Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
To cite this Article Okada, Rei , Nakata, Narutoshi , Spencer† Jr., B. F. , Kasai, Kazuhiko and Kim, Saang Bum(2006)
'RATIONAL POLYNOMIAL APPROXIMATION MODELLING FOR ANALYSIS OF STRUCTURES WITH VE
DAMPERS', Journal of Earthquake Engineering, 10: 1, 97 — 125
To link to this Article: DOI: 10.1080/13632460609350589
URL: http://dx.doi.org/10.1080/13632460609350589
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Journal of Earthquake Engineering, Vol. 10, No. 1 (2006) 97-125
@ Imperial College Press @ Imperial College Press
www.icpress.co.uk
RE1 OKADA
College of Science and Technology, Nihon University
1-8- 14 Kanda Surugadai, Chiyoda-ku, Tokyo, 101-8308, Japan
mkadaQedu. cst.nihon-lr. ac.jp
KAZUHIKO KASAI
Structuml Engineering Research Center, Tokyo Institute of Technology
Nagatsuta Midori-ku Yokohama, Kanagawa 226-8503, Japan
kasaiQserc. titech. ac.jp
Appiication of the RPA method to nonlinear structures is also given. The RPA method
n be effective and efficient for modelling VE damped structure.
is s n ~ n to
1. Introduction
To date. numerous studies have been reported on the modelling of frequency-
dependent phenomena (e.g., the interaction between the soil and structures,
[Venanco-Filho, 19971 the loading for offshore structures [Liagre and Niedzwecki,
20031. em). Many kinds of frequency-dependent materials also have been studied
and tested in the last decade [Dressel, 1998; Ymeri et al., 2001).
Among these frequency-dependent materials, VE (Viscoelastic) materials have
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
received significant recent attention for response control in civil engineering struc-
tures. VE materials are used in vibration control devices, termed VE dampers, for
energy absorption. Although their behaviour is usually considered to be linear, a
key issue regarding VE dampers is the adequate modelling of the frequency- and
temperature-dependence of their properties. Adequate modelling of VE dampers
and \X damped structures is very important for analysis and verification of the
performance of such structural systems.
There are two approaches for modelling VE damped structures. The first one
employs a global damping model based on the direct estimation of the modal p r o p
erties ( e g , natural frequencies, mode shapes, damping ratios) of the VE damped
structure. The global damping model can be used for both time history analysis
and response spectra analysis, but it is generally limited to the case where the pri-
mary structure remains elastic. Chang et al. (19931 adopted this model for the time
history analysis of the VE damped structure based on the conventional modal super-
position methods. Kasai et al. [I9931 discussed the accuracy of the global damping
model with regard to simulation of global responses as well as local responses of
the dampers and the surrounding members.
The second approach uses a local damping model in which the frequency depen-
dence of the VE damper is expressed directly. This model can be used for time
history analysis based on direct integration of the equations of motion; therefore,
it can be used for a structure having nonlinear characteristics.
blodels for the constitutive behaviour of VE material have been proposed by a
number of researchers using fractional derivatives [Chang e t al., 1993; Kasai and
Okuma 20041. Kasai and Okuma [2004] presented a fractional derivative model
using frequency-dependent storage and loss modules. This model can also capture
the temperature-dependence of the properties of the VE damper using the so-
called '-Temperature-frequency equivalence principle". For example, the storage and
loss modulus kom 0.1 Hz to 10 Hz at 0°C are equivalent to those parameters from
3.8 Hz to 380Hz a t 20°C.Likewise, the parameters at 40°C are equivalent to those
from 0.091Hz to 0.910Hz a t 20°C. The accuracy and flexibility of the fractional
RPA Modelling for Analysis of Structures with VED 99
derivative model for VE dampers have been examined and verified for a wide range
of temperatures and excitation frequencies [Chang et al., 19951. Although this model
accurately simulates the damper responses, it can not be analyzed using standard
time history analysis tools, but rather requires implementation of special algorithms
[Chang et al., 19951. Therefore, many approximate methods using conventional
models have been proposed to more simply represent the behaviour of VE dampers.
