See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/229895418

Performance of multiple tuned mass dampers for attenuating undesirable

oscillations of structures under the ground acceleration

Article in Earthquake Engineering & Structural Dynamics · September 2000

DOI: 10.1002/1096-9845(200009)29:9<1405::AID-EQE976>3.0.CO;2-4

CITATIONS READS
214 647

1 author:

Chunxiang Li

52 PUBLICATIONS 1,380 CITATIONS

SEE PROFILE

All content following this page was uploaded by Chunxiang Li on 21 March 2021.

The user has requested enhancement of the downloaded file.

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS
Earthquake Engng Struct. Dyn. 2000; 29:1405–1421

Performance of multiple tuned mass dampers for attenuating

undesirable oscillations of structures under the ground
acceleration

Chunxiang Li ∗
College of Building Engineering and Mechanics, Shanghai Jiaotong University, Shanghai 200030, P.R.China

SUMMARY

Multiple tuned mass dampers (MTMDs) consisting of many tuned mass dampers (TMDs) with a uniform
distribution of natural frequencies are considered for attenuating undesirable vibration of a structure. The
MTMD is manufactured by keeping the sti ness and damping constant and varying the mass. The structure
is represented by its mode-generalized system in the speci c vibration mode being controlled using the mode
reduced-order method. The optimum parameters of the MTMD are investigated to delineate the in uence
of the important parameters on the e ectiveness and robustness of the MTMD by conducting a numerical
searching technique in two directions. The parameters include: the frequency spacing, average damping ratio,
mass ratio and total number. The criterion selected for the optimization is the minimization of the maximum
value of the dynamic magni cation factor (DMF) of the structure with MTMD (i.e. Min.Max.DMF). In this
paper, for the sake of comparison, the MTMD(II), which is made by keeping the mass constant and varying
the sti ness and damping coecient, and a single TMD are also taken into account. It is demonstrated
that the optimum frequency spacing of the MTMD is the same as that of the MTMD(II) and the optimum
average damping ratio of the MTMD is a little larger than that of the MTMD(II). It is also found that the
optimum MTMD is more e ective than the optimum MTMD(II) and the optimum single TMD with equal
mass. Copyright ? 2000 John Wiley & Sons, Ltd.

KEY WORDS: optimum parameters; MTMD; MTMD(II); ground acceleration; Min.Max.DMF

1. INTRODUCTION

In vibration control of structures, the tuned mass damper (TMD) has been accepted as an e ective
passive control device to attenuate undesirable vibration of a structure [1; 2]. The TMD consists of
a mass, a spring and a viscous damper attached to the structure. Its mechanism of mitigating the
structural vibration is to transfer the vibration energy of the structure to the TMD and to dissipate
the energy through the damping of the TMD. In order to enlarge the dissipation energy in the
TMD, it is essential to perform the optimum design of the TMD. That is, the optimum design is

∗ Correspondence to: Chunxiang Li, College of Building Engineering and Mechanics, Shanghai Jiaotong University, 1954
Huashan Road, Shanghai 200030, P.R.China
Contract=grant sponsor: China Postdoctoral Science Foundation

