Smart Base Isolated Benchmark Building

Part I: Problem Denition

Sriram Narasimhan
1
, Student Member ASCE,
Satish Nagarajaiah
2
, Member ASCE,
Erik A. Johnson
3
, Assoc. Member ASCE,
and Henri P. Gavin
4
, Member ASCE
ABSTRACT
This paper presents the benchmark problem denition for seismically excited base-
isolated buildings. The objective of this benchmark study is to provide a well dened
base isolated building with a broad set of carefully chosen parameter sets, performance
measures and guidelines to the participants, so that they can evaluate their control
algorithms. The control algorithms may be passive, active or semi-active. The bench-
mark structure considered is an eight story base isolated building similar to existing
buildings in Los Angeles, California. The base isolation system includes both linear
and nonlinear bearings and control devices. The superstructure is considered to be a
linear elastic system with lateral-torsional behavior. A new nonlinear dynamic analy-
sis program has been developed and made available to facilitate direct comparison of
results of dierent control algorithms.
Keywords: Benchmark, base isolated building, control.
1
Grad. Stud., Dept. of Civil and Env. Eng., Rice Univ., 6100 S. Main Street, Houston, TX
77005; PH 713-348-2690; nsriram@rice.edu
2
Assoc. Prof., Depts. of Civil & Env. Eng. and Mech. Eng. & Mat. Sc., Rice Univ., 6100
S. Main Street, Houston, TX 77005; PH 713-348-6207; nagaraja@rice.edu
3
Assist. Prof., Dept. of Civil and Env. Eng., Univ. of Southern California, Los Angeles,
CA 90089; PH 213-740-0610; johnsonE@usc.edu
4
Assoc. Prof., Dept. of Civil and Env. Eng., Duke Univ., Raleigh, Durham, NC 27708; PH
919-660-5200 hpgavin@duke.edu
INTRODUCTION
Recently well-dened analytical benchmark problems (Caughey 1998; Spencer
et al. 1998a; Spencer et al. 1998b; Ohtori et al. 2003; Yang et al. 2003; Dyke et al.
2003) have been developed for studying response control strategies for building
and bridge structures subjected to seismic and wind excitation, by broad consen-
sus eort of the ASCE structural control committee. The goal of this eort was
to develop benchmark models to provide systematic and standardized means by
which competing control strategies, including devices, algorithms, sensors, etc.
can be evaluated. Carefully dened analytical benchmark problems are an ex-
cellent alternative to expensive experimental benchmark test structures. Due to
eectiveness of the xed base building benchmark eort (Spencer et al. 1998a;
Spencer et al. 1998b; Ohtori et al. 2003; Yang et al. 2003) the ASCE structural
control committee voted to develop a new smart base isolated benchmark prob-
lem. Narasimhan, et al. (2002, 2003) have developed the smart base isolated
benchmark problem, based on input from the ASCE structural control commit-
tee, with capability to model three dierent kinds of base isolation systems: linear
elastomeric systems with low damping or supplemental high damping, frictional
systems, bilinear or nonlinear elastomeric systems or any combination thereof.
The superstructure is assumed to remain linear at all times. A host of control
devices can be considered at the isolation level. No control devices are allowed in
the superstructure.
Base isolation systems, such as sliding and elastomeric bearing systems, reduce
the super-structure response, but with increased base displacements in near-fault
motions. Current practice is to provide non-linear passive dampers to limit the
bearing displacements, however, this increases the forces in the superstructure
and also at the isolation level. Active and semiactive devices present attractive
alternatives to passive non-linear devices. Active and Semi-active control of lin-
2
ear and nonlinear structures using novel devices such as Magneto-Rheological
(MR) dampers, Electro-Rheological dampers and variable stiness systems has
gained signicant attention in the recent years (Spencer and Nagarajaiah 2003).
The eectiveness of structural control strategies and dierent control algorithms
has been demonstrated, by many researchers, experimentally and analytically
(Spencer and Nagarajaiah 2003).
Participants of this benchmark study can propose control strategies for bench-
mark base isolated building and dene, evaluate, and report the results for the
proposed strategy. A set of evaluation criteria have also been developed for the
sake of comparison of various control strategies.
STRUCTURAL MODEL
The benchmark structure is a base-isolated eight-story, steel-braced framed
building, 82.4-m long and 54.3-m wide, similar to existing buildings in Los Ange-
les, California. The oor plan is L-shaped as shown in Fig. 1. The superstructure
bracing is located at the building perimeter. Metal decking and a grid of steel
beams support all concrete oor slabs. The steel superstructure is supported on
a reinforced concrete base slab, which is integral with concrete beams below, and
drop panels below each column location. The isolators are connected between
these drop panels and the footings below as shown in Fig. 1. The superstruc-
ture is modeled as a three dimensional linear elastic system. The superstructure
members, such as beam, column, bracing, and oor slab are modeled in detail.
Floor slabs and the base are assumed to be rigid in plane. The superstructure
and the base are modeled using three master degrees of freedom (DOF) per oor
at the center of mass. The combined model of the superstructure (24 DOF)
and isolation system (3 DOF) consists of 27 degrees of freedom. All twenty four
modes in the xed base case are used in modeling the superstructure. The su-
3
perstructure damping ratio is assumed to be 5% in all xed base modes. The
computed natural periods for the rst nine xed base modes are shown in Table
1. The nominal isolation system consists of 61 friction pendulum bearings and 31
linear elastomeric bearings as shown in Fig. 1. While the nominal model contains
sliding and linear elastomeric bearings, participants may replace them with other
types of bearings.
54.3 m
8
2
.
4

m
19.5 m
4
6
.
2

m
Linear Elastomeric Bearing
Friction Pendulum Bearing
(Not to scale)
N
o
r
t
h
Base Slab
Actuator or Semiactive Device Isolation Bearing
Column
(a)
(b)
(c)
FIG. 1. (a) Isolation Plan, (b) FEM Model of Superstructure and (c) Ele-
vation View with Devices
ISOLATION MODEL
Several isolation elements are included so that any combination of these can
be used to model the isolation system completely. The isolation elements are elas-
4
TABLE 1. Periods
North-South East-West Torsion
Mode 1 2 3 1 2 3 1 2 3
Period 0.78 0.27 0.15 0.89 0.28 0.15 0.66 0.21 0.12
tic, viscous, hysteretic elements for bilinear elastomeric bearings and hysteretic
elements for sliding bearings. The force-displacement characteristics for friction
pendulum, lead rubber bearing and linear isolation bearings is shown in Fig. 2.
The hysteretic elements can be uni-axial or biaxial. The linear elastic and viscous
elements are for modeling linear elastomeric bearings and uid dampers. They
can also be used for modeling bilinear elastomeric isolation systems with corre-
sponding equivalent linear properties, obtained using appropriate linearization
techniques.
The biaxial hysteretic behavior of bilinear elastomeric bearings and/or fric-
tional bearings is modeled using the biaxial interaction equations of Bouc-Wen
model proposed by Park et al. (1986) as follows:
U
y
_

_
z
x
z
y
_

_
=
_

U
x

U
y
_

_
Z
w
_

U
x

U
y
_

_
Z
w
=
_

_
z
2
x
(sgn(

U
x
z
x
) + ) z
x
z
y
(sgn(

U
y
z
y
) + )
z
x
z
y
(sgn(

U
x
z
x
) + ) z
2
y
(sgn(

U
y
z
y
) + )
_

_
(1)
where z
x
and z
y
are dimensionless hysteretic variables that are bounded by values
1, , and are dimensionless quantities, U
x
, U
y
and

U
x
,

U
y
, represent
the displacements and velocities in the x and y directions, respectively, at the
isolation bearing or device and U
y
is the yield displacement. Eq. (1) accounts for
biaxial interaction of both sliding and bilinear hysteretic bearings. When yielding
commences Eq. (1) leads to z
x
= cos and z
y
= sin provided /(+) = 1 with
5
Force
Displacement
Rubber
Steel Plate
Elastomeric Bearing
Force
Displacement
Lead Core
Rubber
Steel Plate
Lead Rubber Bearing
Force
Bearing Material
Spherical Sliding
Surface
Friction Pendulum Bearing
Displacement
k
p
k
e
U
y
k
p
N
FIG. 2. Force-Displacement Characteristics of Bearings
= tan
1
(

U
x
/

U
y
) and resultant velocity

U =
_

U
2
x
+

U
2
y
. The biaxial interaction
can be neglected when the o-diagonal terms of the matrix in Eq. (1) are replaced
by zeros. This results in an uniaxial model with two independent elements in two
orthogonal directions.
The forces, f, mobilized in the elastomeric isolation bearings or devices can
6
be modeled by a elasticviscoplastic model with strain hardening
f
x
= k
p
U
x
+ c
v

