Dynamic stiffness and loss factor measurement of engine rubber mount

by impact test
Lu Ean Ooi, Zaidi Mohd Ripin

School of Mechanical and Aerospace Engineering, Engineering Campus, Universiti Sains Malaysia (USM), 14300 Nibong Tebal, SPS, Pulau Pinang, Malaysia
a r t i c l e i n f o
Article history:
Received 13 August 2010
Accepted 6 December 2010
Available online 13 December 2010
Keywords:
A. Elastomers and rubber
F. Elastic behaviour
H. Selection of material properties
a b s t r a c t
Dynamic stiffness and loss factor for engine rubber mount are important dynamic behaviours to repre-
sent the performance of an engine mount system. The investigation of the dynamic behaviour of engine
mount system using impact technique, particularly in the simultaneous measurement of the dynamic
transfer stiffness and driving point stiffness where the impact hammer replaces the shaker as the source
of excitation is presented in this paper. The results showed that the dynamic driving point stiffness can
only be used to represent the dynamic transfer stiffness for the lower range of frequency. The curve tted
functions of the loss factor obtained from the dynamic driving point stiffness measurement showed lin-
ear dependency on the frequency and the loss factor obtained from the transfer stiffness measurement
showed non-linear dependency on the frequency. Both of the stiffnesses are accurately reproduced by
using these functions. The values of both dynamic stiffnesses obtained from impact technique are vali-
dated with the values obtained from shaker.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
Engine mounts are commonly used to provide vibration attenu-
ation to isolate the vibration source. Accurate dynamic measure-
ment is important to predict the dynamic behaviour of the
engine rubber mount such as stiffness and damping. The existing
damping measurement techniques are generally divided into reso-
nance and non-resonance based methods. Theoretical studies and
denitions exist for each measurement methods. Different location
of the sensors will represent different measured parameters; i.e.
input force and output force. Proper identication of these cap-
tured signals is important because they determine the dynamic
parameters being measurement. The identication of each must
matched the standard denition of the measurement method so
that the correct mathematical model can be selected.
There have been several important work carried out in the past
to improve the understanding and the measurement techniques of
the dynamic behaviour of engine rubber mounts. The concept of
complex stiffness with viscous and hysteresis damping was exqui-
sitely explained by Neumark [1] for a single degree of freedom
(SDOF) system where different cases for damping i.e. harmonic
oscillations and decaying oscillations are re-examined and com-
pared. The measurement of dynamic stiffness of an isolator by
using shaker was presented by Gade et al. [2] by using resonant
and non-resonant methods. The input force and acceleration are
measured to obtain the dynamic stiffness of an isolator. The stiff-
ness measurement at discrete frequencies using shaker was inves-
tigated to model the non-linear elastomeric vibration isolators [3].
The input force fromthe shaker and the deformation of specimen is
used to generate hysteresis loops so that the stiffness of engine
mount is obtained. Dynamic testing to account for non-linear
effects of rubber compounds was conducted by Ramorino et al.
[4] using shaker. The tested frequency range was set up to
1000 Hz. Dynamic modulus was studied instead of dynamic stiff-
ness through the analysis of transmissibility of specimen. Nader
and Ken [5] developed a high frequency testing machine for mea-
suring rubber mount dynamic stiffness up to 5 kHz and the associ-
ated mathematical model. Foumani et al. [6] developed a
technique to optimize the properties of the engine mount in order
to minimize the steering wheel and chassis vibrations. Mundo et al.
[7] had measured and modelled the dynamic stiffness of automo-
tive rubber connections. Ladislav et al. [8] studied the rubber
element for reduction of vibration from railway wheels by deter-
mining the stiffness and phase angle for different preload condi-
tions and force amplitudes using wattmetric method and discrete
Fourier transformation for increase reliability of the data. More re-
cently, Kulik et al. [9] conducted experimental measurement of dy-
namic properties of viscoelastic materials with loaded inertial
mass. Hofer and Lion [10] modelled the frequency and amplitude
dependent properties of carbon black lled rubber. The dynamic
properties are indentied by the storage modulus and loss modu-
lus. A non-linear viscoelastic constitutive model of rubber was
0261-3069/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2010.12.015

Corresponding author. Tel.: +60 4 5941024; fax: +60 4 5941025.

E-mail address: mezaidi@eng.usm.my (Z.M. Ripin).
