You are on page 1of 11

Engineering Structures 30 (2008) 34783488

Contents lists available at ScienceDirect


Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Effect of tuned mass damper on displacement demand of base-isolated structures
Tomoyo Taniguchi
a
, Armen Der Kiureghian
b,c,
, Mikayel Melkumyan
c
a
Tottori University, Tottori, 680-8552, Japan
b
University of California, Berkeley, CA 94720, USA
c
American University of Armenia, Yerevan, Armenia
a r t i c l e i n f o
Article history:
Received 13 July 2007
Received in revised form
30 May 2008
Accepted 30 May 2008
Available online 2 July 2008
Keywords:
Base isolation
Displacement demand
Far-field motions
Near-field motions
Optimal design
Tuned-mass damper
a b s t r a c t
The third author of this paper has previously proposed the installation of a tuned-mass damper (TMD) to
reduce the displacement demandona base isolatedstructure. The TMDconsists of a mass-dashpot-spring
subsystem that is attached to the isolated superstructure, analogous to a pendulum. The present paper
examines the effectiveness of this scheme and determines optimal parameters for the design of the TMD.
Boththe base-isolatedstructure andthe TMDare modeledas single-degree-of-freedom, linear oscillators.
The optimal TMD parameters are determined by considering the response of the base-isolated structure,
with and without the TMD, to a white-noise base acceleration. Such an excitation is representative of
broadband ground motions having a nearly constant intensity over a duration several times longer than
the period of the base-isolated structure. It is found that, under suchanexcitation, a reductionof the order
of 15%25% in the displacement demand of the base-isolated structure can be achieved by adding the
TMD. Next, the responses of an example base-isolated structure with and without an optimally designed
TMD to selected suites of far- and near-field recorded accelerograms are determined. The study shows
that for far-field ground motions the effectiveness of the TMD is more or less similar to that predicted by
the white noise model, whereas for near-field ground motions the effectiveness of the TMD is less, i.e. of
the order of 10% or less. Reasons for this result are described.
2008 Elsevier Ltd. All rights reserved.
1. Introduction
Seismic base isolation has become one of the most effective
technologies in protecting structures against destructive earth-
quakes. By providing flexibility in the base of the structure, the
isolation system absorbs the bulk of the displacement demand of
the earthquake, with the super-structure essentially displacing as
a rigidbody. Furthermore, base isolationdrastically lowers the fun-
damental frequency of the system, putting it outside the dominant
range of input frequencies and, thereby, reduces the acceleration
at floor levels of the structure where sensitive equipment or non-
structural systems may be located. In doing this, the base isolation
system itself undergoes a relatively large displacement. One im-
portant consideration in the design of the base isolation system is
this displacement demand.
Our interest in this paper focuses on a base isolation system
made of laminated rubber bearings. The displacement capacity
of such a system is directly related to its size and damping. One

Corresponding author at: University of California, Berkeley, CA 94720, USA.


