Journal of Sound and Vibration: Zhenlong Xiao, Xingjian Jing, Li Cheng

Journal of Sound and Vibration 332 (2013) 13351354
Contents lists available at SciVerse ScienceDirect
Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi
The transmissibility of vibration isolators with cubic nonlinear

damping under both force and base excitations
Zhenlong Xiao, Xingjian Jing n, Li Cheng
Hong Kong Polytechnic University, Department of Mechanical Engineering, Hung Hom, Kowloon, Hong Kong
a r t i c l e i n f o
abstract
Article history:
Received 12 March 2012
Received in revised form
1 November 2012
Accepted 1 November 2012
Handling Editor: W. Lacarbonara
Available online 24 November 2012
The inuence of a nonlinear damping which is a function of both the velocity and
displacement is investigated for a single degree of freedom (sdof) isolator. The analytical
relationships between the force or displacement transmissibility and the nonlinear
damping coefcient are developed in the frequency domain for the isolator systems
subjected to both force and base excitation. It is theoretically shown that the cubic order
nonlinear damping can produce much better isolation performance, i.e., obvious peak
suppression at resonant frequency and very close transmissibility to system linear
dampingover non-resonant frequencies under both force and base displacement excitations. Moreover, when only the pure cubic order nonlinear damping is used without
linear damping, the force or displacement transmissibility is even better. The results are
compared with the other nonlinear damping terms previously studied in the literature.
Numerical studies are presented to illustrate the results.
& 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Inserting the vibration isolator between the vibration source and the vibration receiver is one of the fundamental ways
to reduce the unwanted vibrations and to protect the equipments from
p disturbance. The basic concept of the vibration
isolator is that, when the frequency of the excitation O is larger than 2o0 , where o0 is the undamped natural frequency
of the isolator, the transmitted force, Ft (or the transmitted displacement, Xt) reaches a value less than the excitation force,
Fi (or the excitation displacement, Xi) [1]. The ratio Ft/Fi and Xt/Xi are denoted as force transmissibility and displacement
transmissibility respectively. There is a well-known dilemma associated with linear viscous damping systems. That is
when the linear damping coefcient
p is increased, the force transmissibility under both base excitation and force excitation
p
when excitation frequency O o 2o0 is further reduced, but the performance when excitation frequency O 4 2o0 is
contrarily deteriorated [2,3]. In order to overcome this dilemma, isolators with nonlinear stiffness and nonlinear damping
have been studied by many authors in exploring the potential nonlinear benets in vibration control [1,49]. Another
reason of the study on nonlinear stiffness and nonlinear damping is that almost all the isolators in practical vibration
systems are inherently nonlinear [10,11]. Therefore, it is important to take into account the existence of the nonlinearity in
order to reach a better isolation performance.
Ravindra and Mallik [4] analyzed the vibration isolators having nonlinearity in both stiffness and damping terms under
both force and base excitations. The transmissibility was obtained by the method of harmonic balance, and the effects of
various types of damping to the transmissibility were also studied. The jump phenomenon was observed in the
Corresponding author.
E-mail addresses: xingjian.jing@googlemail.com, xingjian.jing@polyu.edu.hk (X. Jing).
0022-460X/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jsv.2012.11.001
1336
Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354
transmissibility curve when nonlinear stiffness was introduced in the isolator. Based on nonlinear output frequency
response of the Volterra-class nonlinear systems [5,12,13], nonlinear dampings (which usually are pure functions of
velocity) under force excitation are studied in [6,14,15] for vibration isolators. It was shown that the cubic nonlinear
viscous damping can produce an ideal vibration isolation that only the force transmissibility over the resonant region is
modied but it remains almost unaffected over the non-resonant regions. Milovanovic et al. [16] studied the vibration
isolators with linear and cubic nonlinearities in stiffness and damping terms under based excitation. The inuence of the
nonlinear parameters on the displacement transmissibility was studied, and they presented that the absolute displacement transmissibility of the isolator with cubic damping tends to unity as O-N, which corresponds to a rigidly connected
system.
In the present study, a cubic nonlinear damping (i.e., ( )2(d( )/dt)), which is a function of both the displacement and
velocity, is investigated in vibration isolators. Although the nonlinear damping which is usually a pure function of velocity
such as (d(U)/dt)3 has been studied in the literatureas mentioned before [6,14,15], the nonlinear damping relating to both
the displacement and velocity are not well developed and understood [17]. By using the concept of output frequency
response function, the analytical relationship between the force and absolute displacement transmissibility and the
nonlinear damping coefcient of the vibration isolator are derived. It is theoretically shown that, the introduction of
( )2(d( )/dt) can produce much better vibration isolation performance for the isolator under both force excitation and base
displacement excitation. System equivalent damping can be very high around the resonant frequency but would be similar
to system linear dampingover non-resonant frequencies. Therefore, the transmissibility is signicantly suppressed around
the resonant frequency but remain almost the same as when only the linear damping is used. Moreover, when only the
pure cubic order nonlinear damping term ( )2(d( )/dt) is used in the system with the linear damping coefcient x1 0,
both the force and displacement transmissibilities are even better. This may provide an ideal damping characteristic in
practical applications. Simulation results are provided to illustrate the results.
2. Nonlinear vibration isolators and transmissibility functions

In this section, nonlinear isolators subjected to force excitation and base excitation, are investigated. The nonlinear
isolator is modeled as a parallel combination of a linear spring with stiffness k and a nonlinear damper. The nonlinear
damping force is given as
F nd c

3
dU
dU
d U
c2 U2
c4
dt
dt
dt
(1)
where c is the linear damping coefcient, and c2, c4 are the cubic order nonlinear damping characteristic parameters.
In Fig. 1, the force excitation
Ft Asinot
(2)
is directly exerted on the mass M with amplitude A and frequency o. Fout(t) is the force transmitted to the base, and x1(t) is
the absolute displacement of the mass M.
In Fig. 2, the input base excitation is
ut Asinot
(3)
where A is the amplitude and o is the excitation frequency.
F (t ) = A sin( t)
m
x1 (t )
c
k
d ()
d ()
d ()
+ c () 2
+ c4
2
dt
dt
dt
Fout(t )
Fig. 1. Isolator subjected to force excitation.
1337
x1(t)
c
k
d()
dt
+ c2() 2
d()
dt
d()
+ c4
dt
Fout (t )
u ( t) = Asin(t)
Fig. 2. Isolator subjected to base excitation.
The governing equations and the non-dimensional transmissibility expressions are derived separately in what follows.
2.1. The isolator subjected to force excitation and the force transmissibility
From Fig. 1, the governing equation of the isolator under force excitation can be given as
mx 1 kx1 cx_ 1 c2 x21 x_ 1 c4 x_ 31 Asinot
(4)
c4 x_ 31
(5)
The force ratio Tfr(t) is dened as

T fr t
F out
A
kx1
A
cx_ 1
c2 x21 x_ 1
A
By dening the following non-dimensional parameters:

q
o0 t t, o0 mk , O oo0

z1 t x1 t x1 ot0 , x_ 1 t o0 z_ 1 t,
y1 t
kz1 t
,
A
x 1 t o20 z 1 t
(6)
y2 t T f t,
2
k m
k m3
c2
A
c
4A
x1 p
, b2 p
, b4 pc
5
3
km
the governing equation (4) and the force ratio (5) can be expressed as the following non-dimensional form:
8
< y 1 y1 x1 y_ 1 b2 y2 y_ 1 b4 y_ 3 sinOt
1
1
: y2 y1 x1 y_ 1 b2 y21 y_ 1 b4 y_ 31
(7)
Denote Tf(O) as the force transmissibility of the vibration isolator in terms of the normalized frequency O, it can be
expressed as
T f O 9Y 2 jO9
(8)
where Y2(jO)Y2(jo)9o O, the output spectrum of the second output of system (7).
2.2. The isolator subjected to base excitation, and the force and displacement transmissibility
From Fig. 2, the isolator model under base excitation can be written as
_ x_ 1 c2 ux1 2 u
_ x_ 1 c4 u
_ x_ 1 3
mx 1 kux1 cu
(9)
The force ratio Tfr(t) in this case is denoted by