U
x
+ (k
e
k
p
) U
y
z
x
(2)
f
y
= k
p
U
y
+ c
v

U
y
+ (k
e
k
p
) U
y
z
y
(3)
where k
e
= pre-yield stiness, k
p
= post-yield stiness, c
v
=viscous damping
coecient of the elastomeric bearing or device, U
y
is the yield displacement.
Eq. (1) can also be used to model sliding bearings with at or spherical
sliding surface, by means of a small yield displacement U
y
(because of rigid plastic
behavior and large pre-yield stiness) setting c
v
= 0 and (k
e
k
p
) U
y
= N
f
x
= k
p
U
x
+ Nz
x
(4)
f
y
= k
p
U
y
+ Nz
y
(5)
where is the coecient of friction and N is the average normal force at the
bearing (normal force variation is neglected). In a similar manner other devices
such as nonlinear uid dampers can also be modeled using Eq. (1).
The nominal isolation system consists of 61 friction pendulum bearings and
31 linear elastomeric bearings; however, the participants may replace them with
other types of bearings. Three types of base isolation systems are considered for
control design purposes: (1) linear elastomeric isolation system with low damping;
(2) nonlinear friction isolation system representing friction pendulum system;
and (3) bilinear elastomeric isolation system representing lead-rubber system.
The isolation systems and the corresponding sample control designs have been
described in detail in companion papers by Nagarajaiah and Narasimhan (2004a)
and Erkus and Johnson (2004).
In phase I, the participants need to compare the results of their controllers
7
with the results of the sample active and semiactive controllers presented by
Nagarajaiah and Narasimhan (2004a) for the nominal linear isolation system.
Additionally, they may also compare the results of their controllers with the
sample skyhook controller presented by Nagarajaiah and Narasimhan (2004a).
In phase II the participants need to compare the results of their controllers with
the sample Lyapunov controller and other controllers presented by Nagarajaiah
and Narasimhan (2004b) for nonlinear friction isolation system and/or the sample
controller presented by Erkus and Johnson (2004) in case they consider bilinear
elastomeric isolation system such as lead-rubber isolation system.
THREE DIMENSIONAL NONLINEAR DYNAMIC ANALYSIS
Base isolated buildings are designed such that the superstructure remains
elastic. Hence, in this study the superstructure is modeled by a condensed linear
elastic system. Also, the localized nonlinearities at the isolation level allow con-
densation of the linear superstructure. In addition, the equations of motion are
developed in such a way that the xed-base properties are used for modeling the
linear superstructure. The base and the oors are assumed to be innitely rigid
in plane. The superstructure and the base are modeled using three master de-
grees of freedom (DOF) per oor at the center of mass. Each nonlinear isolation
bearing or device is modeled explicitly using discrete biaxial Bouc-Wen model,
and the forces in the bearings or devices are transformed to the center of mass of
the base using a rigid base slab assumption. All the linear isolation bearings or
devices can be modeled individually or globally by equivalent lumped elements
at the center of mass of the base. The displacement coordinates are shown in
Fig. 3 and the asymmetric model is shown in Fig. 4. The equations of motion
for the elastic superstructure are expressed in the following form:
8
M
nn

U
n1
+C
nn

U
n1
+K
nn
U
n1
= M
nn
R
n3
(

U
g
+

U
b
)
31
(6)
in which, n is three times the number of oors (excluding base), M is the su-
Fixed Reference
m
N
m
b
m
1
m
2
u
b
ug
u
FIG. 3. Displacement Coordinates of the Base Isolated Structure
perstructure mass matrix, C is the superstructure damping matrix in the xed
base case, K is the superstructure stiness matrix in the xed base case and R is
the matrix of earthquake inuence coecients, i.e. the matrix of displacements
and rotation at the center of mass of the oors resulting from a unit translation
in the X and Y directions and unit rotation at the center of mass of the base.
Furthermore,

U,

U and U represent the oor acceleration, velocity and displace-
ment vectors relative to the base,

U
b
is the vector of base acceleration relative to
the ground and

U
g
is the vector of ground acceleration. The control devices are
located at the isolation level only as shown in Fig. 1. The equations of motion
9
for the base are as follows:
R
T
3n
M
nn
_
(

U)
n1
+R
n3
(

U
g
+

U
b
)
31
_
n1
+M
b
33
(

U
g
+

U
b
)
31
+C
b
33

U
b
31
+K
b
33
U
b
31
+f
B
31
+f
c
31
= 0 (7)
in which, M
b
is the diagonal mass matrix of the rigid base, C
b
is the resultant
damping matrix of viscous isolation elements, K
b
is the resultant stiness matrix
of elastic isolation elements, f
B
is the vector containing the nonlinear bearing
forces and f
c
is the vector containing the control forces. Eqn. (6) can be reformu-
lated in the modal domain and the xed base frequencies, damping ratios, and
modes can be used for modeling the superstructure (Nagarajaiah et al. 1991a,b).
Using X = {U
T
U
T
b

U
T

U
T
b
}
T
, the state space equations can be formulated as

X(t) = AX(t) +Bu(t) +B

F
B
(t) +E

U
g
(t) = g(X, u,

U
g
) (8)
A =
_

_
0 I
M
1
K M
1
C
_

_
, E =
_

_
0
M
1
_

_
MR
R
T
MR+M
b
_

_
,
_

_
,
B = B

=
_

_
0
M
1
_

_
0
I
_

_
_

_
,
M =
_

_
M MR
R
T
M R
T
MR+M
b
_

_
, C =
_

_
C 0
0 C
b
_

_
,
K =
_

_
K 0
0 K
b
_

_
, u =
_

_
0
f
c
_

_
, F
B
=
_

_
0
f
B
_

_
.
10
m
b
m
1
m
2
m
i
m
N
x
y
U
x
U
y
U
r
Reference Axis
Center of Resistance
Center of Mass of Base
Isolation Bearing
e
y
i y
i
e
x
i
x
y
x
i
Center of Mass
G
r
o
u
n
d

M
o
t
i
o
n
x
y
z
m
i
Control Devices
FIG. 4. Asymmetric Base Isolated Structure Excited by Bidirectional Ground
Motion
In the above equations, A, B, B

and E are condensed system matrices having 54

states derived from the full three dimensional nite element model. The Eq. (8)
is solved using unconditionally stable Newmarks constant-average acceleration
11
method, which can also be derived from trapezoidal rule given by
X
k+1
= X
k
+
t
2
(g
k
+g
k+1
) (9)
where g
k+1
= g(X
k+1
, u
k+1
,

U
g
(k+1)
). This method is implicit, needing iteration.
The nonlinear forces in the isolation bearings, devices and control forces are up-
dated by solving Eqns. (1) to (5) using the unconditionally stable semi-implicit
Runge-Kutta method (Rosenbrook 1964) suitable for solutions of sti dieren-
tial equations. Then Eqn. (8) is resolved using an iterative predictor-corrector
solution procedure (Nagarajaiah et al. 1991a; Nagarajaiah et al. 1991b) until
equilibrium of nonlinear forces is reached within specied tolerance and conver-
gence is achieved.
MATLAB IMPLEMENTATION
The analytical model is implemented using MATLAB and SIMULINK as
shown in Fig. 5. The analysis program comprises of input data les, a le to
read and assemble the required matrices for input into the nonlinear dynamic
analysis block, which is a SIMULINK based S function program. The full imple-
mentation procedure is also shown in Fig. 6. Additional inputs to the nonlinear
analysis block are the seismic excitation and the control forces provided by the
control devices. The nonlinear response is calculated using a predictor-corrector
algorithm as explained in the earlier sections. All the sensor and control devices
can be modeled in this program as SIMULINK blocks and the outputs of these
models fed into the analysis S-function block.
CONTROL DESIGN
The benchmark study participants are to dene the type, appropriate model,
and location of the sensor(s), control devices, and control algorithms (see Fig.
12








I/O Data and
Struct. Information
Form System Matrices Nonlinear Analysis Model
Control System Components
Sensors Components Devices
Integrate into Benchmark
Framework
Simulate
Submit Results and
Code
FIG. 5. Schematics of MATLAB/SIMULINK Implementation
5). The analysis program will remain invariant to the various control strategies
developed and implemented. The various control strategies can be compared to
one another by having the model and evaluation criteria common to all controllers.
The control devices, sensor devices and the control algorithms can be interfaced
to the structural evaluation model though measurement and device connection
outputs, designated y
m
and y
cd
respectively. The evaluation outputs y
e
are used
for the calculation of performance indices. The outputs y
m
and y
cd
are specied
in the input les provided with the benchmark problem statement.
Sensor Models
The sensors must take the following form

X
s
= g
1
(X
s
, y
m
, u
m
, t) (10)
y
s
= g
2
(X
s
, y
m
, u
m
, v, t) (11)
where X
s
are the states of the sensor, v is the measurement noise vector, u
m
is a
vector of control device continuous time responses and y
s
is the output of sensor
13
Benchmark Problem for Control of Base Isolated Buildings
by
Sriram Narasimhan, Satish Nagarajaiah, Erik Johnson and Henri Gavin
Earthquake
Double Click
to On or Off Devices
yf
yf
1
on
0
off
on/off
f
yf
switch
ym
yf
ys
yf_out
Sensors
ug_x
ug_y
yf
ye
ym
ycd
Nonlinear Analysis
s
i
g
_
1
y
m
y
f
y
e
o
n
/
o
f
f
J9
ye
J1 to J8
[time_record,Y_dir]
[time_record, X_dir]
ys
ys_d
u
Controller
ycd
u
f
yf
Control Devices
Clock
t_out