Tel.: +1 510 642 2469; fax: +1 510 643 8928.
E-mail addresses: t_tomoyo@cv.tottori-u.ac.jp (T. Taniguchi),
adk@ce.Berkeley.edu (A. Der Kiureghian), mmelkumi@aua.am (M. Melkumyan).
option for reducing the displacement demand on the isolation
system is to provide supplemental damping. This, however, may
increase the in-structure accelerations [5]. In this paper we explore
the possibility of using a tuned mass damper to reduce the
displacement demand of the base isolation system.
The third author has been instrumental in initiating the
manufacturing of laminated rubber bearings in Armenia and
using them for retrofitting of existing buildings or for new
construction [8,10]. Today, more than 30 buildings in Armenia
are built, retrofitted or under construction employing the base
isolation technology, mostly using locally manufactured bearings
made of neoprenethus putting Armenia at a top rank in
terms of the number of base-isolated buildings per capita. (The
population of Armenia is around 3 million.) At the present time the
manufacturing technology is capable of producing bearings with
neoprene compounds with low or medium damping. However,
manufacturing of large-size (i.e. larger than 60 cm in diameter),
high-damping (larger than 10%) bearings remains a technological
challenge in Armenia. For this reason, there has been an interest
in exploring alternatives for reducing the displacement demand of
the base-isolation system.
With the above motivation in mind, the third author has
proposed a scheme to reduce the displacement demand of
a base-isolated structure by installing a tuned-mass damper
0141-0296/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2008.05.027
T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488 3479
Fig. 1. Schematic of base-isolated structure with TMD.
(TMD) (denoted as dynamic damper in [7,9]) attached to the
isolation floor. The effectiveness of an appended mass-spring
system in reducing the dynamic response of a structure has been
known for a long time [1,12,11]. Numerous investigations and
implementations of this idea for fixed-base buildings have been
made (see, e.g., [4,13,6,3]). In such buildings, the TMD is usually
placed in an upper floor in order to experience a larger acceleration
for efficiently mobilizing itself and absorbing the energy in the
system. Usually a large portion of the floor or the entire floor of the
building must be devoted to the TMD. In the case of a base-isolated
structure, since the maximum relative displacement occurs at the
level of the bearings, the TMD must be attached immediately
above the isolation system. It can be provided as a mass-spring
subsystem attached either above or below the isolated floor of the
building, as shown in Fig. 1. One can imagine various functions
for such a subsystem, e.g., an exercise room, a swimming pool,
parking space, utilities room, as long as the mass remains relatively
constant in time and large displacement can be accommodated.
As described by Melkumyan [9], the proposed TMD scheme also
has the advantage of increasing the capacity of the base-isolated
building against overturning forces.
This paper aims at examining the effectiveness of the proposed
TMD scheme and determining the set of optimal parameters for
its design. We assume the base-isolated structure can be modeled
as a single-degree-of-freedom (SDOF) oscillator. This essentially
assumes that the superstructure acts as a rigid body, which is a
reasonable assumption for a base-isolated structure. The TMD is
modeled as a SDOF oscillator as well, which is attached to the
base-isolated structure as an appendage. After formulating the
equations of motion and defining key parameters, we consider
the stationary response of the base-isolated structure with and
without the TMD when the system is subjected to a white-
noise base acceleration. Such an excitation is representative of
a broadband ground motion having a nearly constant intensity
over a duration several times longer than the period of the base-
isolated structure. The simplicity of this excitation model allows
us to determine the effectiveness of the TMD in terms of a few
key parameters. This then leads to the identification of optimal
parameters (mass, frequency, damping) of the TMD for a given
base-isolated structure. It is found that the optimally designed
TMD can effect a reduction of 15%25% in the displacement
demand of the isolators. Next, the responses of an example base-
isolated structure with and without an optimally designed TMD
to selected suites of far- and near-field recorded ground motions
are determined. The investigation shows that for far-field ground
motions the effectiveness of the TMDis more or less similar to that
predicted by the white-noise model. Furthermore, the same level
of reduction is achieved in the acceleration response. For near-field
ground motions, on the other hand, the effectiveness of the TMD is
less, i.e., of the order of 10% or less. The reason has to do with the
short, pulse-type nature of near-field ground motions.
Fig. 2. Idealized model of base-isolated structure with TMD.
The practical implementation of a TMD in a base-isolated
building obviously involves additional considerations, such as
provision of space for the displacement of the TMD and cost.
These considerations, admittedly important ones, have not been
addressed in this paper.
2. Equations of motion
Consider the combined system consisting of the base-isolated
structure and the TMD, as shown in Fig. 1. We assume the
base-isolated structure alone behaves approximately as a SDOF
oscillator having an effective mass m
p
, a natural frequency
p
, and
a damping ratio
p
, where the subscript p refers to primary. We
also assume the TMD by itself behaves approximately as a SDOF
oscillator with an effective mass m
s
, a natural frequency
s
, and
a damping ratio
s
, where the subscript s refers to secondary.
The combined systemconsisting of the base-isolated structure (the
primary subsystem) and the TMD(the secondary subsystem) is a 2-
DOF system, as shown in an idealized formin Fig. 2. It is known (see
[2]) that such a composite primarysecondary system is generally
non-classically damped, even when the individual sub-systems are
classically damped. Hence, to properly model the system, account
must be made of the non-classical damping nature of the combined
system.
The equations of motion of the combined system is described
by
M u +C u +Ku = M1 x
g
(t) (1)
where u(t) = [u
p
(t) u
s
(t)]
T
is the vector of displacements
relative to the ground, x
g
(t) is the ground acceleration and
M =
_
m
p
0
0 m
s
_
,
C =
_
2
p

p
m
p
+2
s

s
m
s
2
s

s
m
s
2
s

s
m
s
2
s

s
m
s
_
,
K =
_

2
p
m
p
+
2
s
m
s

2
s
m
s

2
s
m
s

2
s
m
s
_
, 1 =
_
1
1
_
.
(2)
For the subsequent analysis, it is useful to introduce the mass
ratio
=
m
s
m
p
(3)
and the tuning parameter
=

p

ave
(4)
where
ave
= (
p
+
s
)/2 is the average frequency. The mass
ratio describes the size of the TMD; we consider values in the range
= 0.01 to 0.10. The tuning parameter describes the proximity of
the natural frequencies of the two sub-systems. It is well known
that the TMD is more effective when is large and is near zero.
Results are presented below in terms of these two parameters, as
well as the individual system parameters defined earlier.
3480 T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488
Fig. 3. Frequencies of the combined system.
3. Frequencies of the undamped combined system
As mentioned earlier, the combined system is non-classically
damped. Nevertheless, it is insightful to examine its undamped
frequencies. Analytically solving the eigenvalue problem
2
M =
Kfor the undamped system, we obtain the following expressions
for the two natural frequencies of the combined system.

