Vibration in five lectures

Sudipto Mukherjee, IIT Delhi

This lecture is based upon a basic mechanics course and the text by
Ghosh and Mallik. We assume that you understand basic concepts of
degree of freedom as in the figures below having one, two, two and
infinite degree of freedom.

Moreover that if the mass at the end in the figure below is significantly
(factor of 10?) heavier than the beam, it can be approximated by a single
degree of freedom. Otherwise, design using the lumped mass approach
that we will study, would be dicey.
In mechanics, we usually restrict ourselves to studying position (x),
velocity (𝑥̇ ) and acceleration (𝑥̈ ) dependent forces. While the
acceleration dependent term is always linear and proportional to the
mass; that is not necessarily true of the position and velocity dependent
terms, stiffness and damping. Luckily, for small velocities and
displacement, linearity is observed in most mechanical systems.

𝑥̈ 𝑥 𝑥̇

a) Free vibrations: system disturbed from an equilibrium position

(pendulum)
b) Forced vibration: external disturbance independent of the motion
being tracked (eccentric rotating mass)
c) Self-excited vibration: External disturbance is dependent on the
motion of the system (singing of wires)
d) Parametrically excited vibration: system parameters changing with
time.
e) Random vibration: Excitation is not deterministic but only
probabilistically known.
Un-damped free vibration

since Fs=-kx

𝑚𝑥̈ + 𝑘𝑥 = 0
𝑥 (𝑡) = 𝐴 𝑠𝑖𝑛 𝜔𝑛 𝑡 + 𝐵 𝑐𝑜𝑠 𝜔𝑛 𝑡 = 𝑋 𝑐𝑜𝑠(𝜔𝑛 𝑡 + 𝜙)
𝑘
With 𝜔𝑛 = √ being the natural frequency, which is a property of the
𝑚
system

If the motion is in the vertical direction, how does gravity modify the
system? Clearly, the mass will descend by an amount  = mg/k.

Writing the force balance equation in a coordinate system whose origin

is shifted to the new static equilibrium position:

𝑚𝑥̈ = 𝑚𝑔 − 𝑘(𝑥 + 𝛥) = −𝑘𝑥

which is in the same form as earlier with the coordinate shifted!

Energy Conservation Approach

1 1
𝑇 = 𝑚𝑥̇ 2, 𝑉 = −𝑚𝑔𝑥 + 𝑘(𝑥 + 𝛥)2 , with the datum as the
2 2
equilibrium position.
𝑑
Using (𝑇 + 𝑉 ) = 0 we get: (𝑚𝑥̈ + 𝑘𝑥 )𝑥̇ = 0 ,
𝑑𝑡
and only if 𝑥̇ is not equal to zero at some time, we get the same equation
as earlier.
conservative sys means that total energy is constant

This approach is sometime easier than the force balance approach, but is
applicable only for conservative systems.

There are several derivations of free vibration equations in Ghosh and see this on

Mallik that you can investigate (especially the Trifilar suspension which page no
480 of

is perhaps the only experimental method to measure moment of inertia). edge

Other books on vibration will have many more.

One of special interest is the rod on two roller system shown below with
x as the excursion of the rod CG from the midpoint of the roller centres:

𝑎−𝑥 𝑎+𝑥
𝑁1 = 𝑚𝑔 , 𝑁2 = 𝑚𝑔 and writing the equation of motion in the
2𝑎 2𝑎
x direction:
𝑚𝑥̈ = −𝜇(𝑁2 − 𝑁1 )
𝑥̈ + (𝜇𝑚𝑔/𝑎)𝑥 = 0

which is like that of a SDOF system with 𝜔𝑛 = √𝜇𝑔/𝑎!

There is no restoring springlike element, and this is an example of a self-

excited system where the restoring force is generated by the motion of
the rod! This system is used to measure friction coefficients, especially
velocity dependent friction coefficients.
Damping

Energy of real-life systems is not conserved gets dissipated and the rate
of decay of free vibration is a measure of damping. This could be due to:

a) Dry friction between surfaces (Coulomb)

b) Drag between moving body and fluid environment (viscous)
c) Internal friction within a material (hysteretic)
d) Through radiation of energy into environment (wave)

Consider free vibration with viscous damping.

