2006:006 CIV

MA S T ER’S TH E SI S

Analysis of
Structural Damping

Manuel Martínez Rodríguez

MASTER OF SCIENCE PROGRAMME

Mechanical Engineering

Luleå University of Technology

Department of Applied Physics and Mechanical Engineering
Division of Computer Aided Design

2006:006 CIV • ISSN: 1402 - 1617 • ISRN: LTU - EX - - 06/6 - - SE

Analysis of Structural Damping

Manuel Martínez Rodríguez

Supervisor
Dr. Jan-Olov Aidanpää

Division of Computer Aided Design

Dept. Applied Physics and Mechanical Engineering
Luleå University of Technology
Luleå, Sweden
M. Martínez Analysis of Structural Damping

Abstract
One important parameter in the study of dynamic systems is material damping. It is
defined as the ability of a material to absorb vibration by internal friction and convert the
mechanical energy into heat. In spite of its importance in the study of rotating machinery, it
has not been studied deeply because of the difficulty analyzing and testing its different types
and mechanisms.

This master thesis presents a new method to measure damping in mechanical systems.
The mechanical system analyzed in this essay is a rotating shaft.

The thesis consists of two different parts: a theoretical analysis and experimental
work.

Regarding the theoretical analysis, two different methods are explained. The first
method is based on the moments, stresses and strains involved in the system and it is the basis
to design the experimental set up. The second method is based on the energy and it is
proposed to check the validity of the results from the experiment.

On the other hand, two tasks are developed in the experimental work:

• To design the set up for the experiment and assemble it.

• To carry out the experiment with different material in order to check the
validity of the proposed method.

i
M. Martínez Analysis of Structural Damping

Acknowledgements

Firstly, I would like to express my gratitude to Christina Hamsch and Inger Niska
from the International Office for allowing me to stay one semester more studying at Luleå
University of Technology

Secondly, I am sincerely grateful to my supervisor Dr. Jan-Olov Aidanpää for giving

me the opportunity to carry out my master thesis at the Division of Computer Aided Design in
Luleå University of Technology. I am also grateful to him for his help and his kindness when
I had problems to develop the work and for being always friendly with me.

I would like to thank my supervisor at University of Oviedo María Jesús Lamela Rey
for her help with the paperwork, her kindness to answer all my questions and her sensible
advices.

I send a warmly appreciation to the exchange students that I have met during my stay
in Luleå in general and in particular to those that have supported me during the difficulties
along my stay in Sweden.

Finally, I would like to express my most sincere gratitude to my family for their love,
their unconditional support and for giving me the opportunity to graduate.

Luleå, December 2005

Manuel Martínez

ii
M. Martínez Analysis of Structural Damping

Table of contents

CHAPTER 1 – INTRODUCTION 1

1.1 Internal damping 1

1.2 Structural damping 4
1.3 Fluid damping 5
.
CHAPTER 2 – MEASUREMENT OF DAMPING 6

CHAPTER 3 – MEASUREMENT OF DAMPING IN MECHANICAL SYSTEMS 10

3.1 Theoretical Analysis 10

3.1.1 First Method: Loss factor (η) vs. Moments & Forces 12

3.1.2 Second Method: Damping energy 19

3.2 Experimental work 20

CHAPTER 4 – DISCUSSION 26

CHAPTER 5 – CONCLUSION 27

CHAPTER 6 – RECOMMENDATIONS FOR FURTHER WORK 28

iii
M. Martínez Analysis of Structural Damping

Table of symbols

σ Stress

ε Strain

d
(ε max ) Strain rate
dt

E Young’s Modulus

E*, Ê Complex Modulus

I Moment of inertia

Cs

∆U Damping capacity

Tp Peak time

Mp Peak value

PO Percentage overshoot

dv, dh, df Damping capacity per unit volume

ω Frecuency

•
q Relative velocity

ρ Density

Cd It is a function of the Reynold’s number and the geometry

Q Amplification factor

ζ Damping ratio

η Loss factor

γ Viscosity

T Time

R Radius of the shaft

iv
M. Martínez Analysis of Structural Damping

r Distance between the point and the centre of its section

Fx, Fy, Forces

Mx, My Moments

dA Area element

x, y, z Coordinate

Fb Applied force

L Length of the shaft

d Displacement about x of the shaft

δ Deformed position

φ Phase angle

v
M. Martínez Analysis of Structural Damping

Chapter 1. Introduction

Damping is the energy dissipation of a material or system under cyclic stress.

