ME 411 :

Mechanical Vibrations

Engr. Dr. M. Usama Siddiqui

ME Department
DHA Suffa University

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 1

 Topic 5
Harmonic Analysis

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 2

Objectives
After you have finished this topic you will be able to do the following:

• Obtain a periodic motion by a Fourier Series representation

• Calculate Fourier Series coefficients computationally
• Know the Gibbs Phenomenon
• Understand Fourier Transform
• Distinguish between Fourier Series and Fourier Transform

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 3

Fourier Series Representation
Periodic Function

Amplitude
F t   F t   Time


Fourier series representation
a0
x(t ) 
2
 a1 cos(t )  a 2 cos(2t )  a 3 cos(3t )  ⋯   2

 b1 sin(t )  b2 sin(2t )  b3 sin(3t )  …
Conditions for application:
1. Function is periodic F t   F t  
2. Discontinuities are finite within any period
3. Function has a finite number of maxima and minima during any period
T
4. Function is absolutely integrable over any period  x t  dt  
0

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 4

Fourier Series Representation
• Any periodic function of time can be represented by a Fourier series as an infinite sum of sine and cosine
terms.
• If x(t) is a periodic function with period , its Fourier Series representation is given by

= + cos + cos 2 +⋯
2
+ sin + sin 2 +⋯
(1)
= + ( cos + sin )
2

where ω=2π/τ is the fundamental frequency and a0, a1, a2, …, b1, b2, …, are constant coefficients which
are estimated as:
2 2 2
= = cos( ) = sin( )

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 5

Fourier Series Approximation
http://www.falstad.com/fourier/

One Term

Two Terms

Three Terms

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 6

Fourier Series Approximation
http://www.falstad.com/fourier/

Four Terms

Ten Terms

Twenty Terms

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 7

Gibbs Phenomenon
An anomalous behavior observed from a periodic function that is being represented
by Fourier series.

As n increases, the approximation can be seen to improve

everywhere except in the vicinity of the discontinuity, P.
The error in amplitude remains at approximately 9%, even
when → ∞.

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 8

Fourier Series Approximation – Example 1

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 9

Fourier Series Approximation – Example 2
900

800

700

600 x and y-coordinates vs. fictitious time

900
500 800
700

Coordinate value
600
400 500
400
300 300
200
100
200 0
0 50 100 150 200 250 300

100 Fictitious time

x coordinates y coordinates
0
0 100 200 300 400 500 600 700 800 900

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 10

Fourier Series Approximation – Example 2
2
=
If we set = 1,
2
We can calculate, = cos( )

and 2
= sin( )

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 11

Fourier Series Approximation – Example 2

If we set = 5,

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 12

Fourier Series Approximation – Example 2

If we set = 10,

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 13

Fourier Series Approximation – Example 2
Lets vary n from 1 to 100

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 14

Fourier Series Approximation – Example 2

For = 100,

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 15

Bonus Task
• Replicate the above process for any shape (simple shapes like line, circle, triangle, etc. not
included)

• If you can develop the Fourier Series for a unique shape, you will receive bonus on one quiz
or assignment.

• Uniqueness is key. Same solutions  No bonus

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 16

Numerical Estimation of Fourier Series Coef.’s
Recall that a function ( ) can be represented using a Fourier Series as,

= + cos + cos 2 +⋯
2
+ sin + sin 2 +⋯

= + ( cos + sin )
2

where the coefficients are calculated as

2 2 2
= = cos( ) = sin( )

In order to compute the coefficients numerically, we can

• Perform symbolic or numerical integrations on a computer, or
• Estimate the coefficients using a parameter estimation method

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 17

Parameter Estimation - MATLAB
Consider the following example function:
= 10 sin( ) + 5 sin(5 )

Comparing with the following general form,

= + ( cos + sin )
2 2

We can deduce that,

Example function ( )
=2
=1 /
= 10
=5
= 0 for all
= 0 for all except = 1 and =5

Lets compute , , and using MATLAB.

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 18

 Parameter Estimation - MATLAB

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 19

Frequency Spectrum
• Harmonic functions an cos(nωt) or bn sin(nωt) in
Eq. (1) are called the harmonics of order n of
the periodic function x(t).

