 and Agrawal et al.9 Along the lines of semiactive
control, the seismic protection of highway bridges was investigated using the mag-
netorheological (MR) damper by Erkus et al.,10 Liu et al.11 and Sahahrabudhe and
Nagarajaiah.12 Kunde and Jangid13 reviewed the analytical and experimental stud-
ies on the effectiveness of seismic isolation and its implementation in actual bridges.
The seismic response of isolated bridges using lead–rubber bearings was studied by
Jangid.14 Some design formulas were developed and proposed for the supplemental
viscous dampers to highway bridges by Hwang and Tseng.15 The seismic response of
the simply supported base-isolated bridge with different isolators was investigated
by Matsagar and Jangid.16 Soneji and Jangid17,18 studied the effectiveness of seis-
mic isolation and semiactive MR dampers in reducing the seismic response of the
cable-stayed bridges. It is noted that all of the aforementioned studies have focused
on the performance evaluation of base-isolated bridges and little effort has been
November 19, 2008 15:8 WSPC/165-IJSSD 00279
(a)
(b)
Fig. 1. Isolated bridge: (a) general layout, (b) mathematical model.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
The substructure of each bridge consists of rigid abutments and reinforced concrete
piers. The following assumptions are made for the analysis of the isolated bridge
model:
(1) The bridge deck is straight and is supported at discrete locations along its
longitudinal axis by cross diaphragms with a zero skew angle. The bridge deck
is idealized as a rigid body.
(2) The bridge piers are assumed to remain in the elastic state during the earth-
quake excitation. This is a reasonable assumption as the isolation can reduce
the earthquake response in such a way that the structure remains within the
elastic range.
(3) Each pier is modeled as a lumped mass system with a number of segments.
Each segment is connected by a node with one horizontal degree of freedom.
The mass of each segment is lumped into the two adjacent nodes.
(4) Bridge piers are assumed to be rigidly fixed at the foundation level.
(5) The bridge is founded on firm soil or rock, and the earthquake excitations are
assumed to be synchronous at all supports.
(6) The stiffness contribution of nonstructural elements, such as curbs, parapet
walls and wearing coat, is neglected; however, their inertial effect is considered
in the simulation.
(7) Same isolators or dampers are inserted at the piers or abutments.
The above assumptions lead to the mathematical model of the isolated bridge
system [shown in Fig. 1(b)], which was also studied by Ghobarah and Ali,1 Li,19
Pagnini and Solari,20 and Kunde and Jangid.21 The bridge is subjected to the
earthquake ground motion in the longitudinal direction. The governing equations
of motion of the isolated bridge are
where M, C and K are the mass, damping and stiffness matrices, respectively, of
T
the bridge model system; x = {x0 , x1 , x2 , . . . , xn } is the vector of displacement of
the bridge; x0 is the displacement of the bridge deck relative to the ground; xi is the
displacement of the ith node of the bridge pier relative to the ground; ẋ and ẍ denote
the corresponding velocity and acceleration vectors, respectively; D is the location
T
matrix for the forces of the isolator and damper systems; F = {Fa , Fp } is the vector
of forces produced by the control devices; Fa and Fp represent the forces of the
T
control devices at the abutment and the pier level, respectively; r = {1, 1, . . . , 1}
is the influence coefficient vector of size n × 1; and ẍg represents the earthquake
ground acceleration.
Equation (1) can be recast in the state space as a set of first-order equations,
as follows:
ż = Az + B F + Eẍg , (2)
November 19, 2008 15:8 WSPC/165-IJSSD 00279
where z is the state vector, and A, B and E are the system matrices. The state
vector and the system matrices are
x 0 I 0 0
z= , A= , B = , E = . (3)
ẋ −M−1 K −M−1 C −M−1 D −r
The restoring force produced by the isolator, Fbj , and the damper, Fdj , at the
abutment and the pier level is given by
The damping matrix of the pier is not explicitly known, which is constructed
from the assumed modal damping (ξp ) for each mode of vibration using its mode-
shape and frequency. The pier of the bridge is considered to be of uniform cross-
section throughout the height. The fundamental period Tp of the pier that is free
at the top is
2π m̄p h4
Tp = , (5)
(1.875)2 EI
where m̄p is the mass per unit length of the pier, h is the height of the pier, and EI
is the flexural rigidity of the pier. Equation (5) is based on the fundamental period
of a uniform cantilever beam under transverse vibrations.
