Вы находитесь на странице: 1из 44

Swinburne University of Technology

Sarawak Campus
Faculty of Engineering, Computing and
Sciences

The Study of Rolling Element Bearing Faults using Vibration


Analysis

Bachelor of Engineering
(Mechanical)

Philip Chin Kai Wen

May/ 2017
Abstract
Rolling element bearings are abundant and crucial in many machines, they play an
important role in rotating machines. However, bearing failure is one of the main
reasons that cause the breakdown of rotating machines. In most cases, the cost of
bearings themselves is not significant, but the presence of faults in bearings may
result in serious catastrophic consequences which will lead to costly downtime.
Therefore, it is important to detect and identify the bearing faults in advance to avoid
any unnecessary downtime cost. This project addresses the study of diagnosing
rolling element bearing faults using vibration analysis. In this project, different types
of faults are created artificially onto the rolling element bearings and then put to the
test with Fast Fourier Transform (FFT). Analyzation of the vibration spectrums given
by the bearings is shown in this report.

i
Acknowledgement
First of all, I would like to express my deepest gratitude to everyone who has been
providing me support in completing this report. Special appreciation to Dr. Ha How
Ung who has been patiently guiding, assisting and encouraging me throughout the
time of writing this report.

Other than that I would like to thank the lab assistance, Mr Thomas, who gave me
the permission to use all the required materials and equipment to complete the report.

ii
Declaration
I hereby declare that this report entitled “The Study of Rolling Element Bearing
Faults using Vibration Analysis” is the result of my own project work except for
quotations and citations which have been duly acknowledged. I also declare that it
has not been previously or concurrently submitted by any other degree at Swinburne
University of Technology (Sarawak Campus).

Name: Philip Chin Kai Wen

ID: 7434243

Date: 20 May 2017

iii
Table of Contents
List of Figures ............................................................................................................. vi
List of Tables.............................................................................................................. vii
1. Introduction........................................................................................................... 1
1.1 Background ................................................................................................... 1

1.2 Problem Statement ........................................................................................ 2

1.3 Research Aim ................................................................................................ 2

1.4 Research Objective ........................................................................................ 2

1.5 Hypothesis ..................................................................................................... 3

1.6 Research Significant ...................................................................................... 3

1.7 Research Question ......................................................................................... 3

1.8 Research Scope and Assumptions ................................................................. 4

2 Literature Review ................................................................................................. 5


2.2 Types of Bearing Faults ................................................................................ 5

2.2.1 Wear - Abrasive Contamination............................................................. 5


2.2.2 Wear – Bruising and Pitting ................................................................... 5
2.2.3 Wear - Grooving .................................................................................... 5
2.2.4 Etching Corrosion .................................................................................. 5
2.2.5 Inadequate Lubrication........................................................................... 5
2.2.6 Damaged Bearing Cages ........................................................................ 6
2.2.7 Burns from Electric Current ................................................................... 6
2.2.8 Fatigue Spalling ..................................................................................... 6
2.3 Condition Monitoring and Acoustic Emission Response.............................. 7

2.4 Sources of vibration....................................................................................... 8

2.3.1 Roughness of Surface ............................................................................. 8


2.3.2 Waviness ................................................................................................. 9
2.3.3 Discrete Defects of Bearings ................................................................ 10
2.5 Vibration Analysis on Bearings faults ......................................................... 13

2.6 Bearing Fault Frequency ............................................................................. 15

2.5.1 Ball Pass Frequency Inner Race (BPFI) ............................................... 16

iv
2.5.2 Ball Pass Frequency Outer Race (BPFO) ............................................. 16
2.5.3 Fundamental Train Frequency (FTF) ................................................... 16
2.5.4 Ball Spin Frequency (BSF) ................................................................... 16
2.7 Previous Research ....................................................................................... 17

3 Methodology ....................................................................................................... 19
3.1 Methodology Description ............................................................................ 19