Three main types of damping are present in any mechanical system:

• Internal damping
• Structural damping
• Fluid damping

1.1 Internal damping

Internal damping is caused by microstructure defects -impurities, grain
boundaries…-, thermoelastic effects, eddy-current effects in ferromagnetic materials,
dislocation motion in metals and chain motion in polymers. Besides, there are two types
of internal damping: Viscoelastic damping and hysteretic damping. Dealing with the
latter, it must be taken into account that the term, hysteretic damping, is not suitable at all,
for the reason that all types of internal damping are related with hysteretic-loop effects.
The relation between the stress (σ) and the strain (ε) has a hysteretic loop as the figure
(1.1.1) shows.

Figure 1.1.1

The area of the hysteretic loop represents the energy dissipated per unit volume of
material and per stress cycle. The force and the displacement relationship present also a
hysteretic loop.

1
M. Martínez Analysis of Structural Damping

Concerning Viscoelastic damping, the relation between the stress and the strain
is expressed through a linear differential equation with respect to time. The stress of a
Viscoelastic damping material depends on the frequency of variation of the strain, and
therefore on the frequency of motion. What is used to represent Viscoelastic damping is
tree models:

• Kelvin-Voigt model: σ = E ⋅ε + E *⋅
d
(ε ) (1.1.1)
dt
d d
• Maxwel model: σ + cs ⋅
(σ ) = E * ⋅ (ε ) (1.1.2)
dt dt
d d
• Standard linear solid model: σ + c s ⋅ (σ ) = E ⋅ ε + E * ⋅ (ε ) (1.1.3)
dt dt

The most widespread model is the Kelvin-Voigt model as it is the most accurate
for practical purposes. In equation (1.1.1), E is Young’s Modulus and E* is the complex
modulus, which is assumed to be time independent. The term Eε represents the elastic
behaviour of the material and it does not contribute to damping. The term E * ⋅ (ε ) is
d
dt
the responsible of damping. It is defined the damping capacity per unit volume as:

d
dv = E * ∫ (ε ) ⋅ dε (1.1.4)
dt

If the material is subjected to a harmonic excitation, the strain can be expressed as

follows:

ε = ε max ⋅ cos(ωt ) (1.1.5)

By introducing equation (1.1.5) into equation (1.1.4), it is achieved the following

expression for damping capacity per unite volume:

d v = π ⋅ ω ⋅ E * ⋅ε max
2
(1.1.6)

By considering that σ = E ⋅ ε max , the following equation is obtained:

π ⋅ ω ⋅ E * ⋅σ max
2
dv = (1.1.7)
E2

It is possible to say that damping capacity per unit volume for the Kelvin-Voigt
model depends on frequency.

The Kelvin-Voigt model will be explained in detail later because it is the chosen
model to characterize the Viscoelastic behaviour in this paper.

2
M. Martínez Analysis of Structural Damping

As regards hysteretic damping, the stress does not depend considerably on the
frequency of oscillation. What is more, damping capacity per unit volume is independent
of frequency as the following expression shows:

d h = J ⋅ σ max
n
(1.1.8)

For the case n=2, the stress can be expressed by the following expression:

Ê d
σ = E ⋅ε + ⋅ (ε ) (1.1.9)
ω dt

Ê
This expression is equivalent to equation (1.1.1) by considering that E* = .
ω

If the material is subjected to a harmonic excitation where the strain is represented

also by equation (1.1.1), the equation (1.1.9) can be expressed as follows:

π
σ = E ⋅ ε o ⋅ cos(ωt ) + Ê ⋅ ε o ⋅ sin(ωt ) = E ⋅ ε o ⋅ cos(ωt ) + Ê ⋅ ε o ⋅ cos(ωt + ) (1.1.10)
2

It must be taken into account that the stress consists of the elastic component,
which is in phase with strain, and the hysteretic damping component, which is 90º out of
phase.

The response can be expressed as follows:

ε = ε o ⋅ e jωt (1.1.11)

By introducing equation (1.1.11) into equation (1.1.10), a new expression for the
stress is obtained:

σ = ( E + jÊ ) ⋅ ε (1.1.12)

By combining equations (1.1.11) and (1.1.12), an expression, which represents

both Viscoelastic and hysteretic damping behaviour, is obtained:

⎛ Ê⎞ d
σ = E ⋅ ε + ⎜⎜ E * + ⎟⎟ ⋅ (ε ) (1.1.13)
⎝ ω dt⎠

It would be interesting to point out that E, E* and Ê are independent of the

frequency.