3. Control Devices
In this study, laminated rubber bearings are used as the isolators for the bridge. In
addition, the passive viscous damper and three types of semiactive control systems
are considered in the comparative study with regard to the seismic performance
of the bridge. The semiactive dampers considered include the MR damper, the
variable friction damper and the variable stiffness device. The basic characteristics
of these systems are explained in the following subsections.
Fb = cb ẋb + kb xb , (6)
where xb and ẋb are the relative displacement and velocity of the LRB, respectively;
and cb and kb are the damping and stiffness constants of the LRB, respectively.
The stiffness and damping of the LRB are selected to provide the specific values
of the two parameters, namely the isolation time period (Tb ) and damping ratio
November 19, 2008 15:8 WSPC/165-IJSSD 00279
Fb
Steel-plates kb
Rubber
xb
(a)
Piston head
Viscous fluid Piston rod Fd
xd
Orifice
(b)
Fig. 2. Schematic and mathematical model of various control devices: (a) laminated rubber
bearing, (b) viscous damper.
(ξb ), defined as
md
Tb = 2π , (7)
Σkb
Σcb
ξb = , (8)
2md ωb
where Σkb is the total stiffness of the LRB, Σcb is the total effective viscous damping
of the LRB, and ωb = 2π/Tb is the isolation frequency.
(a)
(b)
(c)
Fig. 3. Schematic and mathematical model of various semiactive control systems: (a) MR
damper, (b) variable friction damper, (c) variable stiffness device.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
where z is an evolutionary variable that accounts for the history dependence of the
response; xd is the damper displacement; ẋd is the velocity across the damper; k1
is the accumulator stiffness; c1 is viscous damping at lower velocity in the model
to produce the rolloff; c0 is viscous damping at larger velocity; k0 is the stiffness
at large velocity; yd is the damper piston displacement, and x is the initial dis-
placement of the spring; and α, β, γ, n and A are the parameters called the shape
or characteristic parameters of the model, which control the linearity in unloading
and the smoothness of transition from the pre- to the post-yield region.
The model parameters α, c0 and c1 depend on the voltage to the current driver:
while the filtered voltage u is determined by the applied voltage v to the control
circuit:
u̇ = −η(u − v) (15)
where 1/η is the time constant of this first order filter. A bang–bang control strategy
based on Lyapunov stability theory was used by Spencer et al.23 for controlling MR
dampers. This control algorithm determines the current sent to the damper as
where H(·) is the Heaviside step function and Vmax the maximum input command
voltage to the MR device.
The factor Rf is a ratio of the damper force to the critical friction force, which is
also treated as a gain multiplier. The matrices P, Q and R are the control gains,
written as
kf kc
kh = . (21)
kf + kc
Fd = kh vs xd , (22)
where xd is the relative displacement between the deck and the pier.
For the sake of simplicity and the requirement of little computational effort, a
switching control law is considered. The control force vector (Fd ) can be calculated
from Eq. (22) and the vs expressed for the switching control law is
1 if xd ẋd ≥ 0,
vs = (23)
0 otherwise,
4. Numerical Study
In the numerical study, two bridge models of different cross sections and pier heights
with and without control devices are considered for comparison of the seismic
responses. Each of the bridges is composed of a three-span continuous deck with
reinforced concrete piers and a prestressed concrete box girder. The dynamic prop-
erties of the two bridges are given in Table 1, based on the data of the bridges
studied by Wang et al.25 and Kunde and Jangid.21 The bridge considered for the
first example, referred to as bridge I, has hollow circular piers, and the isolators
are placed at the pier cap. The fundamental period of bridge I is 0.45 s. The bridge
considered for the second example, referred to as bridge II, is a relatively flexible
bridge with oblong piers. The fundamental period of bridge II is 1.5 s. Here, four sets
of semiactive dampers are installed between the deck and the piers or abutments
for exertion of the control forces. For comparison, the response of the bridge deck
to the base motion is computed under the following conditions: (i) bridge without
isolators; (ii) bridge controlled with the isolators, i.e. with the LRB; (iii) bridge
with the isolators and viscous dampers; (iv) bridge with the isolators and semi-
active dampers, i.e. MR damper, variable friction dampers and variable stiffness
devices.
