Oscillation theory

This document gives an overview of the following important subjects:

 Oscillations without damping

 Kinetic and potential energy of a oscillation
 Torsional oscillations
 Parallel and serial constructions of springs
 Oscillations with damping

To train your skills in making calculations about the subjects six exercises are given.
Only the results and answers are indicated.

1
Summary of the most important concepts

Static linear spring behaviour

Linear spring behaviour indicates that the applied force is proportional to the compression,
but in the opposite direction.

This can be indicated with the formula:

F  kz

with:

F force N 
z (vertical) displacement m
N 
k spring rate  
m
A fully linear spring is an idealistic theoretical situation, which in practice never happens. The
spring will break when the force gets too high. Before the spring breaks the spring will be
plastically deformed. During plastic deformation the spring behaviour will no longer be
linear. Linear behaviour only occurs within certain limits of the force range. Normally springs
are specially constructed for linear behaviour within specific applications

Next to real springs this behaviour is also used for modelling, often in combination with a
damped oscillation. A number of vehicle parts act as if they are connected to a (virtual) spring
and damper system.

For instance the elastic behaviour of a tyre can be considered as a spring. To give an
indication for the spring rates used with passenger cars the front springs have generally
speaking a spring rate between 10 and 20 kN/m, while tyres have (virtual) spring rates 10
times higher, 100 kN/m to 200 kN/m.

Performing accurate measurements on cars indicates that the oscillations caused by the road
surface influences the real life spring properties. The spring rates are depending on the
oscillating frequency. Real life is more complex than theory.

F  kz
This formula only indicates the static behaviour of a spring. After applying a certain force the
spring will be compressed over a related distance. After that nothing happens. The system is
in balance.
( With the letter ‘z‘ a vertical displacement is considered but of course any direction is
possible, also horizontal. Later on, dealing with the so called quarter vehicle model, a vertical
displacement will be the focus )

2
Dynamic spring behaviour

When a sudden force thrust is applied upon the spring (which afterwards disappears) the
spring will start to move. The resultant force and displacement will be time dependent

Both have become mathematical functions of time. F (t )   k z (t )

According to Newton’s law: F ma  F (t )  m a (t )

Remember that displacement, velocity and acceleration are defined according to the
differential mathematics (note the different possible notations)

Displacement z (t ) ( vertically )
dz
Velocity v(t )  z ' (t ) or otherwise as v(t )  or v (t )  z
dt
dv d 2 z
Acceleration a(t )  v' (t )  z ' ' (t ) a(t )   2 a (t )  v  z
dt dt

When substituted: m a (t )   k z (t )

This is a so-called second order linear differential equation, which can be written in several
ways.
d 2z
m z" (t )  k z (t )  0 m 2 kz0 m z  k z  0
dt

We will have to find a basic function of z (t ) which will satisfy this differential equation.
In other words, the response as time-dependant behaviour is found.

The trial-and-error method can be an interesting way to try to find a solution for this.
Differentiate twice a trial-function and see what the result will be. When this gives a
mathematical contradiction, you have started with a wrong equation.

Trial 1
Start with an easy one and take m = 1 kg and k = 1 N/m  z  z  0
A quadratic function gives: z (t )  t 2  z ' (t )  2t  z ' ' (t )  2
Substituting gives: 2  t  0 2
 t  2
2

This is not a real number, so z (t )  t 2 is not the solution.

That z (t )  t 2 is not a solution was to be expected. The displacement would only quadratic
increase. Trying other equations like z (t )  t n proves that this type of functions will never
satisfy.