Materials and Design 32 (2011) 18801887
Contents lists available at ScienceDirect
Materials and Design
j our nal homepage: www. el sevi er . com/ l ocat e/ mat des
developed by Ciambella et al. [11] where compression test was
done to obtain material parameters to give better estimation of
the stress response. Hadi et al. [12] developed an indirect method
to measure the input force for power train of a bus in order to
determine the optimal position of engine mount to reduce force
transmission frequency independent stiffness values were used
in their work. It is important to determine the frequency depen-
dent stiffness to give better modelling capabilities and also more
accurately represent the behaviour of the rubber engine mount.
Complex stiffness method is commonly used for damping and
loss factor determination for resilient elements. Some researchers
[1315] adopted this method in their study to further characterize
these elements. The measurement for dynamic stiffness is gener-
ally divided into direct and indirect method. BS ISO 10846 [16] pro-
vided different denitions of the dynamic stiffnesses and the
corresponding measurement set up. There are two different types
of dynamic stiffnesses under direct complex stiffness measure-
ment which are listed in BS ISO 10846. These are dynamic driving
point stiffness (k
11
) and dynamic transfer stiffness (k
21
). BS ISO
10846 listed the experimental measurement set up required for
each dynamic stiffnesses. Shaker or hydraulic actuator is a device
commonly used as vibration exciter in damping measurement
[2,1721]. Direct complex stiffness method has been applied on
an engine mount to evaluate its dynamic behaviour [21]. Dynamic
transfer stiffness was measured and used to estimate the loss fac-
tor. However, the related theoretical explanation was not stated.
The indirect method from BS ISO 10846 was used by Thompson
[22] in the study of resilient rail pad. Some renements on the
method were proposed to complete the application of the indirect
method.
Impact technique was recently developed by Lin et al. [23] to
evaluate the frequency dependent stiffness and loss factor of en-
gine mount where the k
11
was measured. The impact technique
is a simple and powerful technique because it replaces the cumber-
some shaker in the experimental set up for damping measurement.
The impact test can also encourage the measurement to be done
under the real operating conditions i.e. the engine mount which
is installed in a real vehicle because the set up is much easier.
Many researchers [2,3,24] measured k
11
instead of k
21
. How-
ever, the denitions for k
11
and k
21
are different due to the different
location of the sensors as listed in BS ISO 10846. In the actual appli-
cation case for engine rubber mount, it is difcult to mount the
load cell underneath the mount (i.e. below the engine) and it is
much easier to use the impact hammer to characterize the engine
mount (which may be mounted at certain angle). Further develop-
ment in the measurement of the k
21
is important because k
21
can
be used for wider range of frequency and not affected by the mass
between the force transducer and the engine mount [20]. The im-
pact technique developed by Lin et al. [23] was proven only for the
k
11
. The measurement of k
21
is now further developed in this paper
by using impact technique so that the dynamic analysis on the
resilient element or engine mount using impact technique can be
used for both dynamic stiffnesses.
A set up which can measure both the dynamic driving point
stiffness and dynamic transfer stiffness is proposed. The measured
dynamic transfer stiffness is then used to estimate the frequency
dependent stiffness and loss factor of the engine mount. The re-
sults from dynamic transfer stiffness measurement are compared
with the measurement of dynamic driving point stiffness using
the impact test. Mathematical modelling for k
11
and k
21
related
to SDOF system and resilient element is derived clearly to show
the effect of the different dynamic stiffness measurement methods.
The deviation and limitation of k
11
and k
21
using both shaker and
impact hammer are shown experimentally. The engine rubber
mount system is then tested using shaker as source of excitation
to validate the measured data from the impact test method.
2. Theoretical background
According to BS ISO 10846, dynamic driving point stiffness (k
11
)
is the frequency dependent ratio of force on the input side of vibra-
tion isolation to the displacement on the input side with the output
side blocked. As shown in Fig. 1a, dynamic driving point stiffness is
dened as k
11
= F
1
/u
1
; where subscript 1 indicated that the force
and displacement are measured on the input side of the vibration
isolator. The isolator in Fig. 1a is subjected to the input force, F
1
on
the top and the response, u
1
is measured also at the input side (on
the top). The dynamic driving point stiffness, k
11
is mainly deter-
mined by elastic and dissipative force.