1
,
2
=
1

2
_

2
p
+(1 + )
2
s

_
_

2
p
+(1 + )
2
s
_
2
4
2
p

2
s
_1
2
. (5)
Fig. 3 shows plots of the two frequencies of the combined
system, normalized with respect to the frequency of the primary
sub-system, as a function of the tuning parameter for different
mass ratios. The ratio of sub-system frequencies,
s
/
p
, is also
shown for reference. It can be seen that for large negative values,
i.e., for
p

s
, the first mode of the combined system has
a frequency that is close to but lower than that of the primary
sub-system and the second mode has a frequency that is close
to but greater than that of the secondary sub-system. Conversely,
for a large positive value, i.e., for
s

p
, the second
mode of the combined system has a frequency that is close to
but greater than that of the primary sub-system, while the first
mode has a frequency that is close to but smaller than that of
the secondary sub-system. It is reasonable to expect that for these
values of , the mode of the combined systemthat has a frequency
close to that of the primary sub-system dominates its response
in the combined system. For values close to zero, i.e., near
perfect tuning, the two modal frequencies of the combined system
are symmetrically positioned relative to the frequency of the
primary sub-system. In this case, the two modes tend to equally
contribute to the responses of the primary and secondary sub-
systems in the combined system. As we will shortly see, the TMDis
most effective in reducing the displacement demand on the base-
isolated structure when is in the range 0.10.2. In this range, the
second mode of the combined system dominates the response of
the base-isolated structure. It is also notable that the frequencies
of the combined system move further apart from the sub-system
frequencies and from each other as the mass ratio, , increases.
4. Stochastic dynamic analysis
To examine the effectiveness of the TMD in reducing the
displacement demand of the base-isolated structure, we first
consider the stationary response of the combined system to a
zero-mean, broadband stationary stochastic base acceleration.
To properly account for the non-classical damping nature of
the combined system, a frequency-domain approach using the
frequency-response matrix (FRM) (instead of the usual modal
superposition approach) is used. The FRM of the system in (1) is
given by
H() =
_

2
M+iC +K
_
1
. (6)
Let
x
g
x
g
() denote the power spectral density (PSD) of the ground
acceleration. The PSD matrix of the response vector u is then given
by

uu
() = H()M1
x
g
x
g
()1
T
M
T
H()
T
(7)
where the superposed asterisk denotes the complex conjugate.
As measures of the responses of interest, we consider the mean-
squares of the displacements u
p
(t) and u
s
(t) of the primary and
secondary subsystems, respectively. These are given by

2
p
=
_
+

u
p
u
p
()d (8)

2
s
=
_
+

u
s
u
s
()d (9)
where
u
p
u
p
() and
u
s
u
s
() are the diagonal elements of
uu
().
Closed form expressions for the elements of the 2 2 matrix
H()M are derived in Appendix. These are used in (7) to derive
expressions for
u
p
u
p
() and
u
s
u
s
(). Numerical integration is
then used to evaluate (8) and (9).
To investigate the effect of the TMD in reducing the displace-
ment demand of the base-isolated structure, we consider the re-
sponse ratio
p
/
p0
, where
p0
denotes the root-mean-square of
the response of the base isolated structure without the TMD. In
order to make the results independent of the specifics of the in-
put ground motion, the excitation is assumed to be a white-noise
process having a constant PSD
x
g
x
g
() =
0
. This model is a
good approximation for broadband earthquake ground motions
and lightly damped structures. For this model, it is well known that