FIG. 6. SIMULINK Block Diagram for Simulations
in units of volts (Fig. 7). u
m
may consist of device forces, and/or stroke that
may be needed for feedback into the controller.
Control Algorithm
Control algorithms may be designed to work with active or semi-active sys-
tems. The discrete form of the control algorithms may be written as
X
c
k+1
= g
3
(X
c
k
, y
s
k
, k) (12)
u
k
= g
4
(X
c
k
, y
s
k
, k) (13)
14
Replace the contents of this block
with a model of your sensor(s)
2
yf_out
1
ys
Sum1
Sum
K
Sensor
Gains (Ds)
Memory
Measurement
Noise (v)
2
yf
1
ym
FIG. 7. Sensor Model Implementation
1
u
u(k) u(t)
D/A Converter
ys
yf
u
Controller
y(t) y(k)
y(t) y(k)
2
ys_d
1
ys
FIG. 8. Control Algorithm Implementation
where X
c
k
is the discrete state vector at time t = kt, y
s
k
is the discretized sensor
model output and u
k
is the discrete control command from the control algorithm
(Fig. 8).
Control Devices
Control devices may be designed neglecting the dynamics of the devices though
including device dynamics is strongly encouraged. Control devices may be placed
only at the isolation level and their locations are specied in the input les pro-
15
Replace the contents of this block
with a model of your devices
Users can replace the contents with their own devices
2
yf
1
f
Mux
Displacement (X),
displacement (Y),
velocity (X), velocity (Y),
rotation and rotational velocity
at center of mass of the base
MRDAMPER
MR Damper Model
2
u
1
ycd
FIG. 9. Control Devices Implementation
vided with the benchmark problem statement. The device models can be inter-
faced with the building model by including the dynamics of the device (Fig. 9)
as follows

X
cd
= g
5
(X
cd
, y
cd
, u
k
, t) (14)
u = g
6
(X
cd
, y
cd
, u
k
, t) (15)
y
f
= g
7
(X
cd
, y
cd
, u
k
, t) (16)
where the continuous states of the devices are represented by X
cd
. If the dynamics
are neglected, then the model is
u = g
8
(y
cd
, u
k
, t) (17)
y
f
= g
9
(y
cd
, u
k
, t) (18)
EVALUATION CRITERIA
The following nine evaluation criteria are dened for the benchmark problem
based on both maximum and RMS responses of the building. For each control
16
design, these criteria must be evaluated for all the seven earthquakes provided in
two orthogonal directions.
1. Peak base shear (isolation-level) in the controlled structure normalized by
the corresponding shear in the uncontrolled structure,
J
1
(q) =
max
t
V
0
(t, q)
max
t
_
_
_

V
0
(t, q)
_
_
_
2. Peak structure shear (at rst story level) in the controlled structure nor-
malized by the corresponding shear in the uncontrolled structure,
J
2
(q) =
max
t
V
1
(t, q)
max
t
_
_
_

V
1
(t, q)
_
_
_
3. Peak base displacement or isolator deformation in the controlled structure
normalized by the corresponding displacement in the uncontrolled struc-
ture,
J
3
(q) =
max
t,i
d
i
(t, q)
max
t,i
_
_
_

d
i
(t, q)
_
_
_
4. Peak inter-story drift in the controlled structure normalized by the corre-
sponding inter-story drift in the uncontrolled structure,
J
4
(q) =
max
t,f
d
f
(t, q)
max
t,f
_
_
_

d
f
(t, q)
_
_
_
5. Peak absolute oor acceleration in the controlled structure normalized by
the corresponding acceleration in the uncontrolled structure,
J
5
(q) =
max
t,f
a
f
(t, q)
max
t,f
a
f
(t, q)
17
6. Peak force generated by all control devices normalized by the peak base
shear in the controlled structure,
J
6
(q) =
max
t
_
_
_
_

k
F
k
(t, q)
_
_
_
_
max
t
V
0
(t, q)
7. RMS base displacement in the controlled structure normalized by the cor-
responding RMS base displacement in the uncontrolled structure,
J
7
(q) =
max
i

d
(t, q)
max
i

d
(t, q)
8. RMS absolute oor acceleration in the controlled structure normalized by
the corresponding RMS acceleration in the uncontrolled structure,
J
8
(q) =
max
f

a
(t, q)
max
f

a
(t, q)
9. Total energy absorbed by all control devices normalized by energy input
into the controlled structure,
J
9
(q) =

k
_
_
T
q
0
F
k
(t, q)v
k
(t, q)dt
_
_
T
q
0
_
V
0
(t, q)

U
g
(t, q)
_
dt
where, i = isolator number, 1, . . . , N
i
(N
i
= 8); k = device number, 1, . . . , N
d
;
f = oor number, 1, . . . , N
f
; q = earthquake number: 1, . . . , 5; t = time, 0
t T
q
; = inner product; = vector magnitude incorporating NS and EW
components.
EARTHQUAKES
The earthquakes used in this study are both the fault-normal (FN) and fault-
18
parallel (FP) components of Newhall, Sylmar, El Centro, Rinaldi, Kobe, Ji-ji and
Erzinkan as shown in Fig. 10. All the excitations are used at the full intensity
for the evaluation of the performance indices.
CONTROL IMPLEMENTATION AND RESTRICTIONS
1. The outputs that are available for direct measurement are the absolute ac-
celerations at the center of mass of all oor levels and the base. The abso-
lute accelerations consist of two translational and one rotational direction
at each level. In addition, the absolute accelerations and displacements
at all the device locations in two translational directions and the ground
accelerations are available for measurement. Absolute velocity measure-
ments may be added by passing the measured accelerations through a
second order lter.
2. The digitally implemented controller should have a sampling time between
0.001 and 0.01 seconds.
3. The A/D and D/A converters for the digital controller have a 16-bit pre-
cision and a span of 10 Volts.
4. Small RMS noise (at least 0.05 Volts) should be added to the measured
outputs.
5. All control devices should be placed only at the isolation level and no
provision is made for their placement in the superstructure. There is no
limitation on the number of devices that can be placed at the isolation
level.
6. The designer of the controller must justify that the proposed algorithm can
be implemented with the existing hardware and computational resources.
7. The control algorithm has to be stable and closed loop stability has to be
ensured.
19
8. The algorithm must be implemented for all seven earthquakes provided
(i.e., Newhall, Sylmar, El Centro, Rinaldi, Kobe, Ji-ji and Erzinkan) and
results presented in terms of the performance indices in each direction of
the excitation considered.
9. The isolation type can be changed as per the designers requirement. How-
ever, their locations and total number in the building are xed (a total of
92 bearings).
10. The total force generated by the control devices in each direction must be
less (if not at least of the same order) than the total non-linear or/and
linear forces generated by the isolation system.
11. If control dynamics are included, the control signal to each device should
be less than or equal to an absolute maximum of 10 Volts.
12. Participants of this study are required to submit electronically a complete
set of MATLAB les used for their control strategies.
MATLAB AND SIMULINK FILES
Further details, data les, and MATLAB and SIMULINK les can be found
at www.ruf.rice. edu/ nagaraja/baseisolationbenchmark.htm.
CONCLUDING REMARKS
A smart base isolated building benchmark problem with a broad set of care-
fully chosen parameter sets, performance measures and guidelines to the partici-
pants has been dened in this paper. A new nonlinear dynamic analysis program
has been developed and made available to facilitate direct comparison of results
of dierent control algorithms.
Participants of this benchmark study can propose dierent control strategies
for the smart base isolated building problem. Participants can dene, evaluate,
and report the results for the proposed strategy. The control algorithms may be
20
passive, active or semi-active. In phase I, the participants need to compare the
results of their controllers with the results of the sample active and semiactive
controllers presented by Nagarajaiah and Narasimhan (2004a) for the nominal
linear isolation system. Additionally, they may also compare the results of their
controllers with the sample skyhook controller presented by Nagarajaiah and
Narasimhan (2004a). In phase II the participants need to compare the results
of their controllers with the sample Lyapunov controller and other controllers
presented by Nagarajaiah and Narasimhan (2004b) for nonlinear friction isolation
system and/or the sample controller presented by Erkus and Johnson (2004)
in case they consider bilinear elastomeric isolation system such as lead-rubber
isolation system.
ACKNOWLEDGEMENTS
Partial funding provided by the National Science Foundation, CAREER Award
99-96290, and CAREER Award 00-94030, is gratefully acknowledged. The au-
thors would like to thank Professor Bill Spencer, University of Illinois at Urbana-
Champaign and Professor W. D. Iwan, CalTech, for their suggestions.
REFERENCES
Alhan, C. and Gavin, H. P. (2003). Parametric analysis of passive damping in
base isolation. Proc. 16th Eng. Mech. Conf., ASCE, University of Washington.
CDROM.
Caughey, T. K. (1998). the benchmark problem. Earthquake Engrg. & Struct.
Dyn., 27(11), 1125.
Dyke, S. J., Caicedo, J. M., Turan, G., Bergman, L. A., and Hague, S. (2003).
Phase 1: Benchmark control problem for seismic response of cable-stayed
bridges. Journal of Struct. Engrg., ASCE, 129(7), to appear.
21
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20 25 30
1
0
1
A
c
c
.

(
g
)
0 30 60 90
1
0
1
A
c
c
.

(
g
)
0 30 60 90
1
0
1
A
c
c
.

(
g
)
0 5 10 15 20
1
0
1
Time (sec)
A
c
c
.

(
g
)
0 5 10 15 20
1
0
1
Time (sec)
A
c
c
.