The dynamic transfer stiffness (k
21
) is the frequency dependent
ratio of the blocked force on the output side of the vibration isola-
tor to the displacement on the input side. Based on Fig. 1b, dy-
namic transfer stiffness is dened as k
21
= F
2
/u
1
; where subscripts
1 and 2 denote the input and output respectively. Fig. 1b shows
the condition where excitation is made at the input and transmit-
ted force is measured at the blocked output, F
2
. In the case of the
dynamic transfer stiffness, the response is also measured at the in-
put side.
2.1. Dynamic driving point stiffness (k
11
)
The dynamic driving point stiffness can be used to represent the
dynamic behaviour of a SDOF system. For SDOF system with hys-
teretic damping, the equation of motion for this rubber mass sys-
tem can be written as below:
mxt kx1 ig
11
xxt F
1
1
where m is the mass of the system, k(x) is the frequency dependent
stiffness and g
11
(x) is the loss factor. When the system is subjected
to the applied force where
F
1
t F
1
expixt
The response of the system will become
xt H
11
ixF
1
expixt
where H
11
(ix) is the complex receptance function which is dened
as the ratio of the output displacement to the input force of the
system as a function of frequency and can be divided into real
(Re{H
11
(ix)}) and imaginary (Im{H
11
(ix)}) part.
The dynamic driving point stiffness (k
11
) is the reciprocal of the
receptance function
Fig. 1. The concept of (a) dynamic driving point stiffness and (b) dynamic transfer
stiffness for isolator.
L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887 1881
k
11
mx
2
kx1 ig
11
x 2
Phase angle, h obtained in this measurement is the phase lag be-
tween the output (displacement) and the input (force) where
tanh ImfH
11
ixg=RefH
11
ixg 3
The phase angle (h) is related to the loss angle (u) by the equa-
tion below:
g
11
tanu tanh1 x
2
=x
2
n
4
In here, x
n
is the natural frequency of the system and g
11
is the loss
factor obtained from the measurement of the dynamic driving point
stiffness.
2.2. Dynamic transfer stiffness (k
21
)
Dynamic transfer stiffness is normally used in the forced vibra-
tion study of resilient element. The equation of motion for the
resilient element is as below:
cdxt=dt kxt F
2
5
where c is the damping coefcient and k is the stiffness. The stiff-
ness k can be frequency dependent or not. The element is then ex-
cited by using an impact hammer and the transmitted force through
the rubber mount is F
2
(t) = F
2
exp ixt. The response of the system
which is measured at the input point become
x
1
t H
21
ixF
2
expixt 6
The receptance function is given by H
21
(ix) = 1/c(ix) + k.
Since tan u = cx/k, then H
21
(ix) = 1/k(1 + i tan u).
The spectrum of the force and displacement can be measured
independently and used to determine the dynamic transfer stiff-
ness which is the ratio of force to displacement and can be written
as:
k
21
k1 i tanu 7
The element loss angle (u) or sometimes called structural
damping factor is dened by
g
21
tanu ImfH
21
ixg=RefH
21
ixg 8
If the element mass is negligible, the reciprocal of the receptance
function 1/H
21
(ix) is also called the complex stiffness of the ele-
ment. For k
21
, the loss angle, u is also the phase angle, h. However,
for k
11
, the loss angle, u is different from the phase angle, h.
The real part (Re{H(ix)}) and the imaginary part (Im{H(ix)}) of
the receptance function can be obtained from experimental mea-
surement. The phase angle h and u also can be directly obtained
by referring to the location of the sensors. Care should be taken
in locating the sensors in the experiment because the signal from
different location of the system represented different dynamic
behaviour for the system or the resilient element. The selection
of x
n
in the case of dynamic driving point stiffness measurement
must be done carefully because the accuracy for this value will
inuence the accuracy of the damping measurement results. Con-
tinuity across resonant band will occur for the correct value for x
n
[23].
3. Experimental setup
Solid engine rubber to metal mount with a diameter of 15 mm
and 20 mm length is selected as the resilient element test object
for this experimental investigation. The engine rubber mounts
are used for mounting of a 2.72 kg small utility two stroke engine
of a backpack type grass trimmer.
The experimental setup which is used for both dynamic driving
point stiffness and dynamic transfer stiffness are shown in Fig. 2a.