𝑚𝑥̈ + 𝑐𝑥̇ + 𝑘𝑥 = 0

Assuming a solution of the form: x = est, we get:

𝑐 𝑐2 𝑘
𝑚𝑠 2 + 𝑐𝑠 + 𝑘 = 0 which yields 𝑠1.2 = − ±√ 2
−
2𝑚 4𝑚 𝑚

We see the square root, and we are instantly worried about number of
solutions and real vs imaginary solutions.
 a) 𝑐 > 2√𝑘𝑚, over damped, in which case both roots are real and
negative. 𝑥 = 𝐴𝑒 𝑠1𝑡 + 𝐵𝑒 𝑠2𝑡
and does not have periodic
motion.

b) 𝑐𝑐 = 2√𝑘𝑚, critically
damped, in which case both
roots are identical, and the
solution is of the form 𝑥 =
(𝐴 + 𝐵𝑡)𝑒 −2𝑚 (aperiodic) and the ratio, c/cc = , is called the damping
𝑐𝑡

ratio.

c) 𝑐 < 2√𝑘𝑚, under-damped, in which case both roots are imaginary

1
and the solution is periodic and 𝑥 = 𝑋𝑒 −𝜁𝜔𝑛 𝑡 𝑐𝑜𝑠 [(1 − 𝜁 2)2 𝜔𝑛 𝑡 + 𝜙]

The frequency of oscillation is different from n and is called the

damped natural frequency. The amplitude of oscillation decays with
time. The constants X and  depend on the initial conditions.

1
The ratio of successive peaks is, 𝑥𝑥𝑛+1
𝑛
= 𝑒𝑥𝑝 (2𝜋𝜁/(1 − 𝜁 2 )2 ) and

ln(xn/xn+1) =  is called the logarithmic decrement

which is used to measure the damping in the system. Practically,

1
𝑙𝑛 (𝑥 𝑛 ) = 2𝜋𝑝𝜁/(1 − 𝜁 2 )2 = 𝑝𝛿 is used for measurement.
𝑥
𝑛+𝑝
Rotating vector and phase-plane representation

Harmonic quantities are

conveniently represented
in this form with speed
of rotation , and length
of the vector as the
amplitude. The
projection of the vector
on the reference line
represents the
instantaneous value of
the harmonic quantity.

Differentiation multiplies the magnitude by  and rotates the vector by

90o (leading) in the direction of rotation. The vectors can be multiplied
by constants and added vectorially to represent equations.

The other representation is of the phase-plane which has vertical axis as

x and the horizontal axis as 𝑥̇ /𝜔𝑛 . A point at the tip of the vector
represents a particular phase of the motion. Note that the arrowhead on
the plot is such that for a +ve velocity, x increases.
Undamped Vibration with Harmonic Excitation

This solution of a SDF system subjected to a harmonic force Fcos(t) is

used to analyze general vibration problems by using Fourier transforms.

For general force input, compute the

Fourier transform of the input force
and use superposition of solutions for
the general solution.

If we consider 𝑚𝑥̈ + 𝑘𝑥 = 𝐹 𝑐𝑜𝑠 𝜔 𝑡,

the solution is of the form:
𝐹0 /𝑘
𝑥 = 𝐴 𝑠𝑖𝑛 𝜔𝑛 𝑡 + 𝐵 𝑐𝑜𝑠 𝜔𝑛 𝑡 + 𝑐𝑜𝑠 𝜔 𝑡 for  ≠n and
1−(𝜔/𝜔𝑛 )2
𝐹0 𝑡
𝑥 = 𝐴 𝑠𝑖𝑛 𝜔𝑛 𝑡 + 𝐵 𝑐𝑜𝑠 𝜔𝑛 𝑡 + 𝑠𝑖𝑛 𝜔 𝑡 for  =n
2𝑚𝜔𝑛

For zero initial

conditions we see beats
as there are two
frequencies in the final
solution for the first
case.

For the second case, the magnitude of the

frequency at  grows linearly with time
and will finally go to infinity. This is called
resonance.

See
[Link]
[Link]
One needs to remember that all
systems have some damping. So, the
initial free vibration will damp out
eventually and only the steady state
solution will remain in the long run.

For input force of amplitude 𝐹0 ,

𝐹 /𝑘
𝑥 = 0 2 𝑐𝑜𝑠 𝜔 𝑡 for r < 1
1−𝑟
𝐹0 /𝑘
or 𝑥 = 2 𝑐𝑜𝑠(𝜔𝑡 − 𝜋) for r >1.
𝑟 −1
Thus, the response is in phase or
out of phase in the two zones.
Deflection is more than the static
deflection of force 𝐹0 till r > 2

Viscously damped Vibration with Harmonic Excitation

𝑚𝑥̈ + 𝑐𝑥̇ + 𝑘𝑥 = 𝐹0 𝑐𝑜𝑠 𝜔 𝑡

𝑥̈ + 2𝜁𝜔𝑛 𝑥̇ + 𝜔𝑛2 𝑥 = (𝐹0 /𝑚) 𝑐𝑜𝑠 𝜔 𝑡

Which has a solution of the form:

𝑥 = 𝑋 𝑐𝑜𝑠(𝜔𝑡 − 𝜙)with 𝑋 =
𝐹0 /𝑘 2𝜁𝑟
1 and 𝑡𝑎𝑛 𝜙 = 2
[(1−𝑟 2)2 +(2𝜁𝑟)2 ]2 1−𝑟

𝑋 1
Define 𝑀 = = 1
𝐹0 𝑘 [(1−𝑟 2 )2+(2𝜁𝑟)2]2
as the magnification factor and the peak
1
amplitude occurs at 𝜔𝑝 = (1 − 2𝜁 2 )2 𝜔𝑛
At resonance, the phase is /2, with the inertia force being neutralized
by the spring force.

The half power band width is used

to measure the damping for small
damping : 𝜁 = (𝜔2 − 𝜔1 )/(2𝜔𝑛 )
Systems with base excitation, 𝒚 = 𝒀𝒄𝒐𝒔 𝝎𝒕
The base excitation could be an earthquake for example.

𝑚𝑥̈ + 𝑐 (𝑥̇ − 𝑦̇ ) + 𝑘 (𝑥 − 𝑦) = 0

1
1+(2𝜁𝑟)2 2 2𝜁𝑟 3
𝑋 = 𝑌 [(1−𝑟 2)2 2
] and 𝑡𝑎𝑛 𝜙 =
+(2𝜁𝑟) 1−𝑟 2 +(2𝜁𝑟)2

Transmissibility

We now ask how much of pump vibration is

transmitted to the rest of the building.
TR = X/Y in support motion
= FTo/F0 in force excitation
1
1+(2𝜁𝑟)2 2
𝑇𝑅 = [(1−𝑟2 )2 ] in force transmission as
+(2𝜁𝑟)2
well as base motion.
Usually, r is kept less than 5. Note that very low k values lead to large
static deflections and large vibrations even if the force transmitted is
reduced.
Systems with two degrees of freedom
We have seen the nature of vibration of SDOF systems. When
linearized, SDOF system are good models for vibration of many
systems. We need to study 2 DOF systems to design absorbers to control
the induced vibration.

Free Vibrations
For a single degree of freedom system, an arbitrary initial vibration leads
to harmonic oscillation at the natural frequency of the system. In MDF
systems, only specific initial disturbances cause the whole system to
vibrate at single frequencies, which are erroneously called natural
frequencies as there is nothing natural about them. As many such
frequencies as DOF’s.

At these frequencies, the motion is such that the equilibrium positions of

the multiple bodies are achieved simultaneously. The phase difference is
either 0 or π. So, the amplitude of the displacement coordinates have a
fixed ratio, which defines the normal mode and is a system property as
well.

We assume solutions of the form:

x1 = X1 cos ωt, x2 = X2 cos ωt.

where X1 and X2 are signed variables to allow the phase reversal.

𝑚𝑥̈ 1 + 2𝑘𝑥1 − 𝑘𝑥2 = 0 𝑚𝑥̈ 2 + 2𝑘𝑥2 − 𝑘𝑥1 = 0

(2𝑘 − 𝑚𝜔2 )𝑋1 − 𝑘𝑋2 = 0,

−𝑘𝑋1 + (2𝑘 − 𝑚𝜔2 )𝑋2 = 0

An obvious solution is for X1 and X2 to be zero. For non-trivial solutions,

the determinant of the coefficients must be zero, yielding the
characteristic equation:

(2𝑘 − 𝑚𝜔2 )2 − 𝑘 2 = 0 𝑜𝑟 𝑚2 𝜔4 − 4𝑘𝑚𝜔2 + 3𝑘 2 = 0.

𝜔12 = 𝑘/𝑚, 𝜔22 = 3𝑘/𝑚, (11.94)

𝑋1 /𝑋2 = 1 for 𝜔 = 𝜔1 ,
𝑋1 /𝑋2 = −1 for 𝜔 = 𝜔2
are the two normal modes.

In the first mode, the middle spring does not deform, and the natural
frequency is given by √𝑘/𝑚. In the second mode, the midpoint of the
middle spring does not move. So, an equivalent system can be obtained
by considering the midpoint to be fixed there. So two identical systems,
one on the left and one on the right can be created, each like the figure
below:

Point A is called a node and there are in general n -1 nodes for a n DOF system.
The normal modes are given by:
(1) (1) (2) (2)
𝑋1 /𝑋2 = 𝜆(1) , 𝑋1 /𝑋2 = 𝜆(2) .