(
g
)
Northridge
Newhall FN
Northridge
Newhall FP
Northridge
Northridge
Sylmar FN Sylmar FP
Imperial Valley
Imperial Valley
El Centro FN El Centro FP
Northridge
Rinaldi FN
Northridge
Rinaldi FP
Kobe
Kobe NS
Kobe
Kobe EW
Jiji
Jiji068 FN
Jiji068 FP
Jiji
Erzinkan Erzinkan
Erzinkan NS Erzinkan EW
FIG. 10. Time Histories of Earthquake Records. FP - Fault Parallel, FN -
Fault Normal, EW - East West and NS - North South.
Erkus, B. and Johnson, E. A. (2004). Phase II: Smart base isolated benchmark
building part III: A sample controller for bilinear isolation. J. of Structural
Control and Health Monitoring (in review).
MATLAB (2000). The Math Works, Inc., Natick, Massachusetts.
Nagarajaiah, S. and Narasimhan, S. (2004a). Phase I: Controllers for benchmark
base isolated building: Part I. Proc. 4th Int. Workshop on Structural Control
22
and Health Monitoring; also J. of Structural Control and Health Monitoring
(in review), Columbia University.
Nagarajaiah, S. and Narasimhan, S. (2004b). Phase II: Controllers for bench-
mark base isolated building with friction isolation system: Part IV. J. of
Structural Control and Health Monitoring (in preparation).
Nagarajaiah, S., Reinhorn, A. M., and Constantinou, M. C. (1991a). Nonlinear
dynamic analysis of 3-d-base-isolated structures. J. of Str. Engrg., ASCE,
117(7), 20352054.
Nagarajaiah, S., Reinhorn, A. M., and Constantinou, M. C. (1991b). 3D-
BASIS: Nonlinear dynamic analysis of three dimensional base isolated
structures-part 2. Rep. No. NCEER-91-0005, Nat. Ctr. for Earthquake Engrg.
Res., State University of New York, Bualo.
Narasimhan, S., Nagarajaiah, S., Gavin, H., and Johnson, E. (2002). Bench-
mark problem for control of base isolated buildings. Proc. 3rd World Conf. on
structural control, Como, Italy. CDROM.
Ohtori, Y., Christenson, R. E., Spencer, B. F., and Dyke, S. J. (2003). Bench-
mark problems in seismically excited nonlinear buildings. J. Eng. Mech.,
ASCE, 129, to appear, http://cee.uiuc.edu/sstl/.
Park, Y. J., Wen, Y. K., and Ang, A. H. S. (1986). Random vibration of hys-
teritic systems under bi-directional ground motions. Earthquake Engrg. &
Struct. Dyn., 14(4), 543557.
Rosenbrook, H. H. (1964). Some general implicit processes for the numerical
solution of dierential equations. Computer J., 18, 5064.
Spencer, B. F., Dyke, S. J., and S., D. H. (1998a). Benchmark problems in
structural control: part 1 - active mass driver system. Earthquake Engrg. &
Struct. Dyn., 27(11), 11271139.
Spencer, B. F., Dyke, S. J., and S., D. H. (1998b). Benchmark problems in
23
structural control: part 2 - active tendon system. Earthquake Engrg. & Struct.
Dyn., 27(11), 11411147.
Spencer, B. F. and Nagarajaiah, S. (2003). State of the art of structural control.
J. of Struc. Eng., ASCE, 129(7), 845856.
Yang, J. N., Agrawal, A., Samali, B., and Wu, J. C. (2003). A benchmark
problem for response control of wind excited tall buildings. J. of Eng. Mech.,
ASCE, 129, in press; also see Proc. 14th Eng. Mech. Conf., ASCE, UTAustin,
CDROM, 2000, http://www.eng.uci.edu/anil/benchmark.html.
24
Phase I Smart Base Isolated Benchmark Building -
Sample Controllers for Linear Isolation System: Part II
Satish Nagarajaiah
1
, Member ASCE
and Sriram Narasimhan
2
, Student Member ASCE
ABSTRACT
Sample controllers for three dimensional base isolated building benchmark problem
with linear and frictional isolation system are presented in this paper. Kalman lter is
used to estimate the states based on absolute acceleration measurements. Input lters
are used to better inform the controller of the spectral content of the earthquake exci-
tations. Reduced order control oriented model of the benchmark problem with linear
isolation system is developed. H
2
/Linear Quadratic Gaussian controller is presented for
the active case; additionally, a clipped optimal controller is presented for the semiactive
case. Magneto-Rheological uid dampers are used for control in the semiactive case
and actuators are used for control in the active case. The focus of the Phase I study is
on linear isolation system only. However, a preliminary skyhook semiactive controller
is also presented for the benchmark problem with nonlinear friction isolation system.
A more detailed set of sample controllers for nonlinear friction isolation systems will be
presented in Phase II. Computed results of the passive, semiactive, and active cases are
presented. Detailed comparisons of benchmark performance indices for base isolated
structures with linear isolation system and nonlinear friction isolation system, with and
without control, for a set of strong near eld earthquakes are presented. The modeling
and sample control designs demonstrated in this paper can be used to form the basis for
studying wide variety of active and semiactive control strategiesto be developed by the
1
Assoc. Prof., Dept. of Civil and Env. Eng., Rice Univ.,6100 S. Main Street, Houston, TX
77005; PH 713-348-6207; nagaraja@rice.edu
2
Grad. Stud., Dept. of Civil and Env. Eng., Rice Univ., 6100 S. Main Street, Houston, TX
77005; PH 713-348-2690; nsriram@rice.edu
participants in the benchmark studyfor linear and frictional base isolated buildings.
Keywords: Benchmark problem, smart base isolated building, active, semiactive,
H
2
/LQG, skyhook, seismic response control.
INTRODUCTION
Several researchers have studied active and semiactive control of base isolated
structures (Reinhorn et al. 1987, Nagarajaiah et al. 1993, Nagarajaiah 1994,
Reinhorn et al. 1994, Yoshida et al. 1994, Yang et al. 1996, Symans et al.
1999, Yoshida et al. 1999, Nagarajaiah et al. 2000b, Spencer et al. 2000, Sa-
hasrabudhe et al. 2000, Ramallo et al. 2002, Madden et al. 2003). However,
the relative merits of these active and semiactive controllers, as applied to base
isolated structures, has not been investigated by careful comparison on a well
dened benchmark problem. Recently well-dened analytical benchmark prob-
lems (Caughey 1998; Spencer et al. 1998a; Spencer et al. 1998b; Ohtori et al.
2003; Yang et al. 2003; Dyke et al. 2003) have been developed for studying re-
sponse control strategies for building and bridge structures subjected to seismic
and wind excitation, by broad consensus eort of the ASCE structural control
committee. The goal of this eort was to develop benchmark models to pro-
vide systematic and standardized means by which competing control strategies,
including devices, algorithms, sensors, etc. can be evaluated. Carefully dened
analytical benchmark problems are an excellent alternative to expensive experi-
mental benchmark test structures. Due to eectiveness of the xed base building
benchmark eort (Spencer et al. 1998a; Spencer et al. 1998b; Ohtori et al. 2003;
Yang et al. 2003) the ASCE structural control committee voted to develop a
new smart base isolated benchmark problem. Narasimhan, et al. (2002, 2003)
have developed the smart base isolated benchmark problem, based on input from
the ASCE structural control committee, with capability to model three dierent
2
kinds of base isolation systems: linear elastomeric systems with low damping or
supplemental high damping, frictional systems, bilinear or nonlinear elastomeric
systems or any combination thereof. The superstructure is assumed to remain
linear at all times. A host of control devices can be considered at the isolation
level. No control devices are allowed in the superstructure.
This paper presents sample control strategies for benchmark problem with
two types of nominal base isolation systems: (1) linear elastomeric system with
low damping; and (2) nonlinear friction isolation systems. Actuators are used
for active control or Magneto-Rheological (MR) dampers are used for semiactive
control. To illustrate some of the design challenges active and semiactive con-
trol algorithms are presented for the case with linear elastomeric isolation system.
Also, semiactive control algorithms with MR dampers is presented for the nonlin-
ear friction isolation system. The presented active and semiactive sample control
strategies are not meant to be competitive, but are intended to serve as guide
to the participants of the benchmark study. In phase I, the participants need
to compare the results of their controllers with the results of the sample active
and semiactive controllers presented in this paper (Nagarajaiah and Narasimhan
(2004a)) for the nominal linear isolation system. Additionally, they may also
compare the results of their controllers with the sample skyhook controller pre-
sented (Nagarajaiah and Narasimhan (2004a)). In phase II the participants need
to compare the results of their controllers with the sample Lyapunov controller
and other controllers presented by Nagarajaiah and Narasimhan (2004b) for non-
linear friction isolation system and/or the sample controller presented by Erkus
and Johnson (2004) in case they consider bilinear elastomeric isolation system
such as lead-rubber isolation system.
H
2
/LQG sample controller is presented for illustrating the implementation of
active control system in the linear isolation case. In order to illustrate the imple-
3
mentation of semiactive control systems in the linear isolation case, clipped opti-
mal control (Dyke et al. 1996) based on H
2
/LQG method is presented. Skyhook
controller (Karnopp et al. 1974) is presented for illustrating the implementation
of semiactive control in the non-linear friction isolation case.
For the case of H
2
/LQG-active and clipped optimal control design (Yoshioka
et al., 2002), sensors placed on the eighth oor and the isolation level measure
the acceleration responses of two translational (EW and NS) and one rotational
direction. In the case of skyhook control, sensors (accelerometers) are placed at
each of the control device locations, a total of eight in EW and NS directions,
respectively and a sensor to measure ground acceleration.
The H
2
/LQG design is based on a reduced order model (Davison 1966) that
contains 24 states compared to 54 states in a full order model. The frequency
characteristic of the ground excitation is incorporated into the control design
through a shaping lter. The resulting H
2
/LQG controller is designed for the
augmented system consisting of 28 states. The skyhook control for the nonlinear
friction case takes into account both the relative and absolute velocities at the
measured locations. These velocities are computed from acceleration measure-
ments through the use of a higher order lter that approximates an integrator.
NOMINAL LINEAR ELASTOMERIC AND NONLINEAR FRICTION
SYSTEM
Linear elastomeric isolation system consists of 92 low damping elastomeric
bearings. The fundamental period T
b
=3 sec in the linear elastomeric isolation
case. The damping in the linear elastomeric isolation system is considered to
be 3% of critical. Nonlinear isolation system consists of 61 friction pendulum
bearings and 31 linear elastomeric bearingsfor a total of 92 bearings. The fun-
damental period, T
b
=3 sec in the nonlinear friction isolation case. Coecient of
4
friction = 0.06 is considered for the friction bearings. Damping of 3% of critical
is considered for the linear elastomeric bearings. In all cases total of 16 active or
semiactive control devices, 8 in the X and 8 in the Y direction, are placed at the
isolation level.
ACTIVE CONTROL - LINEAR ELASTOMERIC ISOLATION SYSTEM
The state equation (?) is as follows

