Professional Documents
Culture Documents
Variable Dampers For Earthquake Protection of Benchmark Highway Bridges
Variable Dampers For Earthquake Protection of Benchmark Highway Bridges
Variable Dampers For Earthquake Protection of Benchmark Highway Bridges
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0964-1726/18/11/115011)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 129.89.24.43
This content was downloaded on 06/06/2014 at 03:33
E-mail: rsjangid@civil.iitb.ac.in
Abstract
The dynamic behaviour of benchmark highway bridges using variable dampers under six
bidirectional earthquake ground motions is presented. A viscous damper is used as a passive
control device and a variable damper, developed from an MR (magnetorheological) damper, is
used as a semi-active control device. The study is based on the simplified lumped-mass
finite-element model of the 91/5 highway bridge in Southern California. The prime aim of the
study is to investigate the effectiveness of viscous dampers and variable dampers with a
friction-type damping force scheme and a two-step viscous damping force scheme, with
important parametric variation. Velocity-and displacement-dependent models of a variable
damper are used. Numerical simulations are conducted by installing the devices between the
deck and abutments of the bridge. The seismic response of the bridge is compared with the
corresponding uncontrolled case, and controlled by alternative sample control strategies. The
results of the investigation clearly indicate that the base shear and mid-span displacement are
reduced substantially by these protective devices. It is concluded that variable dampers are quite
effective in considerably reducing the large bearing displacements, thereby eliminating the
isolator damages. The control effect of a viscous damper, and the two pre-set damping schemes
of a variable damper in particular, is quite satisfactory in reducing the peak response quantities
of the bridge to a level comparable to or better than that of the sample passive, semi-active and
active controllers.
isolated bridges are therefore more than those of the bridges modelling the damping force with a friction-type damping
with conventional elastomeric bearings. Semi-active protective force scheme and two-step viscous damping force scheme;
systems combine the features of active and passive control (iii) study the effects of variation of large (Cd1 ) and small
to reduce the response to seismic loading. A significant (Cd2 ) damping coefficients of the two-step viscous damping
amount of research and development has been conducted on force scheme and (iv) make a systematic comparison of the
these devices because of their relatively high performance and response of viscous and variable dampers with the sample
low energy requirement. To compare the performance and control strategies provided in the problem definition.
effectiveness of various control systems in protecting bridges
from earthquakes, a benchmark problem on highway bridges 2. The benchmark highway bridge model
has been recently developed by Agrawal et al (2005). In phase
I of the problem, the bridge deck is fixed to the outriggers and The bridge model used for the benchmark study is that of the
in phase II, the bridge deck is isolated from the outriggers. The 91/5 highway overcrossing in Southern California. Seismic
present study is based on phase I. considerations were strongly considered in the design of this
Tan et al (2005) presented sample passive, semi-active bridge, as it is located very close to two major faults and in its
and active control system designs for the seismically excited critical role as a principal overcrossing. A brief description
benchmark highway bridge. Choi et al (2006) developed of the benchmark bridge and its model is presented herein.
a smart passive system, consisting of an MR damper and A detailed description of the bridge can be found in Agrawal
an electromagnetic induction (EMI) system. The results of et al (2005). The superstructure of the bridge consists of a
the numerical simulations show that the control system is two-span continuous cast-in situ pre-stressed concrete (PC)
beneficial in reducing seismic response of the benchmark three-cell box-girder and the substructure is in the form of
highway bridge. The performance of fuzzy logic control PC outriggers. Each span of the bridge is 58.5 m long,
systems for the earthquake protection of the bridge is also spanning a four-lane highway, with two skewed abutments.
investigated (Reigles and Symans 2006, Ali and Ramaswamy The width of the deck is 12.95 m along the east and 15 m
2006). Kawashima and Unjoh (1994) used a displacement- along the west direction. The total mass of the benchmark
dependent damping model of a variable fluid damper. highway bridge is 4237 544 kg and the mass of the deck is
Analytical results and shake table tests of a 30 m long 3278 404 kg. The deck is supported by a 31.4 m long and
bridge indicated that there was significant reduction in deck 6.9 m high PC outrigger, resting on pile foundations. The
displacement and deck acceleration. Spencer et al (1997) ratio of the mass of the piers to the mass of the deck of
demonstrated that MR fluid dampers can be controlled with the bridge is 0.293. In the actual bridge, four conventional
small power supplies and the dynamic range of damping force elastomeric bearings are provided at each abutment and four
level is quite large. Due to high energy dissipation capacity, passive fluid dampers are installed between each abutment and
MR dampers were considered appropriate for developing the deck-end. In the evaluation model, lead rubber bearings
variable dampers. MR dampers of 500 kN capacity have (LRBs) are used in place of the elastomeric bearings. The
been designed and fabricated. Also MR dampers of capacity uncontrolled structure, used as a basis of comparison for the
200–400 kN have been successfully used for civil engineering controlled system, corresponds to the model, isolated with
applications in Japan (Weilian and Jianwei 2009). Variable four LRBs at each deck-end. The model resulting from the
dampers of capacity 500 kN using viscous dampers and MR finite-element formulation has a large number of degrees of
dampers, with a pre-set algorithm of damping force have freedom. To make it manageable for dynamic simulation,
been developed (Kawashima and Unjoh 1994, Ruangrassamee while retaining the fundamental behaviour of the bridge, an
and Kawashima 2001, 2003, Ruangrassamee et al 2004). extensive evaluation model of the bridge with 430 ( N ) degrees
The simulation results for a five-span continuous bridge have of freedom (DOFs) has been developed in ABAQUS. It is
demonstrated their superior performance compared to passive considered to portray the actual dynamics of the bridge and
dampers. Previous investigations have demonstrated the it has been used by many researchers as a basis of comparison
effectiveness of variable dampers in reducing the response of results using various protective systems. Figure 1 shows
quantities of a benchmark cable-stayed bridge (Ruangrassamee the simplified lumped-mass finite-element model considered
and Kawashima 2006). Therefore, it would be interesting for seismic investigations. Transverse (referred to as the x
to investigate the application of protective devices, such as direction) is north–south and the longitudinal (referred to as
viscous dampers and variable dampers, in reducing the mid- the y direction) is east–west. For the purpose of analysis, the
span deck displacement as well as base shear and moments at model is divided into 108 nodes.