2
p0
=
0
/(2
p

3
p
). Furthermore, the response ratio
p
/
p0
is in-
dependent of the intensity
0
of the white noise. In fact, as can be
verifiedfromthe expressions inAppendix, the ratio
p
/
p0
only de-
pends on the mass ratio, , the tuning parameter, , and the damp-
ing ratios
p
and
s
of the two subsystems.
Figs. 4(a)(d) show plots of the response ratio
p
/
p0
as a
function of for the primary damping ratio
p
= 0.05 and
secondary damping ratios
s
= 0.05, 010, 0.15 and 0.20,
respectively. Three curves are shown in each plot for the mass ratio
values = 0.02, 0.05 and 0.10. A fourth curve with diamond-
shaped markers is also shown, which is described below.
First consider Fig. 4(a), which is for
p
=
s
= 0.05. As
one would expect, the effectiveness of the TMD in reducing the
displacement demand on the base-isolated structure depends on
both the mass ratio, , and on the tuning parameter, . For a fixed
, the optimal TMDoccurs at a positive value, which corresponds
to a frequency of the TMD that is smaller than the frequency of the
base-isolated structure. Thus, contrary to the notion of a tuned
mass damper, the optimal reduction in the response of the base-
isolated structure occurs not at = 0 but for 0 < . The locus
of these optimal points for all values is plotted in the figure
(line with diamond markers) and is called the design curve. Note
that the diamond marks on the design curve indicate values
from 0.01 to 0.10 at increments of 0.01. It is seen that the optimal
value of the tuning parameter moves towards greater positive
values with increasing mass ratio. For example, at = 0.05, the
optimal value of the tuning parameter is
opt
= 0.10, which using
(4) corresponds to the optimal TMD frequency of
s
= 0.905
p
,
whereas at = 0.10 the optimal value of the tuning parameter
is
opt
= 0.18, which corresponds to
s
= 0.835
p
. At these
T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488 3481
(a)
s
= 0.05. (b)
s
= 0.10.
(c)
s
= 0.15. (d)
s
= 0.20.
Fig. 4. Reduction in the seismic demand of the base-isolated structure caused by the TMD for
p
= 0.05.
mass ratios, the TMD reduces the demand on the base-isolated
structure by about 21% and 22%, respectively. On the other hand,
the reduction in demand for the optimal TMD with mass ratio =
0.02 is 17%. This confirms the well known result that increasing the
mass ratio increases the effectiveness of the TMD (e.g., see [11]).
Note, however, that even a mass ratio of = 0.01 or 0.02 provides
considerable reduction in the demand.
Figs. 4(b)(d) show results similar to those shown in Fig. 4(a)
but for secondary damping ratios
s
= 0.10, 0.15 and 0.20,
respectively. It is seen that significant improvement in the
effectiveness of the TMD is achieved by increasing its damping
ratio from 0.05 to 0.10. However, further increase in the damping
ratio of the TMD provides marginal improvement or even
diminishes its effectiveness (e.g., compare the curves for = 0.02
for increasing
s
). It appears that the TMD damping ratio
s
= 0.10
is a good choice if the damping ratio of the base-isolated structure
is
p
= 0.05.
Figs. 5(a)(d) show results similar to those described above but
for the primary damping ratio
p
= 0.10. It is seen that, compared
to the case of
p
= 0.05, the effectiveness of the TMD is reduced
to no more than 5%15%. Melkumyan [7] anticipated this effect by
suggesting the use of the TMD in conjunction with low-damping
rubber isolation bearings.
The design curves in Figs. 4 and 5 can be used to select the
frequency of the TMD, for given frequency of the base-isolated
structure, the mass ratio and the two damping ratios, to achieve
the maximum reduction in the displacement demand on the base-
isolated structure. It is important to observe, however, that the
curve for each fixed has a relatively flat bottom. Therefore, if a
small error is made in estimating the frequencies of the primary or
secondary subsystems, the reduction in the response of the base-
isolated structure will not be greatly affected. Roughly speaking,
relative variations

p
and

s
in the two frequencies lead to the
absolute variation


= 2
_

p
+
2

s
in the tuning parameter.
For example, if the primary frequency is known within a 4% error
and the secondary frequency is known within 2% error, then
the estimated error in is around 2
_
0.04
2
+0.02
2
= 0.09.
This formula can be used to estimate the range of variations in
the tuning parameter for given uncertainties in the sub-system
frequencies.
In summary, the following conclusions can be derived from the
above analysis of the stochastic response:
(a) The effectiveness of the TMD in reducing the displacement de-
mand on the base-isolated structure increases with increasing
mass ratio, provided the TMD is optimally tuned;
(b) The optimal TMD always has a frequency smaller than the
frequency of the base-isolated structure;
(c) The TMD is more effective for a lightly damped isolation
system;
(d) For the damping ratio
p
= 0.05 of the base-isolated structure,
a good choice for the damping ratio of the TMD is
s
= 0.10;
higher TMD damping ratios do not significantly improve the
effectiveness of the TMD;
(e) A reduction of 20% or greater in the displacement demand of
the base-isolated structure with damping ratio
p
= 0.05 can
be achieved by use of an optimally designed TMD having a
damping ratio of
s
= 0.10 and mass ratio 0.05 ;
(f) Variations in the order of 2%3% in the frequencies of the
base-isolated structure and an optimally designed TMD have
a negligible influence on the effectiveness of the TMD.
Several of the above conclusions, including (a), (c) and (f), and
the fact that a higher damping for the TMD beyond a certain level
does not increase its effectiveness, are confirmations of well known
results for TMDs in fixed-base buildings.
The above analysis assumes that the TMD responds within
its elastic limit. Since the TMD is nearly tuned to the primary
subsystem, it may experience a large response, which may put it
beyond its yield limit. To investigate this possibility, we examine
the response ratio
s
/
p
, which is only a function of the parameters
, ,
p
and
s
. Fig. 6 shows this response ratio as a function of
3482 T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488
(a)
s
= 0.05. (b)
s
= 0.10.
(c)
s
= 0.15. (d)
s
= 0.20.
Fig. 5. Reduction in the seismic demand of the base-isolated structure caused by the TMD for
p
= 0.10.
Fig. 6. Amplification in the displacement of the TMD for
p
= 0.05 and
s
= 0.10.
the tuning parameter for
p
= 0.05,
s
= 0.10 and the mass
ratio values = 0.02, 0.05 and 0.10. It appears that, depending on
the tuning parameter and the mass ratio, the response of the TMD
can be 24 times larger than that of the base-isolated structure.
This finding clearly calls for a careful design of the TMD and
the space it occupies in the building in order to accommodate
the large displacement demand. Alternatively, one may allow the
TMDto dissipate energy through hysteretic action. This potentially
beneficial effect is not considered in the present study.
5. Time history analysis
The analysis in the preceding section employed a stationary
white-noise process as a model for the ground acceleration. This
is convenient, since for this model the effectiveness of the TMD
can be assessed with the least number of system parameters
and without involving any parameters that characterize the input
excitation. However, one may question whether this idealized
model of the ground motion accurately describes the effectiveness
of the TMD, since real earthquake ground motions are neither
stationary nor have a uniform spectral content, as represented
by the white-noise model. It is well known that, in order for the
TMD to be effective, it is necessary that the energy input into
the system be gradual so that there is time for transfer of energy
from the primary system (the base-isolated structure) into the
secondary system, the TMD. This suggests that the non-stationary
nature of the ground motion may have a strong influence on
the effectiveness of the TMD. For this reasons, it was decided
to examine the effectiveness of the TMD by using time-history
analyses with selected recorded accelerograms. The stand-alone
base-isolated structure considered for this analysis has a frequency
of
p
= rad/s (2.0 s period) and a damping ratio of
p
=
0.05. The TMD is assumed to have the damping ratio
s
=
0.10, the mass ratio = 0.05, and the stand-alone frequency

