See

discussions, stats, and author profiles for this publication at: [Link]

Development of a vibration isolator with dry

Conference Paper · September 2014

CITATION READS

1 131

2 authors:

Rene Boonen Paul Sas

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PhD research project: prognostics of machine elements View project

KU Leuven Noise & Vibration Research Group View project

All content following this page was uploaded by Rene Boonen on 16 November 2014.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Development of a vibration isolator with
dry friction damping

R. Boonen 1 , P. Sas 1
1 KU Leuven, Department of Mechanical Engineering,

Celestijnenlaan 300 B, B-3001, Heverlee, Belgium

Abstract
A new type of suspension component with dry friction damping has been developed. It consist of an Ω-
shaped spring wire which ends can slip in their mountings. By positioning these components under 90o
around the suspended object, the suspension functions in all six degrees of freedom. The suspension compo-
nent will be modelled as a series connection of a spring with a coulomb damper, which in turn is in parallel
with a second spring. This model will be analysed using Hamilton’s equations of motion to deal with the
non-linear phenomena. An experimental setup has been realized whereupon the suspension components are
validated using hammer impulses. The observed response is an efficiently damped sinusoidal motion which
turns into an undamped sinusoid with small amplitude. In the spectrum, the resonance has been limited in
magnitude to 10 dB and the isolation characteristic remaines −40 dB/decade equivalent to the undamped
spring.

1 Introduction

Vibration isolation is a key issue in technical engineering, ranging from buildings to machines until precision
and micro-mechanics. Just a few examples from an quasi unlimited number of applications are air condi-
tioning equipment in buildings, engines and wheel suspension in transport, sensitive optical equipment, hard
drives and so on.
The linear vibration isolator consists of the classical mass-spring-damper configuration [1, 2, 3, 4] where the
undesired vibration has to pass from the environment through the spring-damper component to the suspended
object or vice versa.
The transfer function magnitude of the force transmission through the suspension component is presented
in figure 1 for three different types of suspensions. The vibration isolation occurs at frequencies above the
zero-dB crossing of the transfer function amplitude. The difference between the transfer function amplitude
and the 0-dB is a measure for the isolation performance.

• undamped spring: The isolation characteristic (mangenta line in figure 1) is a −40 dB/decade decline
above the eigenfrequency fn . At the eigenfrenquency, the tranfer function magnitude tends to infinity.
Disturbances occurring at the eigenfrequency will lead to unacceptable amplitudes. This kind of sus-
pension is used for rotating machinery, such as turbines, or for acoustic measurement rooms, where
the suspension resonance will not be excited.

• spring-damper element: The isolation characteristic (red line in figure 1) tends to a −20 dB/decade
decline above the eigenfrequency fn , which is less performant than the undamped spring. This is
caused by the viscous damper in parallel with the spring werethrough the force will be transmitted

545
546 P ROCEEDINGS OF ISMA2014 INCLUDING USD2014

40
undamped spring
spring-damper
30 hysteresis damping

20

magnitude (dB)
10

-10

-20

-30

-40
10-1 100 101
frequency (f/fn)

Figure 1: Force transmission characteristic of different vibration isolator principles.

in the high frequency range. The transition frequency where the isolation characteristic alters from
−40 dB/decade to −20 dB/decade is ft = 21π kc , wherein k is the spring constant and c the damper
constant. At eigenfrenquency, the tranfer function magnitude is limited by the damper. These suspen-
sion components are used in applications where excitation at eigenfrequency can occur, such as wheel
suspensions in vehicles.

• hysteresis damping: Hysteresis damping or internal material damping is dominantly present in elas-
tomeric materials. The damping is expressed as a complex spring constant in the frequency domain.
The isolation characteristic (blue line in figure 1) remains a −40 dB/decade decline above the eigen-
frequency fn . At eigenfrequency, the tranfer function magnitude is limited by the imaginairy part of
the complex spring constant. Rubber suspension elements are widely used in engineering. The disad-
vantages are mainly of technological nature: the elastic modulus is highly temperature dependent and
the material exhibits consideral creep.