the base of the piers of the benchmark highway bridge. The bridge superstructure is represented by 3D beam
This paper investigates the performance of viscous elements and modelled by the B31 element in ABAQUS. Rigid
dampers and variable dampers for the earthquake protection links are used to connect the control devices between the deck-
of the benchmark highway bridge under real earthquake end and abutments. The effects of soil–structure interaction
ground motions through numerical simulations. The analytical at the abutments and approach embankments are taken into
study puts forth a simplified use of the MR damper as consideration (Makris and Zhang 2002). Each column of
a variable damper. The specific objectives of this study the bent is modelled by a B31 element as a stiff column
are to: (i) investigate the performance of viscous dampers; cap, elastic column segments and nonlinear column segments.
(ii) investigate the performance of variable dampers by The bearings are modelled as a shear element by taking the
2
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
3
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 3. (i) Viscous element, (ii) force–displacement loop and (iii) force–velocity relationship of linear viscous damper.
Figure 4. (i) Mathematical model, (ii) force–displacement loop and (iii) force–velocity relationship of friction-type damping force scheme of
variable damper.
4
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
domain is employed. The nonlinear finite-element model of the solution of the equations because the system is non-classically
bridge is considered excited under two horizontal components damped due to the difference in damping in the isolation
of earthquake ground motion, applied along the two orthogonal system as compared to the damping in the superstructure of
directions, acting simultaneously at all supports. The responses the seismically isolated bridge as well as the supplemental
in both directions are considered to be uncoupled and there damping devices. Therefore, the governing equations of
is no interaction of forces. The equations of motion of the motion are solved using the S-function by the Newmark-
evaluation model are expressed in the following matrix form: β iterative method of step-by-step integration, adopting the
[M]{ü(t)} + [C]{u̇(t)} + [K (t)]{u(t)} = −[M]{η}{ü g (t)} average acceleration over a small time interval of t . The
parameters for Newmark’s technique are selected as α = 1/4
+ [b]{F(t)} (5) and = 1/2. The time interval for solving the equations of
{u(t)} = {x 1 , x 2 , x 3 , . . . , x N , y1 , y2 , y3 , . . . , y N } T
(6) motion is taken as 0.002 s. At each time instant, the response
5
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Table 2. Evaluation criteria J1 to J4 for sample passive, semi-active and active control strategy, viscous damper, friction-type damping force
scheme and two-step viscous damping force scheme of variable dampers for North Palm Springs 1986, Chi-Chi 1999, El Centro 1940,
Northridge 1994, Turkey 1999 and Kobe 1995 earthquakes. (Note: bold number indicates minimum value; Max indicates maximum of the six
control strategies.)