Kasai and Okuma [2001]proposed a practical approach for representing the VE
damper with a simple Kelvin model. The model parameters are set at the resonant
frequency of the VE damped structure; good agreement with the fractional deriva-
tive model was obtained in the vicinity of the resonant frequency. However, this
Kelvin model does not necessarily capture the actual behaviour of the VE damper
over a broad range of frequencies. In particular, inaccurate results can occur when
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
the excitation frequency is lower than the resonant frequency. This limitation affects
simulation of a high-frequency VE damped structure under earthquake loading,
since the dominant frequency of the earthquake input can be comparatively low.
Kasai and Okuma [2002], therefore, proposed a method to calibrate the parameters
of the Kelvin model by considering both the resonant frequency and the dominant
excitation frequency.
Use of the Kelvin model was also examined by Soda and Takahashi [1997] and
Huang et al. [1999]. To more accurately represent the behaviour of VE damped
structures, use of models with more damper and spring elements was proposed by
researchers such as Soda and Takahashi [2000] and Kaneko and Nakarnura [1998].
The parameters of these models are typically calculated based on a least squares
approximation of the experimentally obtained storage and loss coefficients of the VE
damper. Soda and Takahashi [I9971 also proposed a Fourier transform approach.
Based on the relationship between the displacement as input and the load as output,
the equivalent stiffness and damping are estimated. In their approach, an arrange-
ment of Kelvin-Voigt elements was used to approximate the damper's frequency
dependence. The number of elements required will depend on the kind of VE mate-
rial and the target frequency range. As with the RPA method proposed in this
paper, the advantage of this class of models is that commercial modelling programs
can be employed to simulate the response of the resulting structural system. The
disadvantage is that the frequency-dependent characteristics of the system that can
be represented limited both in variety and form by the various combinations of the
spring and damper elements.
In this paper, a new approach using a rational polynomial approximation (RPA)
method is introduced for modelling the response of structures that contain discrete
elements with linear frequency-dependent stiffness and damping characteristics.
The RPA method consists of two steps: First, system identification is performed
to obtain a rational polynomial approximation for the system's transfer function.
Then, a time-domain model for the system is realised. The main advantage of the
RPA method is that the resulting model is a system of ordinary differential equa-
tions, facilitating time history analysis of both Linear and nonlinear structures using
100 R. Okada e t al.
where ~ ( s )termed
, here as the target transfer function matrix, is a complex-valued
function of the LapIace variable, s. v(s) and y ( s ) are the Laplace transforms of the
input and output time histories v(t) and y ( t ) , xespectively. The target transfer
function matrix ~ ( s can ) be obtained either experimentally or born appropriate
modelling t e c h n i q u ~ .
In the RPA method, a.n approximate transfer function matrix H(s) is sought
which adequately represent HIS) and takes the following form:
where the kl entry in H(s) represents the transfer function between vl(s) and yk(s)
and can be written s Hkl (s) = %. The numerator and denominator polynomials
can be expressed as
The roots of the denominator polynomial, d ( s ) , are called poles. The magnitudes of
the poles correspond t o the undamped natural frequencies of the system. The roots
of the numerator polynomial, nk'(s), are called zeros of the transfer function Hk[(s).
RPA Modelling for Analysis of Structures with VED 101
of interest.
Kim et ai. [2005] developed an efficient and accurate method for the problem
given in Eq. (4) for the multi-input-multi-output (MIMO) case. Their method com-
bines a linear least square and two nonlinear methods to improve the approximation.
For computational purpose, Eq. (4) is evaluated at N equally spaced discrete points
in the interval w,i, t o w,,,.
where A: B, C:and D are the system matrices; and x is the state vector.
- 0 1 0 .-. 0 - -0-
0 0 1 0
A= , B= 0 ,
0 ... 0 1
--a0 -a1 - - . -aq-2 -aq-l- -1,
C= [ bp bi') . .. b p
b p bl") ... bp)
0 .. .
0 -.. 0
,
102 R. Okada et al.
The relationship between the transfer function By,(s) and the system matrices are
given by
H,, (s) = c (SI - A)-'B + D. (8)
Therefore, the state space realisation based on the RPA method is seen to allow
us to realise a proper transfer function in the form of ordinary differential equations.
These ordinary differential equations can then be investigated by standard system
analysis tools, which is one of the major advantages of the RPA method.
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
. The next sections introduce two approaches, RPAS and RPAD, for frequency
domain system modelling using the RPA method.