The performance of controlled bridge systems is investigated under four real
earthquake ground motions, namely El-Centro, 1940 (PGA = 0.348 g); Kobe,
1995 (PGA = 0.834 g); Loma Prieta, 1989 (PGA = 0.57 g); and Northridge, 1994
(PGA = 0.843 g). The seismic response of the corresponding nonisolated bridge sys-
tem is also analyzed, for comparison purposes. The piers of the nonisolated bridges
are assumed to be connected to the superstructure by a fixed bearing, which allows
rotation but no displacement to occur. The number of nodes considered in the pier
is five (i.e. n = 5). The damping of the pier is taken as ξp = 5% of the critical damp-
ing for all modes of vibration. The bridge isolated by the LRB system is designed to
provide an isolation period of Tb = 2 s and the effective damping ratio is ξb = 10%.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
Fig. 4. Effects of command voltage on peak displacement, deck acceleration, base shear and
control force for bridge I with MR damper.
Fig. 5. Effects of command voltage on peak displacement, deck acceleration, base shear and
control force for bridge II with MR damper.
November 19, 2008 15:8 WSPC/165-IJSSD
00279
Fig. 6. Effects of damper stiffness kd on peak displacement, deck acceleration, base shear and control force for bridge I with SAVFD.
Semiactive Control of Seismically Isolated Bridges
559
November 19, 2008 15:8 WSPC/165-IJSSD 00279
Fig. 7. Effects of the damper stiffness ratio on peak displacement, deck acceleration, base shear
and control force for bridge I with SAVFD.
Fig. 8. Effects of the damper force to critical friction force ratio on peak displacement, deck
acceleration, base shear and control force for bridge II with SAVFD.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
continuously dissipate the seismic energy. From the figures, it is observed that the
isolator displacement reduces mildly with the increase in the Rf value. However,
the deck acceleration and pier base shear decrease for 0.4 < Rf < 0.7, but increase
suddenly when Rf > 0.8. This indicates that the optimum gain multiplier Rf of
the variable friction damper can be considered as 0.6 for both bridges under all
earthquake ground motions. Further, the damper control force increases with the
increase in the gain multiplier Rf , as was expected.
For the SAVSD, a parametric study is conducted by varying the damper stiffness
ratio (αk = kh /kb ) in the range of 0–2. The effect of αk on the seismic response under
all the earthquake excitations is plotted in Figs. 9 and 10 for bridges I and II, respec-
tively. The results show that the isolator displacement reduces with the increase
in the damper stiffness ratio. However, the decrease in the isolator displacement
is not very significant when αk is increased beyond unity. On the other hand, the
deck acceleration and pier base shear increase with the increase in the damper stiff-
ness ratio. Thus, the presence of an SAVSD can reduce the displacement of the
isolator, but increase the deck acceleration and pier base shear. For the present
bridge system and excitations considered, the optimum value of αk can be taken as
unity for minimizing the displacement in the deck or isolator with controlled deck
acceleration and pier base shear.
The performance of the passive viscous damper depends on the damper coef-
ficient. Here, the damping ratio is allowed to vary from ξd = 0 to 20%. Figure 11
shows the effect of the damping ratio of the viscous damper on the seismic response
Fig. 9. Effects of the damper stiffness ratio on peak displacement, deck acceleration, base shear
and control force for bridge I with SAVSD.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
Fig. 10. Effects of the damper stiffness ratio on peak displacement, deck acceleration, base shear
and control force for bridge II with SAVSD.
Fig. 11. Effects of the viscous damper’s damping ratio on peak displacement, deck acceleration,
base shear and control force for bridge I with passive viscous damper.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
of bridge I. As expected, the isolator displacement decreases with the increase in the
damper damping ratio. On the other hand, the deck acceleration and pier base shear
first decrease and then increase with the increase in the damper damping ratio. This
implies that there exists an optimal viscous damping value for which the pier base
shear and deck acceleration attain the minimum. For the selected bridge system
and excitations, the optimum viscous damping ratio is in the vicinity of 10%.
Based on the various optimal parameter values of the control devices, the peak
values of the isolator displacements, deck accelerations, base shear Vb and maximum
control force F computed for bridges I and II are tabulated in Tables 3 and 4,
respectively. These tables indicate that there is significant reduction in the isolator
displacement when the bridge is supplemented by passive or semiactive dampers.