3
M. Martínez Analysis of Structural Damping

1.2 Structural damping

Rubbing friction or contact among different elements in a mechanical system
causes structural damping. Since the dissipation of energy depends on the particular
characteristics of the mechanical system, it is very difficult to define a model that
represents perfectly structural damping. The Coulomb-friction model is as a rule used to
describe energy dissipation caused by rubbing friction. Regarding structural damping
(caused by contact or impacts at joins), energy dissipation is determined by means of the
coefficient of restitution of the two components that are in contact.

Structural damping is usually estimated by means of measuring but the measured

values represent the total damping in the mechanical system. Consequently it is necessary
to estimate the values for the other types of damping and to subtract them from the
measured value in order to obtain a value of structural damping. Structural damping is
much greater than internal damping and it represents a large portion of energy dissipation
in mechanical structures.

As mentioned above, different factors such as rubbing friction or impacts cause

structural damping. The most important form of structural damping is the slip damping.
This form of damping is caused by Coulomb friction at a structural joint. It depends on
many factors such as joint forces or surface properties. Assuming an ideal Coulomb
friction, the damping force at a join can be expressed through the following expression:

⎛•⎞
f = c ⋅ sgn ⎜ q ⎟ (1.2.1)
⎝ ⎠

where:

f = damping force
q = relative displacement at the joint
c= friction parameter
and the signum function is defined by:
sgn(x)= 1 for x ≥ 0
sgn(x)=-1 for x < 0 

4
M. Martínez Analysis of Structural Damping

1.3 Fluid damping

When a material is immersed in a fluid and there is relative motion between the
fluid and the material, as a result the latter is subjected to a drag force. This force causes
an energy dissipation that is known as fluid damping. The following equation expresses
the drag force:

•2
1 ⎛•⎞
f d = ⋅ c d ⋅ ρ ⋅ q ⋅ sgn⎜ q ⎟ (1.3.1)
2 ⎝ ⎠

where:
•
q = Relative velocity
ρ = Density
cd = It is a function of the Reynold’s number and the geometry

Damping capacity per unit volume for fluid damping is:

f d ⋅ dz ⋅ dx ⋅ dq(x, y, z )
Lx Ly

df =
∫∫ ∫
0 0
(1.3.2)
Lx ⋅ L y ⋅ qo

in which Lx and Ly are cross-sectional dimensions of the element in the x and y

direction and qo is a normalizing parameter for relative displacement.

5
M. Martínez Analysis of Structural Damping

Chapter 2. Measurement of damping

From a theoretical point of view there are different methods to measure damping.
These methods are divided in two main groups depending on if the response of the
system is expressed as a function of time or as a function of frequency, i.e. time-response
methods and frequency-response methods.

Logarithmic Decrement Method, Step-Response Method and Hysteretic Loop

Method are time-response methods, whereas Magnification-Factor Method and
Bandwidth Method are frequency-response methods.

Dealing with time-response methods, the Logarithmic Decrement Method is the

most common method to measure damping. In this method, an initial excitation is
applied to a single-degree-of-freedom oscillatory system with viscous damping. As
Figure (2.1) shows, the form of the response is a time decay, which is expressed by the
Formula (2.1).

Figure 2.1 Impulse response of a simple oscillator

y (t ) = y o ⋅ exp(−ζω n t ) ⋅ sin(ω d t ) (2.1)

If the response is known, then it is possible to determine the logarithmic

decrement δ through the formula (2.2).

⎛ ⎞
1
δ = ⋅ Ln⎜⎜ X i ⎟ = 2πζ (2.2)
⎟ 1− ζ 2
r ⎝X i+r ⎠

6
M. Martínez Analysis of Structural Damping

Then the damping ratio ζ is easily calculated with the formula (2.3).

δ
ζ = (2.3)
4π 2 + δ 2

As regards the Step-Response Method, a unit-step excitation is applied to the

single-degree-of-freedom oscillatory system and its time response looks like a typical
step-response curve (see figure 2.2) given by the expression:

1
y (t ) = 1 − ⋅ exp( −ζω n t ) ⋅ sin(ω d t + ϕ ) (2.4)
1−ζ 2

Figure 2.2. A typical step response of a simple oscillator

Damping ratio ( ζ ) can be determined through these tree parameters: peak time
(Tp), peak value (Mp) or percentage overshoot (PO). These tree parameters are easily
obtained from the step-response curve. Formulas (2.5), (2.6) and (2.7) show the relation
between these tree parameters and damping ratio ( ζ ).