Thus, the target transfer function matrix for the entire structure from the ground
acceleration to the relative displacement, ~,,(jw), is given by
The RPAS method employs the RPA scheme to estimate the transfer function for
the entire structure by solving the following problem:
RPA Modelling for Analysi.3 of Stmctvres with VED 103
H,, (jw), the approximate transfer function of the entire structure, is a finite order
rational polynomial in w and can be converted into time domain model using the
state space representation.
Alternatively, the experimentally obtained transfer function for the entire struc-
ture can be used in the RPAS method as the target transfer function ~ ( s ) The .
advantage of the RPAS method is that one model represents the entire structure
with frequency-dependent devices; the resulting time domain model is simple, con-
sisting of a relatively low order model. However, the disadvantage of the RPAS
method is that the application is limited to linear structures with a small number
of DOF.
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
where fd(t) is the devices reaction force; Cd(u)and K d ( w ) are the frequency-
dependent damping and stiffness coefiicients, respectively; and ud(t) and ud(t) are
the diplacement and velocity of the device in local coordinates. Taking the Fourier
transform of Eq. (13) with displacement as input and force as output, the transfer
function can be written as follows:
~ i mf
W--a7
I iW
(jw)~
)
I (7 = 1 . 2 ) .
For example, consider the case when C d ( w ) and K d ( w ) are constant; for 7 = 1,
Eq. (15) is seen to be satisfied.
104 R. Okada et al.
The W A D method employs the RPA scheme for the target transfer function,
I;rfdYd
(jw), of the device by solving the following equation:
The approximate transfer function Hfdyd(jw)is a rational polynomial. The input for
this transfer function is not the displacement ud(t), but rather e.
Therefore, For
7 = 1,yd = ud and HfdVd(jw)= Hf,~,(jw). For 7 = 2, yd = iid and Hfdyd(jw)=
Hfdiid(jw).Because the approximate transfer function obtained in this manner is
proper, it can be converted into the state space form as follows:
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
where yd(t) is the local input for the damper as described previously; fd(t) is the
damper output force; xd(t) is an internal state vector; and Ad, B d , Cd, and Dd are
system matrices for the state space model of the damper.
A model such as is given in Eq. (17) can be developed for multiple frequency
dependent devices and coupled with the model of the structure as shown in Fig. 1.
The model of the structure can take the form:
where w ( t ) is the excitation vector; fd(t) is the vector of the damper forces; yd(t)
is the vector of local coordinate inputs for the device; y,(t) is the output structural
response vector; x,(t) is the internal state vector for the structure; g(x,(t)) describes
the general behaviour of the structure; B,, C,, D,, and E, are the system matrices
for the state. space representation of the structure; and T is the transformation
matrix from the global structural coordinates to the local device coordinates.
To better illustrate this idea, consider the single damper case in a linear, time-
invariant structure. Noting that in this case g(x,(t)) = ~ , x , ( t ) ,combining Eq. (17)
and Eq. (la), and eliminating yd(t) and fd(t), the state space model of the combined
RPAD VE damper
The RPAD method enables the use of standard tools for advanced analysis of the
system. The advantage of the RPAD method is that it does not require the inverse of
a potentially large matrix such as in Eq. ( 1 2 ) for W A S method. Therefore, it is not
limited to the smail number of DOF. Furthermore, RPAD method is not limited
to linear systems and can accommodate vector input excitations. On the other
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
hand, the RPAD method is that resulting model is somewhat more complex, and
the associated time domain analysis will require more computation effort. Overall,
the W A D method has powerful capabilities to perform complex analyses of the
structures with frequency-dependent devices.
Ki ( w ) = G .
1 + (a + b) wQ cos (a7r/2) + As
--
1 + 2awQ cos ( a 7 r / 2 ) + a 2 w 2 ~
= Kd ( w ) , (20)
d
where G is the static shear modulus of the VE material, and a , b, and cr are mate-
rial constants. As and d are the shear area and the thickness of the VE material,
106 R. Okada et al.
respectively. K i (w) and q d (w) are the storage stiffness and the loss factor, respec-
tively, for the VE damper.