Received 30 December 1999

Copyright ? 2000 John Wiley & Sons, Ltd. Accepted 30 March 2000
1406 C. LI

carried out through tuning the natural frequency of the TMD to the structural natural frequency
corresponding to the mode to be controlled and selecting the appropriate value of the damping.
The determination of optimum parameters (i.e. the tuning frequency and the damping ratio) and
the e ectiveness of a TMD to control structural oscillations caused by di erent types of excitations
is now well established [3–9].
One of the disadvantages of a single TMD is its sensitivity to the uctuation in tuning of the
natural frequency of the TMD to the natural frequency of the structure and=or that in the damping
ratio of the TMD [10]. The e ectiveness of a TMD is decreased signi cantly by the o -tuning or
the o -optimum damping in the TMD. That is, a single TMD is not robust at all. Furthermore, the
dynamic characteristics of a structure will change under seismic excitation. This change degrades
the performance of a TMD considerably due to the o set in the tuning of the frequency and=or in
the optimum damping ratio. As a result, the use of more than one tuned mass damper with di erent
dynamic characteristics has been proposed in order to improve the e ectiveness. Iwanami and Seto
[10] proposed dual tuned mass dampers (2TMD) and made research on the optimum design of
2TMD for harmonically forced oscillation of the structure. It was shown in their papers that two
tuned mass dampers are more e ective than a single tuned mass damper. However, the e ectiveness
was not signi cantly improved. Recently, multiple tuned mass dampers (MTMD) were proposed
by Xu and Igusa [11; 12] and also studied by Yamaguchi and Harnpornchai [13], Abe and Fujino
[14], Abe and Igusa [15], Kareem and Kline [16], Jangid [17] and Li [18]. The MTMD approach
is an alternative that consists of a large number of TMDs whose natural frequencies are scattered
around the target structural frequency (i.e. the structural frequency corresponding to the mode to
be controlled). The MTMD promises to be more e ective under excitation frequencies distributed
over a wider band. The MTMD has better performance than a single TMD with the same mass,
and is also more robust to the o -tuning. Also, there exists an optimum frequency spacing for the
MTMD for which e ectiveness of MTMD is maximum.
The view shows that considerable work is carried out on the utilization of TMD and MTMD for
the reduction of the dynamic response of structures. Nevertheless, in the case of seismic excitation,
research in the physical interpretation of the performance of MTMD has been very limited. With
a view to the engineering design and use of MTMD, it is essential to make further research on
the physical interpretation of the dynamic behaviour of a structure with MTMD under important
parametric variations. The present study speci cally addresses this problem. In the study, the
MTMD with uniform sti ness and damping constant is considered. Manufacturing of the MTMD
with each TMD being identical sti ness and damping constant (the mass is varied) is far simpler
than that of the MTMD (referred to as MTMD(II)) with each TMD being di erent sti ness
and damping properties (the mass remains unchanged). In this paper, due to consideration of
a comparison, the MTMD(II) is also taken into account. The response of interest is the relative
displacement with respect to the ground. Under the condition that the ground acceleration excitation
is considered a harmonic excitation in the form of the complex exponential, the absolute value of
the frequency response function of the structure with MTMD is termed the dynamic magni cation
factor (i.e. DMF). The e ectiveness of the MTMD is expressed by the maximum value of the
dynamic magni cation factor of the structure with MTMD (i.e. Max.DMF).
Recently, the steel-structure buildings are developing considerably in Asia, especially in China,
say Shanghai. Due to light-weight and low damping coecient of these buildings, it is easy to adopt
the vibration control technique for the reduction of the displacement response of these structures.
Therefore, it is feasible to suppress the seismic response of structures using MTMD. In this paper,
for the convenience of the engineering design and application of the MTMD, the damping ratio

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1407

Figure 1. The analytical model of the structure with MTMD.

S = 0:02 is selected for the steel-structure buildings according to the Chinese Code-Technical
speci cation for steel structure of tall buildings.
The major objective of this paper is to study the optimum parameters of MTMD for the purpose
of the seismic response mitigation of structures. The parameters are, respectively, the average
damping ratio, frequency spacing, mass ratio, total number of MTMD. In addition, to enhance the
understanding of MTMD behaviour, a study is performed to investigate the interrelation among the
design parameters of MTMD. The e ectiveness and the key optimum parameters of the MTMD
is also discussed in comparison with that of the MTMD(II) and a single TMD to give a reference
to select the MTMD.

2. THE FREQUENCY RESPONSE FUNCTION OF THE STRUCTURE WITH MTMD

In the present paper, MTMD is considered for the control of the speci c vibration mode of a
structure, and the analytical model shown in Figure 1, in which the structure is modeled as a
single-degree-of-freedom (SDOF) system (i.e. the main system or the mode-generalized system
in the mode being controlled) using mode reduced-order method, is adopted. The main system is
characterized by the mode generalized sti ness ks , damping coecient cs and mass ms , respectively.
Each TMD is also modelled as a SDOF system which is set to have di erent dynamic character-
istics. Especially, the natural frequencies of the MTMD are to be equally distributed around the
natural frequency corresponding to the structural vibration mode to be mitigated. As a result, the
total degrees of freedom of this structural system is n + 1. The analysis that follows is based on
this simpli ed system. The equations of motion for the analytical model under ground acceleration
excitation given in Figure 1 can be, respectively, expressed as

ms y s + cs ẏs + ks ys = −ms u g + fMTMD (t) (1)

P
n
fMTMD (t) = [cj (ẏj − ẏs ) + kj (yj − ys )] (2)
j=1

mj (yj − y s ) + cj ( ẏj − ẏs ) + kj (yj − ys ) = −mj y s − mj ug ; j = 1; 2; : : : ; n (3)

in which ys is the displacement of the main system with respect to the ground, yj (j = 1; 2; : : : ; n),
is the displacement of the jth TMD in the MTMD with regard to the ground, ug is the ground

