Available online at www.sciencedirect.

com

Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

www.elsevier.com/locate/cnsns

Dynamic performance analysis of non-linear tuned

vibration absorbers
Jeong-Hoi Koo
a

a,*

, Amit Shukla a, Mehdi Ahmadian

Department of Mechanical and Manufacturing Engineering, Miami University, Oxford, OH 45056, USA
b
Department of Mechanical Engineering, Virginia Tech, Blacksburg, VA 24061, USA
Received 25 February 2007; received in revised form 20 March 2007; accepted 21 March 2007
Available online 3 April 2007

Abstract
The main purpose of this study is to evaluate the dynamic characteristics of a damping controlled semi-active tuned
vibration absorber (TVA). A base-excited, single-degree-of-freedom structure coupled with a TVA model is adopted as
the baseline model for our analysis. A non-model based groundhook control (displacement-based ono control or
ono DBG) was used to control the damping of the TVA. After developing numerical models of a non-linear TVA
along with the ono DBG control and a passive TVA, the optimal tuning parameters of both TVA models were obtained
using an optimization routine. Using the two optimally tuned models, parametric studies were preformed by varying the
values of both on and o-state damping to evaluate the dynamic performances of the TVAs using the peak transmissibility
criteria. The results showed that the peak transmissibility of the semi-active TVA is nearly 21% lower than that of the passive, indicating that the semi-active TVA is more eective in reducing vibration levels. The results further showed that
increases in o-state damping decreases the eectiveness of semi-active TVA in reducing maximum vibration levels.
2007 Elsevier B.V. All rights reserved.
PACS: 05.45.a
Keywords: Tuned vibration absorber; Semi-active control; Optimal tuning; Controllable damping

1. Introduction
Since their invention in the early 1900s, Tuned Vibration Absorbers (TVAs) have been eective engineering
devices that suppress vibrations of machines and structures. A TVA is a vibratory subsystem, normally consisting of a mass, a spring, and a damper, that is mounted on a primary system. The TVA counteracts the
motions of the primary system, absorbing the primary structures vibrations. However, a conventional passive TVA is only eective when it is tuned properly; otherwise, it can magnify the vibration levels of the primary system, hence, the name tuned vibration absorber. In many practical applications, the o-tuning of a
*

Corresponding author. Tel.: +1 513 529 0723; fax: +1 513 529 0717.
E-mail address: koo@muohio.edu (J.-H. Koo).

1007-5704/$ - see front matter 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.cnsns.2007.03.020

1930

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

Fig. 1. Conventional passive TVA model (base-excited model).

TVA often occurs because of system parameter changes over time. To cope with these problems, extensive
studies over the past two decades have been performed to develop new designs and concepts for passive TVAs.
These designs include adaptive, semi-active, and active TVAs. A comprehensive review of congurations,
developments, and applications of TVAs can be found in the papers by Sun et al. [1], Housner et al. [2],
Symans and Constaninou [3], and Soong and Spencer [4].
Fig. 1 shows a conventional TVA model. In this model, the primary structure is coupled with a damped
TVA, and the mass of the structure and absorber are dened by m1 and m2, with their corresponding displacements as x1 and x2, respectively. The absorbers spring (k2) and damper (c2) are mounted on the structure. The
stiness and damping of the structure are represented by k1 and c1, respectively.
In an eort to enhance the performance of the traditional TVAs, this study considers a semi-active TVA by
controlling its damping element. The next section provides an overview of a semi-active TVA, including a nonmodel based control method. After introducing system parameters, an optimal tuning technique is explained.
Following the optimal tuning, parametric studies of the semi-active and equivalent passive TVA models will
be discussed. Finally, a detailed analysis and comparison of the simulation results conclude the work.
2. Semi-active tuned vibration absorbers
The semi-active TVA model (Fig. 2b) replaces a passive damping element with a controllable damping element, which distinguishes it from the conventional passive system (Fig. 2a). A controllable damping element,
such as a magnetorheologoical damper, is the key element for the semi-active TVA.
Fig. 2b is used to derive the dynamic equations of motion for the semi-active model. The equations of
motion that describe this system are

  
  
  

x1
m1 0
x_ 1
x1
c1 c2 c2
k 1 k 2 k 2
c1 x_ in k 1 xin

:
1
x2
0 m2
c2
c2
k 2
k2
x2
0
x_ 2

Fig. 2. Passive TVA versus semi-active TVA; (a) passive model and (b) semi-active model.