⎛ π ⎞
ζ = 1 − ⎜⎜ ⎟⎟ (2.5)
ω
⎝ P n⎠
T

7
M. Martínez Analysis of Structural Damping

1
ζ = (2.6)
1
1+
⎡ Ln(M P − 1) ⎤
2

⎢ π ⎥
⎣ ⎦

1
ζ = (2.7)
1
1+
(
⎡ Ln PO
⎢
⎤
100 ⎥
) 2

⎢ π ⎥
⎣ ⎦

The Hysteretic Loop Method calculates the energy loss per cycle of oscillation
due to steady state harmonic loading. Damping capacity (∆U) is given by the area of the
displacement-force hysteretic loop. Then, the loss factor (η) and the damping ratio ( ζ )
can be easily determined through formula (2.8) and (2.9) respectively.

ΔU
η= (2.8)
2πU max

η = 2ζ (2.9)

The Magnification-Factor Method is a frequency-response method, as it has

been mentioned above. The damping ratio ( ζ ) can be determined, on condition that the
magnitude curve of the frequency-response function is known (see figure 2.2). From this
curve is possible to obtain the amplification factor (Q), which is the magnitude of the
frequency-response function at resonant frequency. Then, damping ratio ( ζ ) can be
easily determined using expression (2.10).

1
Q= (2.10)
2ζ 1 − ζ 2

8
M. Martínez Analysis of Structural Damping

As a final point, the last method to estimate damping from frequency domain is
the Bandwidth Method. This method is also based in the magnitude curve of the
frequency-response function. Bandwidth (Δω) is defined as the width of the frequency-
response magnitude curve when the magnitude is 1 times the peak value. Then,
2
damping ratio can be determined from bandwidth using the expression (2.11):

1 Δω
ζ = (2.11)
2 ωr

Figure 2.2 Bandwidth method of damping measurement

in a single-degree-of-freedom system.

9
M. Martínez Analysis of Structural Damping

Chapter 3. Measurement of damping in

mechanical systems

Although the methods mentioned above are theoretically suitable for measuring
damping, it is actually very difficult to apply them for measuring damping in rotating
machinery. For this reason, there is a lack of experiments to measure damping in rotating
machinery. Consequently, this master thesis tries to design and develop a new experiment
to measure damping in mechanical systems.

As it has been pointed out in the introduction, damping can be expressed through
different parameters such as damping ratio ( ζ ) or the loss factor (η). Therefore, the aim
of the experiment is to determine one of these parameters by means of measuring simple
variables such as forces or displacements. This paper will propose two different methods
to measure the loss factor (η).

This thesis consists of two different parts: a theoretical analysis of the problem
and an experimental work. The theoretical analysis proposes two different methods to
calculate damping. The first method will be used to design the experiment while the
second will be used to check the validity of the measured values from the experiment. In
the experimental part it will explain how the experiment was carried out.

10
M. Martínez Analysis of Structural Damping

3.1 Theoretical analysis

To begin with, it is necessary to set up all the conditions involved in the problem.
The experiment will focus on a rotating shaft that is simply supported. Moreover, the
shaft is subjected to a constant centre load (see figure 3.1.1).

y
Fx

δ ωt
Fy
x

Figure (3.1.1) Moments and forces

One of the most important conditions is the material definition. It must be

assumed that the shaft is made of a Viscoelastic material. Viscoelastic materials show
simultaneously a viscous and an elastic behaviour. Different models can represent
Viscoelastic materials; however in this paper it will only be considered the Kelvin-Voigt
Model. This model represents the behaviour of Viscoelastic material by means of a
dashpot and a spring as the figure 3.1.2 shows.

11
M. Martínez Analysis of Structural Damping

Figure (3.1.2)

The dashpot constrains the spring in order to have the same deformation. The
Kelvin-Voigt model is mathematically expressed by the following expression:

σ = E ⋅ε + γ ⋅
d
(ε ) (3.1.1)
dt

where:

σ = Stress
ε = Strain
γ = Viscosity
t = time

Two different methods are proposed to measure damping, as it has been

mentioned above. The first method tries to relate the loss factor (η) with moments and
forces while the second method tries to relate the loss factor (η) with the energy involved
in the process.