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1408 C. LI

acceleration, mj ; cj and kj represent the mass, damping coecient and sti ness of the jth TMD
in the MTMD, respectively.
Using Laplace transformation, Equation (2) provides a frequency-domain expression FMTMD (s)
in the form
P
n
FMTMD (s) = {(cj s + kj )[Yj (s) − Ys (s)]} (4)
j=1

Through Laplace transformation and with the help of Equation (4), the expression for Equation
(1) in frequency domain are developed as under
P
n
(s2 ms + scs + ks )Ys (s) = −ms Ug (s) + {(c j s + kj )[Yj (s) − Ys (s)]} (5)
j=1

Similarly, after Laplace transformation, the frequency-domain form of Equation (3) can be
written as
s2 mj mj
Yj (s) − Ys (s) = − 2
Ys (s) − 2 Ug (s) (6)
s mj + scj + kj s mj + scj + kj
The transfer function of the structure with MTMD is then obtained by substituting Equation (6)
into Equation (5) to yield
Pn
Ys (s) ms + j=1 mj (cj s + kj )=(s2 mj + scj + kj )
Hys (s) = =− 2 Pn (7)
Ug (s) s ms + scs + ks + j=1 (cj s + kj )mj s2 =(s2 mj + scj + kj )

For generality and convenience, the following parameters are introduced:

ks cs kj cj
!s2 = ; s = ; !j2 = ; j =
ms 2ms !s mj 2mj !j
√
By substituting s = −i! in Equation (7), where i = −1, the frequency response function of
the structure with MTMD can be represented as
Pn
1 + j=1 (mj =ms )(!j2 − i2j !j !)=(!j2 − !2 − i2j !j !)
Hys (−i!) = − 2 Pn (8)
!s − !2 − i2s !s ! − !2 j=1 (mj =ms )(!j2 − i2j !j !)=(!j2 − !2 − i2j !j !)

3. DYNAMIC MAGNIFICATION FACTOR (DMF) OF THE MAIN SYSTEM WITH MTMD

For the simpli cation of the manufacturing of MTMD, it is assumed that the sti ness and damping
coecient of each TMD in the MTMD are same and the natural frequencies of the MTMD are
uniformly distributed. As a result, the MTMD is made by keeping the sti ness, damping constant
and mass variation (i.e. k1 = k2 = · · · = kn = kT ; c1 = c2 = · · · = P
cn = cT ; m1 6= m2 6= · · · =
6 mn ).
n
Let !T be the average frequency of the MTMD (i.e. !T = !
k=1 k =n) and n be the total
number of the MTMD. The increment of the natural frequency of the jth TMD can be given as
 
n+1

!j = !T j − (9)
2 n−1

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1409

where the non-dimensional parameter is the frequency spacing of the MTMD de ned as
!n − !1
= (10)
!T
The natural frequency of the jth TMD is then expressed as
   
n+1
!j = !T +
!j = !T 1 + j − (11)
2 n−1
In the present paper, without any loss of generality, let the average natural frequency of the
MTMD equal to the natural frequency of the structural vibration mode to be suppressed (i.e.
!T = !s ). Then, the ratio of the natural frequency of the jth TMD to the controlled natural
frequency of the structure can be written as
 
!j n+1
rj = =1+ j− (12)
!s 2 n−1
According to the above assumption, the mass and damping ratio of the jth TMD in the MTMD
is, respectively, expressed as
kT
mj = (13)
!j2
cT cT
j = = !j (14)
2mj !j 2kT
The average damping ratio of the MTMD is de ned as
Pn 
j
T = (15)
j=1 n

After some manipulations, Equation (14) leads to the following equation:

Pn 
j cT Pn !
j cT
= = !T (16)
j=1 n 2kT j=1 n 2kT

From Equations (15) and (16), along with the assumption of !T = !s , the expression for the
average damping ratio of the MTMD may be given as
cT
T = !s (17)
2kT
Using Equations (14) and (17), the damping ratio of the jth TMD in the MTMD can be
determined by
j = rj T (18)
The ratio of the total mass of MTMD to the mass of the main system (i.e. the generalized mass
of the structure in the mode that needs to be controlled) is referred to as the mass ratio, i.e.
Pn
j=1 mj
= (19)
ms

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1410 C. LI

The constant sti ness required for each TMD can be evaluated as
ms
kT = Pn 2
(20)
j=1 1=!j