The use of non-linear suspension components is also widely exploited [3, 4, 5], such as non-linear spring
and damping characteristics, the use of mechanisms to introduce non-linear displacement relations and the
use of different types of dampers, such as dry friction. For example, a dry friction damper in parallel with a
spring and a viscous damper has been analysed [5]. The suspension component is very effective to damp the
vibrations. However, at small amplitudes, the dry friction damper blocks and the vibrations are transmitted
to the suspended object without any attenuation.
In this research, a new type of suspension component using dry friction has been developed which is able to
combine the −40 dB/decade isolation characteristic, considerable damping at resonance and robust against
environmental influences such as temperature, in six degrees of freedom.
First, the principle of the suspension component will be discussed. Then, the functioning of the component
will be analysed using an uni-directional model and Hamilton’s method is applied to deal with the non-linear
effects. The model parameters such as the spring constants and the friction will be identified. Then, the
model will be evaluated using a shock excitation and the resulting response will be discussed. Finally, the
model will be evaluated from data measured on an experimental setup.
DAMPING 547

2 Principle of the dry fricton suspension component

The principle of the suspension component is presented in figure 2.

Figure 2: Principle of the dry friction suspension component.

It consist of a spring wire bended in a 3-D Ω-shape. The shape of the wired spring allows some design
freedom to diversify the resonance frequencies in the X-, Y - and Z-direction. The wire is embedded in
the connection blocks. Due to the displacement of the connection blocks as a consequence of a shock or
vibration, reaction torques appear at the connection interfaces. When the vibration amplitudes remain small,
the spring wire remains fixed in the connection blocks and the isolator behaves as an undamped spring. When
these torques overcome the internal dry friction, the connection slips in the interface and vibration energy is
dissipated by the suspension component.

Figure 3: Photograph of the experimental setup with four dry friction suspension components.

Figure 3 shows a photograph of the experimental setup. Four suspension components are located between
the base plate A and the suspended object E. The object has a mass of 0.97 kg. The Ω-shaped spring
wire B is fitted at the suspended object such it can rotate with dry friction. The dimensions of the spring
548 P ROCEEDINGS OF ISMA2014 INCLUDING USD2014

wire are R = 12 mm, a = 20 mm and b = 16 mm. At the base plate side at position C, the friction of
the wire ends can be altered by electro-magnetic coil actuators D. These actuators will allow to optimize
the damping for different disturbing forces using a semi-active control strategy. By positioning the four
suspension components under 90o , the damping mechanism is active in all six degrees of freedom.

3 Analysis of the suspension component

To demonstrate the proof of concept, the analysis is carried out in the height direction as a SDOF system.
The model, presented in figure 4, consists of the suspended mass m, the spring k1 connected in series with

Figure 4: One-dimensional model of the dry friction suspension component.

the dry friction damper with a slipping force Fs which form the damper branche and the spring k2 connected
in parallel with the damper branche. As the suspension component exhibits non-linear behaviour, the method
of Hamilton [6] is used to set up the equations of motion. The method uses a scalar potential function, called
the Hamiltonian, which is the sum of the kinetic and potential energy of the suspended system, expressed
in displacement, momentum and time as independend variables. Taking the derivatives of the Hamiltonian
to displacement and momentum results in the set of first order equations of motion, which can be solved
numerically by any ODE-solver.
The kinetic energy Ek and potential energy Ep2 of the spring k2 are:

Ek = 1
2 m ẋ2 and Ep2 = 1
2 k 2 x2 (1)

The potential energy Ep1 of the spring k1 and the dissipated power Ps in the dry friction damper are depen-
dent of the position x1 which in turn is dependent of the position x of the mass m. In the case that the mass
position is larger than Fk1s , the damper is in motion while the spring length of k1 is constant. So, the potential
energy Ep1 of the spring k1 and the dissipated power Ps in the dry friction damper are:

1 Fs
Ep1 = k1 ( )2 and Ps = Fs sign(ẋ) ẋ (2)
2 k1
wherein the sign-function is −1 if ẋ < 0, 0 if ẋ = 0 and +1 if ẋ > 0.
When the mass displacement becomes sufficiently close to the equilibrium position such that the force in
the damper branche becomes smaller than Fs , the damper blocks and the spring k1 moves. In this case, the
potential energy Ep1 of the spring k1 and the dissipated power Ps in the dry friction damper are:

1 Fs
Ep1 = k1 (x − )2 and Ps = 0 (3)
2 k1

Once the displacement is through the equilibrium position and becomes larger than − Fk1s at the negative side,
the damper is activated again and the spring k1 will be again constant in length. So, the potential energy Ep1
DAMPING 549

of the spring k1 and the dissipated power Ps in the dry friction damper are:

1 Fs
Ep1 = k1 (− )2 and Ps = Fs sign(ẋ) ẋ (4)
2 k1
The transitions between deactivating the damper and activation again takes place at the two positions x = Fk1s
at the positive side and x = − Fk1s at the negative side. This will be expressed by the two switching functions:
Fs Fs
δ1 = δ(x − k1 ) and δ2 = δ(x + k1 ) (5)

The function δ(y) is 0 if y < 0 and +1 if y > 0.

1
Ek = m ẋ2
2
1 1 h Fs i2 (6)
Ep = k2 x2 + k1 x (1 − ∆) −
2 2 k1
Ps = Fs sign(ẋ) ẋ ∆

which are valid for each position x of the suspended mass. The switching function ∆ = δ1 + δ2 .
Once the expressions (6) for the kinetic energy, potential energy and dissipated power are known, the equa-
tions of motion using Hamilton’s method can be set up. The momentum px will be:
∂ Ek
px = = m ẋ (7)
∂ ẋ
The Hamiltonian which is the sum of the kinetic and potential energy, expressed in momentum, displacement
and time will be:
1 p2x 1 1 h Fs i2
H= + k2 x2 + k1 x (1 − ∆) − (8)
2m 2 2 k1
The force expression will be:
∂ Ps ∂ H
ṗx = Fl − −
∂ ẋ ∂x
(9)
h Fs i
= Fl − Fs sign(px ) ∆ − k2 x − k1 x (1 − ∆) − (1 − ∆)
k1
with Fl an external applied force.
Ultimately, the result is a set of two first order differential equations of motion:

 px
ẋ
 =
m (10)
h Fs i
= Fl − Fs sign(px ) ∆ − k2 x − k1 x (1 − ∆) − (1 − ∆)

ṗx

k1
which can be solved by any ODE-solver.

4 Determination of the spring constants

The spring constants k1 and k2 will be determined from a FEM-analysis of the suspension construction as
presented in figure 2. The analysis will be carried out for two cases, one with the wire ends free rotating in
the mounting blocks, the other one with the wire ends fixed.
550 P ROCEEDINGS OF ISMA2014 INCLUDING USD2014

Figure 5: FEM-Analysis of the spring wire of a suspension component. Left: wire ends free; Right: wire
ends fixed.

The first case corresponds to the spring k2 because the spring k1 does not change its length during the motion.
The second case corresponds to the sum of the spring k1 + k2 , as the damper is now blocked and the two
springs are simply in parallel. The resulting deformations of the wire are presented in figure 5.
From these deformations, k2 = 26.6 kN/m and k1 = 52.2 kN/m. The friction force Fs = 14 N is determined
from experiments. Using these data in the equations of motion (10) with an initial displacement x0 = 0 and
an initial velocity of v0 = 1.7 m/s results in the impulse response presented in figure 6.

10
impulse response

5
dispacement (mm)

-5

-10
0 0.1 0.2 0.3 0.4 0.5
time (s)

Figure 6: Simulated displacement of the suspended mass excited by an impulse.

The most interesting phenomena observed in the plot are:

1. In the first 0.2 s of the response, the damper is engaged. The decrease of the displacement
q magnitude
k2
each half period amounts Fk1s . The frequency can be approximated by f = 1
2π m. The damping
mechanism is very efficient.
2. After a transition at 0.2 s, the remaining motion is an undamped vibration with a maximum amplitude
of Fk1s . The frequency is increased compared to the first part of the motion and is determined by
q
f = 21π k1m +k2
.

3. The equilibrium of the undamped vibration does not occur at x = 0. The dry friction damper blocks
somewere between − Fk1s and Fk1s , depending on the initial conditions. In practice, the exact position
DAMPING 551

of the equilibrium will not be predictable. The total magnitude, i.e. the equilibrium position with
thereupon superponed the vibration amplitude will never exceed the interval [− Fk1s , Fk1s ].