3
Trial 2
It is known that with a pure exponential function, differentiating will not influence the
function itself.

The function remains itself: z (t )  et  z ' (t )  et  z ' ' (t )  et

Substitution: et  et  0  2e t  0

This is no solution either, but if there had been a ‘minus-sign’ in e t  e t  0 than yes, we
would have had a solution.
But this solution is in itself not a logical choice, because the displacement would increase
exponentially in time.
(Students who are familiar with complex numbers, will be able to prove that imaginary
exponential equations satisfy)

Trial 3
When we observe the spring behaviour (without damping), it seems to be fully periodical in
time. In the middle a high movement, gradually slowing to a stop at the top and then
accelerating to the middle and again slowly stopping at the bottom. And it keeps repeating
itself. There is one function which complies with this perfectly: the sine or cosine.
Let’s take a sine-function and differentiate it twice:
z (t )  sin t  z ' (t )  cos t  z ' ' (t )   sin t
Substitution gives:  sin t  sin t  0  0 0

The mathematical equation 0 = 0 means that this function satisfies for every t.
Using the cosine gives equal results.
Try proving yourself that every equation of z (t )  A sin t  B cos t also satisfies.

This is called the superposition principle.

It is characteristic for each linear differential equation (also higher order ones). If you have
found two different solutions than all combination of them is also a solution.
This is not valid for non-linear differential equations. Apparently linear differentials are very
special.

General solution
The cosine is identical to the sine only shifted by π/2 rad. using some goniometric formulas
you can change any sin- and cos-combination to only a sine (or cosine) equation. The price to
pay is a phase-shift within the sin-function

In case the position z and velocity v at t=0 sec are unknown then we will have two parameters
in the general solutions as constants.
In case the initial conditions are known, we can determine the value for these two parameters.
As a rule an nth order linear differential equation without initial conditions also gives n
parameters. Just as here.

A periodic motion complying with this sine-shaped oscillation is called a harmonic motion.
Be advised that there are also non-harmonic periodical oscillations, such as a block-signal.

4
The general solution can be written as:

z (t )  A sin 0 t  0 

In which:

A Amplitude m
 rad 
 0 circular frequency  or angular frequency
 s 
 0 initial phase angle rad 

Amplitude
As we started with a spring-mass system with a displacement z in [m] the amplitude indicated
here is also in the same dimension [m]. But as this same oscillation is also valid for other
systems, it can also be indicated in other dimensions for instance electricity [Volt], [Amperes]

Angular frequency
The term angular frequency or circular frequency has to do with a circle, sometimes virtually
but sometimes in reality. Every oscillation can be seen as the projection of a uniform circular
motion on a straight line through the circle (X, Y-axis or under any angle). The dot on the
circle moves uniformly through a phase angle in [rad].
If the dot makes one complete circle in 1 second then the frequency is 1 Hz. The passed
through phase angle is then 2π radians which indicates an angular frequency (also used is
circular frequency) of 2π rad/s.

This gives the relation between frequency and angular velocity (circular frequency).
  2 f

The multiplier 2π has to do with the circumference of a circle converted into a phase angle in
radians.
The projection can be a virtual mathematical one, but also a real one. For instance the up and
down movement of reflectors on the pedals of a bicycle as seen at night from behind.

Natural frequency
When a spring-mass-system is excited, the mass will at first start to move in its natural
frequency. This can also be called the fundamental harmonic, indexed with 0. Also possible is
an oscillation in one of the harmonics, indexed with 1 or 2 etc.

0  2  f 0 , 1  2  f1 , 2  2  f 2 etc.

(For the guitar players; you can play a guitar string with its natural frequency but also in its
second harmonic. This will sound one octave higher (frequency is doubled) and the string is
hardly moving in the middle.)
If a spring-mass system is excited by an external force with a periodical almost equal to the
natural frequency of the system resonance might occur, especially when undamped. This is
something you try to prevent.

5
Phase and phase angle
the sine function, as in the general solution, is always based on an angle in [rad].
This is the total phase angle 0 t   0 which is a linear function in time. It represents the
(virtual) circular movement. The initial phase angle  0 is the location at the time t = 0 sec
The phase is defined by:
phase angle
phase 
2
The passed through phase angle is divided by the movement of one complete circle of 2π
radians.
This results in a dimensionless number indicating how far an oscillation has come (or a phase
shift in relation to another oscillation.)
For instance when the phase is 3.5 the oscillation will have gone through 3 ½ oscillations.
The phase angle will then be 7π.