Response quantity Control strategy N Palm Springs Chi-Chi El Centro Northridge Turkey Kobe Max
Peak base shear, J1 Passive 1.224 0.763 0.636 0.775 0.776 0.860 1.224
Semi-active 0.962 0.842 0.779 0.886 0.904 0.817 0.962
Active 0.950 0.877 0.790 0.896 0.912 0.789 0.950
Viscous 0.977 0.808 0.785 0.827 0.888 0.839 0.977
Friction-type 1.526 0.642 0.661 0.684 0.912 0.995 1.526
2-step viscous 0.998 0.815 0.855 0.844 0.906 0.883 0.998
Peak base moment, J2 Passive 0.633 0.959 0.576 0.960 0.879 0.549 0.960
Semi-active 0.748 0.978 0.708 0.979 0.979 0.666 0.979
Active 0.770 0.966 0.742 0.978 0.978 0.704 0.978
Viscous 0.733 0.967 0.716 0.969 0.977 0.693 0.977
Friction-type 0.791 0.949 0.361 0.938 0.555 0.501 0.949
2-step viscous 0.778 0.965 0.782 0.970 0.981 0.777 0.981
Peak mid-span Passive 0.642 0.713 0.653 0.703 0.587 0.632 0.713
displacement, J3 Semi-active 0.802 0.785 0.775 0.857 0.717 0.663 0.857
Active 0.823 0.799 0.779 0.867 0.746 0.704 0.867
Viscous 0.790 0.763 0.796 0.776 0.718 0.700 0.796
Friction-type 0.533 0.576 0.373 0.568 0.375 0.505 0.576
2-step viscous 0.815 0.837 0.868 0.814 0.799 0.785 0.868
Peak mid-span Passive 1.296 0.954 0.940 0.901 0.937 1.072 1.296
acceleration, J4 Semi-active 0.981 0.876 0.896 0.899 0.801 0.986 0.986
Active 0.794 0.875 0.883 0.844 0.798 0.899 0.899
Viscous 1.021 0.966 0.955 0.923 0.864 0.942 1.021
Friction-type 1.562 0.908 0.956 0.858 1.143 1.234 1.562
2-step viscous 1.039 1.029 1.022 0.957 0.945 0.878 1.039
based on the norm responses, calculated by normalizing the criteria are compared for all earthquakes; however, the time-
norm response quantities by the corresponding norm of the history response for each control scheme is plotted for the
uncontrolled bridge and J15 – J21 are related to the controller. Chi-Chi (1999) earthquake in the longitudinal and transverse
The evaluation criteria are listed in tables 2–5. The normed directions of the bridge. The results for only the dampers
value of the response, denoted ·, is defined as · =
located between the east deck-end and abutments are presented
tf herein, since the dampers on the west side exhibit a similar
1
tf 0
(·)2 dt , where t f is the time required for the response
response. The main response quantities of interest are the
to attenuate. In the present study, 16 evaluation criteria,
base shear in piers ( J1 ), mid-span deck displacement ( J3 ),
namely J1 – J16 , are considered. The evaluation criteria J17
bearing (abutment) deformation ( J5 ) and their corresponding
and J18 are related to the power of the device and are not
norm values ( J9 , J11 and J13 , respectively). The base shear
significant in this study as the power required by the variable
in the piers indicates the forces exerted in the bridge due to
dampers is very low. The number of control devices, J19 , is earthquake ground motion. On the other hand, the relative
16. Since the variable dampers use velocity and displacement displacement of the control devices is crucial from the design
measurements, J20 and J21 related to control resources and point of view of the control system.
sensors are not discussed in the numerical results. In order
to access the performance, the seismic response of the bridge
5.1. Parametric study on viscous damper
with a viscous damper, friction-type damping force scheme
and two-step viscous damping force scheme is compared with The performance of the viscous dampers depends on the
the corresponding uncontrolled (with LRBs) response and the damping coefficient. Likewise, the performance of the variable
sample passive, semi-active and active controllers. Evaluation dampers depends on the damping force, transitional velocity
6
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Table 3. Evaluation criteria J5 to J8 for sample passive, semi-active and active control strategy, viscous damper, friction-type damping force
scheme and two-step viscous damping force scheme of variable dampers for North Palm Springs 1986, Chi-Chi 1999, El Centro 1940,
Northridge 1994, Turkey 1999 and Kobe 1995 earthquakes. (Note: bold number indicates minimum value; Max indicates maximum of the six
control strategies.)
Response quantity Control strategy N Palm Springs Chi-Chi El Centro Northridge Turkey Kobe Max
Peak bearing Passive 0.397 0.671 0.344 0.661 0.542 0.275 0.671
displacement, J5 Semi-active 0.812 0.765 0.566 0.853 0.674 0.512 0.853
Active 0.937 0.803 0.643 0.883 0.714 0.586 0.937
Viscous 0.604 0.725 0.452 0.745 0.680 0.339 0.745
Friction-type 0.170 0.526 0.132 0.511 0.316 0.174 0.526
2-step viscous 0.649 0.797 0.493 0.784 0.752 0.415 0.797
Peak ductility, J6 Passive 0.633 0.598 0.576 0.585 0.194 0.549 0.633
Semi-active 0.748 0.696 0.708 0.828 0.372 0.666 0.828
Active 0.770 0.743 0.742 0.852 0.463 0.704 0.852
Viscous 0.733 0.657 0.716 0.688 0.358 0.693 0.733
Friction-type 0.791 0.412 0.361 0.369 0.122 0.501 0.791
2-step viscous 0.778 0.784 0.782 0.742 0.556 0.777 0.784
Peak dissipated Passive 0 0.226 0 0.333 0 0 0.333
energy, J7 Semi-active 0 0.468 0 0.567 0.236 0 0.567
Active 0 0.512 0 0.624 0.332 0 0.624
Viscous 0 0.408 0 0.436 0.128 0 0.436
Friction-type 0 0.034 0 0.119 0 0 0.119
2-step viscous 0 0.554 0 0.563 0.290 0 0.563
Maximum plastic Passive 0 0.667 0 0.75 0 0 0.75
connections, J8 Semi-active 0 0.667 0 1 0.333 0 1
Active 0 0.667 0 1 0.333 0 1
Viscous 0 0.667 0 1 0.333 0 1
Friction-type 0 0.333 0 0.5 0 0 0.5
2-step viscous 0 0.667 0 1 0.333 0 1
Table 4. Evaluation criteria J9 to J12 for sample passive, semi-active and active control strategy, viscous damper, friction-type damping force
scheme and two-step viscous damping force scheme of variable dampers for North Palm Springs 1986, Chi-Chi 1999, El Centro 1940,
Northridge 1994, Turkey 1999 and Kobe 1995 earthquakes. (Note: bold number indicates minimum value; Max indicates maximum of the six
control strategies.)