s
= 0.910 rad/s (2.2 s period), which corresponds to the
optimally designed value of the tuning parameter,
opt
= 0.094,
as can be seen in Fig. 4b. For this system, a response ratio of 0.75
(i.e. a reduction of 25% in the displacement demand) is expected
from the analysis with the white-noise excitation. The recorded
accelerograms are selected from the PEER strong-motion database
at http://peer.berkeley.edu/smcat/index.html. To properly account
for the non-classical damping nature of the combined system,
direct numerical integral of the equations of motion in (1) is
carried out. The second-order RungeKutta algorithm is used for
this purpose.
Considerable attention was given to the characteristics of
the selected ground motions, which are listed in Table 1. Since
the analysis is linear, and the ratio of responses with and
without the TMD is of interest, the intensity of the motion is
immaterial. However, the temporal evolution of the motion and
its frequency content are important considerations. As proxies
for these characteristics of the recorded motions, we selected
the distance of the recording site from the fault rupture and the
local site condition. The distance from the fault rupture tends
to influence the nonstationary character of the accelerogram. In
T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488 3483
T
a
b
l
e
1
L
i
s
t
o
f
e
a
r
t
h
q
u
a
k
e
s
a
n
d
c
o
m
p
u
t
e
d
r
e
s
p
o
n
s
e
r
a
t
i
o
s
E
a
r
t
h
q
u
a
k
e
R
e
c
o
r
d
c
o
m
p
o
n
e
n
t
U
S
G
S
s
i
t
e
c
l
a
s
s
.
D
i
s
t
a
n
c
e
(
k
m
)
P
G
A
(
g
)
P
G
V
(
c
m
/
s
)
P
G
D
(
c
m
)
A
c
c
e
l
e
r
o
g
r
a
m
(
g
)
R
e
s
p
o
n
s
e
r
a
t
i
o
D
i
s
p
.
(
%
)
A
c
c
.
(
%
)
F
a
r
-
f
i
e
l
d
g
r
o
u
n
d
m
o
t
i
o
n
s
N
o
r
t
h
r
i
d
g
e
1
9
9
4
/
0
1
/
1
7
1
2
:
3
1
N
O
R
T
H
R
1
1
6
0
9
0
B
4
1
.
9
0
.
2
0
8
1
0
.
3
2
.
6
7
1
0
7
1
0
3
L
o
m
a
P
r
i
e
t
a
1
9
8
9
/
1
0
/
1
8
0
0
:
0
5
L
O
M
A
P
S
F
O
0
9
0
C
6
4
.
4
0
.
3
2
9
2
7
.
9
6
.
0
3
9
0
9
2
I
m
p
e
r
i
a
l
V
a
l
l
e
y
1
9
7
9
/
1
0
/
1
5
2
3
:
1
6
I
M
P
V
A
L
L
H
-
D
L
T
3
5
2
C
4
3
.
6
0
.
3
5
1
3
3
.
0
1
9
.
0
2
8
0
7
8
K
o
b
e
1
9
9
5
/
0
1
/
1
6
2
0
:
4
6
K
O
B
E
K
A
K
0
9
0
D
2
6
.
4
0
.
3
4
5
2
7
.
6
9
.
6
7
4
8
0
C
h
i
-
C
h
i
,
T
a
i
w
a
n
1
9
9
9
/
0
9
/
2
0
C
H
I
C
H
I
T
C
U
0
4
7
-
N
B
3
3
.
0
1
0
.
4
1
3
4
0
.
2
2
2
.
2
2
6
0
6
1
K
e
r
n
C
o
u
n
t
y
1
9
5
2
/
0
7
/
2
1
1
1
:
5
3
K
E
R
N
T
A
F
1
1
1
B
4
1
.
0
0
.
1
7
8
1
7
.
5
8
.
9
9
7
6
7
5
N
e
a
r
-
f
i
e
l
d
g
r
o
u
n
d
m
o
t
i
o
n
s
N
o
r
t
h
r
i
d
g
e
1
9
9
4
/
0
1
/
1
7
1
2
:
3
1
N
O
R
T
H
R
N
W
H
0
9
0
C
7
.
1
0
0
.
5
8
3
7
5
.
5
1
7
.
5
7
9
2
9
2
L
o
m
a
P
r
i
e
t
a
1
9
8
9
/
1
0
/
1
8
0
0
:
0
5
L
O
M
A
P
L
G
P
0
0
0
?
6
.
1
0
0
.
5
6
3
9
4
.
8
4
1
.
1
8
9
1
9
1
I
m
p
e
r
i
a
l
V
a
l
l
e
y
1
9
7
9
/
1
0
/
1
5
2
3
:
1
6
I
M
P
V
A
L
L
H
-
E
0
8
1
4
0
C
3
.
8
0
0
.
6
0
2
5
4
.
3
3
2
.
3
2
9
2
9
1
K
o
b
e
1
9
9
5
/
0
1
/
1
6
2
0
:
4
6
K
O
B
E
K
J
M
0
0
0
B
0
.
6
0
0
.
8
2
1
8
1
.
3
1
7
.
6
8
9
3
9
4
C
h
i
-
C
h
i
,
T
a
i
w
a
n
1
9
9
9
/
0
9
/
2
0
C
H
I
C
H
I
T
C
U
1
2
9
-
W
C
1
.
1
8
1
.
0
1
6
0
.
0
5
0
.
1
5
9
7
9
6
T
a
b
a
s
,
I
r
a
n
1
9
7
8
/
0
9
/
1
6
T
A
B
A
S
T
A
B
-
T
R
C
3
.
0
0
0
.
8
5
2
1
2
1
.
4
9
4
.
5
8
7
1
7
2
3484 T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488
(a) Northridge. (b) Loma Prieta.
(c) Imperial Valley. (d) Kobe.
(e) Chi-Chi. (f) Kern county.
Fig. 7. Displacements of the base-isolated structure with (solid line) and without (dashed line) the TMD for far-field ground motions (Unit: m).
particular, near-field ground motions tend to contain a long-period
pulse, the fling, which directly results from the dislocation at
the fault, whereas far-field ground motions tend to have a fairly
smooth transition of temporal and spectral contents. To account
for these effects, two sets of ground motions were selected: near-
field ground motions with distances ranging from 0.6 to 7.1 km
from the closest point on the fault rupture, and far-field ground
motions with distances ranging from26.4 to 64.4 kmfromthe fault
rupture. The local site condition influences the frequency content
of the ground motion. For the selected records, the site condition
is characterized by the USGS classification. Among the records
considered, 4 have classification B (shear-wave velocity in the
range 360750 m/s), 6 have classification C (shear-wave velocity
in the range 180360 m/s), and one has classification D (shear-
wave velocity less than 180 m/s). One record has no classification.
For each record, the ratios of the maximum absolute displacement
andmaximumabsolute accelerationresponses of the base-isolated
structure with the TMD relative to the corresponding responses
of the stand-alone base-isolated structure are computed. These
response ratios along with other characteristics of each earthquake
record are listed in Table 1. Figs. 7 and 8 respectively show
comparisons of the computed displacement responses of the base-
isolated structure with and without the TMD for the six far-
field ground motions and six near-field ground motions, and
Figs. 9 and 10 show similar results for the computed acceleration