The material constants for the VE damper (Sumitomo 3bf acrylic material called
ISDl11) at 20°C are as follows: a = 5.6 x b = 2.10, CY = 0.558, G = 3.92 x
lo4 ~ / and m As/d
~ = 6.06 m, [Kasai et al., 20011. The frequency-dependence of
the damper storage stiffness and loss factor are shown in Fig. 2.
Single-degree-of-freedom (SDOF) and three-degree-of-freedom (3DOF) linear
structures, as well as 3DOF nonlinear structures, are considered in this section. The
structures are based on shear beam modelling with elastic members. Schematics of
the SDOF and 3DOF structures are shown in Fig. 3. The Bouc-Wen hysteresis
model [Wen, 19761 is employed to represent the inelastic elements in the nonlinear
structures. M ,Cfand Kfare the mass, damping and stiffness, respectively, for each
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
storey of the frame without VE damper. For the simple Kelvin model, the damping
C d and stiffness Kd coefficients of the VE damper are selected based on the values at
the resonant frequency of the structure with VE damper [Kasai and Okuma, 20011.
Figure 4 shows the transfer function of the VE damper for the Kelvin model in
comparison with the Kasai model. Because both SDOF and 3DOF structures have
the first natural frequency around 1.OH.z; the Kelvin model is selected to match
Kasai model at 1.0 Hz.
Frequency [Hz]
I M M = 100 (ton)
C, = 25 (kN dm)
K, = 2820 (kNim)
Kf,qf
M =lS(ton)
C, = 3 (kN slm)
K, = 1820 (kN/m)
3.2.1. RPAS
The target transfer function. for the W A S method is constructed as in Eq. (11)
with frequency-dependent parameters for the VE damper shown in Fig. 2, and
the structural model properties as shown in Fig. 3. The poles and zeros resulting
from the RPAS method for SDOF and 3DOF models are listed in Table 1. The
SDOF VE damped structure in the RPAS model is represented by four poles and
two zeros. The 3DOF VE damped structure in RPAS model is represented by eight
poles and six zeros. The system matrices for the SDOF and 3DOF time domain
models are listed in Table 2.
3.2.2. RPAD
The target transfer function of the VE damper for the RPAD method, H F ~ , ~ ~ ( ~ w ) ,
is constructed with frequency-dependent parameters in Fig. 2. Figure 4 shows the
transfer function of the VE damper. The poles and zeros of the VE damper model
are listed in Table l(c). The VE damper is represented by six poles and four zeros.
RPA Modelling for Analysis of Structures with VED 109
The time domain model, the system matrices, of the VE damper model are listed
in Table 2(c). The time domain model of the combined structure-damper system
for the RPAD method can be constructed in Fig. 1 with the VE damper model
obtained here.
0 0 0 0 0 2.66ef010 0
-
.-a 0.1 . $d
-
----- fPAS MDdel
-;
.-0 1.2.
:g OM....................:
u
C
::
a...... *..-
fPAD W e l
,
m
U
C .3.
1.1
p
.................................----
-0.1 .j 1 ':
is
a,
?
'Yi
I:
a -0.2 : a 0.9 .j
4 . 3- 0.8 f
0 2 4 6 8 1 0 0 2 4 ' 6 8 10
Frequency (Hz) Frequency (Hz)
- w Kasai - @d% w
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
@u w *d% w ~asai
0.15, I 0.15, I
..........KeWin mel
0.04
-WAS M e l
-E 0.02,
----- WAD M d e l
d
V)
8
.E 0 .
2
g
.- -0.02 ..
0
-0.04
10 12 14 16 18 20 10 12 14 16 18 20
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
-0.01 I -0.4
0 5 10 15 0 5 10 15
Time [sec] l i m e [sec]
(e) Displacement (El-Centro) (f) Absolute acceleration (El-Centro)
Fig. 6. DiEerence in time history response of SDOF system.
RPA Modelling for Analysis of Structures with VED 113
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
very good agreements with the Kasai model in displacement and acceleration for
the harmonic excitations.
For the random excitations, El-Centro and band-limited white noise, both the
W A S and W A D methods show very good agreement with the Kasai model in
displacement and acceleration, whereas the Kelvin model differs from the Kasai
model in displacement for the band-limited white noise. This poor performance of
the Kelvin model can be expected from the transfer function in Fig. 5.
These results show that both the RPAS and the RPAD methods can effectively
model the behaviour of VE damped structures.