Response quantity Control strategy N Palm Springs Chi-Chi El Centro Northridge Turkey Kobe Max
Norm base Passive 1.031 0.747 0.518 0.709 0.738 0.746 1.031
Shear, J9 Semi-active 0.779 0.846 0.597 0.829 0.840 0.691 0.846
Active 0.743 0.885 0.676 0.867 0.894 0.739 0.894
Viscous 0.853 0.836 0.655 0.795 0.838 0.729 0.853
Friction-type 1.245 0.525 0.541 0.584 0.827 0.828 1.245
2-step viscous 0.824 0.882 0.729 0.830 0.896 0.804 0.896
Norm base moment, J10 Passive 0.529 0.747 0.393 0.718 0.366 0.515 0.747
Semi-active 0.662 0.798 0.559 0.838 0.502 0.656 0.838
Active 0.696 0.834 0.643 0.878 0.532 0.713 0.878
Viscous 0.682 0.789 0.623 0.809 0.511 0.689 0.809
Friction-type 0.615 0.572 0.274 0.751 0.229 0.435 0.751
2-step viscous 0.770 0.839 0.697 0.850 0.689 0.762 0.850
Norm mid-span Passive 0.557 0.645 0.410 0.650 0.442 0.544 0.650
displacement, J11 Semi-active 0.683 0.753 0.580 0.777 0.573 0.675 0.777
Active 0.703 0.784 0.656 0.805 0.607 0.729 0.805
Viscous 0.708 0.731 0.642 0.724 0.568 0.718 0.731
Friction-type 0.414 0.417 0.235 0.516 0.273 0.358 0.516
2-step viscous 0.790 0.790 0.717 0.784 0.643 0.792 0.792
Norm mid-span Passive 1.017 0.765 0.743 0.777 1.008 1.059 1.059
acceleration, J12 Semi-active 0.789 0.807 0.690 0.818 0.809 0.837 0.837
Active 0.723 0.791 0.685 0.796 0.795 0.798 0.798
Viscous 0.890 0.842 0.734 0.830 0.899 0.896 0.899
Friction-type 1.231 0.630 0.793 0.731 1.152 1.206 1.231
2-step viscous 0.873 0.924 0.820 0.896 0.903 0.925 0.925
and the maximum and minimum damping coefficient. Hence, linear viscous damper is varied from 600 to 1600 kN s m−1 .
extensive numerical simulations are conducted to determine Figure 6 shows variation of peak base shear, mid-span
the optimal parameters. The damping coefficient c of a deck displacement, bearing (abutment) deformation and their
7
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Table 5. Evaluation criteria J13 to J16 for sample passive, semi-active and active control strategy, viscous damper, friction-type damping force
scheme and two-step viscous damping force scheme of variable dampers for North Palm Springs 1986, Chi-Chi 1999, El Centro 1940,
Northridge 1994, Turkey 1999 and Kobe 1995 earthquakes. (Note: bold number indicates minimum value; Max indicates maximum of the six
control strategies.)
Response quantity Control strategy N Palm Springs Chi-Chi El Centro Northridge Turkey Kobe Max
Norm bearing Passive 0.251 0.616 0.248 0.616 0.291 0.196 0.616
displacement, J13 Semi-active 0.454 0.746 0.393 0.774 0.405 0.355 0.774
Active 0.483 0.784 0.484 0.821 0.521 0.472 0.821
Viscous 0.333 0.710 0.416 0.700 0.384 0.278 0.710
Friction-type 0.097 0.386 0.111 0.474 0.170 0.104 0.474
2-step viscous 0.355 0.766 0.465 0.758 0.442 0.323 0.766
Norm ductility, J14 Passive 0.529 0.657 0.393 0.994 0.035 0.515 0.994
Semi-active 0.662 0.693 0.559 0.771 0.220 0.656 0.771
Active 0.696 0.648 0.643 0.827 0.239 0.713 0.827
Viscous 0.682 0.618 0.623 0.616 0.223 0.689 0.689
Friction-type 0.615 0.681 0.274 1.218 0.022 0.435 1.218
2-step viscous 0.770 0.506 0.697 0.717 0.521 0.762 0.770
Maximum control Passive 0.012 0.024 0.010 0.022 0.017 0.013 0.024
force, J15 Semi-active 0.011 0.023 0.008 0.023 0.016 0.010 0.023
Active 0.010 0.024 0.006 0.023 0.015 0.008 0.024
Viscous 0.009 0.022 0.006 0.020 0.014 0.009 0.020
Friction-type 0.024 0.024 0.021 0.024 0.024 0.024 0.024
2-step viscous 0.011 0.024 0.007 0.021 0.016 0.010 0.024
Maximum device Passive 0.382 0.642 0.317 0.602 0.537 0.271 0.642
stroke, J16 Semi-active 0.782 0.732 0.521 0.777 0.669 0.505 0.782
Active 0.902 0.769 0.592 0.804 0.708 0.578 0.902
Viscous 0.582 0.694 0.416 0.679 0.674 0.334 0.694
Friction-type 0.164 0.504 0.121 0.465 0.313 0.172 0.504
2-step viscous 0.624 0.763 0.453 0.714 0.745 0.409 0.763
corresponding norm values. Base shear decreases with force scheme is considered as 1000 kN. This is also equal to
increase in the damping coefficient for all earthquakes, except the control force capacity of MR dampers used in the design
for North Palm Springs. The mid-span displacement and of sample semi-active controllers. Considering 16 control
the bearing displacement are found to reduce as the damping devices, the total damping force provided is 48.8% of the
coefficient increases. Large capacity viscous dampers are weight of the deck. Further, the transitional velocity u̇ t is varied
capable of achieving a greater level of vibration reduction. from 0.05 to 1 m s−1 . The results of the study are presented in
However, for c > 600 kN s m−1 , the control force exceeds figure 8. The trend of the results indicates that the base shear
the capacity of the device (1000 kN). For c = 600 kN s m−1 or response decreases drastically with an increase in the u̇ t for the