This equation may be rewritten as

kT  
= Pn 2 =!2
= Pn 2
(21)
ms !s2 !
j=1 s j j=1 1=rj

Then, the frequency response function of the structure with MTMD can be represented explicitly
by substituting Equations (12)–(21) into Equation (8) as
!
 Pn 2 2 2
1 + Pn 2 j=1 (1 − i2T )=(rj −  − i2T rj )
j=1 1=r j
!s2 Hys (−i!) = − ! (22)
2 P
  n 2 2
1 − 2 − i2s  − Pn 2
2
j=1 (1 − i2T )=(rj −  − i2T rj )
j=1 1=rj

in which  = !=!s is the ratio of the external excitation frequency to the controlled natural fre-
quency of the structure.
Here the dynamic magni cation factor (DMF) of the main system (i.e. the mode-generalized
system in the mode to be controlled) with MTMD is explicitly de ned as

DMF = |!s2 Hys (−i!)| (23)

Substituting Equation (22) into Equation (23), the dynamic magni cation factor (DMF) can be
computed by
s
R 2e () + Im2 ()
DMF = (24)
R2e () + Im2 ()

in which

!
 Pn (r 2 − 2 ) + 42 2 r 2
j T j
R e () = 1 + Pn 2 2 − 2 )2 + 42 2 r 4
j=1 1=rj (r
j=1 j T j
!
 P
n 23 T
Im () = Pn 2 2
j=1 1=rj j=1 (rj − 2 )2 + 42 2T rj4
!
2  Pn (r 2 − 2 ) + 42 2 r 2
2 j T j
Re () = 1 −  − Pn 2 2 − 2 )2 + 42 2 r 4
j=1 1=rj (r
j=1 j T j
" ! #
2  P
n 23 T
Im () = − 2s  + Pn 2 2 2 2 2 2 4
j=1 1=rj j=1 (rj −  ) + 4 T rj

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1411

4. RESEARCH ON THE OPTIMUM PARAMETERS AND ROBUSTNESS OF MTMD

The minimization of the maximum value of the dynamic magni cation factor of the structure with
MTMD (i.e. Min.Max.DMF) is used as the criterion for searching for the optimum parameters of
the MTMD. Then, it can be said that the MTMD which gives the value of Min.Max.DMF is the
optimum MTMD. It is also indicated that the optimim MTMD which gives the smaller value of
Min.Max.DMF is a more e ective MTMD.

4.1. Search for the optimum parameters of MTMD in the case of assuming the di erent
values of the average damping ratio T
Using the criterion of the optimization procedure selected, based on the analytically calculated
results, the optimum frequency spacing opt and the values of Min.Max.DMF corresponding to the
given values of the average damping ratio T are illustrated in Table I with the di erent values
of the mass ratio and total numbers of the MTMD. Figures 2 and 3 depict the relation curves
between Max.DMF and the frequency spacing under the condition of assuming the di erent
values of the average damping ratio for the di erent values of the mass ratio and total numbers
of the MTMD. The ordinate and abscissa in each gure are the maximum dynamic magni cation
factor (Max.DMF) and the frequency spacing ( ), respectively.
From the trends and the results as shown in Table I and Figures 2 and 3, the following con-
clusions or suggestions may be drawn:
(1) For a given value of mass ratio of MTMD, there exist the optimum values of frequency
spacing and average damping ratio with respect to the MTMD for the di erent total numbers. That
is, there exists an optimum MTMD, which provides the value of Min.Max.DMF, for a given mass
ratio and total number.
(2) For a given value of mass ratio of MTMD, increasing the total number of the MTMD
increases the value of the optimum frequency spacing. That is, the natural frequencies of the
MTMD are uniformly distributed widely in the vicinity of the controlled frequency of the structure.
In other words, the MTMD with a large number of TMDs can promote its robustness.
(3) For a given value of mass ratio of MTMD, with the increasing of the total number n, the
values of the optimum average damping ratio and Min.Max.DMF decrease (that is, the e ectiveness
of MTMD becomes better). However, when the total number of the MTMD is equal to or beyond
11, with the exception of the case of  = 0:01 and n = 21, the reduction of the optimum average
damping ratio tends to saturate, and the increase in the optimum frequency spacing remains almost
invariant. It is also noted that, the decrease in the value of Min.Max.DMF is not signi cant. These
mean that, the improvement in the e ectiveness and robustness of the MTMD is not signi cant in
this case. Consequently, when designing the MTMD, the total number less than or equal to 11 is
usually considered.
(4) For a given total number of MTMD, with the increasing of the value of mass ratio, the
optimum values of the frequency spacing and average damping ratio increase but the value of
Min.Max.DMF decrease. That is, the e ectiveness and robustness become better.
(5) If the total number of MTMD is equal to or beyond 11, with increase of the mass ratio,
say  = 0:03 when the value of frequency spacing is set to be larger than the optimum one, the
increase of Max.DMF is not signi cant. It is also indicated that setting a smaller value of frequency
spacing than the optimum one, the increase of Max.DMF is signi cant.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1412 C. LI