!G
P i
. ..............................-!
-0.5~ I
0 2 4 6 8 1 0
Frequency (Hz) Frequency (Hz)
@d2u w - QdZ w h s a i
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
-15l I
0 2 4 6 8 1 0
Frequency (Hz)
(a) Displacement (1st floor) (b) Absolute acceleration (1st floor)
-0.4 -
0 2 4 6 8
Frquency (Hz)
@U w-@u w ffisai
1 0
-0.4~
0
10,
2
4
6
'
4 6 0
Frequency (Hz)
w-odzu w %ai
1
I
0
-0.4 I 4.4
0 2 4 6 8 1 0 0 2 4 6 8 1 0
Frequency (Hz) frequency (Hz)
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
w - ad2u w ffisai
-101 I -101 I
0 2 4 8 8 1 0 0 2 4 6 8 1 0
Frequency (Hz) Frequency (Hz)
(e) Displacement (3rd Boor) (f) Absolute acceleration (3rd floor)
Fig. 7. (Continued)
...........Kelvin Model
- 0.005
RPAS Model
RPAD Model
(a) 1st floor displacement (0.3 Hz) (b) 1st floor absolute acceleration (0.3 Hz)
Fig. 8. Difference in time history response of 3DOF system (Sinusoidal excitation).
116 R. Okada et al.
-0.01 1 I -0.03 I
10 12 14 16 18 20 10 12 14 16 18 20
lime [sec] lime [sec]
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
(c) 3rd floor displacement (0.3 Hz) (d) 3rd floor absolute acceleration (0.3 Hz)
- -0.05
10 12 14 16 18 20 10 12 14 16 18 20
Time [sec] Time [sec]
(e) 1st floor displacement (1.0Hz) (f) 1st floor absolute acceleration (1.0 Hz)
-0.01
10 12 14 16
Time [sec]
(g) 3rd floor displacement (1.0 Hz)
18 20
I -0.4
10
- 12 14 16
Time [sec]
18 20
harmonic and random excitations. For the harmonic excitation at 1.0 Hz, all the
models show very good agreement in both displacement and acceleration.
Similar trends are seen in the random excitation cases (see Fig. 9). For both El-
Centro NS record and band-limited white noise, the RPAS and the RPAD methods
show very good correspondence with the Kasai model; the Kelvin model shows a
slight difference.
The RPAS and the RPAD methods have been shown to capture the behaviour
of the fractional derivative model very well in the frequency and time domain for
both SDOF and 3DOF systems.
0 5 10 15 20
l i m e [sec]
(a) 1st floor displacement (El-Centro) (b) 1st floor absolute acceleration (El-Centro)
0 5 10 15 20 0 5 10 15 20
Time [s ec] Time [ s e
(c) 3rd floor displacement (El-Centro) (d) 3rd floor absolute acceleration (El-Centro)
Fig. 9. Difference in time history response of 3DOF system (Random excitation)
118 R. Okada et al.
I
10 12 14 16 18 20
lime [sec] lime [sec]
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
( e ) 1st floor displacement (White noise) (f) 1st floor absolute acceleration (White noise)
-0.01
10 12 14 16 18
lime [sec]
(g) 3rd floor displacement (White noise) (h) 3rd floor absolute acceleration (White noise)
Fig. 9. (Continued)
l5
i i i i i
I I I I I
I l l
-I0-- I I
A--L-l--.
I I ,I
-15 " " 'I
I I I
1 5 I 1 I i Ii iI i I i / l 5 I1 Ii i I i I i Ii
I I I I I
-15 " " '
-0.1-0.05 0 0.05 0.1
Displacement (rn) Displacement (m) Displacement (m)
7 : I
i
I I
;
I
;
I
0.15
- 0.1
- 0.05
E
C
5 0
-m
0.05
b
4.1
0 5 10 0 5 10 0 5 10
Time (sec) Time (sec) Time (sec)
I I I
I I I
I I I
0 5 10' 0 5 10 0 5 10
Time (sac) Time (sec) Time (sec)
4. Conclusions
In this paper, the RPAS and the RPAD methods are introduced as the frequency-
dependent system identification tools. Both methods are based on the approx-
imation of the model in hequency domain, transfer function, in the rational
polynomials. The time domain models are obtained from the realisation of the
approximate proper transfer functions in the ordinary differential equation form.