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

1931

Assuming
x1 X 1 est ;

2a

st

x2 X 2 e ;
xin X in est :

2b
2c

where s = jx and x is the driving frequency. Substituting Eqs. (2a)(2c) into Eq. (1) yields

  

X1
c1 s k 1 X in
m1 s2 c1 c2 s k 1 k 2
c2 s  k 2

:
X2
0
c2 s  k 2
m2 s2 c2 s k 2
Using Cramers Rule, the amplitudes X1 and X2 can be solved for as


 c1 s k 1 X in
c2 s  k 2 


0
m2 s2 c2 s k 2  c1 s k 1 X in m2 s2 c2 s k 2

;
X1
det A
det A


 m1 s2 c2 s k 1 k 2 c1 s k 1 X in 



 c s k X c s k
c2 s  k 2
0
1
1
in 2
2
;
X2

det A
det A

where detA = (m1s2 + c1s + c2s + k1 + k2)(m2s2 + c2s + k2)  (c2s + k2)2.
Therefore, the transmissibility equations become
X1
c1 s k 1 m2 s2 c2 s k 2

;
X in m1 s2 c1 c2 s k 1 k 2 m2 s2 c2 s k 2  c2 s k 2 2
X2
c1 s k 1 c2 s k 2

:
X in m1 s2 c1 c2 s k 1 k 2 m2 s2 c2 s k 2  c2 s k 2 2

6
7

The above transmissibility equations will be used in the numerical simulation study of the semi-active TVA
in a later section.
Fig. 3 illustrates how a controllable damper can provide a wide range of damping force. At a given velocity,
Vg, the corresponding damper force for the passive damper is a constant force, Fpa. On the other hand, the
controllable damper oers a damper force ranging from Fo to Fon. This controllable damper, which provides
a wide dynamic force range, can signicantly improve the dynamic performance of the semi-active TVA with
proper control methods.
To eectively control this semi-active damping, this study uses the displacement-based, ono groundhook
(ono DBG). This is because it was identied as a suitable control policy in the previous simulation work by
the authors [5]. The ono DBG control adjusts the damping level of the semi-active damping element based

Fig. 3. Semi-active damping.

1932

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

on the product of the primary systems displacement and the relative velocity across the damper. The ono
DBG groundhook control policy can be written as
x1 v12 P 0; ccontrollable con ;
x1 v12 < 0; ccontrollable coff ;

where x1 is the displacement of the structure and v12 represents the relative velocity across the damper.
3. Optimal tuning of the TVAs
This section presents the simulation parameters and the optimal tuning of simulation models. The tuned
systems provide for an equal comparison of system performance. The simulation models are tuned by using
Table 1
System parameters
Parameter

Value

Structure mass (m1)

TVA mass (m2)
Structure stiness (k1)
Structure damping ratio (f1)

260.8 kg
14.5 kg
334259 N/m
3%

Table 2
Numerical parameters used in Simulink models
Simulation input

Initial frequency

Target frequency

Simulation time

Equation solver

Fixed-step size

Chirp

0.5 Hz

10 Hz

150 s

RungeKutta

1/100

Fig. 4. Flow chart for optimization routine.