This setup consists of three engine mounts, impact hammer
(Kistler, type: 9724A5000), an accelerometer (Kistler, type:
8776A50), analyzer (LMS spectral testing), and load cell (Kistler,
type: 9272). A preload mass of 2.72 kg is mounted on the engine
mount to apply the preload similar to engine mass in the real oper-
ating conditions. As shown in Fig. 2a, the load cell is placed at the
bottom of rubber mount. It is used to measure the blocked output
force of system. The hammer is used to excite the test specimen
and the accelerometer is mounted at the bottom of preload plate
and used to record the input side acceleration. Calibration of sen-
sors is done before each measurement.
Fig. 2b shows the experimental set up where the shaker is used.
The impact hammer is now replaced by the electromagnetic shaker
as the excitation source. The force transducer is connected to the
shaker via a stringer. This force transducer is to measure the input
force from the shaker recorded using LMS software (spectral test-
ing). This analyzer is also used to drive the shaker and also re-
corded data from both the shaker and hammer in order to ensure
that the measurement results as comparable.
This setup provides a quick way to measure the driving point
stiffness and the transfer stiffness at the same time without mak-
ing any changes on the setup. However, the output base of the sys-
Fig. 2. Experimental setup for k
11
and k
21
measurement by using (a) impact and (b) shaker.
1882 L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887
tem must be xed so that the output force measured can be con-
sidered as blocked force. This setup can also be used to measure
the force transmission of the engine rubber mounts. The resulting
signals captured from the hammer, force transducer, load cell and
accelerometer are then analyzed by using LMS software (spectral
testing). Load cell is connected to channel 1, hammer or force
transducer is connected to channel 2 and accelerometer is con-
nected to channel 3. The signals are all exponentially weighted.
The frequency response function (acceleration/force) is estimated
using multi channels fast Fourier transformation calculation. Five
averages are taken where the average type is linear average. The
unity value in coherence function for all frequency shows the lin-
ear dependency between force and displacement signals.
For dynamic driving point stiffness, the excitation force from
hammer is recorded and the input side acceleration of the preload
plate is measured. Both signals are captured at the centre of the
plate. The ratio of these signals and its phase lag are recorded
and subsequently used to calculate the loss angle. For the dynamic
transfer stiffness, the output blocked force measured using load
cell mounted at the bottom of the engine rubber mounts. The foun-
dation of the system must be conrmed to be rigid enough to pro-
vide the blocked output for the system. For the case of the shaker
excitation, the input force from shaker is recorded by the force
transducer which is connected to the shaker via stringer. By refer-
ring this captured signals to the mathematical formulation in Sec-
tion 2, F
1
is the force from either shaker or hammer, u
1
is from the
accelerometer and the F
2
is the blocked force from load cell.
4. Results and discussion
The receptance function of the system is measured and re-
corded in magnitude, real and imaginary part. These data are then
used to calculate the dynamic driving point stiffness (k
11
), dynamic
transfer stiffness (k
21
) and loss factor (tan u) by using the equa-
tions shown in Section 2. The measured receptance function for
the preload mass of 2.72 kg and 0.90 kg are shown in Fig. 3. The
data were obtained directly from measurement without curve t-
ting. The tested frequency range is from 0 Hz to 200 Hz. However,
the results shown are only from 10 Hz due to poor coherence be-
low 10 Hz. This is due to the slender shape of the engine mount
which caused exing of the mount at low frequency. The real (Re{-
H(ix)}) and imaginary (Im{H(ix)}) parts of the receptance function
are also recorded from an online function and used to calculate the
loss factor.
The natural frequency of the system is indicated by the peak va-
lue of the receptance function as shown in Fig. 3. In this system
with the preload mass of 2.72 kg, the natural frequency of the sys-
tem is 91 Hz. The small irregular peak at around 72 Hz is caused by
the plates of preload mass. For fabrication purpose, this 2.72 kg
preload mass consists of three separate plates (as in Fig. 2a). The
plates are bolted together. Some rocking motion may have oc-
curred between the plates and this generated this small irregular
peak. This was conrmed by running the test with the preload
mass reduced to only one plate of 0.90 kg. Fig. 3 shows the recep-
tance function for preload mass with a single plate. The result
showed that the irregular peak at around 72 Hz has disappeared.
At the same time, since the mass of the preload is reduced, the res-
onant frequency of the system is shifted to 145 Hz.