Period
Instead of frequency one can also use the dimension period T.
1
It is the reciprocal of the frequency: T 
f

Influence of mass and spring rate

There is a relation between the angular velocity, the oscillating mass and the spring rate,
which is to be expected. A stiffer spring will oscillate with a higher frequency than a softer
spring and how higher the mass the slower the movement.

Starting with differentiating the displacement formula, using the chain rule we can derive the
formula for velocity and doing it once again, the acceleration

displacement z (t )  A sin 0 t  0 

velocity v(t )  z ' (t )  A0 cos 0 t  0 
acceleration a(t )  v' (t )  z ' ' (t )   A02 sin 0 t  0 

Substituting in the second order differential equation gives:

m z" (t )  k z (t )  0  A m 02 sin 0 t   0   A k sin 0 t   0   0

 
Reordering: m 0  k   A sin 0 t   0    0
2

A multiplication of two factors is equal zero. That is possible if the individual factors are
equal zero. Taking the first factor gives:
k
 m 02  k  0  m 02  k  0 
m

Or when written in frequency f or period T

1 k m
f0  T0  2
2 m k

6
Energy of an undamped oscillation
In an undamped oscillation only potential and kinetic energy are important.
Kinetic energy E k  12 m v becomes time dependent. E k  12 m v (t ) 
2 2

With position (or displacement) x ,substituting v(t )  x' (t )  A 0 cos 0 t  0  gives

Ek  m  A0 cos 0 t   0   m A 2 02 cos 2 0 t   0 

1 2 1
2 2

Potential energy is the work the spring can deliver while moving back to its rest position.

Work is defined as W  F  x Where F is a function of position x : F (x)

We have to look at this system in very small differentials  W  F ( x)  x .

Which can be written as: d W  F ( x) dx By integrating this differential from starting point
x0 to x=0 we get the potential energy

0
E p   F ( x) dx 
x0 x0
0

 k x dx   12 k x 2 0
x0
 0   12 k x02  12 kx02
So it full depends on the starting position x0. Let’s make x0 time dependent x(t)
 E p  12 k x 2

Substituting x(t )  A sin 0 t  0  gives:

Ep  k  A sin 0 t  0   k A2 sin 2 0 t  0 

1 2 1
2 2

Total energy Et

Et  E p  E k

Et  1
2 A2 k sin 2 0 t  0   1
2 A2 m 02cos 2 0 t  0 

Using: k  m 02

Et  1
2 m 02 A2  sin 2 0 t   0   cos 2 0 t   0  

And using the trigonometric formula: sin 2   cos 2   1

Also valid for: sin 0 t   0   cos 0 t   0   1

2 2

The total energy becomes:

Et  12 m 02 A2  12 k A2

7
The time dependency is gone from the equation and only fixed parameters are left. This
means, that given a certain spring rate and amplitude the total energy is always constant.
This is the law of conservation of energy

Which is logical here, as there is only kinetic and potential energy which is continuously
exchanging itself? Added together the total (undamped) energy is always constant.

Torsional oscillations
Harmonic oscillators occurring in a number of areas of engineering are equivalent in the sense
that their mathematical models are identical. Below is a table showing analogous quantities in
mechanical translational and rotational oscillator systems. If analogous parameters on the
same line in the table are given numerically equal values, the behaviour of the oscillators—
their output waveform, resonant frequency, damping factor, etc.—are the same.

TRANSLATIONAL OSCILLATIONS ROTATIONAL OSCILLATIONS

position, location, displacement x angle 

velocity v  x angular velocity   

acceleration a  v  x angular acceleration   

force F torque M

mass m moment of inertia J

Newton’s law F  m a M J

spring rate F   k x torsion constant M    

k 
resonance 0  0 
m J
m J
T0  2 T0  2
k 

The moment of inertia may have the same quantity as mass has, but mass doesn’t change
when changing directions of translations. The moment of inertia can and does change when
changing the rotational axis.

8
The moment of inertia changes with the location and orientation of the rotation axis. If you
know the inertia with the rotation axis through the centre of gravity, you can calculate it for a
different location.