Figure 2. E ectiveness of MTMD in Max.DMF versus frequency spacing for  = 0:01 and di erent average
damping ratios: (a) 3TMD, (b) 5TMD, (c) 9TMD, (d) 11TMD, (e) 13TMD and (f) 21TMD.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1413

Figure 3. E ectiveness of MTMD in Max.DMF versus frequency spacing for  = 0:03 and di erent average
damping ratios: (a) 3TMD, (b) 5TMD, (c) 9TMD, (d) 11TMD, (e) 13TMD and (f) 21TMD.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1414 C. LI

Table I. The optimum parameters of MTMD corresponding to the given values of the average damping ratio T .

The total The total mass ratio  = 0:01 The total mass ratio  = 0:03
number
of The average The The average The
MTMD damping ratio Min.Max.DMF frequency damping ratio Min.Max.DMF frequency
n T spacing T spacing

3 0.01 12.80 0.12 0.03 7.90 0.20

0.02 9.95 0.11 0.04 7.14 0.19
0.03 9.41 0.11 0.05 6.68 0.19
0.04 9.25 0.10 0.06 6.54 0.18
0.05 9.53 0.09 0.07 6.60 0.17
0.06 10.00 0.08 0.08 6.68 0.17
5 0.01 10.54 0.15 0.02 7.34 0.25
0.02 8.91 0.14 0.03 6.56 0.25
0.03 8.88 0.13 0.04 6.25 0.24
0.04 9.17 0.12 0.05 6.22 0.23
0.05 9.58 0.11 0.06 6.30 0.21
0.06 10.02 0.10 0.07 6.48 0.21
9 0.01 9.36 0.17 0.01 7.78 0.29
0.02 8.68 0.15 0.02 6.23 0.29
0.03 8.79 0.14 0.03 6.07 0.28
0.04 9.20 0.13 0.04 6.01 0.26
0.05 9.61 0.12 0.05 6.14 0.25
0.06 10.03 0.11 0.06 6.30 0.24
11 0.005 10.47 0.18 0.01 7.07 0.31
0.01 8.79 0.17 0.02 6.09 0.29
0.02 8.71 0.16 0.03 6.08 0.29
0.03 8.93 0.15 0.04 6.06 0.27
0.04 9.24 0.14 0.05 6.10 0.25
0.05 9.63 0.13 0.06 6.29 0.24
13 0.005 9.72 0.19 0.01 6.76 0.31
0.01 8.87 0.18 0.02 6.02 0.30
0.02 8.64 0.16 0.03 5.91 0.28
0.03 8.89 0.15 0.04 6.01 0.27
0.04 9.22 0.14 0.05 6.14 0.26
0.05 9.63 0.13 0.06 6.30 0.25
21 0.0005 7.87 0.20 0.01 6.18 0.33
0.001 7.84 0.20 0.02 5.96 0.30
0.005 8.45 0.20 0.03 5.92 0.29
0.01 8.61 0.18 0.04 6.03 0.27
0.02 8.65 0.16 0.05 6.12 0.26
0.03 8.80 0.15 0.06 6.31 0.25

4.2. Search for the optimum parameters of MTMD in the case of assuming the di erent
values of the frequency spacing
Based on the criterion selected, using a numerical searching technique, the optimum average damp-
ing ratio T opt and the values of Min.Max.DMF corresponding to the given values of the frequency