T fr t
F out
kA
1
A
ux1
c
kA
_ x_ 1
u
c2
kA
_ x_ 1
ux1 2 u
c4
kA
_ x_ 1 3
u
(10)
Denote the relative displacement x of the isolator as

x x1 u
Then Eqs. (9,10) can be rewritten as
8
< mx kx cx_ c2 x2 x_ c4 x_ 3 mAo2 sinot
: Tf r 1 x
A
c _
x c2 x2 x_ c4 x_ 3
kA
kA
kA
(11)
(12)
1338
Dene the following non-dimensional parameters:

q
o0 t t, o0 mk , O oo0

zt xt x ot0 , x_ t o0 z_ t,
y1 t zt ,
A
x t o20 z t
y2 t T f t,
2
c2
A
c
x1 p
, b2 p
, b4
km
km
(13)
p 2
c4p
kA
m3
Then Eq. (12) can be rewritten as the following non-dimensional form:

8
< y 1 y1 x1 y_ 1 b2 y2 y_ 1 b4 y_ 3 O2 sinOt
1
: y2 y1 x1 y_ 1 b2 y21 y_ 1 b4 y_ 31
(14)
Similar to Section 2.1, the force transmissibility of the vibration isolator in terms of the normalized frequency O, Tf(O),
can also be expressed as
T f O 9Y 2 jO9
(15)
where Y2(jO) is the output spectrum Y2(jo) of the second output of system (14) evaluated at frequency o O.
While comparing the equations in (7) with the equations in (14), it can be seen that the base displacement excitation is
equivalent to a force excitation when the strength of disturbing force is proportional to the square of exciting frequency,
and the force transmissibility under the force excitation and that under the base displacement excitation have the same
expression.
In the following, the absolute displacement transmissibility is derived. In this case, the governing equation is given in
Eq. (9), and the displacement ratio is dened by
x1
T dr t x
(16)
A
Then Eq. (9) can be rewritten as
2
_
_ 1 c2 A u1 x2 u_ 1 x_ c4 A u_ 1 x_ 3
mx kx cxku
1 cu
(17)
where x is the displacement ratio and u1 is given by

u1
u
A
sinot
(18)
Using the following non-dimensional variables:

q
k
O oo0
m,

yt xt x ot0 , x_ t o0 y_ t, x t o20 y t

u2 t u1 t u1 ot0 sinOt, u_ 1 t o0 u_ 2 t
p 2
2
c2
A
kA
c
x1 p
, b2 p
, b4 c4p
3
o0 t t, o0
km
km
(19)
Eq. (17) can be written into the following non-dimensional form:

_
_ 2 b2 u2 y2 u_ 2 y_ b4 u_ 2 y_ 3
y y x1 yu
2 x1 u
(20)
Thus, the displacement transmissibility of the vibration isolator in terms of the normalized frequency O, Td(O), can be
expressed as
T d O 9X jOo0 9 9YjO9
(21)
where Y(jO) is the output spectrum Y(jo) of system (20) evaluated at frequency o O.
In the next section, an explicit and analytical relationship between the force or displacement transmissibility and the
nonlinear damping coefcients b2 and b4 will be developed in the frequency domain for the nonlinear isolators.
3. The force and displacement transmissibility in the frequency domain
The nonlinear output frequency response concept is recently proposed [5,12,13] for the frequency domain study of the
nonlinear Volterra systems, which represent a wide classes of nonlinear systems whose input and output can be expressed
as the Volterra series around the equilibrium. One of the advantages of this concept is that it can give an explicit analytical
relationship between the output frequency response and the parameters of the nonlinear systems which can be described
by differential equation models. Therefore, the analytical relationships between the force transmissibility Tf(O)or
displacement transmissibility Td(O) and the nonlinear damping coefcients b2 and b4 are established in this section.
1339
3.1. The force transmissibility Tf(O)

From systems (7) and (14), the base displacement excitation is basically equivalent to a force excitation, and the force
ratio under base displacement excitation has the same form as that under force excitation. In this section, the force
transmissibility in the frequency domain under both two excitation types will be developed.
According to [15,18,19], systems (7) and (14) can be regarded as a one-input-two-output system, the output spectra
can be obtained as
Y J jo
N
X
1 X
n1
n12
o1 on o
HJn jo1 ,. . .,jon U o1 U on ,
J 1,2
(22)
where HJn jo1 ,. . .,jon is the nth order generalized frequency response function (GFRF) between the input and the Jth
output of the system, N is the maximum order of nonlinearity in the Volterra series expansion of the system outputs. U(oi)
is the input Fourier transform. For system (7),
8
>
< j when oi O, i 1,. . .,n
otherwise
(23a)
U oi 0
>
: j when o O, i 1,. . .,n
i
For system (14),
8
2
>
< jO when oi O,
U oi 0
>
: 2
jO when oi O,
i 1,. . .,n
(23-b)
otherwise
i 1,. . .,n
Eq. (22) involves the computation of the nth order GFRFs HJn jo1 ,. . .,jon . The explicit expression and its derivation for
o1 ,. . .,jon can be referred to Appendix A. With this result, the output spectrum Y2(jo) of the second output of system
(7) and system (14) can be written as
H2n j
Y 2 jo P0 jo P10 job2 P 11 job4
n
X
n-m m
P nm job2 b4
m0

N=2
X
N=2m m
b4
PN=2m job2
(24)
m0
where
P 0 jo H21 joUjo
(25)
Qn-m 3m
Nn
2n
1
jozlk 1 jozl jn
Y
X
k1
1 jo2 X
k k
1
H
j
o
U
Pnm jo 2n
Q
i
i
1
n-1
o1 o2n 1 o
z
z
2 Ljo
z1
i1
i 1 Ljoli 1 joli jn
i
(26)
The denition of L(jo) is given in Appendix A. Eq. (24) presents an analytical relationship between the second output
spectrum and the nonlinear characteristic parameters b2 and b4. According to Eqs. (8) and (15), the force transmissibility
can be given by

N=2

n
X X

nm m
T f O 9Y 2 jO9 P 0 jO
(27)
Pnm jOb2 b4

n1m0
It can be seen that Eqs. (25)(27) are explicit functions of the input and rst-order GFRF. Substituting Eq. (23a) into Eqs.
(25) and (26), for system (7) (nonlinear isolator subjected to force excitation) it can be obtained that
P 0 jO
P nm jO
1
2n
jO2
2n
j1 jx1 O
LjO
Nn
X
X
2
9LjO9 LjO
o1 o2n 1 O
(28)
Qn-m 3m
z1
k1
Qn1
i1
jozlk 1 jozl
Ljo
li 1
jo
jnk
z

li jni

(29)
where oi 2 fO, Og,i 1,. . .,2n 1.

Similarly for system (14) (the nonlinear isolator subjected to base displacement excitation), it can be obtained that
P 0 jO
jO2 1 jx1 O
LjO
(30)
1340
Pnm jO
1
2n
jO4n 4
2n
Nn
X
X
2
9LjO9 LjO
o1 o2n 1 O
z1
Qn-m 3m
k1
Qn-1
i1
jozlk 1 jozl
Ljo
li 1
jnk
z

li jni
jo
(31)
where oi 2 fO, Og,i 1,. . .,2n 1.

From Eq. (27), when there is no nonlinear damping coefcients, i.e., b2 0 and b4 0, the nonlinear sdof vibration
isolatorbecomes a linear isolator. The force transmissibility will be easy to obtain.
For system (7), the linear isolator is under force excitation,
v

u
2
j1 jx1 O u
u 1 x1 O
(32)
T f O 9Y 2 jO9 9P0 jO9
t

2
LjO
2
1O2 x1 O
For system (14), the linear isolator is under base displacement excitation,
v

u
2
jO2 1 jx O
u
1 x1 O

2u
1
T f O 9Y 2 jO9 9P 0 jO9
O t

2

LjO
2
1O2 x1 O
(33)
3.2. The displacement transmissibility Td(O)

In this case, the displacement transmissibility is given by Eq. (21).In order to obtain the output spectrum Y(jo) of
system (20), Eq. (20) is expanded as
_
_ 2 b2 u22 u_ 2 b4 u_ 32 2b2 yu2 u_ 2 b2 yu
_ 22
y y x1 yu
2 x1 u
_ 2 3b4 y_ 2 u_ 2 b2 y2 y_ b4 y_ 3 0
3b4 y_ u_ 22 b2 y2 u_ 2 2b2 yyu
(34)
According to Ref. [20], the coefcients of the nonlinear differential Eq. (34) can be expressed in the general form as
C 1,0 2 1,C 1,0 1 x1 ,C 1,0 0 1,C 0,1 1 x1 ,C 0,1 0 1
C 0,3 0,0,1 b2 ,C 0,3 1,1,1 b4
C 1,2 0,0,1 2b2 ,C 1,2 1,0,0 b2 ,C 1,2 1,1,1 3b4 else C p,q U 0
C 2,1 0,0,1 b2 ,C 2,1 0,1,0 2b2 ,C 2,1 1,1,1 3b4
(35)
C 3,0 0,0,1 b2 ,C 3,0 1,1,1 b4 ,