lower, the control force for all earthquakes is within the device North Palm Springs, Turkey and Kobe earthquakes. Further,
capacity. Hence for most effective use of a linear viscous the base shear increases marginally for the Chi-Chi, El Centro
damper, the optimum damping coefficient c is considered to and Northridge earthquakes. However, a lower value of u̇ t is
be 600 kN s m−1 . favourable for all earthquakes, considering the displacement
response. To study the force–velocity variation, graphs were
5.2. Parametric study on friction-type damping force scheme plotted for u̇ t = 0.05, 0.25, 0.50, 0.75 and 1 m s−1 for all
of variable damper earthquakes. It is observed that, for 0.25 u̇ t 1.0, the
A parametric study is carried out to investigate the effect of graphs do not follow the required pattern, as shown in figure 4,
variation in the friction damping force Fd and the transitional for all earthquakes. Hence by increasing u̇ t beyond 0.25 m s−1
velocity u̇ t of the friction-type damping force scheme of a does not give the required variation of force with velocity.
variable damper on the response quantities of interest. Initially, As the control force is less than 1000 kN for this range of
u̇ t is assumed as 0.2 m s−1 and the friction-type damping force transitional velocities, the horizontal portion of the graph is
of the damper is varied from 100 to 1000 kN, i.e. the total absent and the graphs are similar to the force–velocity variation
damping force is varied from about 0.3% to 3% of the deck of linear viscous dampers. The exact pattern of graphs is
weight, respectively. The results are presented in figure 7. It obtained for u̇ t = 0.15 and 0.20 m s−1 . Thus, the optimized
is observed that the pier base shear reduces with an increase parameters for the friction-type damping force scheme of a
in the damping force for all earthquakes, except for North variable damper are judiciously selected as Fd = 1000 kN and
Palm Springs. Further, it can be seen that, with an increase u̇ t = 0.15 m s−1 , which yield a maximum reduction in the
in the damping force, the displacement response reduces displacement of the deck and abutment bearings of the bridge,
significantly. Hence, for further numerical simulations, the without hampering the significant gain achieved in the base
damping force of a variable damper for a friction-type damping shear response.
8
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 6. Variation of J1 , J3 , J5 , J9 , J11 and J13 with damping coefficient of viscous damper.
5.3. Parametric study on two-step viscous damping force in figure 10. However, to provide a smaller damping force
scheme of variable damper when the stroke and velocity of the damper have opposite
signs, the ratio should be higher. Force–displacement loops are
For the two-step viscous damping scheme of a variable damper,
plotted for Cd1 = 600 kN s m−1 and different Cd2 . To get an
different combinations of Cd1 (large damping coefficient) and
optimal compromise between the response reduction, energy
Cd2 (small damping coefficient) are considered. Three values
dissipation and the damper capacity, Cd1 = 600 kN s m−1 and
of ( CCd1d2 ) are considered, namely 2, 3 and 4. The response
is found to improve for higher values of Cd1 , as it yields a Cd2 = 300 kN s m−1 are considered to be the optimum.
maximum damping force: however, the control force exceeds The results of the analytical parametric study performed
the device capacity for Cd1 > 650 kN s m−1 . At Cd1 = on the viscous damper, friction-type damping force scheme and
650 kN s m−1 , the control force of the variable damper is two-step viscous damping force scheme of a variable damper
less than 1000 kN for all earthquakes. Further trials to prove that these devices are beneficial for seismic response
decide Cd2 are carried out with Cd1 = 600 kN s m−1 . Six reduction of the benchmark highway bridge. Optimum
different values of the ratio ( CCd1d2 ) are considered. The results selection of the device parameters largely governs the response
of the selected evaluation criteria are presented in figure 9. of the bridge. For all control strategies, the parameters should
The criteria are found to increase with the increase in the be selected to achieve two objectives. First, the response of
ratio. A smaller value of the ratio is preferred for improved the bridge is required to remain elastic under the action of
response and also for maximum energy dissipation, as depicted small earthquakes, wind and braking forces. Second, during
9
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 7. Variation of J1 , J3 , J5 , J9 , J11 and J13 with damping force Fd of friction-type damping force scheme of variable damper.