The general solution is obtained as a linear combination of these two motions:

(1) (2)
𝑥1 = 𝜆(1) 𝑋2 cos(𝜔1 𝑡 + 𝜙1) + 𝜆(2) 𝑋2 cos(𝜔2 𝑡 + 𝜙2),
 (1) (2)
𝑥2 = 𝑋2 cos(𝜔1 𝑡 + 𝜙1 ) + 𝑋2 cos(𝜔2 𝑡 + 𝜙2 ) (11.96)

(1) (2)
Which has four unknown constants, 𝑋2 , 𝑋2 , 𝜙1, 𝜙2 , to be determined from IC’s.

Forced vibration with harmonic excitation

𝑚1 𝑥̈1 + 𝑘1 𝑥1 + 𝑘2 (𝑥1 − 𝑥2 ) = 𝐹0 cos 𝜔𝑡 (11.97)

𝑚2 𝑥̈ 2 − 𝑘2 (𝑥1 − 𝑥2 ) = 0.

Ignoring the transient part:

𝑥1 = 𝑋1 𝑐𝑜𝑠 𝜔𝑡, 𝑥2 = 𝑋2 𝑐𝑜𝑠 𝜔𝑡 (11.98)

(𝑘1 + 𝑘2 − 𝑚1 𝜔2 )𝑋1 − 𝑘2 𝑋2 = 𝐹0 , −𝑘2 𝑋1 + (𝑘2 − 𝑚2 𝜔2 )𝑋2 = 0.

𝐹0 (𝑘2 −𝑚2 𝜔2 )
𝑋1 = (11.99a)
(𝑘1 +𝑘2 −𝑚1 𝜔2 )(𝑘2 −𝑚2 𝜔2 )−𝑘22
𝐹0 𝑘2
𝑋2 = . (11.99b)
(𝑘1 +𝑘2 −𝑚1 𝜔2 )(𝑘2 −𝑚2 𝜔2 )−𝑘22

Characteristic equation for the system is obtained by setting the

denominator to zero:
(𝑘1 + 𝑘2 − 𝑚1 𝜔2 )(𝑘2 − 𝑚2 𝜔2 ) − 𝑘22 = 𝑚1 𝑚2 (𝜔2 − 𝜔12 )(𝜔2 − 𝜔22 ).
Undamped Vibration Absorbers (Neutralizers)

X1 = 0, when 𝑘2 = 𝑚2𝜔2 opens interesting possibilities in controlling the

vibration of undamped SDOF under the action of a forcing frequency near ωn
(large vibrations).

𝐹0 (𝑘𝑎 −𝑚𝑎 𝜔2 )
𝑋= 2 (11.100a)
(𝑘+𝑘𝑎 −𝑚𝜔2 )(𝑘𝑎 −𝑚𝑎 𝜔2 )−𝑘𝑎
𝐹0 𝑘𝑎
𝑋𝑎 = 2 (11.100b)
(𝑘+𝑘𝑎 −𝑚𝜔2 )(𝑘𝑎 −𝑚𝑎𝜔2 )−𝑘𝑎

𝜔 = 𝜔𝑛𝑎 = (𝑘𝑎 /𝑚𝑎 )1/2 (11.102)

Leads to:
𝑋𝑎 = −𝐹0 /𝑘𝑎 and zero motion of the primary mass! However, this is effective
only if the forcing frequency is constant and the system is tuned.

A small deviation in forcing

frequency from ωnp can make the
amplitude larger than without the
absorber. So, the interval between
𝜔1 and 𝜔2 , which is dependent only
on the mass ratio ν = ma/m , must be
as wide as possible.

Determine 𝜔1 and 𝜔2 from zero

crossing of the amplitude of the mass.
(𝑘 + 𝑘𝑎 − 𝑚𝜔2 )(𝑘𝑎 − 𝑚𝑎 𝜔2 ) − 𝑘𝑎2 = 0
𝑘𝑎 𝑘 𝑘𝑎 𝑘𝑘𝑎
𝜔4 − ( + + ) 𝜔2 + =0
𝑚𝑎 𝑚 𝑚 𝑚𝑚𝑎
𝜔 𝜔
( )4 − (2 + 𝑣)( )2 + 1 = 0 (11.103)
𝜔𝑛𝑎 𝜔𝑛𝑎
Using ω/ωna = χ,
1 1
1
𝜒1 = [(4 + 𝑣)2 + √𝜈], 𝜒2 = 1/2[(4 + 𝑣)2 − √𝜈]. (11.105)
2
𝜔2 − 𝜔1 = 𝜔𝑛𝑎 √𝜈. (11.106)