In order to obtain the output spectrum Y(jo) of system (34), the following propositions are given.
Proposition 1. The (2n1)th GFRF H2n 1 jo1 ,. . .,jo2n 1 of system (34) when there exists only one nonlinear term with
nonlinear coefcient Cp,q(l1,l2,l3), where, p q 3,p,q 0, ,3,l1 ,l2 ,l3 0,1 , under the assumption that there is m terms of 1,
i.e. (3-m) terns of 0, in l1,l2,l3, can be determined as
Qmn
z
z
Nn,p p-1
n 1
X
Y
C np,q l1 ,l2 ,l3

k 1 jolk 1 jolk jnk
(36)
H11 joji Qn-1
H12n 1 jo1 ,. . .,jo2n 1
z
z
Ljo1 jo2n 1 z 1 i 1
i 1 Ljoli 1 joli jn
i
Proof. See Appendix B &.
When only one nonlinear term is introduced in Eq. (34), Proposition 1 states that the high order GFRF can be represented
as a straightforward function of the nonlinear coefcient and the linear transfer function. So when the linear transfer
function and the nonlinear coefcient introduced are known, the high order GFRF can be obtained directly according to Eq.
(36). Proposition 1 also shows how the output nonlinear degree p and the order of the derivative of both input and output
affect the high order GFRF.
Proposition 2. The (2n1)th GFRF H2n 1 jo1 ,. . .,jo2n 1 of system (34) when there exists L nonlinear terms with nonlinear
coefcients C pi ,qi l1 ,l2 ,l3 , where pi qi 3,pi ,qi 0,. . .,3,i 1,. . .,nl , l1,l2,l3 0,1, under the assumption that there is mi terms of
1, i.e. (3-mi) terms of 0, in l1,l2,l3 for C pi ,qi l1 ,l2 ,l3 , can be determined as:
nL
P
C pj ,qj
X1
,L,njnL
n nL nji n-1
Ljo1 jo2n 1
S
i 1
P
0
1
PnL
Q ni L 1 njmi mi
N n,pj pj 1
pi 1nji 1
nL
jozlk 1 jozl jn
iY
1
X
Y

k1
nji
k k C
1
n1 B
0,
C pi ,qi li1 ,li2 ,li3
H1 joki
@
A
Qn1
z
z
z1
i1
k1
i 1 Ljoli 1 joli jn
i
P L
PnL
where nji njmi, iaj, njj 1njmj, ni
1 nji n1,
i 1 njmi n.
H12n 1 jo1 , ,jo2n 1
j1
(37)
1341
Proof. See Appendix C &.

Proposition 2 considers the case that more than one nonlinear term exist in Eq. (34), which is an extension of Proposition
1. It implies how the nonlinear coefcient C pi qi , the output nonlinear degrees pi and the order of the derivate of both input
and output li1,li2,li3 inuence the high order GFRF analytically.
System (20) or (34) can be seen as a single-input-single-output system, so the output spectrum can be obtained via Eq.
(22) with J1 using the input stated in Eq. (19) or its Fourier transformin Eq. (23a). The (2n 1)th order GFRF can be
obtained from Proposition 2. Therefore the output spectrum can be written as
Yjo P0 jo
0
B
B
B
B
n1@
n
X
N=2
X
njm1 ,...,njmnL 0 ,
nL
P
njmi n
1
!C
nL
Y
C
n
C
C pijmi
,qi li1 ,li2 ,li3 P nk jo C
A
i1
(38)
i 1
P 0 jo
2nQ
1
P nk jo
i1
Ujo X
Ljo
2n
Nn,pj
o1 o2n 1 o
1 jox1
U jo
Ljo
PnL
i 1
z1
pY
i 1njmi 1
(39)
PnL

H11 joki
i 1
njmi mi
k1
Qn1
k1
i1
jozlk 1 jozl
Ljo
li 1
jo
jnk
z

li jni
(40)
According to Eq. (21), the displacement transmissibility can be obtained by substituting Eq. (23a) into Eqs. (39) and (40):

0
1

!
N=2
nL
n

X
Y
XB
njmi

C
T d O 9Y jO9 P 0 jO
C pi ,qi li1 ,li2 ,li3 P nk jO A
(41)
@

PnL
n1
i1

n
n
njm1 ,...,njmnL 0,
jmi
i 1
where
P 0 jO
Pnk jO
1 j
22n LjO
Nn,pj
o1 o2n 1 O
z1
PnL
i 1
j1 jOx1
LjO
pY
i 1njmi 1

H11 joki
k1
PnL
i 1
(42)
njmi mi
k1
Qn1
i1
jozlk 1 jozl
Ljozl 1 jozl
i
2 fO,Og
jni

jnk
oki , ozlk i , ozli i

(43)
4. The effects of nonlinear damping coefcients on vibration isolation

In Section 3, the force transmissibility under both force and base excitations and the absolute displacement
transmissibility under base displacement excitation are derived. All of them have an explicit analytical relationship with
the nonlinear damping coefcients introduced. The following results can be obtained.
Proposition 3. For force excitation, the nonlinear damping ( )2(d( )/dt) can produce the following performance in force
transmissibility:
(I) When O b1 or O 51,
T O 9P 0 jO9
s
1 x1 O2
1O2 2 x1 O2
(44)
(II) When O E1, there exists a b such that the force transmissibility can be expressed in an alternating series with respect to the
nonlinear coefcient b2 if 0o b2 o b. The force transmissibility can therefore be suppressed by exploiting the properties of
alternating series.
Proof. See Appendix D &.
Proposition 3 indicates that the nonlinear damping term ( )2(d( )/dt) has almost no effect on the force transmissibility
over the non-resonant frequency regions where the frequency is much lower or much higher than the resonant frequency,
while the force transmissibility is obviously suppressed at the resonant frequency due to the introduction of the nonlinear
damping term under force excitation.
Proposition 4. When the isolator is under base displacement excitation, the force transmissibility with the nonlinear damping
(d( )/dt)3 is dramatically deteriorated thigh frequency, while the nonlinear damping ( )2(d( )/dt) can make the force
1342
transmissibility very close to the low-damping linear referenced case at the same frequency. The introduction of any one of these
two cubic degree nonlinear terms can both make the force transmissibility very close to the low-damping linear referenced case
at low frequency:
(I) When O 51, both ( )2(d( )/dt) and (d( )/dt)3 can make
2
T O 9P0 jO9 O
1O2 2 x1 O2
(II) When O b1, only ( )2(d( )/dt) term can make

2
T O 9P0 jO9 O
s
1 x1 O2
s
1 x1 O2
1O2 2 x1 O2
(45)
(46)
While the nonlinear term (d( )/dt)3 will make
v
u
2
u
1 x1 O
T Oc9P0 jO9 O t
2
2
1O2 x1 O
2u
(47)
(III) When O E1, there exists a b such that the force transmissibility can be expressed in an alternating series with respect to the
nonlinear coefcient b2(b4) if 0 o b2 o b(0o b4 o b). The force transmissibility can therefore be suppressed by exploiting the
properties of alternating series.
Proof. See Appendix E &.
Proposition 4 shows that both the two nonlinear damping terms can signicantly reduce the force transmissibility over
the resonant frequency and remain the force transmissibility almost unaffected at low frequency. However, at high
frequency the nonlinear term (d( )/dt)3 dramatically increases the force transmissibility while the nonlinear damping
term ( )2(d( )/dt) keeps the force transmissibility very close to the low-damping linear referenced case. These indicates
that the nonlinear damping term which is a function of both displacement and velocity produce much better force
transmissibility performance than the nonlinear damping term which is only velocity-dependent under base displacement
excitation.
For the displacement transmissibility under base displacement excitation, it is very similar to the force transmissibility
discussed above.
Proposition 5. Consider the displacement transmissibility under base displacement excitation. The performance at high
frequency with the nonlinear damping (d( )/dt)3 is dramatically deteriorated, while the nonlinear damping (U)2(d(U)/dt) can
make the displacement transmissibility very close to the low-damping linear referenced case over this frequency region. Both of
these two cubic order nonlinearities can make the displacement transmissibility very close to the low-damping linear case aglow
frequency:
(I) When O 51, both ( )2(d( )/dt) and (d( )/dt)3 can make s
1 x1 O2
T O 9P0 jO9
1O2 2 x1 O2
(II) When O b1, only ( )2(d( )/dt) term can make
T O 9P0 jO9
s
1 x1 O2
1O2 2 x1 O2
(48)
(49)
while the nonlinear term (d( )/dt)3 will make
q
2
1 x1 O
T Oc9P 0 jO9 r
2
2
2
x1 O
1O
(III) When O E1, there exists a b 40 such that

dTO2
dTO2
o0 or
o0
db2
db4
if 0 o b2 o b (or 0o b4 o b)
Proof. See Appendix F
&.
(50)
!
(51)
1343
5. Simulation studies and discussions