severe seismic events, the level of the forces and displacements It is observed that there is no increase in the pier base
is required to remain within acceptable limits to ensure that shear response in both directions. Performance of a viscous
the bridge continues to function. It is observed that a variable damper and a two-step viscous damping force scheme is
damper with a simple controlling mechanism of damping comparable in both directions, with about 20% reduction in
force, based on the stroke and velocity of the device, is the response. A friction-type damping force scheme performs
capable of controlling the displacement response of the deck well in the longitudinal direction, with 31% reduction in the
and bearings at abutments, while simultaneously limiting the base shear. A similar trend of results is observed for the
base shear response of the piers. mid-span displacement response, as depicted from figure 12.
The response in the transverse direction is reduced to 80%
5.4. Time-history analyses by all three control strategies. However, in the longitudinal
The results of time-history analyses along the transverse and direction, the mid-span displacement is substantially (42%)
longitudinal directions are presented in figure 11 for a viscous reduced due to a friction-type damping force scheme of
damper, friction-type damping force scheme and two-step a variable damper, compared to 22% for the two viscous
viscous damping force scheme of a variable damper under strategies. It is observed from the time variation plots shown
the Chi-Chi, 1999 earthquake. The results obtained with in figure 13 that the reduction in the displacement response
an LRB isolation system only (uncontrolled response) are of the bearings at the abutments is remarkable for the bridge
plotted in the same graph for the purpose of comparison. with a friction-type damping force scheme of a variable
10
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 8. Variation of J1 , J3 , J5 , J9 , J11 and J13 with transitional velocity u̇ t of friction-type damping force scheme of variable damper.
damper. The bearing displacements are dramatically reduced force scheme it is 0.08 m with a control force of 400 kN. The
to only 7% in the transverse direction and to 55% in the dampers oriented along the longitudinal direction of the bridge
longitudinal direction. The reduction by viscous strategies is require a control force in the range of 900–1000 kN, with a
50% in the transverse direction and 23% in the longitudinal device stroke in the range of 0.16–0.24 m. From the size of the
direction. hysteresis loops it is observed that dampers oriented along both
directions dissipate substantial amounts of seismic energy.
5.5. Force–deformation behaviour
The force–deformation behaviour in the transverse and 5.6. Force–velocity variation
longitudinal directions for a viscous damper, friction-type
damping force scheme and two-step viscous damping force The force–velocity variations are shown in figure 15. For
scheme of a variable damper is plotted in figure 14. The energy a viscous damper, the variation of damping force with the
dissipating capability is characterized by the plot of damper velocity of the damper is linear in both directions. However,
force versus deformation across the device, i.e. the hysteresis the friction-type damping force scheme shows linear variation
loop. The stroke of the damper in the transverse direction for in the transverse direction and, as per the control law shown
a viscous damper is 0.05 m with a control force of 263 kN, in figure 4, in the longitudinal direction. For the two-step
for a friction-type damping force scheme it is 0.01 m, with a viscous force scheme, the loops are bow-shaped, with a clip
control force of 540 kN, and for a two-step viscous damping at the centre.
11
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
12
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 10. Force–displacement loops for C d1 = 600 kN s m−1 and Cd2 = 150, 200 and 300 kN s m−1 for two-step viscous damping force
scheme of variable damper.
the pounding of the deck with the abutments. Also, the of mid-span deck acceleration. The dampers are found to be
reduction in peak bearing displacement prevents the isolation more effective in reducing bearing displacement, as compared
damage arising due to instability at large displacements during to the mid-span acceleration. The friction-type damping
earthquakes. The significant response reduction obtained force scheme of a variable damper is not very effective in
confirms the effectiveness of the friction-type damping force reducing the mid-span acceleration. Only for the near-fault
scheme. The results of analytical investigations clearly indicate earthquakes Chi-Chi and Northridge does it give improved
that the performance of the friction-type damping force scheme results over the passive controller. This is because the rapid
is much superior, compared to the sample control strategies. fluctuation of damping force according to the control law
Thus, variable dampers can be effectively used for large increases the dynamic response of the bridge, causing an
displacement response reduction of the benchmark highway increase in the absolute acceleration. With the two-step viscous
bridge, with little or no increase in the pier base shear. damping force scheme, the mid-span acceleration is found
Further, viscous dampers provide a better response to increase marginally for the selected damping coefficient
compared to the two-step viscous damping force scheme values. Acceleration response is not greatly influenced by
and equivalent response to the semi-active and active control the variable dampers. Thus, with the two schemes of
systems. The trend of results obtained also shows the seismic variable damper, the earthquake forces transmitted to the
response reduction, particularly in reducing the excessive bridge system can be reduced at the expense of increasing mid-
displacements at the isolation level associated with an increase span acceleration.
13
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 11. Time history of base shear along the longitudinal and transverse direction with a viscous damper, friction-type damping force
scheme and two-step viscous damping force scheme of a variable damper under the Chi-Chi, 1999 earthquake motion.