3.2 Flow Chart ................................................................................................... 21

3.3 Apparatus and Materials .............................................................................. 22

3.3.1 Bruel & Kjaer Lan I/F 3560C 4CH 25kHZ Sound Vibration Acoustic
FFT Analyzer ...................................................................................................... 22
3.3.2 Bruel & Kjaer Accelerometer (Type 4508 B 001) ............................... 22
3.3.3 Bruel & Kjaer AO-0038-D-030 Cable.................................................. 23
3.3.4 Single motor test rig (Gunt TM170 Balancing Apparatus) .................. 23
3.3.5 NTN 127 Bearings ................................................................................ 24
3.4 Experimental Setup ..................................................................................... 25

3.4.1 Bearing Preparation .............................................................................. 25


3.4.2 Experimental Platform Configuration .................................................. 26
3.5 Characteristic Defect Frequencies ............................................................... 26

4 Results and Discussion ....................................................................................... 28


4.1 Healthy Bearing ........................................................................................... 28

4.2 Bearing with Cracked Inner Race................................................................ 29

4.3 Bearing with Cracked Outer Race ............................................................... 30

4.4 Bearing with Damaged Ball ....................................................................... 31

5. Conclusion .......................................................................................................... 32
6. Recommendations for Future Work ................................................................... 33
References .................................................................................................................. 34
Gantt Chart ................................................................................................................. 36

v
List of Figures

Figure 1: Percent film vs lambda ratio (Λ) (S.J. Lacey, 2008) .................................... 9

Figure 2: Vibration caused by the wavy raceway (T. Momono et.al, 1999).............. 10

Figure 3: Vibration Signal of a healthy bearing (S.J. Lacey, 2008) .......................... 11

Figure 4: Vibration Signal of a Faulty Bearing (S.J. Lacey, 2008) ........................... 11

Figure 5: The effect of bearing raceway defect on the positions of bearing ball (J. Liu

et.al, 2012).................................................................................................................. 12

Figure 6: Amplitude modulation around the defective area (S.J. Lacey, 2008) ........ 12

Figure 7: Bruel & Ksaer FFT Analyzer ..................................................................... 22

Figure 8: Bruel & Kjaer Accelerometer ..................................................................... 22

Figure 9: Cable connector cable ................................................................................. 23

Figure 10: Single rotor test rig (www.gunt.de) .......................................................... 23

Figure 11: NTN 627 Bearing ..................................................................................... 24

Figure 12: Defect on bearing inner race ..................................................................... 25

Figure 13: Defect on bearing outer race ..................................................................... 25

Figure 14: Spall on bearing ball ................................................................................. 25

Figure 15: Test rig configuration ............................................................................... 26

Figure 16: Location of Accelerometer ....................................................................... 26

Figure 17: Vibration Spectrum of Healthy Bearing ................................................... 28

Figure 18: Vibration Spectrum of Bearing with Cracked Inner Race ........................ 29

Figure 19: Vibration Spectrum of Bearing with Cracked Outer Race ....................... 30

Figure 20: Vibration Spectrum of Bearing with Damaged Ball ................................ 31

vi
List of Tables

Table 1: Dimension of 627 Bearing ........................................................................... 24

Table 2: Characteristic Defect Frequency of 627 Bearing ......................................... 27

vii
1. Introduction
1.1 Background

Rolling element bearings can be considered as the key elements in machinery,


especially in the ones that exhibit rotational motion. However, the failures in them
may damage the whole system to an unmanageable level. A typical rolling element
bearing consists of an outer and inner raceway with a set of rolling elements or balls
located in between, while a cage is installed to keep the balls in place. Rolling
element bearings faults may occur in the raceways, the balls or even the cage, for
example, scratches, cracks, on the surface of a raceway.

Many bearings fail prematurely due to several factors that contribute in bearing
malfunction. It is usually not easy to determine the exact cause but most likely they
are due to the presence of foreign matter such as dirt in bearings, improper mounting,
bearing misalignment, bearing corrosion or improper bearing lubrication. All these
factors bring in issues to machinery like loud noise and vibration.