The following simulations using the RungeKutta method are given to verify the theoretical results above.
Fig. 3 represents the force transmissibility for the isolator subjected to force excitation under different linear damping
coefcients and different cubic order nonlinear damping terms. The solid line and dash line represent the force
transmissibility when only linear damping coefcient is introduced in the isolator where x1 0.1 and x1 0.325
respectively. The dot line shows the performance when the linear damping coefcient and cubic order damping term
(d( )/dt)3 are introduced in the isolator where x1 0.1 and b4 0.028, while the star line represents the performance when
linear damping coefcient and cubic order nonlinear term ( )2(d( )/dt) are introduced in the isolator where x1 0.1 and
b2 0.4. In Fig. 3, it is shown that the star line and the dot line are almost superimposed on the solid line at both low
frequency and high frequency, and have the force transmissibility much smaller than the solid line around the resonant
frequency. So it is clear that the cubic order nonlinear damping terms ( )2(d( )/dt) and (d( )/dt)3 can both produce the
ideal isolation performance, that is, the force transmissibility over the resonant frequency is obviously suppressed while
keeping almost unaffected over the non-resonant regions. The dash line is presented as a reference, from which it can be
seen that in order to get the same force transmissibility in the resonant frequency with that when the cubic order
nonlinear term is introduced in the isolator, the linear damping coefcient x1 is increased from 0.1 to 0.325, and then the
force transmissibility at high frequency increases obviously.
In Fig. 4 the force transmissibility under base displacement excitation produced by the cubic order nonlinear term
( )2(d( )/dt) (where the linear damping coefcient x1 0.1 and the cubic order nonlinear coefcient b2 0.1) is presented
in star line. The force transmissibility has been signicantly suppressed over the resonant frequency and over the nonresonant regions remains very close to the performance presented in solid line when only linear damping coefcient is
introduced where x1 0.1. The performance when cubic order damping term (d( )/dt)3 is introduced in the isolator where
x1 0.1 and b4 0.03 is shown in dot line, from which it can be seen that the force transmissibility at high frequency is
dramatically deteriorated compared with that when there only exists the linear damping coefcient where x1 0.1. The
performance deterioration at high frequency limits the practical use of the cubic order damping term (d( )/dt)3 although
the force transmissibility is also suppressed over the resonant frequency. The dash line is also presented as a reference
where only the linear damping term with the coefcient x1 0.306 is introduced, from which it can be seen that in order to
reach the similar force transmissibility to the case that the cubic order nonlinear damping term ( )2(d( )/dt) is introduced
at resonant frequency, the force transmissibility increases obviously at high frequency.
In Fig. 5 the star line stands for the absolute displacement transmissibility when the cubic order nonlinear term
( )2(d( )/dt) is introduced where the linear damping coefcient x1 0.1 and the cubic order nonlinear damping coefcient
b2 0.1. The transmissibility is signicantly suppressed at the resonant frequency and over the non-resonant frequency it
remains very close to the solid line which represents the linear damping case with linear damping coefcient x1 0.1. The
absolute displacement transmissibility produced by the cubic order damping term (d( )/dt)3 with x1 0.1 and b4 0.03 is
shown in dot line, which tends to be 0 dB as O tends to be innity, and this corresponds to a rigidly-connected system. The
dash solid line shows that when the linear damping coefcient is increased from 0.1 to 0.306 in order to achieve the
similar absolute displacement transmissibility over the resonant frequency to the case that the cubic order nonlinear
terms are introduced in the isolator, the performance at high frequency increases obviously.
force transmissibility Tf ,20*log10(|Y2(j)|)
30
1=0.1;2=0;4=0
20
1=0.325,2=0,4=0
1=0.1,2=0.4,4=0
10
1=0.1,2=0,4=0.028
0
-10
-20
-30
-40
-50
-60
-0.5
0.5
1
frequency,log10()
1.5
Fig. 3. The force transmissibility for an sdof isolator subjected to force excitation with different linear damping coefcients and different cubic order
nonlinear terms.
1344
From Figs. 3 to 5, the cubic order nonlinear term ( )2(d( )/dt) can produce ideal isolation performance, that is, signicant
vibration suppression over the resonant frequency and low damping effect over non-resonant frequencies, under both force and
base displacement excitations. In order to have a more straightforward insight into the nonlinear mechanism in vibration
suppression, the equivalent damping coefcients of the vibration isolator system under different cases are provided in Fig. 6. As
shown in Fig. 6, the cubic order nonlinear terms ( )2(d( )/dt) and (d( )/dt)3 have equivalent linear damping coefcient very close
to 0.306 at resonant frequency. Therefore, these two nonlinear terms have the transmissibility very close to the case when the
linear coefcient x1 0.306 at resonant frequency as presented in Figs. 4 and 5. At high frequency, the nonlinear term (d( )/dt)3
has a very large equivalent linear damping coefcient, which corresponds to the deteriorated isolation performance in Figs. 4 and
5, but the equivalent linear damping coefcient of the nonlinear term ( )2(d( )/dt) remains very small, which is close to 0.1. The
better isolation performance of the nonlinear term ( )2(d( )/dt) at high frequency presented in Figs. 4 and 5 is therefore produced
by this small equivalent damping effect. It is known that the ideal isolation performance requires the damping coefcient to be
larger at resonant frequency but smaller at high frequency. Fig. 6 shows that the nonlinear damping characteristic ( )2(d( )/dt)
can achieve this objective much better than the other cases.
In what follows, the case when only the cubic order nonlinear damping term ( )2(d( )/dt) is introduced in the isolator
with the linear damping coefcient x1beingzero, is studied since it has much better damping effect over all frequencies as
discussed above.
In Fig. 7 the star line represents the force transmissibility for the isolator with only the cubic order nonlinear term
( )2(d( )/dt) and b2 0.1subjected to the base displacement excitation. The dash line indicates the case that the linear
80
1=0.1;2=0;4=0
60
1=0.306,2=0;4=0
1=0.1,2=0.1,4=0
40
1=0.1,2=0,4=0.03
20
-20
-40
-1
-0.5
0
0.5
1
frequency,log10()
1.5
Fig. 4. The force transmissibility for the isolator subjected to base displacement excitation with different linear damping coefcients and different cubic
order nonlinear terms.
30
1=0.1;2=0;4=0
displacement transmissibility
Td ,20*log10(|Y(j)|)
20
1=0.306,2=0;4=0
1=0.1,2=0.1,4=0
10
1=0.1,2=0,4=0.03
0
-10
-20
-30
-40
-50
-60
-1
-0.5
0
0.5
1
frequency,log10()
1.5
Fig. 5. The absolute displacement transmissibility for the isolator subjected to base displacement excitation with different linear damping coefcients
and under different cubic order nonlinear terms.
1345
Equivalent linear
damping ,20*log10(|F nd(j)|/|V(j)|)
10
5
1=0.1;2=0;4=0
1=0.306,2=0;4=0
1=0.1,2=0.1,4=0
1=0.1,2=0,4=0.03
-5
-10
-15
-20
-25
-1
-0.8 -0.6 -0.4 -0.2