The moment of inertia of a point mass m rotating at a distance r from the axis of rotation is:

J  mr2

In case you have more point masses rotating around the same point, the whole can be added.
Just like you do with masses.
n
J  m1 r12  m2 r22  m3 r32  m4 r42  m5 r52 ...........   mi ri 2
i 1

In a spatially extended object, it is of course (in practice) impossible to carry out an addition
of all rotating molecules (atoms). It’s more convenient to proceed with an integration of all
mass elements dm. This gives:

n
J m r
i 1
i i
2
 J r
2
dm

If the specific mass of the object is the same all over, then the mass element can also be
described as a volume element dV with specific mass ρ in front of the integration sign.

J r dm   r 2  dV    r 2 dV
2

If the object has a complex structure, this is more easily said than done.
First a clever mathematical parameterization has to be found for the structure and only then
can the integration be performed.
If the shape is more symmetrical, this calculation can become a lot easier.

Example
Calculate the moment of inertia of a circular disc with height h with the axis of rotation
perpendicular to the disc and through the centre of gravity.
(This type of object is frequently used in technology, for instance wheels, gearwheels etc)
The volume is surface x height.
The surface is a circle which is created by integrating 2πrdr from r = 0 to r = R
See if you come to the same calculation and result.
R
J    r 2 dV    r 2 2 r h dr   2 h
0 
0
R
r 3 dr   2 h r
1
4
4 R
0
  2 h 14 R 4

 
J   2 h 14 R 4  12   hR 4  12   R 2 h  R 2  12 mR 2

The total mass is ρV with V = (surface x height) = πR2·h

There follows the simple formula: J  12 mR
2

9
Different objects
Likewise you can calculate the moment of inertia for other shapes and objects
(rod, cube, sphere, etc.) Below is an overview.

Thin-walled cylinder with radius R

and with longitudinal axis = rotation axis J  mR 2

Hollow cylinder with inner radius R1

outer radius R2 and longitudinal axis = rotation axis 
J  12 m R12  R22 
Rod with length L, rotation axis through
the centre of gravity and perpendicular to the rod J  121 m L2

Rectangular box with sides a and b

Rotation axis through the centre of gravity and
perpendicular to the box 
J  121 m a 2  b 2 
Thin ring with radius R and
a tangent line as rotation axis J  32 m R 2

Sphere with radius R with the rotation axis

through the centre of gravity J  52 m R 2

This list is far from complete, but covers most systems. More exotic shapes are not generally
used.

Torsion constant
The torsion constant or the coefficient of torsion spring functions in the same way as the
‘spring rate‘ does with the statical mass-spring behaviour. If one twists a rod along its
longitudinal axis, you create static counter torque, which works in the opposite direction and
is proportional to the applied torque.
This counter torque is proportional to the angular displacement from the equilibrium position.
 Nm 
M     with torsion constant  in  
 rad 
When jolted with a torque a torsional oscillation will occur in which the natural frequency is
dependent on the torsion constant and the moment of inertia.

 1 
0  0  2 f 0 f0 
J 2 J

10
Parallel and series springs
Springs can be installed in series and parallel or as a combination of both. The whole system
can be seen as a new system with just one replacement spring. The spring rate for this
replacement spring is to be calculated.

C1
C1 C2

M M

C2

Let’s start with the serial installation. Consider what is understood with this. There’s only one
connection point and the springs are installed in line. The mass at the end of the last spring
can only freely move in one direction.
Take at first an installation with only two springs (see left-hand side drawing). For each
spring the equation F   k x is valid : F1   k1 x1 and F2   k 2 x2
The total displacement is equal to xt  x1  x2
Because of Newton’s law, action = - reaction, the force working on each spring is equal to the
total force acting on the system Ft  F1  F2
F
With x   substituted in xt  x1  x2
k
F F F Ft Ft Ft 1 1 1
Gives:   t   1  2      
kt k1 k 2 kt k1 k 2 kt k1 k 2

With a parallel installation (see right-hand side drawing) force and displacement act
mathematically in an opposite way. The displacement for each spring is equal and the forces
have to be added up xt  x1  x2 and Ft  F1  F2

Substituting this in F   k x gives

 kt xt   k1 x1  k2 x2  kt xt  k1 xt  k2 xt  kt  k1  k2

Remark: When a mass is clamped with springs between two fixed sides, this is also a parallel
installation, although it looks like a serial installation. The two fixed sides can be considered
as one side only.