X(t) = AX(t) +Bu(t) +E

U
g
(t) = g(X, u,

U
g
) (1)
where X = {U
T
U
T
b

U
T

U
T
b
}
T
A =
_

_
0 I
M
1
K M
1
C
_

_
, B = B

=
_

_
0
M
1
_

_
0
I
_

_
_

_
, (2)
E =
_

_
0
M
1
_

_
MR
R
T
MR+M
b
_

_
_

_
.
M =
_

_
M
nn
M
nn
R
n3
R
T
3n
M
nn
R
T
3n
M
nn
R
n3
+M
33
_

_
, C =
_

_
C
nn
0
0 C
b
33
_

_
,
K =
_

_
K
nn
0
0 K
b
33
_

_
, u =
_

_
0
f
c
31
_

_
. (3)
In the above equations, A, B, B

and E are condensed system matrices having

54 states derived from the full three dimensional nite element model. M is the
superstructure mass matrix, C is the superstructure damping matrix in the xed
base case, K is the superstructure stiness matrix in the xed base case, M
b
is
5
the mass matrix of the rigid base, C
b
is the resultant damping matrix of viscous
isolation elements, K
b
is the resultant stiness matrix of elastic isolation elements
and f
c
is the vector containing control forces. R is the matrix of earthquake
inuence coecients, i.e. the matrix of displacements and rotation at the center
of mass of the oors resulting from a unit translation in the X and Y directions
and unit rotation at the center of mass of the base. Furthermore,

U,

U and
U represent the oor acceleration, velocity and displacement vectors relative to
the base,

U
b
is the vector of base acceleration relative to the ground and

U
g
is
the vector of ground acceleration. Note B

F
B
in equation (8) in Part 1 is zero
since the forces from the isolation system are accounted for by the linear isolation
global stiness and damping terms in the A matrix.
For the controller implementation, this model is further reduced to 24 states
using model reduction techniques (Davison 1966) as compared to 54 states in a
full order model. The retained states correspond to the displacement and velocity
on the eighth oor, fth oor, rst oor and base. The reduction is accomplished
by constructing a matrix of lower order that has the same dominant eigenvalues
and eigenvectors as the original system (Davison 1966). The state space equations
can be formulated as
x
r
= A
r
x
r
+B
r
u +E
r

U
g
(4)
z
r
= C
zr
x
r
+D
zr
u +F
zr

U
g
(5)
y
mr
= C
mr
x
r
+D
mr
u +F
mr

U
g
+v
r
(6)
where A
r
, B
r
and E
r
are the system matrices, z
r
is the regulated output vec-
tor which is obtained by choosing the appropriate mapping matrices, C
zr
, D
zr
and F
zr
. y
mr
is the measurement vector obtained by choosing matrices C
mr
,
D
mr
and F
mr
appropriately. v
r
is the measurement noise vector. The mea-
6
sured outputs are the responses of the eighth oor, base and ground denoted by
y
mr
= [ x
8ax
x
8ay
x
8a
x
bax
x
bay
x
ba
u
gx
u
gy
]
T
and the outputs to be regu-
lated include the inter-story drifts and base displacements at the farthest corner
of the building and absolute accelerations for all degrees of freedom given by
z
r
= [x
b
x
i
x
i1
x
b
x
i
]
T
; where, i denotes the oor under consideration.
To better inform the controller of the frequency characteristics of earthquakes,
the input excitation is modeled as a ltered white noise. The shaping lter is given
by
x
f
= A
f
x
f
+B
f
w

U
g
= C
f
x
f
(7)
where w is the white noise excitation and x
f
are the states of the shaping lter.
Combining Eq. (7) with Eq. (4), the augmented system becomes
_

_
x
r
x
f
_

_
=
_

_
A
r
E
r
C
f
0 A
f
_

_
_

_
x
r
x
f
_

_
+
_

_
B
r
0
_

_
u +
_

_
0
B
f
_

_
w (8)
from which the augmented equations can be written as
x
a
= A
a
x
a
+B
a
u +E
a
w (9)
y
a
= C
ya
x
a
+D
ya
u +F
ya

U
g
+v
a
(10)
z
a
= C
za
x
a
+D
za
u +F
za

U
g
(11)
where the matrices A
a
, B
a
and E
a
are augmented system matrices. C
ya
, D
ya
,
F
ya
, C
za
, D
za
and F
za
are mapping matrices of appropriate dimensions. v
a
is the
measurement noise vector. As shown in Fig. 1, the new shaping lter (eq. 12)
models the ground excitation characteristics (other models such as Kanai-Tajimi
lter or pulse lter developed by Agrawal and Zhou (2004) may also be used to
7
model ground excitation) closely.
F(s) =
4
g

g
s
s
2
+ 2
g

g
s +
2
g
, where,
g
= 2 rad/sec,
g
= 0.3. (12)
10
1
10
0
10
1
10
3
10
2
10
1
10
0
10
1
10
2
NewhallFN
SylmarFN
KobeNS
RinaldiFN
Filter
P
o
w
e
r

S
p
e
c
t
r
a
l

D
e
n
s
i
t
y

(
(
m
/
s
2
)
2
/
H
z
)
Frequency (Hz)
FIG. 1. Excitation lter and power spectral density of earthquakes
The measured responses contain identically distributed RMS noise of 0.14
Volts and they are modelled as Gaussian rectangular pulse processes with a pulse
width of 0.005 seconds. The sensor gains are given by (10/9.81)[I]V/(m/sec
2
),
where I is of order 8. Assuming the independence of ground excitation and
measurement noises, cost function that weights the regulated outputs and the
control forces is given as
J = lim

E
__

0
_
z
T
a
Qz
a
+u
T
Ru
_
dt
_
(13)
8
where R is an identity matrix that weights the control forces and Q is a diagonal
matrix of the form 10
3
I
5454
that weights the regulated outputs. The separation
principle allows the control and estimation problems to be treated independently
(for linear systems only). The control law takes the form
u = K
a
x
a
(14)
where, K
a
is the full state feedback gain matrix and x
a
is the Kalman lter
estimate of the state vector based on the augmented model. The block diagram
for the augmented controller is shown in Fig. 2.
Filter
Plant Model

Kalman Filter
Observer
-K
a
Excitation
x
a
z
a
v
y
a
+
+
u
H
2
/LQG Controller
w
Augmented Plant Model
FIG. 2. Design of H
2
/LQG Controller
Calculations of K
a
and the Kalman estimator gains are performed using MAT-
LAB control toolbox functions lqry and lqe2. Calculations to determine the dis-
crete time compensator are performed using the function c2dm in MATLAB.
9
SEMIACTIVE CONTROL - LINEAR ELASTOMERIC ISOLATION
SYSTEM
In order to illustrate the application of semi active control system using MR
dampers (Spencer et al. 1997; Carlson and Charzan 1994), clipped optimal con-
trol strategy (Dyke et al. 1996) based on H
2
/LQG method is presented. The
clipped-optimal control approach involves the design of a controller for an active
system with the desired optimal control force being generated by an MR damper
according to the voltage control law,
u = V
max
H ({f
c
f} f) (15)
where V
max
is the maximum voltage, H is the heaviside function, f
c
is the optimal
force required as per Eq. (14) and f
MR
is the force that is generated by the
MR damper. For better performance, the control signal generated by the digital
controller is passed through a low pass lter before it is commanded to the device.
The lter is given by,
= + u (16)
where, the lter parameter, = 10 rad/sec. The force generated by the damper is
a function of the voltage supplied. The damper is modeled using a spring, a dash
pot and hysteretic element in parallel as shown in Fig. 3. The force generated
by the damper is given by
f
MR
= (z)f() + C

U
b
+ kU
b
(17)
where, =
a
+
b
, C = C
a
+ C
b
, f() is a function of voltage , supplied to
the MR damper. The hysteresis variable z is obtained by solving the dierential
10
FIG. 3. MR Damper Model
0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6
3000
2000
1000
0
1000
2000
3000
Displacement (m)
F
o
r
c
e

(
k
N
)
Passive Off
Passive On
FIG. 4. Force Displacement Rela-
tionship of MR Damper
equation:
Y
i
z
i
+