0
0.2 0.4
frequency,log10()
0.6
0.8
Fig. 6. The equivalent linear damping coefcient of the isolator subjected to base displacement excitation with different linear damping and different
cubic order nonlinear damping terms.
30
20
10
0
1=0.1;2=0;4=0
-10
1=0,2=0.1,4=0
1=0.22,2=0;4=0
-20
-30
-40
-1
-0.5
0
0.5
1
frequency,log10()
1.5
Fig. 7. The force transmissibility for an isolator subjected to base displacement excitation with the pure cubic order nonlinear damping term ( )2(d( )/dt).
damping coefcient is increased from 0.1 to 0.22 in order to obtain the same force transmissibility as that when only the
cubic order nonlinear term is introduced in the isolator. It can be seen that when the linear damping coefcient is
increased to suppress the transmissibility around the resonant frequency, the transmissibility at high frequency increases
obviously. This is the famous dilemma in vibration isolations. Comparing the star line with the solid line, it is evidently
that the cubic order nonlinear term ( )2(d( )/dt) can overcome the dilemma and produce an ideal isolation performance
that the transmissibility is suppressed at both resonant frequency and high frequency and keep almost unaffected at low
frequency.
In Fig. 8 the absolute displacement transmissibility for the isolator subjected to base displacement excitation when only
the cubic order nonlinear term ( )2(d( )/dt) is introduced in the isolator with b2 0.1 is presented in star line, compared
with the cases with linear damping coefcient x1 0.1 and x1 0.306 respectively. Similarly, the cubic order nonlinear
term ( )2(d( )/dt) can also overcome the dilemma in absolute displacement transmissibility and produce a much better
isolation performance.
6. Conclusions
In this paper, the inuence of a cubic order nonlinear damping term ( )2(d( )/dt) is studied for an sdof isolator system.
It is shown that the proposed nonlinear damping can overcome the dilemma in vibration isolation that when the linear
1346
40
displacement transmissibility
Td ,20*log10(|Y(j)|)
1=0.1;2=0;4=0
1=0,2=0.1,4=0
20
1=0.306,2=0;4=0
0
-20
-40
-60
-80
-1
-0.5
0
0.5
1
frequency,log10()
1.5
Fig. 8. The absolute displacement transmissibility for an isolator subjected to base displacement excitation with pure cubic order nonlinear term
( )2(d( )/dt) and with pure linear damping.
damping coefcient is increased to reduce the transmissibility over the resonant frequency it at the same time increases
the transmissibility at high frequency. The force transmissibility and the absolute displacement transmissibility for the
isolator subjected to both force excitation and base displacement excitation when cubic order nonlinear terms are
introduced are established, from which the analytical and explicit relationship between the transmissibility and the cubic
order nonlinear coefcients can be obtained. Much better isolation performance is produced with the cubic order
nonlinear damping term ( )2(d( )/dt), and numerical studies are given to verify the theoretical results. The following
conclusions can be made:
(I) The cubic order nonlinear damping term ( )2(d( )/dt) can produce better isolation performance for an isolator
under both force excitation and base displacement excitations. While the known cubic nonlinear damping which is
only a pure function of velocity is limited in vibration control subjected to base excitations. This may imply that the
optimal nonlinear damping could be dependent not only on velocity but also on displacement.
(II) The proposed cubic order nonlinear damping term can obviously suppress the transmissibility over the resonant
frequency and remains very close to the low-damping linear referenced case over the non-resonant frequency
regions. It can demonstrate even better performance when the system linear damping is zero. This provides a fairly
ideal damping characteristic in practical applications.
(III) The nonlinear frequency domain method adopted in this study provides a powerful tool for the analysis and design
of nonlinear damping systems. It can provide a straightforward expression for the relationship between nonlinear
output spectrum and any characteristic parameters which dene the nonlinearity of the system and thus facilitate
the nonlinear analysis and design. Further study will focus on more general optimal analysis and design of
nonlinear stiffness and damping characteristics in vibration control.
Acknowledgment
The authors would like to gratefully acknowledge the support from the GRF project (Ref. 517810) of Hong Kong RGC,
Department General Research Funds and Competitive Research Grants of Hong Kong Polytechnic University.
Appendix A
The specic expression of HJn jo1 ,. . .,jon can be obtained according to [15] considering the one-input-two-output
nonlinear differential model as follows:
H21 jo1 1 jo1 x1 H11 jo1
(A-1)
H2n jo1 ,. . .,jon jo1 jon 2 H1n jo1 ,. . .,jon n 2,. . .,N
(A-2)
H11 jo1
1
Ljo1
(A-3)
H13 jo1 ,jo2 ,jo3 b2

Q3
b4
1347
Q3
1
i 1 H 1 joi
jo1
Ljo1 jo2 jo3
3
H11 joi Y
joi
Ljo1 jo2 jo3 i 1
i1
(A-4)
H12n jo1 ,. . .,jo2n 0, n 1,. . .,bN=2c

H12n 1 jo1 , ,jo2n 1
n
X
bnm
bm
2
4
m0
Q2n 1
Nn
X
H11 joi
Ljo1 jo2n 1 z 1
Qnm 3m
k1
i1
(A-5)
jozlk 1 jozl
Qn1
i 1 Ljo
li 1
jnk
z

li jni
jo

n 1,. . ., N=2
(A-6)
where
jni 2 f3,5,. . .,2n1g
i 1,. . .,n1,n Z 2
jnk 2 f1,3,. . .,2n1g

oZ 2 fo1 ,. . ., o2n 1 g,
i 1,. . .,n1,j 1,. . .,jni ,n Z2
oZ 2 fo1 ,. . ., o2n 1 g,
k 1,. . .,n1,j 1,. . .,jk ,n Z2
k 1,. . .,n1,n Z 2
li j
n
1
Y
Ljozli 1 jozl jn 1
i
i1
(A-7)
lk j
for n 1
Nn is an n dependent integer:
Ljo1 jon 1 x1 jo1 jon jo1 jon 2
(A-8)
Appendix B. Proof of Proposition 1

According to Ref. [12], C p,q l1 ,l2 ,l3 ,p q 3,p,q 0,. . .,3,l1 ,l2 ,l3 0,1 will only exist in the (2n 1)th GFRF (generalize
frequency response function) H2n 1 jo1 ,. . .,jo2n 1 , where n 1,2,. . ., this means that the even order GFRF
H2n jo1 ,. . .,jo2n 0. Assume that all the GFRF whose order less than (2n 1) all satisfy proposition 1. The (2n 1)th
GFRF can be obtained [13]:
H12n 1 jo1 ,. . .,jo2n 1

l

C p,q l1 ,l2 ,l3
jo2n 1q 1 p 1 jo2n 1 lp q 3 H12n 1q,p jo1 , ,jo2n 1q
Ljo1 jo2n 1
p
Y
2n1
X

H12n 1-q,p jo1 ,. . .,jo2n 1q
r 1 rp 1;
(B-1)

l
H1ri joX 1 ,. . .,joX ri joX 1 joX ri i
r i 2n 1q i 1
Pi1
where X x 1 r x .

According to the assumption, Hri joX 1 ,. . .,joX ri can be expressed as
Qmnri
Nnr ,p p1nr 1
nr 1
jozlk 1 joz nri

1
1
k1
X
Y

C
l
,l
,l
l k jk
p,q 1 2 3
1
1

Hri jo1 ,. . .,jonri
H1 joji Qn 1
ri
z
L jo1 jori z 1
Ljo joz n
i1
i1
li 1
li ji
ri
where r i 2nri 1. Then,

p
Y

l
i1

nr
C p,qi l1 ,l2 ,l3 j
p
Y
i1
Pp
C p,q
n
i 1 ri
Qp
oX 1 joX ri

L joX 1 joX ri
l1 ,l2 ,l3 i 1 joX 1 joX ri

Qp
L joX 1 joX ri
i1
li
li
NP
p
N nr
,p p-1nr 1
i
Yi
X
z1
nr ,p
i
p1
i1
p
P
Qmnri
k1
joji Qn -1
ri
i 1
z1
i1
jozlk 1 joz nri

l k jk
Ljozl 1 joz nri

i1
l i ji
Qm
nr i p
i 1
H11
Pp
i 1
nr i
k1

H11 joji
jozlk 1 joz Pp
lk
P
Q pi 1 nri p
i1
jk
i 1
nr
Ljozl 1 joz Pp
i
li
ji
i 1

i
nr

i
1348
Because
p
X
r i 2n 1q 2
i1
p
X
nr i p
i1
then,
p
X
nri n1
i1
p
Y

l
i1
Nn,p p1
n 1
Y
X
C n-1
p,q l1 ,l2 ,l3
z1
i1
Qmn1

Qp
li
k1
jozlk 1 joz Pp
lk
jk
i 1
joX 1 joX ri
Qn1p
H11 joi Qpi 1
L joX 1 joX ri
Ljozl 1 joz Pp
i1
i1
i
li
ji
i 1
nr

i
nr
i
(B-2)
Substitute Eq. (B-2) into Eq. (B-1), then the Proposition 1 holds.
Appendix C. Proof of Proposition 2
Similar to the Proof of Proposition 1,
nL
X
C pj ,qj l1 ,l2 ,l3
Ljo1 jo2n 1
j1