The RPAS method is directly applied to the transfer function of the entire struc-
ture including the frequency-dependent device. This approach is quite powerful in
modelling of as-built structures based on measured response data. It can be used for
the assessment of the existing structures. The RPAD method employs the rational
polynomial approximation to the frequency device. The entire structural model is
obtained by combining the realisation of structure and the dampers. This approach
enables the detail analysis of the VE damped structures and is not limited to the
linear system.
As a case study, viscoelastic (VE) damped structures were considered. The VE
damper is one of the chalIenging frequency-dependent models, which a simple Kelvin
model cannot fully represent. Analytical result shows that both the RPAS and the
RPAD methods can capture the behaviour of the Kasai's fractional derivative model
very well over a wide frequency range in both the SDOF and 3DOF linear structures.
Time domain analysis shows very good agreement with the fractional derivative
model for harmonic and random excitations. It is proved that the RPA method
is very effective for modelling of the frequency-dependent systems both in fre-
quency and time domain analysis. Furthermore, simulation of nonlinear structures
with VE dampers is successfully performed to show the applicability of the RPA
method.
The RPA method results in a physically realisable model that could be r e p
resented by ordinary differential equations of finite order. The RPA method can
capture the modelling, such as in fractional derivative method, in wide frequency
122 R. Okada et al.
range in the general representation. The RPA method also allows standard time-
step integration methods to be employed for system analysis. Therefore, the RPA
method is very effective for the modelling and analysis of such frequency-dependent
systems.
T h e Matlab code for calculating the responses for the examples in Sec. 3 can be
downloaded at: http://cee.uiuc.edu/sstl/RPA/.
References
Bagley, R. L. and Torvik, P. J. [1983] "A theoretical basis for the application of fractional
calculus to viscoelasticity," Journal of Rheology 27, 201-210.
Chang, K. C., Soong, T. T., Eai, M. L. and Nielsen, E. J. [1993] "Development of a
design procedure for structures with added viscoelastic dampers," ATC-17-1 Seminar,
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
Soda, S., Wada, J., Hirata, Y. and Yamanaka, H. [I9941 Visc~elasticdamper. Cyclic
loading test to construct mechanical model," J. Stmct. Constr. Eng. 457,29-36.
Soda, S. and Takahashi, Y. [1997] "Quantification of frequency-dependent property of
visco-elastic damper," J. Stmct. Constr. Eng. 498,43-49.
Soda, S.and Takahashi, Y. [2000] "Performance based seismic design of building structures
with viscoelastic dampers," 12WCEE 1236.
Venancio-Filho, F. et al. [I9971 "Soil-structure interaction anaiysis of NPP containments:
Substructure and frequency domain methods," Nuclear Engineering and Design 174,
165-176.
Wen, -Y. K. [I9761 "Method of random vibration of hysteretic systems," J. Engineering
Mechanics 102, 249-263.
Ymeri, H.: Nauwelaers, B. and Maex, K. [2001] "Frequency-dependent mutual resistance
and inductance formulas for coupled IC interconnects on an Si-Si02 Substrate," The
VLSI Journal 30, 133-141.
Downloaded By: [INFLIBNET India Order] At: 09:46 7 October 2010
f d ( t ) : Damper force
fd(t): Damper force vector
x d ( t ) : Internal state vector for damper model
x,(t): Internal state vector for structural model
GdUd ( j w ) : Transfer function hom damper displacement to
damper force
A d , Bd, C d and D d : System matrices of the damper model
. A,, B,, C,, D, and E,: System matrices of the entire structural model
g ( x , ( t ) ) : Differential equation matrix
T: Transformation matrix from structural coordinate to
damper coordinate
G: Static shear modulus of VE material
a, b and a: the VE material constants
As and d: the shear area and the thickness of the VE material,
respectively
KA ( w ) : Storage stiffness for the VE damper
q d ( w ) : LOSSfactor for VE damper
Appendix
To simulate the responses of the structure based on Kasai constitutive rule for VE
damper directly, the following procedure is employed.
(i) Calculate the frequency domain output response:
(ii) Compute the inverse Fourier transform of the frequency domain output
response to get the time domain output response of elastic VE damped structure.