responses.
T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488 3485
(a) Northridge. (b) Loma Prieta.
(c) Imperial Valley. (d) Kobe.
(e) Chi-Chi. (f) Tabas.
Fig. 8. Displacements of the base-isolated structure with (solid line) and without (dashed line) the TMD for near-field ground motions (Unit: m).
The following observations can be made from the computed
response ratios listed in Table 1 and the time-history results in
Figs. 710: It is first noted in Table 1 that in all but one case
(far-field Northridge record) reductions in the displacement and
acceleration responses of the base-isolated structure are achieved
by adding the TMD. Secondly, the response ratios are nearly the
same for the displacement and acceleration responses for each
ground motion. This indicates that the TMD has similar influences
on the displacement and acceleration responses of the base-
isolated structure. Thus, the reductioninthe displacement demand
is not achieved at the expense of increasing the acceleration
response, as may be the case with other alternatives, such as the
provision of supplemental damping [5]. Thirdly, no correlation is
observed between the response ratios and the site classifications.
Evidently, the site classification is too crude a measure to have a
direct relation with the effectiveness of the TMD.
For the far-field ground motions (the first six rows in Table 1),
the response ratio is around the expected 75% for four out
of the six records. Among these records, the ones of Imperial
Valley, Kobe and KernCounty earthquakes have nearly stationary
characters during their respective strong motions phases. The Chi-
Chi earthquake has a distinctly nonstationary behavior. However,
the largest pulses in the accelerogram occur after a period of
gradual increase in the intensity of the motion (see Figs. 7 and 9
bottom left). This gives opportunity to the TMD to be mobilized
and absorb energy from the primary structure. For the far-field
Loma Prieta record, the response ratios are 0.90 and 0.92. The small
reduction in the response can be attributed to the fact that the peak
3486 T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488
(a) Northridge. (b) Loma Prieta.
(c) Imperial Valley. (d) Kobe.
(e) Chi-Chi. (f) Kern county.
Fig. 9. Accelerations of the base-isolated structure with (solid line) and without (dashed line) the TMD for far-field ground motions (Unit: g).
responses to this record are primarily due to a large acceleration
pulse happening around 11s (see Figs. 7 and 9 top right). Since
this large acceleration pulse occurs early in the excitation, the
effectiveness of the TMD is similar to the case of near-field ground
motions described below. The most puzzling result is that of
the far-field Northridge record, which appears to have a fairly
stationary character during its strong-motion phase, but the effect
of the TMD in this case is a slight amplification of the response of
the base-isolated structure. This case is further examined below.
For the near-field ground motions (the last six rows in Table 1),
the reductions in the displacement and acceleration responses of
the base-isolated structure are less than 10% in all but the case of
Tabas record. We attribute this to the fact that, with the exception
of Tabas, all these motions contain large acceleration pulses early
in their time histories, which produce the peak responses of the
base-isolated structure (see Figs. 8 and 10). As mentioned earlier,
this is a typical characteristic of near-field ground motions, which
are directly affected by the fault slip with little influence from
the dispersive effect of waves traveling long distances. For such
records, there is no time for transfer of energy from the base-
isolated structure into the TMD and, hence, the TMD does not
become fully mobilized to achieve its effectiveness, as predicted
by the stationary analysis in the preceding section. Although this
phenomenon is generally known, we have not found a quantitative
analysis of its effect in the TMD literature, particularly in relation
to near-field ground motions. In contrast to the other near-field
T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488 3487
(a) Northridge. (b) Loma Prieta.
(c) Imperial Valley. (d) Kobe.
(e) Chi-Chi. (f) Tabas.
Fig. 10. Accelerations of the base-isolated structure with (solid line) and without (dashed line) the TMD for near-field ground motions (Unit: g).
records, since large acceleration pulses in the Tabas record occur
in the middle of the record, there is enough time for the TMD to
be mobilized to achieve its effectiveness (see Figs. 8 and 10 bottom
right). Inthis sense, the Tabas recordbehaves like a far-fieldground
motion, while the far-field Loma Prieta record behaves like a near-
field ground motion.
We now return to examine the case of the far-field Northridge
record. As mentioned earlier, this record has a nearly stationary
strong-motion phase, yet the response ratios for both the
displacement and acceleration responses are greater than 1. That