H12n 1qj ,pj jo1 ,. . .,jo2n 1qj
x 1 r x .H r k
nL
X
pj
Y
2n1
X
r 1P
r k 2n 1qj

l
H1rk joX 1 ,. . .,joX rk joX 1 joX rk k
r pj 1 k 1

joX 1 ,. . .,joX rk can be expressed as

H1rk joX 1 ,. . .,joX rk
P
0
1
PnL
Q ni L 1 nrkmi mi z
N nrk ,pkj pkj 1
jol 1 joz n
pki 1nrki 1
nrk
1
nL
i
1
k 1

P
P
Q
Q
l
j
k
B
k k C
Qnrk 1
C npirki,qi li1 ,li2 ,li3
H11 joki
@
A
Ljozl 1 joz n
rk1,...,nrkn 0
z1
i1
k1
i 1
li j
i
L
nPn
i
Pk1
where X
H12n 1 jo1 ,. . .,jo2n 1

lp 1

l
j
jo2n 1qj 1
jo2n 1 pj qj 3 H12n 1qj ,pj jo1 ,. . .,jo2n 1-qj
L
n
nrk 1
i 1 rki
C pkj ,qkj

L joX 1 joX rk
kj 1
where rk 2nrk 1,
PnL
i1
nrki 1 nrk ,
PnL
i1
nrkmi nrk , nrkmj nrkj 1, nrkmi nrki, iaj. Then,

pj
Y

k1
Ppj
n
k 1 rki
C
li1 ,li2 ,li3
p
,q
C pkj ,qkj Qpj i 1 i i

k
k
L
j
o

joX rk
X1
k1
k1

Pp P
Ppj PnL
Ppj
Q kj 1 ni L 1 nrkmi mi
pkj 1
pi 1nrki pj
k 1
k 1
i 1
k
jozlk 1 jozl jn
Y

k1
k k
H11 joki
Ppj
Q k 1 nrk pj
z
z
k1
Ljol 1 jol jn
i1
i
i i
pj
NP
P
nrk ,
pkj
X
z1
P QnL
Because
pj
X
k1
rk
pj
X
2nrk 1 2
k1
pj
X
nrk pj 2n 1qj
k1
so
pj
X
k1
nrk n1
1349
Denote
pj
X
pj
X
nrki n0ji
k1
pj
Y

k1
Qpj
n0ji
Nn,pj
X
i 1 C pi ,qi li1 ,li2 ,li3
C
k 1 pkjk ,qkjk

Qpj
L joX 1 joX rk
k1
Ppj
P QnL
nrkmi n0jmi
k1
k 1
pkj 1
k
P
nL
pi 1n0ji pj
Y i1
z1

H11 joki
P
Q ni L 1 n0jmi mi
k1
k1
jozlk 1 jozl
Qn-1-pj
Ljo
i1
li 1
jo
jnk
z

li jni
Assume that jk j, then

C pkj
nL
Y
,qkj
nL
Y
n0
C piji,qi li1 ,li2 ,li3 C pj ,qj
i1
n0
C piji,qi li1 ,li2 ,li3
i1
nL
Y
n00
i1
nL
nL
nL
X

X
X

pi 1 n0ji pj pj 1
pi 1 n0ji pj
pi 1 n00ji pj
pkjk 1
i1
i1
i1
where n00jj n0jj 1, n00ji n00ji , iaj. Because nrkmj nrkj 1, nrkmi nrki, iaj, it can be obtained that
nL
X
nL
X
n0ji pj
i1
Assume that there are ki terms of C pi ,qi in
Qpj
k1
C pkj
PnL
,qkj
i1
pj
n0ji ki
n0jmi
i1
ki pj , then
pj
X
nrki ki
k1
nrkmi n0jmi
k1
so,
pj
Y
0
n1
X
nj1 0,...,njnL 0;
PnL
n n1
i1 ji

k1
PnL
n,pj
nL
X
BY
nji
B
C
l
,l
,l
pi ,qi i1 i2 i3
@
i 1
z1
i1
pi 1nji pj
Y

H11 joki
P
Q ni L 1 n0jmi mi
k1
Qn1
k1
i1
jozlk 1 jozl
1
C
C
A
n
k jk
Ljozl 1 jozl
i
jni
where nji n0ji ki n0jmi , then,

jo2n 1qj 1
lp
j 1
jo2n 1 pj qj 3
pj
Y

l
H1rk joX 1 ,. . .,joX rk joX 1 joX rk k
k1
n1
X
PnL
nj1 0,...,njnL 0;
i1
pj
Y

k1
PnL
nj1 0,...,njnL 0;
joX 1 joX rk
0
nL
BY
n1
X
i1
nji n1
B
@
nj1 0,...,njnL 0;
PnL
n n1
i1 ji
lk
Nn,pj
i 1
i 1
lp
pi 1nji pj
Y
PnL
i 1
i 1
n0jmi mi
k1
z
z
i 1 Ljoli 1 jol
j 1
k1
nrk
pj
nL
X
X
k1i1
nrkmi
pj
nL X
X

H11 joki
i 1
n0jmi mi mj
i1

H11 joki
Ljozli 1 jozl
P
Q ni L 1 n0jmi mi
k1
Qn-1
i1
i1k1
nL
X
i1
n0jmi
nL
X
i1
jozlk 1 jozl
1
C
C
A
n
k jk
k1
Qn1
k1
nrkmi
jo2n 1 pj qj 3
PnL
pi 1nji pj
Y
jni
jnk
nji n1
jni
jozlk 1 jozl
1
C
C
A
n
k jk
Ljozl 1 jozl
where njmi n0jmi nji , miamj, njmj n0jmj 1 njj 1 :

pj
X
jozlk 1 jozl
Qn-1
k1
z1
i1
PnL

H11 joki
jo2n 1qj 1
PnL
n,pj
nL
X
BY
nji
B
C
l
,l
,l
pi ,qi i1 i2 i3
@
pi 1nji pj
Y
k1
z1
i1
0
n1
X
PnL
z1
i1
nji n1
n,pj
nL
X
BY
n
@
jni
1350
so,
nL
X
njmi
i1
This completes the proof
nL
X
n0jmi 1 n
i1
&.
Appendix D. Proof of Proposition 3

(I) When only the nonlinear term ( )2(d( )/dt) is introduced in the isolation system, i.e. b2a0 and b4 0, only the term
Pn0(jO) is needed to be considered in the force transmissibility under force excitation, Eq. (26):

Qn
z
z

Nn
X
X
1
k 1 jolk 1 jolk jn
jO2
k

9P n0 jO9 2n
Q

2n
n1
o1 o2n 1 O
z
z
2 9LjO9 LjO2
z1
i 1 Ljoli 1 joli jn
k
O2
r
!2n 2

2
2
2
1O
x1 O
22n
Qn
Nn
X
X
o1 o2n 1 O
ozlk 1 jozlk jn
k 1 j
z 1 Qn1
i1
li 1
L1 jo
jo
z
li jnk

x1 jozl 1 jozl
n
i jk
when O 51,
9P n0 jO9 r
1
2n
O2
Nn Y
n
X
X
o1 o2n 1 O
z1k1
9jozlk 1 jozl jn 9
k
1
22n
On 2 c1n
where
c1n

z
Nn Y
n joz

X
lk 1 jolk jnk

o1 o2n 1 O

O
z1
k1
is a bounded constant which is an n dependent but independent of O. So when Oo1,

1
9Pn0 jO9 r 2n On 2 c1n 0 for n 1,2,. . ., N=2
2
when O b1
9P n0 jO9 r
1
2n
Qn
Nn
X
O2
r
!2n 2

2
2
2
x1 O
1O
k 1 j
ozlk 1 jozlk jn
z 1 Qn1
i1
lk 1
z
lk jnk

9L1 jo
jo
x1 jozl 1 jozl jn 9
k
k k

z
Nn Y
2
n X
n joz

j
o
n
X
1 O
O
1 1
lk jk
lk 1
2n 4n 4 2n2

2n 5n c2n
o1 o2n 1 O

O
2 O
2
O
O
z1k1
o1 o2n 1 O
where
c2n

Nn Y
n joz
jozl jn
X
lk 1
k k

o1 o2n 1 O

O
z1
k1
is a bounded constant which is a n dependent but independent of O. So when O b1,

9P n0 jO9 r
So conclusion (I) of Proposition 3 holds.