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

1933

a numerical optimization technique, which is necessary for the models because of the presence of damping in
the primary system and the non-linearity of semi-active control. The goal of the optimization is to nd the
TVA parameters, such as on/o-damping ratios (fon and foff ), and stiness (k2), which generate the best performance of each model. The optimization routine uses a minimization of the maximum value of transmissibility in a frequency range of interest as an objective function.
3.1. System parameters
The simulation models are developed by coupling the controller and the mathematical models that were
discussed earlier. The actual realization of these simulation models is done with Matlab. The Simulink toolbox
within Matlab is used to build block diagrams for each of these simulation models. Table 1 shows the system
parameters of simulations, and Table 2 shows a summary of the simulation parameters used in the Simulink.
3.2. Optimization routine
Fig. 4 shows the ow chart for the optimization technique. The execution of the optimization routine
involves three steps. In the main program, the system parameters are dened, along with the initial values
and ranges of the simulation parameters. The main program calls the Simulink numerical models, which
are responsible for generating the peak transmissibilities. These peak transmissibilities are then sent to the
optimizer, fmincon.m, where the minimum values of the peak transmissibilities are returned, along with
the corresponding simulation parameters. Table 3 summarizes the initial values and parameter ranges for this
optimization.
Table 3
Summary of initial values and parameter ranges
Model

Parameter

Initial value

Parameter range

Passive

Damping ratio
TVA stiness (N/m)

0.1
13 185

0:1 6 fpa 6 0:7

10 000 6 k a pa 6 100 000

Ono DBG

On-state damping ratio

O-state damping ratio
TVA stiness (N/m)

0.7
0.07
12 449.6

0:1 6 fon od 6 0:7

0:01 6 foff od 6 0:07
10 000 6 k a od 6 100 000

Fig. 5. Optimization results for baseline models.

1934

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

Fig. 5 shows the transmissibility plots of the optimally tuned TVAs. The ono DBG controlled semi-active
TVA outperforms the passive system in reducing the peak transmissibility. The semi-active TVA achieved
nearly 21% of more reduction of the maximum vibrations levels over the passive TVA. Table 4 shows a summary of the values of baseline parameters. These results will serve as the baseline values for parametric studies
in the subsequent sections.
4. Dynamic characteristics of the TVAs
This section contains the results of the parametric studies performed on the baseline numerical models.
These parametric studies provide an understanding of the dynamics of the TVAs as their parameters change.
It uses transmissibility and phase plots to evaluate each TVAs performance. The phase angle analysis adds
valuable explanations in analyzing the results.
4.1. Performance analysis of the passive TVA
Fig. 6a shows the transmissibility, or the ratio between the output and the input displacement of the primary structure, for a passive TVA as the damping ratio (f2) changes from 0.0 to 0.9. When the damping ratio
Table 4
Summary of optimal parameters for baseline models
Parameters

Passive

Semi-active

TVA stiness (N/m)

On-state damping ratio
O-state damping ratio
Passive reduction (%)

13 185
0.145
N/A
N/A

12 397
0.7
0.07
20.96

Fig. 6. Eect of damping ratio on passive TVA dynamic performance: (a) transmissibility (X 1 =X in ); (b) phase angles between the TVA and
the structure.

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

1935

is 0.0, there are two large peaks, and a complete isolation occurs at the valley. Increasing the damping ratio to
its baseline value lowers these resonant peaks and widens the valley between the two peaks. However, this has
a negative eect; it raises the valley oor. Further increasing the damping ratio above its baseline value gradually causes the two peaks become a single peak. This means that the structure and the TVA become strongly
coupled, and function nearly as a single mass, eectively negating any benets of the TVA.
Fig. 6b shows the phase angles between the TVA mass and the structure mass as the damping ratio
increases. The phase plots support the above discussion. When the damping ratio is at its baseline value,
the phase angle around the tuned frequency is relatively close to 90; the TVA eectively counteracts the
motions of the structure mass, achieving the minimum transmissibility over the entire frequency range of interest. However, as the damping ratio increases, the phase angle decreases. When the damping ratio is 0.9, the
phase angle drops to about 30, indicating that the two masses have become strongly coupled. Thus, a single
resonant peak forms in the transmissibility plot at this high damping ratio. For the passive system, excessively
increasing the damping ratio results in a coupling of the TVA, and eectively renders it useless in reducing the
vibrations of the structure mass.