Figure 12. Time history of mid-span displacement along the longitudinal and transverse directions with viscous damper, friction-type
damping force scheme and two-step viscous damping force scheme of variable damper under the Chi-Chi, 1999 earthquake motion.
Similar observations of a reduction in seismic response improved/comparable performance over the semi-active and
can be made from table 3 for evaluation criteria J5 to active controllers. A smaller ductility factor indicates reduction
J8 . The friction-type damping force scheme of a variable in the curvature of the central bent columns. Performance of
damper substantially reduces the peak bearing displacement the two-step viscous damping force scheme is comparable to
for all earthquakes. For near-fault earthquakes, namely Chi- that of the active controllers. Minimum dissipated energy of
Chi and Northridge, the bearing displacement is reduced to curvatures at the end of members and the minimum number of
half and to one-third for the Turkey earthquake. Viscous plastic hinges formed indicate that the moment at any section
dampers and the two-step viscous damping force scheme show of piers of the bridge is less than the fully plastic moment.
14
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Bearing displacement (m)
Bearing displacement (m)
Figure 13. Time history of bearing displacement along the longitudinal and transverse directions with viscous damper, friction-type damping
force scheme and two-step viscous damping force scheme of variable damper under the Chi-Chi, 1999 earthquake motion.
Comparison of evaluation criteria J9 to J12 is presented as to the viscous dampers and the two-step viscous damping
in table 4. The friction-type damping force scheme works force scheme. The maximum response reduction is 88%,
well in reducing the norm base shear. The two-step viscous compared to 42% of active, 50% of semi-active and 73% of
damping force scheme shows almost the same reduction for passive controllers.
all earthquakes. For the norm base moment, the friction- The last column of tables 2–5 presents the maximum
type damping force scheme shows improved response over the response of the control strategies. It is observed that the
semi-active and active strategies. Viscous dampers perform performance of the variable dampers is superior to that of
marginally superior to the active controllers. For norm the active controllers in terms of peak as well as normed
mid-span displacement, the performance of the friction-type response reductions, namely in peak bearing displacement
damping force scheme is much superior to all the sample J5 , peak ductility J6 , peak dissipated energy J7 , norm base
control strategies, with 50%–75% response reduction implying moment J10 , norm mid-span displacement J11 , norm bearing
that the friction-type damping force scheme is effective for the displacement J13 and maximum device stroke J16 . The
aseismic design of bridges. The two-step viscous damping peak bearing displacement is smallest using the friction-type
force scheme of variable dampers performs equally well for damping scheme of a variable damper and it is a maximum
all earthquakes. Variable dampers are not very effective in using active controllers. The force requirement J15 , in the
reducing the norm mid-span acceleration. Viscous dampers case of variable dampers, is the same as that for the active
and the two-step viscous scheme show marginal reduction in controller. Variable dampers require only a minimal power
the norm mid-span acceleration. supply and can be operated by a battery; on the other hand,
Table 5 present evaluation criteria J13 to J16 . For an active controller requires an external power supply. This
improving the norm bearing displacement, the friction-type is a significant advantage of variable dampers over active
damping force scheme performs very well. The amount of controllers. Also, their performance is better than the passive
displacement is substantially smaller to have any possible viscous dampers in reducing displacement response quantities.
impact of the deck with the abutments in the longitudinal The results of the investigation demonstrate that viscous
direction or with the bent piers in the transverse direction. dampers and variable dampers with the friction-type damping
Viscous dampers perform well, compared to the semi-active force scheme and two-step viscous damping force scheme are
and active controllers. From the trend of results it is observed capable of controlling the peak displacement response of the
that the two-step viscous damping force scheme is more bridge, isolated with LRBs, and thus reducing the length of
efficient than the active controller in reducing the norm bearing expansion joints. Furthermore, except for the North Palm
displacement. Maximum control force is the least by viscous Springs earthquake, the peak base shear is also controlled
dampers and the maximum device stroke is the smallest by effectively. In general, the evaluation criteria are smaller than
the friction-type damping force scheme. There is substantial one, indicating that the control strategies presented herein are
response reduction, compared to the sample controllers as well capable of reducing the response of the benchmark highway
15
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
(i)
(ii)
(iii)
Figure 14. Force–displacement loop in the longitudinal and transverse directions for (i) viscous damper, (ii) friction-type damping force
scheme and (iii) two-step viscous damping force scheme of variable damper.
bridge for a wide variety of earthquake records. In addition, set control force schemes for a variable damper, subjected to
the effectiveness of the control strategies for the isolated strong earthquake ground motions, is investigated. The seismic
benchmark highway bridge is significantly influenced by the response is studied under two horizontal components of six
time history of an earthquake excitation. The results have real earthquake ground motions. The seismic responses of
clearly shown that the viscous and variable dampers are the bridge with the friction-type damping force scheme and
quite effective in protecting the integrity of the benchmark two-step viscous damping force scheme are evaluated using
highway bridge subjected to near-field earthquakes. Hence, standard numerical techniques and the developed Simulink
the use of variable dampers as supplemental damping devices models. The comparison of seismic response of the bridge
largely solves the problem of superstructure displacement of controlled with viscous and variable dampers and the sample
an isolated benchmark highway bridge, along with controlling controllers is made in order to verify their effectiveness.
the seismically induced forces in the bridge. The results of the simulations show the advantage of using
seismic isolation combined with passive viscous dampers and
semi-active variable dampers. Variable dampers are able
6. Conclusions to reduce the isolator and mid-span displacements, without
greatly increasing the mid-span accelerations. From the trend
The response of a simplified model of the 91/5 highway bridge of the results of the present study the following conclusions are
in Southern California with a viscous damper and two pre- drawn:
16
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Figure 15. Force–velocity variation for viscous damper, friction-type damping force scheme and two-step viscous damping force scheme of
variable damper.