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1415

Table II. The optimum parameters of MTMD corresponding to the given values of the frequency spacing .
The total The total mass ratio  = 0:01 The total mass ratio  = 0:03
number
of The The average The The average
MTMD frequency Min.Max.DMF damping ratio frequency Min.Max.DMF damping ratio
n spacing T spacing T
3 0.14 10.43 0.036 0.22 7.16 0.062
0.12 9.78 0.033 0.20 6.82 0.060
0.10 9.18 0.037 0.18 6.50 0.062
0.08 9.84 0.046 0.16 6.75 0.080
0.06 10.31 0.059 0.14 7.17 0.089
5 0.17 9.94 0.027 0.27 6.74 0.042
0.15 9.29 0.023 0.25 6.41 0.039
0.13 8.75 0.027 0.23 6.14 0.045
0.11 9.35 0.041 0.21 6.30 0.060
0.09 9.94 0.054 0.19 6.70 0.072
9 0.19 9.66 0.017 0.30 6.33 0.026
0.17 9.04 0.015 0.28 6.01 0.025
0.15 8.62 0.022 0.26 5.97 0.038
0.13 9.13 0.035 0.24 6.16 0.054
0.11 9.68 0.043 0.22 6.53 0.066
11 0.20 9.69 0.015 0.31 6.28 0.022
0.18 9.08 0.013 0.29 5.99 0.022
0.16 8.57 0.016 0.27 5.92 0.033
0.14 8.89 0.031 0.25 6.09 0.049
0.12 9.48 0.041 0.23 6.45 0.059
13 0.20 9.47 0.012 0.32 6.31 0.021
0.18 8.85 0.011 0.30 6.02 0.019
0.16 8.50 0.017 0.28 5.91 0.030
0.14 8.96 0.031 0.26 6.01 0.044
0.12 9.53 0.041 0.24 6.33 0.057
21 0.24 9.72 0.006 0.33 6.12 0.013
0.22 8.46 0.004 0.31 5.92 0.017
0.20 7.69 0.003 0.29 5.88 0.028
0.18 8.51 0.007 0.27 5.99 0.042
0.16 8.60 0.021 0.25 6.29 0.056

spacing are shown in Table II with the di erent mass ratio and di erent total numbers of the
MTMD. Figures 4 and 5 plot the interrelation curves between Max.DMF and the average damping
ratio T for the di erent values of the average damping ratio, mass ratio and total numbers of the
MTMD.
The major conclusions obtained through Table II and Figures 4 and 5 are as follows:
(1) For a given mass ratio, there exist the optimum frequency spacing and average damping
ratio with respect to the di erent total numbers selected. That is, the optimum design of MTMD
may be performed for a given mass ratio and total number. This conclusion is the same as that
in the above (see Section 4.1).

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1416 C. LI

Figure 4. E ect of average damping ratio on e ectiveness of MTMD with  = 0:01 and di erent frequency
spacings: (a) 3TMD, (b) 5TMD, (c) 11TMD.

(2) For a given mass ratio, with increase of the total number of MTMD, the optimum average
damping ratio and Min.Max.DMF keeps decreasing and the optimum frequency spacing increases.
This trends in decreasing or increasing are the same as those in the above (see Section 4.1).
(3) When the value of the average damping ratio is set to be larger than the optimum one,
the increase in Max.DMF is not signi cant. It is also indicated that setting a smaller value of the

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1417

Figure 5. E ect of average damping ratio on e ectiveness of MTMD with  = 0:03 and di erent frequency
spacings: (a) 3TMD, (b) 5TMD, (c) 11TMD .

average damping ratio than the optimum one, the increase of Max.DMF is signi cant. Especially,
when the value of the average damping ratio is lower than the optimum one for all cases, the
value in Max.DMF is increase rapidly and signi cantly with the further reduction of the average
damping ratio.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1418 C. LI

Figure 6. E ect of average damping ratio on e ectiveness of MTMD with each set of the total number and
the optimum frequency spacing: (a)  = 0:01 and (b)  = 0:03.

(4) The results presented in Tables I and II show that the optimum parameters of the MTMD
(i.e. T opt and opt ) and the values of Min.Max.DMF, which are obtained through the given values
of the average damping ratio and frequency spacing, respectively, are almost same.

4.3. The sensitivity of Max.DMF to the average damping ratio T

Figure 6 shows the variation of Max.DMF against the average damping ratio T for each set of
the total number and the optimum frequency spacing under the condition of the di erent values
of mass ratio.
From Figure 6, it can be seen that as the total number of the MTMD increases, Max.DMF
decreases. This reduction indicates that the e ectiveness of the MTMD gets better. However, it
can be very clearly seen from Figure 6 that the reduction of Max.DMF remains almost invariant
when the total number of the MTMD is equal to or beyond 11 with the exception of the case
 = 0:01 and n = 21. It is also noted that when the average damping ratio is set to be smaller
than the optimum value, Max.DMF is increased rapidly and signi cantly with the decrease of the
average damping ratio from the optimum value, as shown in Figure 6. That is, under this condition,
Max.DMF is very sensitive to the average damping ratio. On the contrary, if the damping ratio is
xed to be larger than the optimum value, with the increase of the average damping ratio from the
optimum one, the change in Max.DMF is little. That is, in this case, Max.DMF is not sensitive
to the average damping ratio.