12
M. Martínez Analysis of Structural Damping

3.1.1 First Method: Loss factor (η) vs. Moments & Forces

During the experiment, the shaft must be kept in a constant deformed position (δ),
as the figure 3.1.1 shows, and rotating with the frequency ω. The strain in any point will
change according to the following expression:

ε max
ε (r , t ) = ⋅ r ⋅ sin (ω ⋅ t ) (3.1.1.1)
R

where:

R = Radius of the shaft

r = Distance between the point and the centre of its section
t = Time
ω = Rotating frequency
ε max = Maximum strain

Deriving the equation (3.1.1.1), it is obtained the strain rate, which induces a
stationary stress state similar to the bending stress state but rotated 90 degrees. The
equation (3.1.1.2) expresses the strain rate:

ε
ε (r , t ) = max ⋅ r ⋅ ω ⋅ cos(ω ⋅ t )
d
(3.1.12)
dt R

On the other hand, the bending moment about x can be calculated from the
following expression:

R2 − x2
R
σ max ⋅ y
Mx = ∫ ∫
− R− R 2 − x 2
R
⋅ y ⋅ dy ⋅ dx (3.1.1.13)

The result of the integration is:

1
Mx = ⋅ σ max ⋅ R 3 ⋅ π (3.1.1.14)
4

Mx can also be expressed through the following expression:

13
M. Martínez Analysis of Structural Damping

1
Mx = ⋅ σ y max ⋅ R 3 ⋅ π (3.1.1.15)
4

Assuming that:

σ = σ x +σ y = γ ⋅
d
(ε ) + E ⋅ ε (3.1.1.16)
dt

Then, it would be conclude that Mx is found as follows:

1
Mx = ⋅ E ⋅ ε max ⋅ R 3 ⋅ π (3.1.1.17)
4

In the same way than with Mx, it is possible to reason in order to obtain My:

R2 − y2
R
σ max ⋅ x
My = ∫ ∫
− R− R − y
2 2
R
⋅ x ⋅ dx ⋅ dy (3.1.1.18)

The result of the integration is:

1
My = ⋅ σ max ⋅ R 3 ⋅ π (3.1.1.19)
4

Equation (3.1.1.19) can be expressed just as equation (3.1.1.20) shows:

1
My = ⋅ σ x max ⋅ R 3 ⋅ π (3.1.1.20)
4

Reasoning in a similar way than with the moment about x, it is obtained an

expression that depends on the viscosity, the strain rate and the radius (3.1.1.20).

1
⋅ γ ⋅ (ε max ) ⋅ R 3 ⋅ π
d
My = (3.1.1.21)
4 dt

14
M. Martínez Analysis of Structural Damping

At this point, an expression that relates a moment with the viscosity has been
obtained. However, the aim is to get an expression that relates a moment or a force with
the loss factor (η). Thus, the next step will be to relate the viscosity with the loss factor
(η). It will be used the complex modulus to relate both parameters.

As it has been mentioned above, Viscoelastic materials show

simultaneously a viscous and an elastic behaviour. It means that their modulus must be
represented by a complex quantity, which possesses both stored and dissipative energy
component. This complex modulus can be expressed in different ways depending on the
model used to describe the system. On one hand, one of the hysteretic models defines the
complex modulus through the following expression:

E* = E (1 + iη ) = E + i ⋅η ⋅ E (3.1.1.22)

On the other hand, the hysteretic model can also be applied when a material is
subjected to cyclic loading. In this case, it is necessary to assume that both the time
history of the stress cycles and the history of the deformation are harmonic. Moreover,
the strain will be delayed in time by a phase angle φ, which is considered independent
from the frequency. Under these conditions the time histories of stress, strain and strain
rate can be expressed as follows:

σ = σ o ⋅ cos(ω ⋅ t ) = σ o ⋅ e iω ⋅t (3.1.1.23)

ε = ε o ⋅ cos(ω ⋅ t − ϕ ) = ε o ⋅ e − i⋅ϕ ⋅ e i⋅ω ⋅t (3.1.1.24)

d
(ε ) = ε o ⋅ e −i⋅ϕ ⋅ e i⋅ω⋅t ⋅ i ⋅ ω (3.1.1.25)
dt

where:

ω: frequency of the cyclic loading

t: time
φ: phase angle
σo: maximum stress
εo: maximum strain

The complex modulus relates stress and strain according to the following
expression:

σ
E* = (3.1.1.26)
ε

15
M. Martínez Analysis of Structural Damping

By introducing equations (3.1.1.23) and (3.1.1.24) into equation (3.1.1.26), the

following expression for the complex modulus is obtained:

σo
E* = (3.1.1.27)
ε o ⋅ e −i⋅ϕ

Now, it is necessary to relate σo with εo. By introducing the equations (3.1.1.24)

and (3.1.1.25) into equation (3.1.1), the following expression is obtained:

σ = E ⋅ (ε o ⋅ e − i⋅ϕ ⋅ e i⋅ω ⋅t ) + γ ⋅ e − i⋅ϕ ⋅ e i⋅ω ⋅t ⋅ i ⋅ ω (3.1.1.28)

As equations (3.1.1.23) and (3.1.1.28) define the same parameter –stress- it is

possible to state that:

( )
E ⋅ ε o ⋅ e − i⋅ϕ ⋅ e i⋅ω ⋅t + γ ⋅ e − i⋅ϕ ⋅ e i⋅ω ⋅t ⋅ i ⋅ ω = σ o ⋅ e iω ⋅t (3.1.1.29)

Consequently, σo can be expressed through the following expression:

σ o = ε o ⋅ e − i ⋅ϕ ⋅ (E + i ⋅ ω ⋅ γ ) (3.1.1.30)

If equation (3.1.1.30) is introduced into equation (3.1.1.27), it is obtained an

expression, which defines the complex modulus:

E* = E + i ⋅ γ ⋅ ω (3.1.1.31)

If equations (3.1.1.22) and (3.1.1.31) are compared, it is possible to assert that:

γ ⋅ ω = E ⋅η (3.1.1.32)

Therefore, viscosity can be expressed as follows:

E ⋅η
γ = (3.1.1.33)
ω

The equation (3.1.1.33) is especially interesting because it shows that viscosity

depends on the frequency.

After relating the viscosity with the loss factor (η), the last step would be to relate
the loss factor (η) with a force instead of a moment because measuring moments can be
more complicated that measuring forces.
At this point, it would be possible to express My as function of the loss factor (η)
by introducing equation (3.1.1.33) into equation (3.1.1.21).

16
M. Martínez Analysis of Structural Damping

1 E ⋅η d
My = ⋅ ⋅ (ε max ) ⋅ R 3 ⋅ π (3.1.1.34)
4 ω dt

Applying the solid mechanics theory to the shaft shown in figure (3.1.1), it is
possible to state that:

4
Fx = ⋅My (3.1.1.35)
L

4
Fy = ⋅Mx (3.1.1.36)
L

where L is the length of the shaft.

One of the parameters of equation (3.1.1.34), εmáx, is unknown; hence it is

necessary to determine it. It is possible to express εmáx as function of Fy. The value of Fy
is known because it is the force that is introduced to keep the shaft in a constant deformed
position (δ). Introducing equation (3.1.1.17) into equation (3.1.1.26), it can be obtained
the next expression:

1 L
⋅ E ⋅ ε max ⋅ R 3 ⋅ π = ⋅ Fy (3.1.1.37)
4 4

By working out the value of εmáx from equation (3.1.1.37), the following
expression is obtained:

L ⋅ Fy
ε max = (3.1.1.38)
E ⋅ R3 ⋅ π

Finally, the last thing would be to relate the loss factor (η) with the parameters
that are known or that can be easily measurable. By introducing the equation (3.1.1.34)
into equation (3.1.1.35), the following expression is obtained:

17
M. Martínez Analysis of Structural Damping

4 1 E ⋅η d
Fx = ⋅ ⋅ (ε max ) ⋅ R 3 ⋅ π (3.1.1.39)
L 4 ω dt

If equation (3.1.1.38) is introduced into equation (3.1.12),

d
(ε max ) is expressed
dt
through the following:

L ⋅ Fy
(ε max ) = E ⋅ R ⋅ π ⋅ r ⋅ ω ⋅ cos(ω ⋅ t )
d 3
(3.1.1.40)
dt R

Thus, Fx can be expressed through the following expression:

L ⋅ Fy
4 1 E ⋅η E ⋅ R 3 ⋅ π
Fx = ⋅ ⋅ r ⋅ ω ⋅ cos(ω ⋅ t ) ⋅ R 3 ⋅ π (3.1.1.41)
L4 ω R

By simplifying equation (3.1.1.41), the following expression is obtained:

⋅ Fy ⋅ η ⋅ cos(ω ⋅ t )
r
Fx = (3.1.1.42)
R

As Fx and Fy will be measured in the outer part of the shaft, it is possible to state
that:
Fx = Fy ⋅η ⋅ cos(ω ⋅ t ) (3.1.1.43)

As the value of Fx is fluctuating in time, it is possible to assert that the maximum

value of Fx is given by the following expression:

Fx = Fy ⋅η (3.1.1.44)

As a result, the loss factor (η) can be calculated as function of Fy and the
maximum value of Fx:
F
η= x (3.1.1.45)
Fy