5 Experimental evaluation

The experimental setup, presented in figure 3, has been realized. The measurements have been carried out
using an impulse hammer and an accelerometer in the centre of the suspended mass.
First, the acceleration response of the mass is recorded in time domain and plotted in figure 7 in red line,
together with the second derivative of the displacement signal presented in figure 6 in blue line. The spring
constants k1 and k2 have been derived from the FEM-analysis, the friction force has been estimated from the
experiment and used in the simulation. The friction force can be adapted on the experimental setup using the
electro-magnetic actuators in a range between 5 N and 54 N.

simulated
measured
400

200
accelleration (m/s2)

-200

-400

0 0.1 0.2 0.3 0.4 0.5

Figure 7: Comparison of the simulated with the measured acceleration of the suspended mass excited by an
impulse.
20
simulated
measured
10

0
magnitude (dB)

-10

-20

-30

-40

-50

-60
1 2
10 10
frequency (Hz)

Figure 8: Spectra of the simulated and the measured displacements corresponding to the time signals pre-
sented in figure 7.
552 P ROCEEDINGS OF ISMA2014 INCLUDING USD2014

20
measured
-40 dB/decade line
10

magnitude (dB)
-10

-20

-30

-40

-50

-60
101 102
frequency (Hz)

Figure 9: Spectrum of the acceleration measured through the suspension compared with a −40 dB/decade
line.

The measured response exhibits the same behaviour as the predicted one. The acceleration amplitude de-
creases quasi linearly until the amplitude becomes to small to engage the friction damper. Then, the vibration
amplitude remains constant. In the experimental setup of course, this amplitude will become zero after some
seconds. The eigenfrequency of the setup in the undamped situation is quasi equal to the simulated one.
The difference in frequency between simulation and experiment with active friction damper is due to the
assumption that the spring ends are free which is not fully correct. Nevertheless, this assumption is sufficient
in the design stage of the suspension component.
The measured frequency spectrum, presented in figure 8 in red line, has been averaged over 10 hammer
hits. The simulated spectrum is calculated from the simulated acceleration signal presented in figure 7. The
resonance amplitude is limited to about 10 dB, while the slope of the isolation part of the spectrum remains
−40dB/decade. Also, a measurement of the acceleration of the suspended mass has been carried out whereby
the base plate is excited by hammer hits. The spectrum, averaged over 10 hammer hits, is presented in figure 9
and is compared to a −40 dB/decade line. It demonstrates that the isolation characteristic is equivalent to the
undamped spring.

6 Conclusion

A new type of suspension component has been developed. It consist of an Ω-shaped spring wire which ends
can slip in their mountings. By positioning these components under 90o around the suspended object, the
suspension functions in all six degrees of freedom. A model of the suspension component has been set up
and analysed using Hamilton’s equations of motion to deal with the non-linear phenomena. It is based on a
series connection of a spring with a coulomb damper, which in turn is connected in parallel with a second
spring. An experimental setup has been realized whereupon the suspension components are validated using
hammer impulses. The observed response is an efficiently damped sinusoidal motion which turns into an
undamped sinusoid with small amplitude. In the spectrum, the resonance has been limited in magnitude to
10 dB and the isolation characteristic remaines −40 dB/decade equivalent to the undamped spring.
DAMPING 553

References

[1] J.P. Den Hartog, Mechanical vibrations, McGraw-Hill, New York, (1956).

[2] G. Genta, Vibration dynamics and control, Springer, New York, (2008).

[3] R. Ibrahim, Recent advances in nonlinear passive vibration isolators, Journal of Sound and Vibration,
Vol. 314, (2008), pp. 371-452.

[4] C. Ho, Z. Lang, S. A. Billings, Design of vibration isolators by exploiting the beneficial effects of stiffness
and damping non-linearities, Journal of Sound and Vibration, Vol. 333, (2014), pp. 2489-2505.

[5] M. S. Hundal, Response of a base excited system with coulomb and viscous friction, Journal of Sound
and Vibration, Vol. 64, No. 3, (1979), pp. 371-378.

[6] C. Lanczos, The variational principles of mechanics, Dover publications, New York, (1986).
554 P ROCEEDINGS OF ISMA2014 INCLUDING USD2014

View pu