(II) The proof is given in Theorem 3 of Ref. [17].
1
2n
5n
c2n 0

for n 1,2,. . ., N=2
1351
Appendix E. Proof of Proposition 4

When only ( )2(d( )/dt) term is introduced in the isolator, only Pn0(jO) is needed to be considered in the force
transmissibility under base displacement excitation. Similarly, only P0n(jO) is needed to be consider when only (d( )/dt)3 is
introduced:

Qn
z
z

Nn
X
X
1
k 1 jolk 1 jolk jn
jO4n 4
O4n 4
k

r 1
9P n0 jO9 2n
2n
Q
r
!2n 2

n1
2n
o1 o2n 1 O
z
z

2
2
2 LjO LjO2
z1
i 1 Ljolk 1 jolk jn
2
k
1O2 x1 O
Qn
Nn
X
X
o1 o2n 1 O
k 1 j
z 1 Qn1
i1
lk 1
L1 jo
ozlk 1 jozlk jn
jo
z
lk jnk

x1 jozl 1 jozl
n
k jk

Q3n
z
z

Nn
X
X
1
k 1 jolk 1 jolk jn
jO4n 4
k

9P 0n jO9 2n
Q

2n
n1
o1 o2n 1 O
z
z
2 9LjO9 LjO2
z1
i 1 Ljolk 1 jol jn
k
22n
o1 o2n 1 O
r
!2n 2

2
2
1O2 x1 O
Q3n
Nn
X
k 1 j
z 1 Qn1
i1
O4n 4
lk 1
L1 jo
ozlk 1 jozlk jn
jo
z
lk jnk

x1 jozl 1 jozl
n
k jk
(I) When O 51
9P n0 jO9 r
O4n 4
2n
Nn Y
n
X
X
o1 o2n 1 O
9jozlk 1 jozl jn 9
k
z1k1
1
22n
O5n 4 c1n
where
c1n

z
Nn Y
n joz

X
lk 1 jolk jnk

o1 o2n 1 O

O
z1
k1
is a bounded constant which is a n dependent but independent of O. So when O 51,

1
9P n0 jO9 r 2n O5n 4 c1n 0 for n 1,2,. . ., N=2
2
and
9P 0n jO9 r
1
2n
O4n 4
Nn Y
3n
X
X
o1 o2n 1 O
z1k1
9jozlk 1 jozl jn 9
k
1
22n
O7n 4 c10 n
where
0
c1 n

z
Nn Y
3n joz

X
lk 1 jolk jnk

o1 o2n 1 O

O
z1
k1
is a bounded constant which is a n dependent but independent of O. So when Oo1,

1
9P 0n jO9 r 2n O7n 4 c1n 0 for n 1,2,. . ., N=2
2
(II) When O b1,
9P n0 jO9
4n 4
1 O
O
22n O4n 4 O2n2
X
o1 o2n 1 O
22n On2

Nn Y
n joz
jozl jn
X
lk 1
k k

O
z1k1
c1n
1352
where

z
Nn Y
n joz

X
lk 1 jolk jnk

o1 o2n 1 O

O
z1
c1n
k1
is a bounded constant which is an n dependent but independent O So when O b1,

While

Nn Y
3n joz
jozl jn
X
1 O4n 4 O3n X
1 n2 0
lk 1
k k
9P 0n jO9 r 2n 4n 4 2n2
c1 n

2n O
o
1 o2n 1 O

O
2 O
2
O
z1k1
where

z
Nn Y

3n joz
X
lk 1 jolk jnk

o1 o2n 1 O

O
z1
c10 n
k1
is a bounded constant which is dependent on n but independent of O. So when O b1, the upper limit of 9P0n(jO)9 is
proportional to the square of the exciting frequency, and then the force transmissibility when (d( )/dt)3 term is
introduced in the isolator under base excitation in high frequency is much larger than that in linear case.
(III) The proof is given in Ref. [17].
Appendix F. Proof of Proposition 5

P

PnL
Q ni L 1 njmi mi

N n,pj
pi 1njmi 1
1 j X
jozlk 1 jozl jn
i 1 Y
X

k
1

k k
1
9Pnk jO9 2n
H1 joki

Qn1
o1 o2n 1 O
z
z
2 LjO

z1
k1
i 1 Ljolk 1 jolk jn

k
P
P
Q ni L 1 njmi mi
Q ni L 1 pi 1njmi 1 1
N n,pj
jozlk 1 jozl jn
X k1
9H1 jO9 X
1
k k
k1
r 2n
Q
n1
o1 o2n 1 O
z
z
9LjO9
2
z1
i 1 Ljolk 1 jolk jn
k
q PnL pi 1njmi 1
P
i 1
Q ni L 1 njmi mi
2
Nn,pj
1 Ox1
jozlk 1 jozl jn
X
X k1
1
k k
PnL
2n
Q
!
n1
o
O
z
z
r
1
2n 1
pi 1njmi 2
2

Lj
o

j
o
n
i
1

2
z1
i1
lk 1
l k jk
2
1O2 Ox1
(I) When O 51
9P nk jO9
1
22n
PnL
n m
i 1 jmi i
PnL
N n,pj
o1 o2n 1 O
1
22n
PnL
z1
n m
i 1 jmi i
z

z
Ynjmi mi jolk 1 jolk jnk

O
i 1
k1
c1n
where
c1n
X
o1 o2n 1 O
Nn,pj
z1
PnL
z

z
Ynjmi mi jolk 1 jolk jnk

O
i 1
k1
is a bounded constant which is dependent on n but independent of O. So when O 51,

1 PnL
9Pnk jO9 r 2n O i 1 njmi mi c1n 0 for n 1,2,. . ., N=2
2
(II) When O b1, 9Pnk(jO)9 is considered when only ( )2(d( )/dt) term or only (d( )/dt)3 is introduced in the isolator
respectively. The upper limits of 9Pnk(jO)9 under different conditions are presented in Table F1 where
PnL

N n,pj
z
n m
jozl jn
1 jmi i jo
X i Y
1 X
lk 1
k k
C_const n 2n

o1 o2n 1 O

O
2
z1
k1
is a bounded constant which is dependent on n but independent of O.
1353
Table F1
Upper limit.
pi 0, n 1
pi 1
pi 2
pi 3
( )2(d( )/dt)
(d( )/dt)3
(1/O)C_const(1)
[1/(On 1)]C_const(n)
[1/(O2n 1)]C_const(n)
[1/(O3n 1)]C_const(n)
OC_const(1)
[On 1]C_const(n)
[1/O]C_const(n)
[1/(On 1)]C_const(n)
In the table above, the second column is the upper limit when only ( )2(d( )/dt) is introduced, and the third column is
the upper limit when only (d( )/dt)3 is introduced. The second row represents the upper limit of the pure cubic input
nonlinearity, and the third to fth rows represent the upper limit when all the n terms of nonlinear coefcients having
the same degree of output nonlinearity pi. It can be seen from the second column that the upper limit of the pure cubic
input nonlinearity is proportional to1/O, which tends to zero the slowest while comparing to the third to fth row. So
the absolute displacement transmissibility when only ( )2(d( )/dt) is introduced in the isolator is very close to that in
the linear case, i.e., only a little larger than the linear absolute displacement transmissibility.
In the third column, the upper limit of the cubic pure input nonlinearity is proportional to the frequency O, and the
upper limit when all the nonlinear coefcients are composed of C1,2(1,1,1)is proportional to On 1, so the absolute
displacement transmissibility in high frequency when only (d( )/dt)3 term is introduced in the isolator under base
excitation is much larger than that in linear case.
(III) Consider rstly that only ( )2(d( )/dt) is introduced in the isolator:
2
0
13
!
bN=2c
nL
n
X
Y
X B
njmi
6
C7
2
C pi ,qi li1 ,li2 ,li3 P nk jO A5
TO 4P0 jO
@
PnL
n1
i
1
n n
n ,...,njmnL 0,
i 1 jmi
2
0 jm1
13
!
bN=2c
nL
n
X
Y
X
n
6
B
C7
4P 0 jO
C pijmi
@
,qi li1 ,li2 ,li3 P nk jO A5
PnL
n1
i1
njm1 ,...,njmnL 0,
2bN=2c
X
n
X
n0 q0
i 1
nL
Y
n
X
jm1 ,...,njmn 0
L
nPn
L
n
q,
i 1 jmi
PnL
njmi n
njmi n0jmi
C pi ,qi
li1 ,li2 ,li3 P qk1 jOPnqk2 jO
i1
n0
n-q
i 1 jmi
and
2bN=2c
X
n
X
n2 q0
dTO2
ReP 0 jOP 1 jO
db2
"
#

n
nL
L
X
njmi n0jmi dC p ,q Y
n
X
jm1 ,,njmn 0
L
nPn
L
n
q,
i 1 jmi
PnL
i1
C pi ,qi
db2
njmi n0jmi
C pi ,qi
li1 ,li2 ,li3 P qk1 jOPn-qk2 jO
i1
n0
n-q
i 1 jmi