is, for this ground motion, attaching the optimally designed
TMD actually enhances both the displacement and acceleration
demands on the base-isolated structure. To understand the reason
for this behavior, we examine the response spectra of the selected
far-field ground motions. Fig. 11 shows the 5% damping response
spectra for these motions, all normalized by their values at
0.5 Hz, which is the frequency of the base-isolated structure
without the TMD. A thicker line highlights the spectrum for the
Northridge record. It can be seen that the Northridge record has
a sharp dominant peak at around 0.4 Hz frequency, which has
a much larger value than the spectral displacements at all other
frequencies. Thus, the energy in this motion is mostly concentrated
around 0.4 Hz frequency. As can be verified in Fig. 3, this frequency
happens to coincide with the first undamped modal frequency of
the base-isolatedstructure withthe TMD. Thus, by adding the TMD,
the first mode of the combined system, i.e. the mode dominated
3488 T. Taniguchi et al. / Engineering Structures 30 (2008) 34783488
Fig. 11. Normalized response spectra of far-field ground motions for 5% damping.
by the displacement of the TMD, is subjected to this large spectral
amplitude, resulting in an amplification of the response of the
base-isolated structure relative to its response without the TMD.
In other words, in this case, because of resonance of the TMD with
the input excitation, the TMD further excites the base-isolated
structure instead of absorbing its energy. It is noted that none of
the other response spectra in Fig. 11 has this particular feature.
The response ratios predicted based on stochastic analysis
shown in Figs. 4 and 5 represent ensemble averages. The
effectiveness of the TMD for individual realizations of the
stochastic ground motion would, of course, vary around these
averages. Therefore, in examining the time-history results, it is
appropriate to consider the averages over the selected samples.
Considering all six samples, the average response ratios for the
far-field ground motions are 0.81 for both the displacement and
acceleration responses. If the Northridge record is not included,
the averages reduce to 0.76 and 0.77, respectively, which are in
line with that predicted by the stochastic analysis. For near-field
ground motions, the average response ratios are around 0.89 for
both displacement and acceleration responses. If the Tabas record
is excluded, both averages are around 0.93. It is clear that the
TMD is not effective in reducing the demand on the base-isolated
structure for near-field ground motions.
6. Conclusions
The effectiveness of a TMD to reduce the seismic demand on a
base-isolated structure is investigated. Using stochastic dynamic
analysis based on a white-noise model of the ground motion, the
optimal parameters of the TMD that maximally reduce the seismic
demand on the base-isolated structure are determined. This
investigation reveals that, depending on the mass, damping and
frequency characteristics of the TMD, the displacement demand
on the base isolated structure can be reduced by 15%25%. It is
shown that the TMD is more effective for lightly damped isolators.
Furthermore, the effectiveness of the TMD increases with its mass,
but not necessarily with its damping.
To account for the nonstationary and non-white nature of
ground motions, a series of time history analyses with far- and
near-field recorded ground motions are carried out. The analyses
showthat for far-fieldgroundmotions the effectiveness of the TMD
is in concordance with the predictions of the stochastic analysis,
except for one particular record, which happens to have a sharp
spectral peak in resonance with the TMD. Importantly, the TMD
produces virtually identical reductions in the displacement and
acceleration demands of the base-isolated structure. For near-field
ground motions, the effectiveness of the TMD is no more than
7%10%. The reason is that for such motions the peak response
usually is due to a large pulse early in the record, so that sufficient
time is unavailable for the TMD to be mobilized.
One may question the viability and cost-effectiveness of
installing a TMD with a mass ratio as large as 5% or more of the
building mass to effect a reduction of no more than 15%25% in the
displacement demand of a base isolation system. However, if the
TMD is designed as an integral part of the base-isolated building
and it serves a useful function, then such a scheme may prove to
be both beneficial and economical. We note that, even though the
TMDmay experience large displacements, these motions will have
low frequency and, hence, will be tolerable by humans and certain
equipment (similar to wind-induced motions in top floors of tall
buildings). In any case, the results presented in this paper provide
valuable information to any engineer contemplating the use of a
TMD in a base-isolated building.
Appendix
The product of the frequency response matrix and the mass
matrix in (7) is given by
H()M
=
1
D
_