Fig. 4 shows the comparison between k
21
and k
11
which are
measured using the impact test method where k
21
is frequency
dependent and increases with the frequency. The trend of this dy-
namic transfer stiffness is similar to the one reported by Nadeau
and Champoux [20] where the shaker was used as the source of
vibration exciter. It increases slightly from 600 kN/m to 800 kN/
m. This value is lower compared to the results reported by Nadeau
and Champoux [20] due to the different type of engine mount and
the shape of engine mount which has a signicant effect on the
damping and stiffness capacity of engine mount [7,20]. The dy-
namic driving point stiffness is also successfully measured by using
the impact test. The measured dynamic driving point showed a
drop in the resonant frequency which is also reported by Morison
et al. [21]. The authors explained that the dynamic stiffness is al-
most constant at the lower frequency and begin to drop at resonant
frequency. This is caused by the system resonance effect.
The results in Fig. 4 also showed that both dynamic stiffness
values are close to each other in the low frequency range. It started
to deviate signicantly from 30 Hz. In this case at frequencies be-
low 30 Hz, the dynamic driving point stiffness can be used to rep-
resent dynamic transfer stiffness. This is caused by the inertial
force at lower frequencies are small compared to elastic force in
the system [21]. However, the limit of this frequency range is
determined by the properties of the test rig. The approximation
of dynamic driving point stiffness and dynamic transfer stiffness
is no longer valid if the inertial force of the mass (moving mass
or internal mass of resilient element) is high enough compared
to the elastic forces especially at high frequency. This frequency
limit has also been reported by Morison et al. [21] by using resil-
ient element fastening as specimen and shaker as source of excita-
tion. However, in this paper this frequency range is experimentally
tested by changing the attached mass from 2.72 kg to 0.90 kg to
look at the effect of preload mass. It is benecial for some testing
which is only interested in low frequency range since the measure-
ment of dynamic driving point stiffness is easier and faster com-
pared to dynamic transfer stiffness in real application case.
However, in most cases, the dynamic transfer stiffness is preferred
Fig. 3. Measured receptance function for system. Fig. 4. Comparison of k
11
and k
21
(m = 2.72 kg).
L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887 1883
to the driving point stiffness since the driving point stiffness is case
sensitive to the mass between force transducer and engine mount
[21].
One of the causes for the deviation between k
11
and k
21
is the
mass between the test isolator and the input force transducer
[15]. This mass will cause some bias error in the input force mea-
surement and limit the frequency range for the correct measure-
ment of the dynamic driving point stiffness (k
11
). The error is
signicant especially for large mass. It means that the measure-
ment for k
11
is valid only when the mass between the test isolator
and the input force transducer is small enough to be considered as
mass less system. In order to look into more detail on this effect,
the tests for dynamic driving point stiffness and dynamic transfer
stiffness measurement are repeated by reducing the mass between
the test isolator and the input force transducer to 0.90 kg. Similar
experimental procedures were applied. The result after mass
reduction is shown in Fig. 5 where the dynamic driving point stiff-
ness and the dynamic transfer stiffness begin to deviate from
60 Hz. The low frequency range which the dynamic driving point
stiffness can be used to represent dynamic transfer stiffness be-
comes wider which is from 30 Hz to 60 Hz. The results also indi-
cated that the frequency range will become wider if the mass
become smaller. Since the mass of the system was changed, the
natural frequency where the minimum dynamic driving point stiff-
ness occurred became higher. Although the mass is changed, the
values for stiffness still remain the same for the dynamic transfer
stiffness. For dynamic driving point stiffness, the values are higher
when the preload mass is 2.72 kg especially at higher frequency
where the inertia effect takes place.
Although the preload mass has changed, the values of the
dynamic transfer stiffness still remain the same within 600
800 kN/m range. However, for dynamic driving point stiffness,
the values are within the range of 04000 kN/m for the preload
mass of 2.72 kg and this range is reduced to 01000 kN/m when
the preload mass is changed to 0.90 kg. This means that higher pre-
load mass produced high values of dynamic driving point stiffness.
This is because the preload mass dominated the contributions to
the value of dynamic driving point stiffness and the stiffness dom-
inated the response at low frequencies [14,25].
The loss factors which are calculated from dynamic driving
point stiffness and dynamic transfer stiffness by using Eqs. (4)
and (8) are shown in Figs. 6 and 7. The loss factor (g
11
) of the en-
gine rubber mount system calculated from the dynamic driving
point stiffness showed large variation of the damping curve at res-
onance as shown in Fig. 6. Similar nding was reported by Lin et al.