18
M. Martínez Analysis of Structural Damping

3.1.2 Second Method: Damping energy

Assuming that a harmonic load, Fb sin(ωt), at the centre of a simply supported

beam of length L, the damping energy can be integrated according to:

Elastic energy: We = ∫∫ Fb ⋅ ε ⋅ dz ⋅ dA (3.1.2.1)

⎛d ⎞
Damping energy: Wd = ∫∫∫ Fb ⋅ ⎜ ε ⎟ ⋅ dz ⋅ dA ⋅ dt (3.1.2.2)
⎝ dt ⎠

Where:

dA: area element

z: coordinate along the shaft
Fb: applied force
ε: strain

Using the formulas above the damping energy over a period becomes:

2⋅π 2
R2 − y 2 ⎛ y ⋅ z ⋅ Fb ⎞
⋅ ω ⋅ cos(ω ⋅ t )⎟ ⋅ dx ⋅ dy ⋅ dz ⋅ dt (3.1.2.3)
R L
Wd = ∫ ω ∫ ∫ 2 ∫
γ ⋅⎜
⎝ E⋅I
2
0 −R − R −y 0
⎠

The solution of the integration is given by the following expression:

1
⋅ L3 ⋅ π 2 ⋅ ω ⋅ γ ⋅ Fb ⋅ R 4
2
Wd = (3.1.2.4)
12 ⋅ E ⋅ I 2

The maximum elastic energy is expressed through the following expression:

2
R R2 − y2 L ⎛ y ⋅ z ⋅ Fb ⎞
We = ∫ ∫ ∫ ⎜ ⎟ ⋅ dz ⋅ dx ⋅ dy (3.1.2.5)
−R − R2 − y2 0
⎝ E⋅I ⎠

Having solved the integration above, an expression for the elastic energy is
obtained:

1
⋅ L3 ⋅ Fb ⋅ R 4 ⋅ π
2
Wd = (3.1.2.6)
12 ⋅ E ⋅ I 2

19
M. Martínez Analysis of Structural Damping

The loss factor (η) is defined as the ratio of damping energy loss per
radian divided with the peak strain energy (elastic) as follows:

Wd
η= (3.1.2.7)
2 ⋅ π ⋅ We

By introducing equations (3.1.2.5) and (3.1.2.6) into equation (3.1.2.7), an

expression for the loss factor (η) is obtained:

ω ⋅γ
η= (3.1.2.8)
2⋅E

20
M. Martínez Analysis of Structural Damping

3.2. Experimental work

This master thesis was initially conceived as a research where the experimental
work should have been very important. However, the poor results during experiments
changed the aim of this master thesis. Since it was not possible to carry out successful
experiments to confirm the validity of the proposed method, it was decided that this thesis
should analyse why the experiment was a failure.

First of all, it will be described the setup of the experiment. It consists basically of
a simply supported beam that is connected through a coupler to an electric motor. The
beam is supported by two ball bearings that provides only translational constrains. This
means that reaction forces may be induced but not moments. This group of elements is
mounted over a metallic frame as shows figure 3.2.1.

Figure 3.2.1

As it has been mentioned in the theoretical analysis part, the loss factor (η) can be
calculated as function of Fy and the maximum value of Fx. It is important to remember
that Fy is the introduced force to keep the shaft in a constant deformed position (δ). The
easiest way to get the loss factor (η) would be to measure Fx and Fy directly. However,

21
M. Martínez Analysis of Structural Damping

faced with the shortage of resources, it was decided to obtain Fy through a direct
measurement and Fx through an indirect measurement.

As regards Fy, it is necessary to apply a force to the shaft and, at the same time,
measure it. To achieve this purpose was used a dynamometer. The dynamometer was
attached to the frame and the beam as figure 3.2.2 shows.

Figure 3.2.2

Nevertheless, it was especially difficult to attach the dynamometer to the beam

because the beam should be rotating during the experiment. Using a ball bearing, as
figure 3.2.3 shows, the problem is solved. The bearing was attached to the dynamometer
through a metallic wire.

22
M. Martínez Analysis of Structural Damping

Figure 3.2.3

The following picture shows the dynamometer and the device that was used to
adjust the introduced force.