U
bi

z
i
|z
i
| +

U
bi
z
2
i


U
bi
= 0 (18)
where U
b
is the displacement experienced by the MR damper. The parameter
Y
i
is the yield displacement of the hysteretic element, , ,
a
,
b
, C
a
and C
b
are constants. The force displacement characteristics of one of the eight MR
Dampers subjected to harmonic excitation is shown in Fig. 4 for both passive-on
and passive-o cases.
SEMIACTIVE SKYHOOK CONTROL - NONLINEAR FRICTION
ISOLATION SYSTEM
Skyhook control (Karnopp et al. 1974) is used to illustrate the semiactive
control strategy for the nonlinear friction isolation case. The skyhook control
algorithm is given by
C(t) =
_

_
C
max
u
a
u
b
> 0
0 or C
min
u
a
u
b
< 0
(19)
Here, C
min
is the minimum damping coecient, C
max
is the maximum damping
11
coecient of the damper, u
a
is the absolute velocity and u
b
is the relative velocity.
The velocities are computed from acceleration measurements at the eight device
locations through the use of a higher order lter which approximates an integra-
tor. The measured outputs at the eight MR damper locations are represented as
y
mf
= [ x
dev1
x
dev2
x
dev3
x
dev4
x
dev5
x
dev6
x
dev7
x
dev8
]
T
. Notice that the mea-
sured outputs in the linear elastomeric isolation system and the friction isolation
system are not the same. Participants may choose any realistic set of measure-
ments that is possible with the currently available technology for designing their
control strategy.
EVALUATION OF SAMPLE CONTROL DESIGNS
The results of the evaluations for three dierent control designs are presented
in Tables 1 to 11. The results presented in Tables 1 to 6 are for the fault normal
(FN) component and the fault parallel (FP) components acting in two perpen-
dicular directions; the evaluation is reported in terms of the performance indices
described in the denition paper (Narasimhan et al. 2003). The uncontrolled
response quantities are presented in Tables 8 to 11 for the fault normal (FN)
component and the fault parallel (FP) component acting in two perpendicular
directions. Time history responses in the NS direction for Newhall earthquake
FN and FP components acting on the benchmark building are shown in Fig. 5.
The force displacement loops for the MR damper and the isolation bearings (lin-
ear and frictional) for both clipped optimal and skyhook control is shown in Fig.
6. The maximum corner drifts normalized by their corresponding uncontrolled
values are shown in Table 7 for the three dierent control designs.
The results of active control of the benchmark problem with linear elastomeric
isolation system are summarized in Tables 1 and 2. Actuators are used apply the
active control forces to the base of the structure. In this control strategy most
12
of the response quantities are reduced substantially from the uncontrolled cases.
The benet of active control strategy is the reduction of base displacements and
shears of upto 25% without increase in drift or accelerations.
The results of clipped optimal control strategy for the benchmark problem
with linear elastomeric isolation system are presented in Tables 3 and 4. The semi-
active force is applied to the base of the structure by sixteen MR Dampers, eight
in the X and eight in the Y direction. Fourteen of the MR Dampers are located
in the periphery of the base slab and two near the center of mass of the base
slab. Performance indices are presented in Tables 3 and 4 for both passive and
semiactive control cases. The advantage of semi active clipped optimal control is
evident in signicant reductions in base displacements as compared to the active
control case. The reduction in base displacements is about 25 to 50% compared
to 10 to 25% for active case. The reductions are achieved at the cost of increased
oor accelerations and inter-story drifts; however, the increases in the semiactive
control case are less than that of the passive damping case. This increase is
observed mostly at higher oors. The performance of controlled case is better
than the passive case in terms of achieving the good reduction in base drifts with a
correspondingly lower increase in oor accelerations and story drifts. It is worth
noting that the sample controllers presented are not meant to be competitive
and hence, further reductions are possible even for base displacements in the
semiactive control case with better control algorithms.
The results of skyhook control algorithm for the benchmark problem with
non-linear friction isolation system are presented in Tables 5 and 6. Performance
indices are presented in Tables 5 and 6 for both passive and skyhook control cases.
The Skyhook controller performs better than the passive case in all earthquakes.
The base and structural shear remains at the level of uncontrolled structure for
all earthquakes except El Centro and Kobe for the controlled case. There is an in-
13
crease in the base and structure shear for El Centro and Kobe earthquakes. The
reduction in maximum base displacements is between 15 to 50%. The results
of the passive case are better than the controlled case for peak base displace-
ments in most cases; however, the inter-story drifts in the controlled case are
signicantly better than the passive case. For both the passive and controlled
cases, the inter-story drifts are higher than the uncontrolled case. The peak ac-
celerations increased for both controlled and passive cases with the magnitude of
increase much higher for the passive case in all excitations. It is worth noting
that further reductions are possible in the semiactive control case with better
control algorithms.
CONCLUDING REMARKS
In phase I, the participants need to compare the results of their controllers with
the results of the sample active and semiactive controllers presented in this paper
for the nominal linear isolation system. Additionally, they also need to compare
the results of their controllers with the sample controller presented in this paper
for the nonlinear friction isolation system in case they consider nonlinear friction
isolation systems. In phase II the participants need to compare the results of their
controllers with the sample Lyapunov controller and other controllers presented
by Nagarajaiah and Narasimhan (2004b) for nonlinear friction isolation system
and/or the sample controller presented by Erkus and Johnson (2004) in case
they consider bilinear elastomeric isolation system such as lead-rubber isolation
system. The control algorithms presented in this paper are for illustration and
design purposes only and are not meant to be competitive. Participants may use
more competitive control design for their own control strategies.
The models and data for the base isolated benchmark problem are available
in a set of MATLAB m-les and C executables. They can be accessed from
14
0 2 4 6 8 10 12 14 16 18 20
0.5
0
0.5
D
i
s
p
.

(
m
)
0 2 4 6 8 10 12 14 16 18 20
5
0
5
A
c
c
.

(
m
/
s
2
)
0 2 4 6 8 10 12 14 16 18 20
0.5
0
0.5
0 2 4 6 8 10 12 14 16 18 20
10
0
10
0 2 4 6 8 10 12 14 16 18 20
0.5
0
0.5
0 2 4 6 8 10 12 14 16 18 20
10
0
10
Time (sec)
D
i
s
p
.

(
m
)
D
i
s
p
.

(
m
)
A
c
c
.

(
m
/
s
2
)
A
c
c
.

(
m
/
s
2
)
(a1)
(a2)
(b1)
(b2)
(c1)
(c2)
Active Control
Active Control
Clipped Optimal
Clipped Optimal
Skyhook
Skyhook
FIG. 5. Time history responses, both controlled (red,thick) and uncontrolled
(blue,thin), at the center of mass of the base in the NS direction for Newhall
earthquake FN-X and FP-Y components acting on the benchmark building:
(a-1,a-2) Base displacement and top oor acceleration responses for active
control, linear elastomeric system; (b-1,b-2) Base displacement and top
oor acceleration responses for clipped optimal control, linear elastomeric
system; (c-1,c-2) Base displacement and top oor acceleration responses
for skyhook control, friction isolation system.
http://www.ruf. rice.edu/ nagaraja/baseisolationbenchmark.htm.
ACKNOWLEDGEMENTS
Partial funding provided by the National Science Foundation, NSF-CAREER
15
0.4 0.2 0 0.2 0.4
1500
1000
500
0
500
1000
1500
Isolator Disp. (m)
M
R
D

F
o
r
c
e

(
k
N
)
0.4 0.2 0 0.2 0.4
500
0
500
I
s
o
l
a
t
o
r

F
o
r
c
e

(
k
N
)
0.2 0.1 0 0.1 0.2
1500
1000
500
0
500
1000
1500
0.2 0.1 0 0.1 0.2
500
0
500
Isolator Disp. (m)
M
R
D