[23]. This large variation or discontinuity at resonance is caused by
the real part of the measured receptance function at resonance is
not exactly cancelled by the zeros in the numerator. Since there
is a small irregular peak at the measured receptance function, sim-
ilar large variation at resonance also occurred at that frequency.
However, these large variations in the resonant frequency did not
happen for the case where loss factor (g
21
) is calculated from dy-
namic transfer stiffness as shown in Fig. 7.
The loss factor (g
21
) curve in Fig. 7 showed the frequency
dependent behaviour with the value in the range of 0.010.18.
Similar observation was also reported by Gade et al. [2] and Na-
deau and Champoux [20]. Nadeau and Champoux [20] explained
that there were some uncertainties in loss factor curve which make
the loss factor curve not smooth compared to the stiffness curve.
This uncertainty was caused by the sensitivity of the loss factor
to the boundary conditions and also to the rigid body modes. Gade
et al. [2] also proved that there is irregular peak in loss factor due
to the stinger resonance.
5. Polynomial curve tting of loss factor
The polynomial curve tting approach is applied to the mea-
sured loss factor fromboth dynamic driving point stiffness and also
the dynamic transfer stiffness. The curve tted function shows that
the loss factor (g
11
) is a linear function of frequency as shown in
Fig. 6.
The function of curve tted loss factor (g
11
) from polynomial
curve tting is Fig. 5. Comparison of k
11
and k
21
(m = 0.90 kg).
Fig. 6. Loss factor (g
11
) calculated from dynamic driving point stiffness (k
11
) and
the polynomial curve tting with m = 2.72 kg.
Fig. 7. Loss factor (g
21
) calculated from dynamic transfer stiffness (k
21
) and the
polynomial curve tting with m = 2.72 kg.
1884 L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887
g
11
f 0:000617f 0:03389 9
This function can be used to predict the loss factor at any fre-
quency within the frequency range of 0200 Hz. The loss factor
at resonant frequency is then evaluated by using the half power
bandwidth method or quality factor method based on Fig. 3 for
the mass of 2.72 kg. The loss factor at resonant frequency is equal
to
g Dx=x
n
0:0934 10
where Dx is the difference between the frequencies where the fre-
quency response amplitude at the particular frequency is equal to
1=

2
p
of the peak amplitude. As a comparison, the curve tted func-
tion in equation (9) is also used to calculate the loss factor at reso-
nant frequency, x
n
= 91 Hz. The loss factor at resonant frequency
obtained from Eq. (9) is, g = 0.0900. The difference between the val-
ues obtained from Eqs. (9) and (10) is only around 3.78%. This dif-
ference is small and acceptable.
In order to complete the study of the dynamic properties of en-
gine rubber mount, the frequency dependent stiffness is computed
by using the same equation as used by Lin et al. [23] as below:
kx RefHixg=jHixj
2
1 r
2
11
In here, r = x/x
n
where x
n
is the natural frequency of the system
(91 Hz). The result indicated that the stiffness of the system be-
haves as frequency dependent rather than a constant stiffness.
The frequency dependent stiffness (k(x)) of engine rubber
mount is also curve tted to get the relationship as a function of
frequency as sown in Fig. 8. In this gure, piecewise polynomial
curve tting is used so that the stiffness for different frequency
range can be predicted accurately [23]. This calculated frequency
dependent stiffness is divided into three regions: (a) below reso-
nant frequency range, (b) within resonant frequency range and also
(c) above resonant frequency range.
5.1. Below resonant frequency range
The selected frequency range used for curve tting purpose is
from 10 to 67 Hz. Data below 10 Hz is not included due to the poor
coherence. The upper frequency range is determined based on
where the amplitude of the response is around half of the maxi-
mum response amplitude [23]. The curve tted function is as
below:
k
1
f 0:01677f
2
600 12
This equation can be used to predict the static stiffness value
which is 600 kN/m.
5.2. Within resonant frequency range
The selected frequency range for this section is from 68 to
95 Hz. The curve tted function is obtained as below:
k
2
f 0:0564f
2
422 13
Care should be taken to ensure the continuity at the frequency
range limits which are 68 Hz and 95 Hz.