Conclusion:
1 1 1
Serial installation   parallel installation kt  k1  k 2
kt k1 k2
Although it has nothing to do with each other, you can compare this with electrical systems.
With electrical resistors the replacement value is the exact opposite, but identical with
capacitors (coincidentally, the C of capacitance of a capacitor is identical to the C of the
spring rate.)

11
Damped oscillation

In real oscillators friction or dampers slow down the movement. In most systems the frictional
forces are proportional to the velocity.

Fs k
xs

m
c
F
Fd xd

For the damped mass-spring system we can create the next equations:

Fs   k xs

dxd
Fd   c
dt
In this case the spring and damper are parallel installed, which is normally also the case.
Then:

F  Fs  Fd en xs  xd  x

Substituting there follows:

F   k x  cx  mx 
mx  cx  kx  0

Time t
Amplitude A
Start phase angle 0
k
Undamped circle frequency  0  2 f 0 
m
Damped circle frequency d  0 1   2
c
Damping ratio   
2 mk

The next equation is valid for the displacement x(t). It satisfies the differential equation.
 
x(t )  Ae  0 t sin  d t   0 
 

12
For the calculation of the velocity at time t, you must differentiate x(t).
Make use of the product and chain rule for differentiating functions.
Pay attention to the positions of the damped and undamped circle frequencies.

   
   0  Ae   0 t sin  d t   0   d A e  0 t cos  d t   0 
dx
v(t ) 
dt    

dx     
v(t )   A e  0 t    0 sin  d t   0   d cos  d t   0  
dt     

The undamped oscillation is in reality practically non-existent. In the mathematical model the
undamped oscillation remains unchanged indefinitely once a mass has been moved outside its
equilibrium. A more realistic model is the damped oscillation. When a mass has been moved
out of its equilibrium it loses a little energy during each oscillation cycle. This energy is
dissipated as heat. Generally speaking the energy diminishes in time as an exponential
function.
The displacement of a damped oscillation over time depends on the damping ratio.
When the damping ratio is less than 1 it looks as below:

Is the damping ratio greater (but still less) than 1 the oscillation will stop sooner.

Becomes the damping ratio even higher and reaches the ‘critical damping’ value, then the
displacement reaches ‘0’ very quickly, without moving through its equilibrium (see graph
below).
Becomes the factor even greater, then the oscillation is ‘overdamped’

13
In some cases a certain overshoot is wanted. For instance with analogue measuring
instruments just 1 time overshoot is wanted for acceptable reaction speed and a quickly
reached stable readout. In those cases the damping is about 0.7 times the critical damping
factor.

Summarized:

c
Damping ratio   
2 mk

 =0 undamped
The system oscillates in its natural frequency ω0 without experiencing
decay of its amplitude. ( only pure sin- or cos-function )

 <1 underdamped
The system oscillates (at reduced frequency ωd compared to the
undamped case) with the amplitude gradually decreasing to zero.

 =1 critically damped
The system returns to equilibrium as quickly as possible without
oscillating. ( it’s a theoretical case )

 >1 overdamped
The system returns (exponentially decays) to equilibrium without
oscillating.

14
Exercise 1

An undamped mass-spring system with a spring rate of 18500 N/m and a mass of 300 kg
performs a harmonic oscillation with amplitude of 3 cm
The phase angle at time t = 0 sec is 5 rad
Calculate:

a) Circle frequency, frequency and period

b) Displacement and phase at time t = 1 sec

c) Velocity and acceleration at time t = 2 sec

Exercise 2

A mass-spring system with a spring rate of 12 kN/m performs an undamped harmonic

oscillation. The displacement x(t) in cm as a function of time t in sec behaves as follows:

x(t) = 3 sin ( 8 t + 2 )

a) What is the value of the circle frequency and the phase on time t = 0 sec?

b) Determine the phase angle at time t = 0 sec and t = 1 sec?

c) Determine the kinetic energy at time t = 1 sec

d) Determine the potential energy at time t = 1 sec

e) What is the total energy at time t = 2 sec and on t = 5 sec ?