F
o
r
c
e

(
k
N
)
Isolator Disp. (m) Isolator Disp. (m)
(a) (b)
(c) (d)
I
s
o
l
a
t
o
r

F
o
r
c
e

(
k
N
)
FIG. 6. (a) Clipped-optimal - MR Damper Force-displacement for Newhall
- Y direction, (b) Clipped-optimal - Linear Bearing Force-displacement for
Newhall - Y Direction, (c) Skyhook - MR Damper Force-displacement for
Kobe - Y Direction and (d) Skyhook - Frictional Force-displacement for
Kobe - Y Direction. Both passive on (blue, dashed) and controlled (red,
solid) are shown; earthquake FP - X and FN - Y and the bearing and device
location corresponding to 45.
Grant 9996290 is gratefully acknowledged. The authors would like to thank
Professor Bill Spencer, University of Illinois at Urbana-Champaign for his sug-
gestions.
REFERENCES
Agrawal, A. and Zhou, X. (2004). A novel active controller design based on
H
2
/LQG algorithm with pulse lters. Proc. 4th Int. Workshop on Structural
Control and Health Monitoring, Columbia University.
16
TABLE 1. Results for Active Control (FP - X and FN - Y)
J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall 0.86 0.88 0.80 0.84 0.87 0.14 0.68 0.79 0.44
Slymar 0.93 0.94 0.93 0.95 0.95 0.13 0.72 0.84 0.45
El Centro 0.97 0.95 0.83 0.81 0.82 0.10 0.78 0.70 0.37
Rinaldi 0.97 0.96 0.95 0.94 0.97 0.13 0.73 0.74 0.46
Kobe 0.84 0.85 0.79 0.82 0.90 0.10 0.75 0.69 0.41
Jiji 0.90 0.89 0.85 0.89 0.90 0.09 0.81 0.87 0.34
Erzinkan 0.99 1.02 0.79 0.85 0.99 0.11 0.80 0.78 0.45
TABLE 2. Results for Active Control (FN - X and FP -Y)
J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall 0.80 0.86 0.75 0.85 0.84 0.15 0.72 0.72 0.43
Sylmar 0.96 0.94 0.91 0.92 0.92 0.13 0.71 0.73 0.54
El Centro 0.99 0.98 0.93 0.97 1.02 0.09 0.80 0.77 0.37
Rinaldi 0.90 0.94 0.74 0.95 0.94 0.10 0.67 0.63 0.48
Kobe 0.98 0.98 0.83 0.96 0.94 0.10 0.87 0.79 0.27
Jiji 0.86 0.85 0.79 0.84 0.84 0.12 0.78 0.75 0.41
Erzinkan 0.90 0.90 0.74 0.91 0.92 0.12 0.73 0.71 0.52
Carlson, J. D. and Charzan, M. J. (1994). Magnetorheological uid dampers. U.S.
Patent No. 5,277,281.
Caughey, T. K. (1998). The benchmark problem. Earthquake Engrg. & Struct.
Dyn., 27(11), 1125.
Davison, E. J. (1966). A method for simplifying linear dynamic systems. IEEE
Transactions on Automatic Control, Vol. AC-11(1), 93101.
Dyke, S. J., Caicedo, J. M., Turan, G., Bergman, L. A., and Hague, S. (2003).
Phase 1: Benchmark control problem for seismic response of cable-stayed
bridges. Journal of Struct. Engrg., ASCE, 129(7), to appear.
Dyke, S. J., Spencer Jr., B. F., Sain, M. K., and Carlson, J. D. (1996). Seismic re-
sponse reduction using magnetorheological dampers. Proceedings of the IFAC
17
TABLE 3. Results for Clipped Optimal Control (FP - X and FN - Y)
Earthquake Case J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall Passive 0.91 0.95 0.51 1.30 2.49 0.34 0.25 1.07 0.89
Control 0.97 1.02 0.56 1.04 1.49 0.30 0.33 0.89 0.79
Sylmar Passive 0.90 0.93 0.66 0.81 1.48 0.25 0.40 0.82 0.86
Control 0.90 0.91 0.73 0.87 1.16 0.24 0.45 0.74 0.81
El Centro Passive 0.73 0.87 0.14 1.22 2.86 0.67 0.09 1.61 0.82
Control 1.25 1.24 0.54 1.26 1.61 0.38 0.42 0.76 0.65
Rinaldi Passive 0.95 0.96 0.50 0.97 1.12 0.29 0.27 0.83 0.86
Control 1.04 1.02 0.60 0.96 1.01 0.27 0.38 0.71 0.77
Kobe Passive 0.84 0.84 0.36 1.19 2.34 0.39 0.16 1.14 0.87
Control 1.04 1.03 0.52 1.00 1.63 0.28 0.26 0.73 0.73
Jiji Passive 0.83 0.82 0.65 0.86 0.92 0.17 0.42 0.82 0.70
Control 0.84 0.84 0.65 0.86 0.87 0.17 0.46 0.72 0.64
Erzinkan Passive 0.94 0.95 0.49 0.85 1.21 0.26 0.32 0.60 0.87
Control 0.93 0.93 0.47 0.86 1.23 0.25 0.34 0.63 0.80
TABLE 4. Results for Clipped Optimal Control (FP - Y and FN - X)
Earthquake Case J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall Passive 0.83 0.93 0.51 1.32 1.85 0.33 0.34 1.05 0.89
Control 0.88 0.92 0.55 1.24 1.40 0.30 0.42 0.84 0.80
Sylmar Passive 0.79 0.78 0.68 0.80 1.25 0.25 0.46 0.67 0.85
Control 0.80 0.79 0.74 0.79 0.92 0.23 0.51 0.61 0.81
El Centro Passive 0.73 0.93 0.19 2.18 3.45 0.69 0.12 2.00 0.81
Control 1.25 1.24 0.65 1.37 2.08 0.37 0.42 0.92 0.69
Rinaldi Passive 0.88 0.93 0.53 0.93 1.12 0.28 0.24 0.58 0.87
Control 0.98 1.01 0.62 0.99 1.02 0.26 0.30 0.47 0.78
Kobe Passive 0.96 1.00 0.40 1.30 2.24 0.41 0.20 1.44 0.87
Control 1.15 1.20 0.52 1.33 1.47 0.30 0.38 0.98 0.72
Jiji Passive 0.74 0.74 0.63 0.75 0.77 0.17 0.40 0.74 0.70
Control 0.74 0.73 0.63 0.73 0.80 0.17 0.46 0.61 0.64
Erzinkan Passive 0.85 0.85 0.51 0.95 1.13 0.25 0.29 0.48 0.87
Control 0.84 0.83 0.50 0.89 1.14 0.24 0.32 0.52 0.79
18
TABLE 5. Results for Skyhook Control (FP - X and FN - Y)
Index J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall Passive 1.04 1.06 0.67 1.37 1.22 0.32 0.60 1.46 0.52
Control 0.86 0.87 0.75 1.18 1.13 0.34 0.66 1.35 0.42
Sylmar Passive 0.99 1.02 0.74 1.25 1.81 0.25 0.57 1.41 0.51
Control 0.92 0.99 0.76 1.05 1.59 0.25 0.61 1.29 0.44
El Centro Passive 1.43 1.38 0.45 1.68 1.34 0.39 1.19 1.46 0.56
Control 1.22 1.19 0.56 1.22 1.22 0.45 1.05 1.35 0.44
Rinaldi Passive 1.05 1.04 0.77 1.20 2.13 0.28 0.82 1.71 0.52
Control 0.93 0.95 0.85 1.08 1.67 0.29 0.76 1.53 0.44
Kobe Passive 1.12 1.43 0.62 1.69 1.78 0.35 0.70 1.54 0.50
Control 1.04 1.36 0.57 1.41 1.58 0.36 0.65 1.38 0.43
Jiji Passive 0.81 0.82 0.67 0.98 1.38 0.18 0.58 1.43 0.46
Control 0.86 0.86 0.72 0.96 1.27 0.17 0.69 1.39 0.39
Erzinkan Passive 1.04 1.07 0.74 1.16 1.66 0.25 0.68 1.28 0.52
Control 0.91 0.92 0.71 1.01 1.38 0.28 0.63 1.16 0.46
TABLE 6. Results for Skyhook Control (FP - Y and FN - X)
Index J1 J2 J3 J4 J5 J6 J7 J8 J9
Newhall Passive 1.02 1.06 0.70 1.36 1.64 0.30 0.61 1.44 0.51
Control 0.89 0.91 0.79 1.11 1.27 0.33 0.72 1.26 0.42
Sylmar Passive 1.02 1.07 0.80 1.09 1.96 0.23 0.59 1.28 0.51
Control 0.94 1.00 0.77 0.92 1.56 0.25 0.58 1.17 0.45
El Centro Passive 1.46 1.51 0.51 1.51 1.45 0.41 1.11 1.40 0.56
Control 1.43 1.21 0.77 1.15 1.21 0.38 0.97 1.27 0.43
Rinaldi Passive 1.04 1.01 0.75 1.07 1.44 0.27 0.88 1.63 0.52
Control 0.91 0.90 0.85 1.00 1.36 0.29 0.80 1.66 0.44
Kobe Passive 1.19 1.29 0.57 1.70 1.74 0.33 0.75 1.53 0.50
Control 1.08 1.14 0.51 1.25 1.49 0.33 0.76 1.37 0.42
Jiji Passive 0.79 0.81 0.63 0.85 1.07 0.18 0.54 1.28 0.46
Control 0.86 0.86 0.70 0.89 0.98 0.17 0.67 1.13 0.39
Erzinkan Passive 1.10 1.02 0.64 1.05 1.46 0.24 0.58 1.16 0.53
Control 0.96 0.95 0.61 1.00 1.58 0.26 0.53 1.04 0.47
19
TABLE 7. Results for Corner Drifts (Normalized by Uncontrolled Values)
Index Active Control Clipped Optimal Skyhook Control
Newhall FP-X 0.85 1.28 1.10
FN-X 0.81 1.11 1.03
Sylmar FP-X 1.09 0.96 0.94
FN-X 0.97 0.90 0.91
El Centro FP-X 0.82 0.93 1.08
FN-X 0.96 1.30 1.04
Rinaldi FP-X 1.01 0.89 1.18
FN-X 1.00 0.93 1.18
Kobe FP-X 0.89 1.07 1.31
FN-X 1.00 1.10 1.21
Jiji FP-X 0.83 0.78 1.00
FN-X 0.91 0.92 0.93
Erzinkan FP-X 0.90 0.73 1.03
FN-X 0.89 0.94 1.07
TABLE 8. Linear Isolation System - Uncontrolled Response Quantities (FP
- X and FN - Y)
Newhall Sylmar El Centro Rinaldi Kobe Jiji Erzinkan
Peak Base Shear
(Norm. by W

)
0.180 0.272 0.098 0.238 0.171 0.488 0.250
Peak Str. Shear
(Norm. by W

)
0.150 0.228 0.084 0.207 0.143 0.408 0.204
Peak Isolator
Deformation (m)
0.583 0.731 0.486 0.855 0.637 1.490 1.051
Peak I.S Drift
(Norm. by h

)
0.002 0.003 0.001 0.003 0.002 0.004 0.003
Peak Absolute
Acc. (g

)
0.231 0.318 0.135 0.329 0.191 0.509 0.267
RMS Disp.
(m)
0.281 0.297 0.201 0.333 0.264 0.344 0.471
RMS Acc.
(g