5.3. Above resonant frequency range
The selected frequency range for this section is from 96 to
200 Hz. The curve tted function at this section shows the linear
trend of stiffness as a function of frequency as below:
k
3
f 0:1203f 950 14
These three curve tted function are combined to give the fre-
quency dependent stiffness for the whole range of the tested fre-
quency of 0200 Hz as shown in Fig. 8. This gure shows the
frequency dependent stiffness curves from Eq. (11) and from the
curve tting (Eqs. (12)(14)) which agrees well with experimental
data.
The loss factor (g
21
) calculated from the dynamic transfer stiff-
ness approach is also curve tted to obtain the predicted dynamic
behaviour as a function of the frequency as shown in Fig. 7. The g-
ure shows that the curve tted function of loss factor (g
21
) is non-
linear which is different from the linear behaviour of the loss factor
(g
11
) as shown in Fig. 6.
The curve tted function is
g
21
f 6:485 10
6
f
2
9:409 10
4
0:06546 15
The function of the curve tted loss factor (g
11
) and frequency
dependent stiffness are veried by comparing the measured dy-
namic driving point stiffness to the reproduction of dynamic driv-
ing point stiffness using the predicted results (Eqs. (9), (12), (13),
and (14)). The comparison is shown in Fig. 9. The results showed
good correlation between the measured and predicted function.
In order to conrm that the values of k
11
obtained using the im-
pact technique is applicable to other preload mass condition, the
result of the preload mass of 0.90 kg is also included in Fig. 9. In
this gure, the predicted k
11
curves for both m = 2.72 kg and
m = 0.90 kg showed good agreement with the curves obtained
experimental. In both cases, the R
2
value is 0.95 for each case.
Fig. 8. Polynomial curve tting for frequency dependent stiffness, k(x).
Fig. 9. Comparison between measured and predicted dynamic driving point
stiffness, k
11
.
L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887 1885
6. Verication of the impact test method by shaker
The results from the impact test are veried by using electro-
magnetic shaker as the excitation source. Similar experimental set-
up is used and the impact hammer is now replaced with the
electromagnetic shaker. The shaker is driven by swept frequency
signals from 0 to 200 Hz with an interval of 1 Hz. The results for
the dynamic driving point stiffness measured by the impact test
and shaker excitation are shown in Fig. 10. It is evident from
Fig. 10 that the results obtained from both the impact test and also
the shaker excitations agree well with correlation index of 0.98 in
the frequency range of interest. Similar observation can be made
on the dynamic transfer stiffness measurement as shown in
Fig. 11. In Fig. 11, the results for the dynamic transfer stiffness
measurement by using impact hammer and shaker showed some
deviations at initial lower frequencies (below 20 Hz) with the aver-
age deviation of 245.9 kN/m over the range of 600800 kN/m.
However, these two lines converge from 20 Hz and above with
the average deviation of 52.3 kN/m.
7. Conclusion
The development of impact technique for the measurement of
dynamic transfer stiffness is presented. This technique provides
an alternative way for dynamic characterization of engine rubber
mount. The advantages of the impact technique are the relatively
simple experimental setup and can be used for measuring the dy-
namic properties of engine mount in a real vehicle. The dynamic
properties of engine rubber mounts which is represented by the
dynamic driving point stiffness and dynamic transfer stiffness
can be measured directly based on the location of the
sensors. The results showed that for the engine mount with the
preload mass of 2.72 kg, the dynamic driving point stiffness can
be used to represent the dynamic transfer stiffness for frequencies
below 30 Hz. This range can be extended to 60 Hz with reduced
preloaded mass 0.90 kg between the test isolator and input force
transducer. Polynomial curve tting is applied on the measure-
ment results to obtain the loss factor as a function of frequency.
This function can accurately predict the loss factor at resonant fre-
quency with the error of 3.78% compared to half power bandwidth
method.
Acknowledgements
The authors would like to gratefully acknowledge Universiti
Sains Malaysia (USM) research grant [A/C 6013360] and the Fel-
lowship for the nancial support. The authors also acknowledged
Mr. Wan Amri and Mr. Baharum for their assistance in the experi-
mental setup.
References
[1] Neumark S. Concept of complex stiffness applied to problems of oscillations
with viscous and hysteresis damping. Aeronautical Research Council reports
and memoranda no. 3269; 1962.