Figure 3.2.4

23
M. Martínez Analysis of Structural Damping

About Fx, it was considered that it could be obtained by relating it with Fy through
geometry. Assuming that L is the constant distance between the beam and the place
where the dynamometer is attached to frame and d is the displacement of the beam about
x, then if is possible to obtain the value of Fx by applying geometry (see figure 3.2.5)

Fx d

L
α α

FT
Fy
Figure 3.2.5

From the figure xxx it is possible to infer the followings expression:

d Fx
tgα = = (3.2.1)
L Fy

As it was demonstrated previously in the theoretical part, the loss factor (η) can be
expressed through the following equation:

Fx
η= (3.2.2)
Fy

By identifying terms in equations (3.2.1) and (3.2.2), it is possible to assert that:

d
η= (3.2.3)
L

24
M. Martínez Analysis of Structural Damping

Surprisingly, the loss factor (η) can be easily obtained by measuring only a length
(L) and a displacement (d). As it was said previously, the length is given by the geometry
of the experimental set up. So, it is only necessary to measure the displacement about x of
the beam (d).

Analyzing the best way to measure this displacement and the available equipment,
it was decided that the best option was to use a dial indicator that is shown in figure 3.2.6.

Figure 3.2.6

The range of the dial indicator is from 0 to 10 mm and its graduation 0.01 mm. To
hold the dial indicator it was used a magnetic base indicator holder.

25
M. Martínez Analysis of Structural Damping

Chapter 4. Discussion
This part should show and discuss data from the experiments. However, the
experiments failed because of the lack of equipment. Therefore, next it will be analyzed
the different reasons that make impossible to obtain good results.

Firstly, it must be taken into account that the forces involved in the problem as
well as stresses or strains are expressed through trigonometric functions. It means that
they have a periodic and fluctuating behaviour and thus, the displacement (d) will behave
in the same way. Since the dial indicator was analogical, it was very difficult to determine
with precision the range of fluctuating displacements because their values changed too
fast.

Secondly, it was observed that the vibration on the beam was greater than the
initially expected. It could be due to the fact that the beam had initially a permanent bend
shape, what caused a higher level of vibration and thus larger displacements of the beam.
It means that it is complicate to evaluate the amount of the displacement (d) that is the
main reason for this additional vibration and to the normal vibration.

Thirdly, it was assumed that the length L was constant but actually it was not. As
it was pointed out above, the deformed position (δ) of the beam was obtained by
attaching the beam to the frame through a dynamometer, a metallic wire and a tightener.
It is well known that the dynamometer is a device that is based on the deformation of a
spring and that this deformation is proportional to the applied force (Newton’s Law).
During the experiment it was observed that there were small displacements, both vertical
and horizontal, in the group composed of the dynamometer, the wire and the tightener. It
was due to the vibration transmitted from the rotating shaft to the frame.

26
M. Martínez Analysis of Structural Damping

Chapter 5. Conclusions

This master thesis has established a new theoretical model to measure damping in
mechanical systems.

The theoretical analysis indicates that it is possible to measure damping according

to the proposed experiment.

The experiment failed due to the lack of suitable equipment. Nevertheless, it

should be possible to carry out successful experiments using better facilities.

Finally, it is considered that this master thesis could be a good point of departure
for further works.

27
M. Martínez Analysis of Structural Damping

Chapter 6. Recommendations for further work

Since the proposed method could not be proved, the most part of the
recommendations are related with the experimental work.

Firstly, the dial indicator should be digital because it should be very useful to
connect it to a computer in order to analysis how the displacement (d) depends on the
frequency or when the displacement (d) is maximum.

Secondly, the displacement (d) could be measured by a laser displacement sensor

or by a non-contact inductive displacement sensor in order to get more accurate results.

Thirdly, to reduce vibration problems, the shaft should be attached to the frame
through a rigid element. In this way, it could be assumed that the length (L) is really
constant.

To conclude, the theoretical analysis of this thesis is based on a single degree of

freedom system model. Using a multi-degree of freedom system, the results should give
better and more realistic results.

28
M. Martínez Analysis of Structural Damping

References
[1] Silva, Clarence W. de, Vibration: Fundamentals and Practice, Boca Raton,
FL: CRC Press, cop. [2000]

[2] Cyril M. Harris, Shock and Vibration Handbook, New York: McGraw-
Hill, cop. [1996]

[3] Reeves, Charles W., Machine & Systems Condition Monitoring Series,
Oxford: Coxmoor, [1998]

[4] Giancarlo Genta, Vibration of Structures and Machines, Maple-Vail Book

Manufacturing Group, York, PA, [1993]

[5] Vance, John M., Rotordynamics of Turbomachinery, John Wiley & Sons,
Inc, [1988]

[6] [Online]. Available:

http://www.roushind.com/news_downloads/white_papers/Insight.pdf

[7] [Online]. Available:

http://www.mfg.mtu.edu/cyberman/machtool/machtool/vibration/damping.
html

29