2
s

2
+2i
s

s

_

2
s
+2i
s

2
s
+2i
s

s

2
p
+
2
s

2
+2i
_

p
+
s

s
_
_
(A.1)
where D is
D =
2
_

2
p

2
s
(1 +) 4
p

s
_
+
2
p

2
s
+2i
_

s
_

s
+
p

s
_

2
_

s
(1 +) +
p

p
__
. (A.2)
References
[1] Den Hartog JP. Mechanical vibrations. New York (NY): McGraw-Hill; 1947.
[2] Igusa T, Der KiureghianA. Dynamic characterization of two degree-of-freedom
equipment-structure systems. J Eng Mech ASCE 1985;111(1):119.
[3] Inaudi J, Kelly JM. Mass dampers using friction-dissipating devices. J Eng Mech
ASCE 1995;121.
[4] Kaynia AM, Veneziano D, Biggs JM. Seismic effectiveness of tuned mass
dampers. J Struct Eng ASCE 1981;107(8):146584.
[5] Kelly JM. The role of damping in seismic isolation. Earthq Eng Struct Dyn 1999;
28:320.
[6] Khachian EE, Melkumyan MG, Khlgatyan ZM. Method of seismic protection
of multistoried buildings. In: Proceeding of UNESCO international seminar on
Spitak-88 earthquake. 1989.
[7] Melkumyan MG. The state-of-the-art in structural control in Armenia and
proposal on application of the dynamic dampers for seismically isolated
buildings. In: Proceedings of the third international workshop on structural
control for civil and infrastructure engineering, 2000.
[8] Melkumyan MG. State-of-the-art on application, R&D and design rules for
seismic isolation of civil structures in Armenia. In: Proceedings of the 8th
world seminar on seismic isolation, energy dissipation and active vibration
control of structures. 2003.
[9] Melkumyan MG. First application of the dynamic damper in the design of
seismically isolateddwelling house. In: LeschR, Irschik H, Krommer M, editors,
Proceeding of the third European conference on structural control. vol. I, Mini
symposium M6. 2004.
[10] Melkumyan MG. Current situation in application of seismic isolation
technologies in Armenia. In: Proceedings of international conference on the
250th anniversary of the 1755 Lisbon earthquake, 2005.
[11] Reed FE. Dynamic vibration absorbers and auxiliary mass dampers. Shock
and vibration handbook, vol. 1. New York (NY): McGraw-Hill, Inc.; 1961
[Chapter 6].
[12] Suzuki T. Response of elevated water tanks (2nd report). J Arch Institute Japan
1950;6:2932 [In Japanese].
[13] Warburton GB. Optimum absorber parameters for various combinations of
response and excitation parameters. Earthquake Eng Struct Dynam 1982;10:
381491.

You might also like