(1) There exist optimum values of parameters for the viscous References
damper, friction-type damping force scheme and two-step
viscous damping force scheme of a variable damper for Agrawal A K, Tan P, Nagarajaiah S and Zhang J 2005 Benchmark
controlling the base shear, mid-span displacement and structural control problem for a seismically excited highway
bearing displacement response. bridge, part I: problem definition http://www-ce.engr.ccny.
cuny.edu/People/Agrawal,2005
(2) The supplemental damping in the benchmark highway Ali S F and Ramaswamy A 2006 Benchmark control problem for
bridge using semi-active variable dampers helps to reduce highway bridge based on FLC Proc. ASCE Structures Congr.
the displacement response substantially. However, the 2006 (St. Louis, MI, May 2006)
acceleration response of the deck increases marginally. Choi K, Jung H, Cho S and Lee I W 2006 Application of smart
(3) There is very little or no increase in the pier base shear passive damping system using MR damper to highway bridge
benchmark problem MOVIC 2006: The 8th Int. Conf. on Motion
response of the bridge with variable dampers, compared and Vibration Control (Korea, Aug. 2006)
to that of the isolation system alone and the response by Hyung-Jo J, Spencer B F Jr and Lee M 2003 Control of seismically
the sample controllers. excited cable-stayed bridge employing magnetorheological fluid
(4) The friction-type damping force scheme of variable dampers J. Struct. Eng. 129 873–83
dampers is found to be significantly controlling the Kawashima K and Unjoh S 1994 Seismic response control of bridges
by variable dampers J. Struct. Eng., ASCE 120 2583–601
peak displacement response of the deck and abutment Kunde M C and Jangid R S 2003 Seismic behaviour of isolated
bearings, while simultaneously limiting the pier base bridges: a state-of-the-art review Electron. J. Struct. Eng. 3
shear response. Average reduction in the peak bearing 140–70
displacement is substantial, compared with the active Makris N and Zhang J 2002 Seismic response analysis of highway
controllers. overcrossings including soil-structure interaction Earthq. Eng.
Struct. Dyn. 31 1967–91
(5) The two-step viscous damping force scheme decreases the Reigles D G and Symans M D 2006 Response modification of
deck response and the hysteretic behaviour of the piers. highway bridge benchmark structure using supervisory fuzzy
However, the friction-type damping force scheme yields control of smart seismic isolation system Proc. ASCE Structures
more reduction in the response than the two-step viscous Congr. 2006 (St. Louis, MO, May 2006)
Ruangrassamee A and Kawashima K 2001 Experimental study on
damping force scheme.
semi-active control of bridges with use of magnetorheological
(6) The variable dampers with both the friction-type damping damper J. Struct. Eng., JSCE A 47 639–50
force scheme and two-step viscous damping force scheme Ruangrassamee A and Kawashima K 2002 Seismic response control
are effective in reducing seismic response, confirming that of a benchmark cable-stayed bridge by variable dampers Proc.
MR dampers can be effectively used to realize the two pre- Am. Control Conf. (Anchorage, AK) pp 3064–9
Ruangrassamee A and Kawashima K 2003 Control of nonlinear
set control force schemes.
bridge response with pounding effect by variable dampers Eng.
(7) The friction-type damping force scheme tends to provide Struct. 25 593–606
the best overall performance, minimizing both the mid- Ruangrassamee A and Kawashima K 2006 Seismic response control
span displacement and pier base shear, without causing of a cable-stayed bridge by variable dampers J. Earthq. Eng.
large displacements in the isolation devices. 10 153–65
17
Smart Mater. Struct. 18 (2009) 115011 S N Madhekar and R S Jangid
Ruangrassamee A, Nakamura G and Kawashima K 2004 A model Tan P, Agrawal A K, Nagarajaiah S and Zhang J 2005 Benchmark
test and analysis on seismic response control of bridges by MR structural control problem for a seismically excited highway
dampers Proc. 13th World Conf. on Earthquake Engineering bridge, part II: sample control designs http://www-ce.engr.ccny.
cuny.edu/People/Agrawal,2005
(Vancouver, BC) Paper No.893
Weilian Q U and Jianwei T U 2009 Theoretical and experimental
Spencer B, Dyke S J, Saint M K and Carlson J 1997 study on seismic response control on top of Three-Gorges ship
Phenomenological model of a magnetorheological damper lift towers using magnetorheological intelligent isolation system
J. Eng. Mech., ASCE 123 230–8 and its key technique Front. Archit. Civil Eng. China 3 32–41
18