A reduction of the order of 50% in the isolator displacement was observed for
the isolated bridges with semiactive damper systems. In some cases the presence
of the supplemental passive or semiactive dampers can slightly increase the deck
acceleration and base shear. However, this increase in the deck acceleration and base
Fig. 12. Force displacement diagrams of bridge I controlled by different semiactive devices sub-
jected to the El-Centro earthquake.
seismic event. Such a system may be designed and used with current conventional
isolated bridge technology. Further, it was demonstrated that the MR dampers are
more capable of controlling the peak isolator displacement of the bridges, thereby
reducing the required length of expansion joints in the bridge structures.
5. Conclusions
The effectiveness and performance of various semiactive systems installed in high-
way bridges subjected to different types of earthquakes has been investigated. The
dampers considered included the semiactive MR damper variable friction dampers
and the variable stiffness devices. The performance of the seismically isolated bridge
November 19, 2008 15:8 WSPC/165-IJSSD 00279
Fig. 13. Time history displacements of the deck of bridge I controlled by semiactive systems
subjected to the El-Centro earthquake.
Fig. 14. Time history accelerations of the deck of bridge I controlled by semiactive systems
subjected to the El-Centro earthquake.
November 19, 2008 15:8 WSPC/165-IJSSD 00279
with the semiactive dampers was compared with the bridge that was not isolated,
isolated, or isolated with passive viscous damper systems. A parametric study was
conducted to find the optimal parametric values for the semiactive control devices
for minimizing the structural response. From the trends of the numerical studies
conducted for the two bridges with different control devices, the following conclu-
sions may be drawn:
(1) Passive and semiactive control devices are very effective in controlling the earth-
quake response of highway bridges. They can also be used for retrofitting of
existing bridges.
(2) The displacement responses of the isolators in a seismically isolated bridge can
be significantly reduced by the use of semiactive dampers with marginal or no
increase in the deck acceleration and pier base shear.
(3) For a given isolated bridge system and excitation, there exists an optimal com-
mand voltage for the MR damper and the gain multiplier of the variable friction
damper, for which the deck acceleration and pier base shear attain the min-
imum value. However, the isolator displacement goes on decreasing with the
increase in the command voltage for the MR damper and the gain multiplier
of the variable friction damper.
(4) The performance of the semiactive control devices for the control of seismically
isolated bridges is better than that of the passive viscous damper system in
terms of peak response reduction.
(5) Among the semiactive devices, the performance of the MR damper and the
variable friction damper in response control of isolated bridges is better than
that of the stiffness devices.
References
1. A. Ghobarah and H. M. Ali, Seismic performance of highway bridges, Eng. Struct.
10(3) (1988) 157–166.
2. N. Yang and J. C. Yu, Hybrid control of seismic-excited bridge structures, Earthq.
Eng. Struct. Dynam. 24(11) (1995) 1437–1451.
3. K. Kawashima and S. Unjoh, Seismic response control of bridges by variable dampers,
J. Struct. Eng. 120(9) (1994) 2583–2601.
4. M. Symans and S. W. Kelly, Fuzzy logic control of bridge structures using intelligent
semi-active seismic isolation system, Earthq. Eng. Struct. Dynam. 28(1) (1999) 37–60.
5. W. N. Patten, J. Sun and G. Li, Field test of an intelligent stiffener for bridges at the
I-35 Walnut Creek Bridge, Earthq. Eng. Struct. Dynam. 28(2) (1999) 109–126.
6. S. Nagarajaiah and K. Ferrell, Stability of elastomeric seismic isolation bearings,
J. Struct. Eng. 125(9) (1999) 946–954.
7. J. S. Hwang, J. D. Wu, T. C. Pan and G. Yang, A mathematical hysteretic model for
elastomeric isolation bearings, Earthq. Eng. Struct. Dynam. 31(4) (2002) 771–789.
8. W. L. He, A. K. Agrawal and K. Mahmoud, Control of seismically excited cable-
stayed bridge using resetting semi-active stiffness dampers, J. Bridge Eng. 6(6) (2001)
376–384.
November 19, 2008 15:8 WSPC/165-IJSSD 00279