)
0.074 0.098 0.055 0.098 0.082 0.100 0.131
20
TABLE 9. Linear Isolation System - Uncontrolled Response Quantities (FP
- Y and FN - X)
Newhall Sylmar El Centro Rinaldi Kobe Jiji Erzinkan
Peak Base Shear
(Norm. by W

)
0.200 0.314 0.096 0.258 0.142 0.557 0.287
Peak Str. Shear
(Norm. by W

)
0.164 0.267 0.081 0.215 0.124 0.468 0.240
Peak Isolator
Deformation (m)
0.593 0.763 0.325 0.791 0.602 1.549 0.955
Peak I.S Drift
(Norm. by h

)
0.003 0.004 0.001 0.003 0.002 0.006 0.003
Peak Absolute
Acc. (g

)
0.295 0.363 0.111 0.352 0.205 0.598 0.302
RMS Disp.
(m)
0.210 0.306 0.148 0.344 0.222 0.360 0.482
RMS Acc.
(g

)
0.076 0.132 0.042 0.140 0.061 0.126 0.178

W: Weight of the structure (202,000 kN); h: Average story height (4.04 m); g:
Acceleration due to gravity (9.81 m/sec
2
)
TABLE 10. Friction Isolation System - Uncontrolled Response Quantities
(FP - X and FN - Y)
Newhall Sylmar El Centro Rinaldi Kobe Jiji Erzinkan
Peak Base Shear
(Norm. by W

)
0.170 0.245 0.083 0.228 0.140 0.424 0.220
Peak Str. Shear
(Norm. by W

)
0.151 0.213 0.086 0.199 0.110 0.352 0.180
Peak Isolator
Deformation (m)
0.300 0.489 0.108 0.433 0.252 1.095 0.512
Peak I.S Drift
(Norm. by h

)
0.002 0.003 0.002 0.003 0.002 0.004 0.003
Peak Absolute
Acc. (g

)
0.572 0.355 0.402 0.338 0.360 0.478 0.270
RMS Disp.
(m)
0.071 0.124 0.023 0.084 0.050 0.166 0.151
RMS Acc.
(g

)
0.080 0.076 0.079 0.071 0.083 0.071 0.074
21
TABLE 11. Friction Isolation System - Uncontrolled Response Quantities
(FP - Y and FN - X)
Newhall Sylmar El Centro Rinaldi Kobe Jiji Erzinkan
Peak Base Shear
(Norm. by W

)
0.177 0.251 0.080 0.232 0.140 0.428 0.225
Peak Str. Shear
(Norm. by W

)
0.162 0.205 0.087 0.201 0.119 0.363 0.195
Peak Isolator
Deformation (m)
0.292 0.495 0.085 0.428 0.281 1.188 0.563
Peak I.S Drift
(Norm. by h

)
0.003 0.003 0.002 0.003 0.002 0.005 0.003
Peak Absolute
Acc. (g

)
0.464 0.434 0.375 0.377 0.368 0.493 0.319
RMS Disp.
(m)
0.068 0.133 0.020 0.084 0.051 0.182 0.168
RMS Acc.
(g

)
0.081 0.079 0.079 0.068 0.082 0.076 0.082

W: Weight of the structure (202,000 kN); h: Average story height (4.04 m); g:
Acceleration due to gravity (9.81 m/sec
2
)
World Congress, Vol. L, San Francisco, California, June 30-July 5. 145150.
Erkus, B. and Johnson, E. A. (2004). Phase II: Smart base isolated benchmark
building part III: A sample controller for bilinear isolation. J. of Structural
Control and Health Monitoring (in review).
Karnopp, D., Crosby, M. J., and Harwood, R. A. (1974). Vibration control using
semi-active force generators. J. of Engrg. for Industry, ASME, 96(2), 619626.
Madden, G., Wongprasert, N., and Symans, M. D. (2003). Analytical and numer-
ical study of a smart base isolation system for seismic protection of buildings.
Comput. Aided Civ. Infrastruct. Engrg., 18, 1930.
Makris, N. (1997). Rigidity-plasticity-viscocity: Can electrorheological dampers
protect base-isolated structures from near-source earthquakes. Earthquake En-
grg. & Struct. Dyn., 26, 571591.
22
MATLAB (2000). The Math Works, Inc., Natick, Massachusetts.
Nagarajaiah, S. (1994). Fuzzy controller for structures with hybrid isolation
system. Proc. 1st World Conf. on Struc. Control, Vol. TA2, Los Angeles, CA.
6776.
Nagarajaiah, S. and Narasimhan, S. (2004a). Phase I: Controllers for benchmark
base isolated building: Part I. Proc. 4th Int. Workshop on Structural Control
and Health Monitoring; also J. of Structural Control and Health Monitoring
(in review), Columbia University.
Nagarajaiah, S. and Narasimhan, S. (2004b). Phase II: Controllers for bench-
mark base isolated building with friction isolation system: Part IV. J. of
Structural Control and Health Monitoring (in preparation).
Nagarajaiah, S., Reinhorn, A. M., and Constantinou, M. C. (1991a). Nonlinear
dynamic analysis of 3-d-base-isolated structures. J. of Str. Engrg., ASCE,
117(7), 20352054.
Nagarajaiah, S., Reinhorn, A. M., and Constantinou, M. C. (1991b). 3D-
BASIS: Nonlinear dynamic analysis of three dimensional base isolated
structures-part 2. Rep. No. NCEER-91-0005, Nat. Ctr. for Earthquake Engrg.
Res., State University of New York, Bualo.
Nagarajaiah, S., Riley, M. A., and Reinhorn, A. (1993). Control of sliding-
isolated bridge with absolute acceleration feedback. J. Eng. Mech., ASCE,
119(11), 23172332.
Nagarajaiah, S., Sahasrabudhe, S., and Iyer, R. (1994). Seismic response of slid-
ing isolated bridges with MR dampers. Proc. American Control Conference.
CDROM.
Nagarajaiah, S. and Sun, X. (2000). Response of base-isolated USC hospital
building in northridge earthquake. J. Struct. Engrg., ASCE, 126(10), 1177
1186.
23
Ohtori, Y., Christenson, R. E., Spencer, B. F., and Dyke, S. J. (2003). Bench-
mark problems in seismically excited nonlinear buildings. J. Eng. Mech.,
ASCE, 129, to appear, http://cee.uiuc.edu/sstl/.
Ramallo, J., Johnson, E. A., and Spencer, B. (2002). Smart base isolation sys-
tems. J. Eng. Mech., ASCE, 128(10), 10881099.
Reinhorn, A. and Riley, M. (1994). Control of bridge vibrations with hybrid
devices. Proc. 1st World Conf. on Struc. Control, Vol. TA2, Los Angeles, CA.
5059.
Reinhorn, A., Soong, T., and Wen, C. Y. (1987). Base-isolated structures with
active control. Proc. ASME PVP Conf., Vol. PVP-127, ASME, San Diego,
CA. 413420.
Sahasrabudhe, S., Nagarajaiah, S., and Hard, C. (2000). Experimental study of
sliding isolated bridges with smart dampers subjected to near source ground
motion. Proc. 13th Engrg. Mech. Conference, Austin, Texas. CDROM.
Spencer, B. F., Dyke, S. J., and Deoskar, H. S. (1998a). Benchmark problems
in structural control: part 1 - active mass driver system. Earthquake Engrg.
& Struct. Dyn., 27(11), 11271139.
Spencer, B. F., Dyke, S. J., and Deoskar, H. S. (1998b). Benchmark problems in
structural control: part 2 - active tendon system. Earthquake Engrg. & Struct.
Dyn., 27(11), 11411147.
Spencer, B. F., Dyke, S. J., Sain, M. K., and Carlson, J. D. (1997). Phenomeno-
logical model of a magnetorheological damper. J. Eng. Mech., 123(3), 230238.
Spencer, B. F., Johnson, E. A., and C., R. J. (2000). Smart isolation for seismic
control. JSME International Joural, Series C, 43(3), 704711.
Stengel, R. (1994). Optimal Control and Estimation. Dover Publications, New
York.
Symans, M. D. and Kelly, S. W. (1999). Fuzzy logic control of bridge structures
24
using intelligent semi-active seismic isolation. Earthquake Engrg. & Struct.
Dyn., 28, 3760.
Yang, J. N., Agrawal, A., Samali, B., and Wu, J. C. (2003). A benchmark
problem for response control of wind excited tall buildings. J. of Eng. Mech.,
ASCE, 129, in press; also see Proc. 14th Eng. Mech. Conf., ASCE, UTAustin,
CDROM, 2000, http://www.eng.uci.edu/anil/benchmark.html.
Yang, J. N., Wu, J., Reinhorn, A., and Riley, M. (1996). Control of sliding-
isolated buildings using sliding-mode control. J. of Struc. Engrg., ASCE, 122,
179186.
Yoshida, K., Kang, S., and Kim, T. (1994). LQG control and H

control of
vibration isolation for MDOF systems. Proc. 1st World Conf. on Struc. Con-
trol, Vol. TA4, Los Angeles, CA. 4352.
Yoshida, K., Yoshida, S., and Takeda, Y. (1999). Semi-active control of base
isolation usign feedforward information of disturbance. Proc. 2nd World Conf.
on Struc. Control, Vol. 1, Kyoto, Japan. 377386.
Yoshioka, H., Ramallo, J. C., and Spencer, B. F. (2002). Smart base isolation
systems employing magnetorheological dampers. J. of Engrg. Mech., ASCE,
128(5), 540551.
25