• The Harmonic of order n has a period of τ/n.

• Harmonics plotted as vertical lines on a diagram

of amplitude (an and bn) versus frequency (nω),
is called the frequency spectrum.

ω = Fundamental Frequency nω = Harmonics are multiples of the

fundamental frequency (or a first harmonic)

Frequency Spectrum of a typical periodic function of time

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 20

Time vs. Frequency Domain

Initial conditions are lost in frequency

domain. Often considered unnecessary
in many practical applications.

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 21

Time vs. Frequency Domain
Harmonic
Components

Original function Amplitudes of

in time domain Harmonic components

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 22

Fourier Transform
• Periodic Functions  Fourier Series
• Aperiodic Functions  Fourier Transforms
What will be the time period of an aperiodic function?

=∞ for aperiodic functions

• Fourier Series

= + ( cos + sin )
2

= | ℤ

• Fourier Transform

= is the Fourier Transform

• Gives vibration amplitudes at different frequencies
Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 23
Fast Fourier Transform (FFT)
• A set of algorithms that are used to efficiently calculate Fourier transforms of discrete data
• Very general  Work with periodic functions as well
Example: http://www.falstad.com/fourier/
https://www.szynalski.com/tone-generator/
Capture top 4 frequencies
(Note: visually inspect for other peaks
FFT using a cell-phone app FFT Spectrum Analyzer (Install it) as well)

Change Amplitude axis length

Change frequency axis length

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 24

FFT Example 1 - Analyzing single tones

200 Hz Sine Wave

https://szynalski.com/tone#200,v1

200 Hz Square Wave

https://szynalski.com/tone#200,squ,v1

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 25

FFT Example 1 - Analyzing single tones

FFT of Square Wave

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 26

FFT Example 1 - Small Experiment
Use your newly installed app to find the frequencies of the sounds being played.
Vary the frequency axis length to identify the largest peak

Tone 1 Tone 2

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 27

FFT Example 2 - Analyzing signals with noise
• FFT can help separate main signal from noise

• Let’s build a simple waveform with and use an

FFT for vibration analysis.

• This constructed waveform will consist

of three different frequency components: 22
Hz, 60 Hz, and 100 Hz with amplitudes of 1g,
2g, and 1.5g respectively

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 28

FFT Example 3 – Vibrations in an IC Engine

• 4-cylinders 4-strokes engine

• Idling at constant rpm

• Vibration data for a 50 ms

time period
30 Hz = 1800 rpm

15 Hz = 900 rpm

Reference
https://blog.mide.com/vibration-analysis-fft-psd-and-spectrogram
Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 29
FFT Example 3 – Vibrations in an IC Engine

• FFT fails if the frequencies

of the vibrations vary

• In case of our IC engine,

the engine sat off for a
period of time, idled, then
the engine was revved
before letting it idle again
and finally turning it off

Reference
https://blog.mide.com/vibration-analysis-fft-psd-and-spectrogram

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 30

FFT Example 3 – Vibrations in an IC Engine

• In case of varying frequencies, we make a

spectrogram to visualize the variation of
frequencies during operation.

• For our engine example

Reference
https://blog.mide.com/vibration-analysis-fft-psd-and-spectrogram
Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 31
FFT Example 4
Spectrogram example:
https://musiclab.chromeexperiments.com/Spectrogram/

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 32

Complex Data  FFT  Basic Ingredients
• If earthquake vibrations can be separated into "ingredients" (vibrations of different speeds & amplitudes),
buildings can be designed to avoid interacting with the strongest ones.

• If sound waves can be separated into ingredients (bass and treble frequencies), we can boost the parts we
care about, and hide the ones we don't. The crackle of random noise can be removed. Maybe similar "sound
recipes" can be compared (music recognition services compare recipes, not the raw audio clips).

• If computer data can be represented with oscillating patterns, perhaps the least-important ones can be
ignored. This "lossy compression" can drastically shrink file sizes (and why JPEG and MP3 files are much
smaller than raw .bmp or .wav files).

Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 33

 QUESTIONS … ...?
Dr. M. U. Siddiqui ME-411 Mech. Vibrations Slide 34