4.2. Performance analysis of the semi-active TVA

For the case of ono DBG controlled semi-active TVA, the on-state damping ratio (fon ) varies from 0.1 to
0.9, with an increment of 0.2 while the o-state damping ratio is xed at the baseline value. Unlike the passive
and the velocity-based systems, the two resonant peaks decrease as the on-state damping ratio increases, without raising the isolation valley (see Fig. 7a). This indicates that the ono DBG control keeps the TVA and the
structure masses decoupled at the tuned frequency, enabling the TVA to eectively counteract the motions of
the structure at a high on-state damping ratio. This is because the ono DBG control policy ensures the

Fig. 7. Eect of on-state damping ratio of the performance of semi-active TVA with ono displacement-based groundhook control: (a)
transmissibility (X 1 =X in ); (b) phase angles between the TVA and the structure.

1936

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

Fig. 8. Eect of o-state damping ratio on transmissibility for the semi-active TVA.

minimum (o-state) damping ratio at the valley, independent of the on-state damping ratio, preventing lockup (coupling of the two bodies). This result is one of the key benets of this semi-active system.
Fig. 7b shows the phase angle changes of the ono DBG system as the on-state damping ratio increases.
The phase angle of at the tuned frequency (valley) stayed close to 90, regardless of the on-state damping ratio.
Thus, the ono DBG TVA achieved the minimum transmissibility. Moreover, the phase angles at the two
resonant frequencies approach 90 as the on-state damping ratio increases, reducing the resonant peaks, as
shown in Fig. 7a.
To evaluate the eect of changing the o-state damping ratio (foff ), it varied from 0.05 to 0.13. Fig. 8 shows
transmissibility plots of the ono DBG TVA as foff varied. When the o-state damping ratio increases, the
valley oor also increases. Moreover, the amplitudes of the two peaks grow, and they tend to become one.
This observation indicates that high o-state damping ratios negate the performance of the ono DBG
TVA. This analysis suggests that the o-state damping ratio should be tuned at its optimal value in order
to oer its maximum performance gains.
5. Conclusions
In this study, the dynamic performance of a damping controlled non-linear TVA and its equivalent passive
TVA were evaluated. A non-model based groundhook control called ono DBG control policy was
adopted for damping control of the semi-active TVA. Using an optimization routine, the semi-active TVA
and its equivalent passive TVA were optimally tuned for equal evaluation. The performances of each of
the optimized cases are then compared using the peak transmissibility criteria. The results indicate that the
semi-active TVA outperforms the passive in reducing the peak transmissibility. The semi-active reduced more
than 20% of the maximum vibrations levels as compared its passive counterpart. Furthermore, the results suggest that the o-state damping ratio of the semi-active TVA should be tuned at its optimal value in order to
oer its maximum performance gains.
References
[1] Sun JQ, Jolly MR, Norris MA. Passive, adaptive and active tuned vibration absorbers a survey. In: Transaction of the ASME, 50th
anniversary of the design engineering division, vol. 117; 1995. p. 23442.

J.-H. Koo et al. / Communications in Nonlinear Science and Numerical Simulation 13 (2008) 19291937

1937

[2] Housner GW, Bergman LA, Caughey TK, Chassiakos AG, Claus RO, Masri SF, et al. Structural control: past, present, and future. J
Eng Mech 1997;123(9):897971.
[3] Symans MD, Constantinou MC. Semi-active control systems for seismic protection of structures: a state-of-art review. J Eng Struct
1999;21:46987.
[4] Soong TT, Spencer Jr BF. Supplemental energy dissipation: state-of-the-art and state-of-the-practice. J Eng Struct 2002;24:24359.
[5] Koo J-H, Ahmadian M, Setareh M, Murray T. In search of suitable control methods for semi-active tuned vibration absorbers. J Vib
Cont 2004;10(2):16374.