2
4
3
when O E1, P0(j) (1jx1)/x1, P 1 j 1=x1 1=4x1 j 1=2x1 3=2x1
!
1
1
1
ReP 0 jP1 j
3 5 o0
2x1 2x
4x1
1
Therefore, when O E1, there must exist a b 40 such that

dTO2
o0
db2
if 0 o b2 o b
When only the nonlinear term (d( )/dt)3 is introduced in the isolator, the proof can be done by following the same
procedure. Then Proposition 5 holds.
References
[1]
[2]
[3]
[4]
[5]
[6]
R.A. Ibrahim, Recent advances in nonlinear passive vibration isolators, Journal of Sound and Vibration 314 (2008) 371452.
L. Meirovitch, Fundamentals of vibrations, McGraw-Hill, New York, 2001, pp. 110130.
S.S. Rao, F.F. Yap, Mechanical Vibrations, Addison-Wesley, New York, 1995, pp. 235244.
B. Ravindra, A.K. Mallik, Performance of non-linear vibration isolators under harmonic excitation, Journal of Sound and Vibration 170 (1994) 325337.
Z.Q. Lang, S.A. Billings, R. Yue, J. Li, Output frequency response function of nonlinear Volterra systems, Automatica 43 (2007) 805816.
Z.Q. Lang, X.J. Jing, S.A. Billings, G.R. Tomlinson, Z.K. Peng, Theoretical study of the effects of nonlinear viscous damping on vibration isolation of sdof
systems, Journal of Sound and Vibration 323 (2009) 352365.
1354
[7] G.N. Jazar, R. Houim, A. Narimani, M.F. Golnaraghi, Frequency response and jump avoidance in a nonlinear passive engine mount, Journal of Vibration
and Control 12 (2006) 12051237.
[8] G. Popov, S. Sankar, Modelling and analysis of non-linear orice type damping in vibration isolators, Journal of Sound and Vibration 183 (1995)
751764.
[9] P. Yang, J. Yang, J. Ding, Dynamic transmissibility of a complex nonlinear coupling isolator, Tsinghua Science and Technology 11 (2006) 538542.
[10] E.V. Isolators, On the modelling of non-linear elastomeric vibration isolators, Journal of Sound and Vibration 219 (1999) 239253.
[11] C.M. Richards, R. Singh, Characterization of rubber isolator nonlinearities in the context of single- and multi-degree-of-freedom experimental
systems, Journal of Sound and Vibration 247 (2001) 807834.
[12] X.J. Jing, Z.Q. Lang, S.A. Billings, Output frequency response function-based analysis for nonlinear Volterra systems, Mechanical Systems and Signal
Processing 22 (2008) 102120.
[13] X.J. Jing, Z.Q. Lang, S.A. Billings, G.R. Tomlinson, The parametric characteristic of frequency response functions for nonlinear systems, International
Journal of Control 79 (2006) 15521564.
[14] X.J. Jing, Z.Q. Lang, S.A. Billings, G.R. Tomlinson, Frequency domain analysis for suppression of output vibration from periodic disturbance using
nonlinearities, Journal of Sound and Vibration 314 (2008) 536557.
[15] X. Jing, Z. Lang, Frequency domain analysis of a dimensionless cubic nonlinear damping system subject to harmonic input, Nonlinear Dynamics 58
(2009) 469485.
[16] Z. Milovanovic, I. Kovacic, M.J. Brennan, On the displacement transmissibility of a base excited viscously damped nonlinear vibration isolator, Journal
of Vibration and Acoustics 131 (2009) 054502054507.
[17] X.J. Jing, Z.Q. Lang, S.A. Billings, Nonlinear inuence in the frequency domain: alternating series, Systems and Control Letters 60 (2011) 295309.
[18] A.K. Swain, S.A. Billings, Generalized frequency response function matrix for MIMO non-linear systems, International Journal of Control 74 (2001)
829844.
[19] X.J. Jing, Z.Q. Lang, S.A. Billings, Frequency domain analysis for non-linear Volterra systems with a general non-linear output function, International
Journal of Control 81 (2008) 235251.
[20] S.A. Billings, J.C. Peyton Jones, Mapping non-linear integro-differential equations into the frequency domain, International Journal of Control 52
(1990) 863879.

Journal of Sound and Vibration: Zhenlong Xiao, Xingjian Jing, Li Cheng

Transféré par

Informations du document

Description originale:

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Journal of Sound and Vibration: Zhenlong Xiao, Xingjian Jing, Li Cheng

Transféré par

Droits d'auteur :

Formats disponibles

Journal of Sound and Vibration 332 (2013) 13351354

Contents lists available at SciVerse ScienceDirect

Journal of Sound and Vibration

The transmissibility of vibration isolators with cubic nonlinear

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

2. Nonlinear vibration isolators and transmissibility functions

where A is the amplitude and o is the excitation frequency.

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

The force ratio Tfr(t) is dened as

By dening the following non-dimensional parameters:

The force ratio Tfr(t) in this case is denoted by

Denote the relative displacement x of the isolator as

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

Dene the following non-dimensional parameters:

Then Eq. (12) can be rewritten as the following non-dimensional form:

where x is the displacement ratio and u1 is given by

Using the following non-dimensional variables:

Eq. (17) can be written into the following non-dimensional form:

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

3.1. The force transmissibility Tf(O)

HJn jo1 ,. . .,jon U o1 U on ,

Y 2 jo P0 jo P10 job2 P 11 job4

where oi 2 fO, Og,i 1,. . .,2n 1.

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

where oi 2 fO, Og,i 1,. . .,2n 1.

3.2. The displacement transmissibility Td(O)

C 3,0 0,0,1 b2 ,C 3,0 1,1,1 b4 ,

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

Proof. See Appendix C &.

4. The effects of nonlinear damping coefcients on vibration isolation

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

(II) When O b1, only ( )2(d( )/dt) term can make

While the nonlinear term (d( )/dt)3 will make

while the nonlinear term (d( )/dt)3 will make

(III) When O E1, there exists a b 40 such that

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

5. Simulation studies and discussions

force transmissibility Tf ,20*log10(|Y2(j)|)

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

force transmissibility Tf ,20*log10(|Y2(j)|)

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

-0.8 -0.6 -0.4 -0.2

force transmissibility Tf ,20*log10(|Y2(j)|)

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

H13 jo1 ,jo2 ,jo3 b2

H12n jo1 ,. . .,jo2n 0, n 1,. . .,bN=2c

jnk 2 f1,3,. . .,2n1g

i 1,. . .,n1,j 1,. . .,jni ,n Z2

k 1,. . .,n1,j 1,. . .,jk ,n Z2

Appendix B. Proof of Proposition 1

where r i 2nri 1. Then,

l1 ,l2 ,l3 i 1 joX 1 joX ri

jozlk 1 joz  nri

Ljozl 1 joz  nri

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

C pj ,qj l1 ,l2 ,l3

H12n 1 jo1 ,. . .,jo2n 1

nrkmi nrk , nrkmj nrkj 1, nrkmi nrki, iaj. Then,

Z. Xiao et al. / Journal of Sound and Vibration 332 (2013) 13351354

Assume that jk j, then

C piji,qi li1 ,li2 ,li3 C pj ,qj

jozlk 1 joz nri

Ljozl 1 joz nri