Exercise 3

In a laboratory engineers are testing a torsion spring application in an automotive construction.

Prototypes are tested and the first measurements are performed on a torsion spring attached to a
circular metal ring.
The ring has a radius of 9 cm. The rotation point is also the centre of mass.
In these first tests the mass of the spring can be neglected in relation to the mass of the ring.
A static enforced torque of 57 Nm causes the ring to turn 30°.
When acting under an impulse the dynamic behaviour can be studied. The resultant
(undamped) torsion movement has a frequency of 6 Hz.

Determine:

a) The torsion constant of this torsion spring

b) The moment of inertia of the ring

c) The mass of the ring

15
Exercise 4

At a test department of TNO the test engineers perform some measurements. They test a mass-
spring system with a spring rate of 195 N/cm and a mass of 380 kg. The system is jolted with a
sudden force.

a) Determine the undamped natural frequency.

The amplitude of the resultant harmonic movement is 2 cm.

The phase angle at time t = 0 sec is 1.5 rad. The oscillation can at start be seen as undamped.
Determine for time t = 3 sec:

b) the acceleration c) potential energy d) kinetic energy

Determine for time t = 5 sec

e) The total amount of energy of the system.

Exercise 5

Given a damped mass-spring system with a mass of 320 kg, a spring rate of 12.7 kN/m and a
damping constant of 5100 kg/s

a) Determine the damping ratio.

b) What type of damping is applicable here?

The mass and spring rate remains unchanged. But the designers want to change the damping
in such a way, that the damping ratio becomes 0.29.

c) To what value do they have to change the damping constant and what type of damping is
then applicable?

At time t = 0 sec the initial phase angle is 0.9 rad and the x(0) = 5 cm.

Determine:

d) The undamped natural frequency

e) The damped natural frequency

f) The displacement and velocity at time t = 0.5 sec

16
Exercise 6

Given three springs with the following spring rates

k1 = 800 N/m k2 = 1500 N/m k3 = 2800 N/m

These springs are tested in two different configurations in which at the end a mass of 25 kg
is attached.
Determine the replacement (total) spring rate, the period and natural frequency in case the
springs are installed as indicated:

a) Parallel

b) Serial

Formulas

d 2x dx dv d 2 x
m 2
 c  kx  0 a  x(t )  A sin 0t   0 
dt dt dt dt 2

1 m
 t    0 t   0
k
02  T  2
m f k

c
E k  12 m v 2 E p  12 k u 2  
2 mk

n
J
J m r
i 1
i i
2
J  r dm   r  dV
2 2
T  2

V V

J sphere  52 m R 2 J wheel  12 m R 2 J rod  121 m L2

 
d  0 1   2 x(t )  Ae   0 t sin  d t  0 
 

dx     
v(t )   A e  0 t    0 sin  d t   0   d cos  d t   0  
dt     

17
Answers

Exercise 1

a) 7.85 rad/s 1.25 Hz 0.8 s

b) 0.85 cm 2.05

c) -6.6 cm/s -1.8 m/s2

Exercise 2 Exercise 3

a) 8 rad/s 0.32 a) 109 Nm/rad

b) 2 rad 10 rad b) 765 kgcm2 ≈ 0.08 kgm2

c) 3.8 J c) 9.5 kg

d) 1.6 J

e) 5.4 J 5.4 J

Exercise 4 Exercise 5

a) 1.14 Hz a) 1.26

b) 0.9 m/s2 b) above critical damping (overdamped)

c) 2.8 J c) 1169 kg/s

d) 1.1 J d) 1 Hz

e) 3.9 J e) 0.96 Hz

f) -1.8 cm 7.78 cm/s

Exercise 6

a) 5100 N/m 0.4 s 2.3 Hz

b) 440 N/m 1.5 s 0.7 Hz

18