4.4. The sensitivity of Max.DMF to the frequency spacing

Figure 7 depicts the correlation curves between Max.DMF and the frequency spacing for each
set of the total number and the optimum average damping ratio with the di erent values of mass
ratio.
As is clearly shown in Figure 7, the frequency spacing of the MTMD signi cantly in uences
the e ectiveness of the MTMD. There exists an optimum value of the frequency spacing which
provides maximum e ectiveness of the MTMD (that is, there exists Min.Max.DMF) for each
set of the total number and the optimum average damping ratio for a given mass ratio. This
optimum value of the frequency spacing increases with the increase of the total number and

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1419

Figure 7. E ectiveness of MTMD in Max.DMF versus frequency spacing for each set of the total number and
the optimum average damping ratio: (a)  = 0:01 and (b)  = 0:03.

mass ratio of the MTMD. Figure 7 suggests that the value of the optimum frequency spacing
nearly keeps invariant when the total number of the MTMD is equal to or beyond 11 with the
exception of the case of  = 0:01 and n = 21. The mass ratio has no in uence on this trend.
Figure 7 also shows that when the frequency spacing is set to be smaller than the optimum value,
with increase of the total number of the MTMD, Max.DMF is increased and this means that the
e ectiveness of the MTMD with the large values of the total number is sensitive to the frequency
spacing. On the contrary, when the frequency ratio is xed to be larger than the optimum value,
with increase of the total number, Max.DMF decreases but this decrease is not signi cant (with
the exception of the case of  = 0:01 and n = 21), as depicted in Figure 7. Therefore, it can be
said that the frequency spacing has little in uence on the e ectiveness of the MTMD in this case.

5. COMPARISON OF THE EFFECTIVENESS AND THE KEY PARAMETERS WITH

THE OPTIMUM MTMD(II) AND SINGLE TMD

In this section, the e ectiveness and the key parameters of the optimum MTMD is discussed in
comparison with the optimum MTMD(II) and single TMD. Based on the above criteria de ned
for the optimality, the optimum parameters of MTMD, MTMD(II) and TMD and corresponding
Min.Max.DMF for di erent total numbers of MTMD and MTMD(II) and the di erent mass ratios
are obtained through numerical analysis and shown in Table III. It should be noted that the mass
ratio and the damping ratio of each TMD in the MTMD(II) are set to be constant (i.e. T and
T ). The MTMD(II) is manufactured by varying the sti ness, damping coecient and keeping the
mass constant (i.e. k1 6= k2 6= · · · =
6 kn ; c1 6= c2 6= · · · =
6 cn ; m1 = m2 = · · · = mn ). Therefore, the
mass ratio  and the damping ratio T represents, respectively, the total mass ratio (i.e.  = nT )
and the damping ratio of each TMD in the MTMD(II). For the MTMD(II), the average damping
ratio is equal to the damping ratio of each TMD. Tuning frequency ratio f indicates the ratio of
the natural frequency of TMD to the controlled natural frequency of the structure. For the sake of
comparison, tuning frequency ratio f is taken as 1. As can be seen from Table III, the optimum

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
1420 C. LI

Table III. The e ectiveness and the key parameters of the optimum MTMD, MTMD(II) and single TMD.

MTMD MTMD(II) TMD

 n opt T opt Min.Max.DMF opt T opt Min.Max.DMF f T opt Min.Max.DMF

0.01 1 1.00 0.074 10.94

3 0.10 0.04 9.25 0.10 0.04 9.63
5 0.13 0.03 8.88 0.13 0.02 9.30
11 0.16 0.02 8.71 0.15 0.02 8.94
0.03 1 1.00 0.114 8.37
3 0.18 0.06 6.54 0.18 0.05 7.06
5 0.23 0.05 6.22 0.23 0.04 6.84
11 0.27 0.04 6.06 0.27 0.02 6.42

frequency spacing of the MTMD is the same as that of the MTMD(II) and the optimum damping
ratio are much smaller for the MTMD and MTMD(II) than for a single TMD, which can be one
of the advantages of the MTMD and MTMD(II). Table III also demonstrates that the optimum
MTMD is more e ective than the optimum MTMD(II) and the optimum single TMD and the
optimum average damping ratio is a little larger for the MTMD than for the MTMD(II).