[2] Gade S, Zaveri K, Konstantin HH, Herlufsen H. Damping measurements from
impulse response functions from resonance and non-resonance excitation
techniques. Tech Rev 1994;2:2844.
[3] Mallik AK, Kher V, Pirit M, Hatwal H. On the modeling of non-linear
elastomeric vibration isolators. J Sound Vib 1999;219(2):23953.
[4] Ramorino G, Vetturi D, Cambiaghi D, Pegoretti A, Ricco T. Development in
dynamic testing of rubber compounds: assessment of non-linear effects. Polym
Test 2003;2:6817.
[5] Nader V, Ken LS. High frequency testing of rubber mounts. ISA Trans
2002;41:14554.
[6] Foumani MS, Khajepour A, Durali M. Optimization of engine mount
characteristics using experimental/numerical analysis. J Vib Control
2003;9:112139.
[7] Mundo D, Mas P, Clausi D. Dynamic characterization and numerical modelling
of automotive rubber connections. Proc IMechE D: J Automob Eng
2006;220:42534.
[8] Ladislav P, Ludek P, Frantisek V, Jan C. Laboratory measurement of stiffness and
damping of rubber element. Eng Mech 2007;14(1/2):1322.
[9] Kulik VM, Semenov BN, Boiko AV, Seoudi BM, Chun HH, Lee I. Measurement of
dynamic properties of viscoelastic materials. Exp Mech 2009;49:41725.
[10] Hofer P, Lion A. Modelling of frequency and amplitude dependent
material properties of lled reinforced rubber. J Mech Phys Solids
2009;57:50020.
[11] Ciambella J, Paolone A, Vidoli S. A comparison of nonlinear integral based
viscoelastic models through compression tests on lled rubber. Mech Mater
2010;42:93244.
[12] Hadi AE, Martin B, Loredo A, Jego E. Vibration reduction on city buses:
determination of optimal position of engine mounts. Mech Syst Signal Process
2010;24:2198209.
[13] Joseph RM, Kirsten AB, Scott CP. Complex stiffness measurement of vibration
damped structural elements. Int Modal Anal Conf 2000;4062(2):3917.
[14] Mohan DR, Scott G, Dave G. Measurement of dynamic parameters of
automotive exhaust hangers. SAE technical paper series; 2001. 01NVC-121.
[15] Bloss B, Mohan DR. Measurement of damping in structures by the power input
method. Exp Tech 2002(May/June):302.
[16] BS ISO 10846 1-5. Acoustic and vibration laboratory measurement of vibro-
acoustic transfer properties of resilient elements; 2009.
[17] Gibson RF, Plunkett R. A forced-vibration technique for measurement of
material damping. Exp Mech 1977(August):297302.
[18] Abdulhadi H. Stiffness and damping coefcients of rubber. Ing Arch
1985;55:4217.
[19] Soula M, Vinh T, Chevalier Y, Beda T, Esteoule C. Measurements of isothermal
complex moduli of viscoelastic materials over a large range of frequencies. J
Sound Vib 1997;205(2):16784.
[20] Nadeau S, Champoux Y. Application of the direct complex stiffness method to
engine mounts. Exp Tech 2000(May/June):213.
[21] Morison C, Wang A, Bewes O. Methods for measuring the dynamic stiffness of
resilient fastenings for low frequency vibration isolation of railways, their
Fig. 10. Dynamic driving point stiffness (k
11
) by impact technique and shaker.
Fig. 11. Dynamic transfer stiffness (k
21
) by impact technique and shaker.
1886 L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887
problems and possible solutions. J Low Freq Noise, Vib Active Control
2005;24(2):10716.
[22] Thompson DJ. Developments of the indirect method for measuring the high
frequency dynamic stiffness of resilient elements. J Sound Vib 1998;213(1):
16988.
[23] Lin TR, Farag NH, Pan J. Evaluation of frequency dependent rubber mount
stiffness and damping by impact test. Appl Acoust 2005;66:82944.
[24] Zhang J, Richards CM. Parameter identication of analytical and experimental
rubber isolators represented by Maxwell models. Mech Syst Signal Process
2007;21:281432.
[25] Kari L. On the dynamic stiffness of preloaded vibration isolators in the audible
frequency range: modelling and experiments. J Acoust Soc Am 2003;113(4):
190921.
L.E. Ooi, Z.M. Ripin/ Materials and Design 32 (2011) 18801887 1887