6. CONCLUSIONS

Using numerical searching technique in two directions, the optimum parameters of the MTMD for
the damped steel structure, whose damping ratio is taken as 0.02 according to the Chinese Code-
Technical speci cation for steel structure of tall buildings for the convenience of the engineering
design and use, are investigated in detail. The criterion selected for the optimum searching is the
minimization of the maximum value of the dynamic magni cation factor (DMF) of the structure
with the MTMD under the ground acceleration. It is demonstrated that (1) the optimum average
damping ratio of the MTMD decreases with the increase of the total number of the MTMD and
increases with the increase of the mass ratio, (2) the optimum frequency spacing of the MTMD
increases with the increase of both the total number and mass ratio, (3) with the increase in the total
number of the MTMD beyond a certain value (in this case n = 11), the key optimum parameters
(i.e. opt and T opt ) and Min.Max.DMF remain almost invariant, (4) the optimum parameters and
Min.Max.DMF obtained by searching in two directions are nearly same, (5) the sensitivity of
Max.DMF to the key parameters (i.e. the frequency spacing and the average damping ratio T )
are studied to give a further reference to the parameter selection. It is clari ed that the optimum
frequency spacing of the MTMD is the same as that of the MTMD(II) and the optimum average
damping ratio of the MTMD is a little larger than that of the MTMD(II). It is also shown that the
optimum MTMD is more e ective than the optimum MTMD(II) and the optimum single TMD
with equal mass. Hence, the MTMD is preferably suitable for attenuating undesirable oscillations
of structures under the ground acceleration.

ACKNOWLEDGEMENTS

The writer gratefully acknowledges the nancial support provided by China Postdoctoral Science Foundation.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421
 PERFORMANCE OF MULTIPLE TUNED MASS DAMPERS 1421

REFERENCES
1. Brock JE. A note on the damped vibration absorber. Journal of Applied Mechanics ASME 1946; 13: A-284.
2. Den Hartog JP. Mechanical Vibrations (4th edn). McGraw Hill: New York, 1956.
3. Fujino Y, Abe M. Design formulas for tuned mass dampers based on a perturbation technique. Earthquake Engineering
and Structural Dynamics 1993; 22: 833 – 854.
4. Thompson AG. Optimum tuning and damping of a dynamic vibration absorber applied to a force excited and damped
primary system. Journal of Sound and Vibration 1981; 77: 403– 415.
5. Tsai HC, Lin GC. Optimum tuned mass dampers for minimizing steady-state response of support excited and damped
system. Earthquake Engineering and Structural Dynamics 1993; 22: 957– 973.
6. Warburton GB. Optimum absorber parameters for various combinations of response and excitation parameters.
Earthquake Engineering and Structural Dynamics 1982; 10: 381– 401.
7. Warburton GB, Ayorinde EO. Optimum absorber parameters for simple systems. Earthquake Engineering and
Structural Dynamics 1980; 8: 197–217.
8. Sadek F, Mohraz B, Taylor AW, Chung RM. A method of estimating the parameters of tuned mass dampers for
seismic applications. Earthquake Engineering and Structural Dynamics 1997; 26: 617– 635.
9. Chunxiang Li. TMD control research and design of tall buildings subjected to seismic excitation. Journal of Shanghai
Jiaotong University 1999; 33: 746 –749 (in Chinese).
10. Iwanami K, Seto K. Optimum design of dual tuned mass dampers and their e ectiveness. Proceedings of the JSME(C)
1984; 50(449): 44 –52 (in Japanese).
11. Xu K, Igusa T. Dynamic characteristics of multiple substructures with closely spaced frequencies. Earthquake
Engineering and Structural Dynamics 1992; 21: 1059 –1070.
12. Igusa T, Xu K. Vibration control using multiple tuned mass damper. Journal of Sound and Vibration 1994; 175:
491–503.
13. Yamaguchi H, Harnpornchai N. Fundamental characteristics of multiple tuned mass dampers for suppressing
harmonically forced oscillations. Earthquake Engineering and Structural Dynamics 1993; 22: 51– 62.
14. Abe M, Fujino Y. Dynamic characterization of multiple tuned mass dampers and some design formulas. Earthquake
Engineering and Structural Dynamics 1994; 23: 813– 835.
15. Abe M, Igusa T. Tuned mass dampers for structures with closely spaced natural frequencies. Earthquake Engineering
and Structural Dynamics 1995; 24: 247–261.
16. Kareem A, Kline S. Performance of multiple mass dampers under random loading. Journal of Structural
Engineering ASCE 1995; 121: 348 –361.
17. Jangid RS. Dynamic characteristics of structures with multiple tuned mass dampers. Journal of Structural Engineering
and Mechanics 1995; 3: 497–509.
18. Chunxiang Li. Research on the optimum theories and methods for wind and earthquake resistant control design of
high-rise steel structures. Ph.D. Thesis, Tongji University, Department of Civil Engineering, Shanghai, 200092, 1998.

Copyright ? 2000 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2000; 29:1405–1421

View publication stats