In today’s world, noise and vibration have become a critical aspect in all types of
machinery and equipment, it is well received that quiet stable running is synonymous
with quality (Lacey 2008). Many bearing manufacturers have utilized vibration
analysis as a method for determining the quality of bearings.

A major advantage of vibration analysis is that it can identify the developing


problems that occur in rolling element bearings at an earlier stage, before the
problems escalate, resulting in unwanted downtime (Dong, W 2009).

Condition monitoring of rolling element bearings can be divided into three main
stages, detection, diagnosis and prognosis (Howard, I.M 1994). Detection is simply
determining that there is a fault or mechanical change that occurs in a bearing.
Diagnosis is to determine the location of bearing faults and the type of faults.
Prognosis on the other hand involves in estimating the remaining life of the faulty
bearings.

In the project, a discussion will be made based on the study of rolling element
bearing faults using vibration analysis which includes the mode of bearing failures,

1
how the bearing faults influence bearing dynamics, the measurement of vibration and
the detection of bearing faults from the vibration characteristics.

1.2 Problem Statement

Rolling element bearings are among the most critical machine components that can
be found in different industries. Throughout the years, bearings have been
undergoing different kinds of improvement either in their design, materials and
lubrication technology, as a result, bearings have proven to be long lasting and
reliable when properly utilised.

However, bearings do fail at some point of a time, and its failure is always associated
with significant damage to machine parts and can be considered as one the main
reasons that causes the breakdown of machines. To avoid this from happening,
condition monitoring philosophy of bearings using vibration analysis method has
been gaining wide acceptance throughout every industry. The problem in condition
of bearing is that bearings are usually located very deep inside a machine and are
quite inaccessible. Besides that, bearings are also not built to be taken apart for visual
inspection, thus making it even harder to identify the bearings’ condition. Hence, an
alternative method is needed for the purpose.

1.3 Research Aim

The aim of this research is to study the faults of rolling element bearings using
vibration analysis.

1.4 Research Objective

1. To apply of vibration monitoring method on rolling element bearings and study


its capabilities.

2
2. To detect and diagnose rolling element bearings faults using vibration analysis
method.

1.5 Hypothesis

1. Vibration amplitude will be higher with faulty bearings compared to healthy


ones.

2. Types of bearing faults can be determined by the amplitude of vibration at a


specific frequency.

1.6 Research Significant

The research is carried out to determine bearing faults using vibration analysis and to
study how the faults in the bearing can affect the vibration spectrum of a bearing.

1.7 Research Question

This research project will focus mainly on the detection of different types of bearings
faults using vibration analysis method. As bearing faults are often the main
contributors to machine failure, therefore investigation of bearing failures is done to
avoid any catastrophic damage to rotating machines. There are several questions that
need to be considered:

1. Is it possible to determine the types of defects of bearing through vibration


analysis?

2. How do we diagnose the faults present in rolling element bearings?

• How do we determine the types of defects? By observing the amplitude of


vibration?

3. Is there any machines or software that is capable of detecting bearings faults?

3
• What machine or software?

• Are the results obtained consistent?

4. What are the factors that might affect the outcome of the experiment?

• The speed of rotation?

• How will different conditions of bearings affect the vibration


characteristics?

1.8 Research Scope and Assumptions

In the research, literature review will be done to study more about the relationship
between the bearing faults and its vibration spectrum.

Vibration analysis will be done on several of NTN 627 bearings. Different faults will
be created artificially to the bearings for the purpose.

Some assumptions are:

1. The shaft where the bearings are fitted to is perfectly balanced


2. All the bearings are perfect and there are no manufacturing defects on the
bearings out of the box.

4
2 Literature Review

2.2 Types of Bearing Faults

2.2.1 Wear - Abrasive Contamination


This mode of damage is created by the presence of foreign particles within the
bearing. Some of the examples of these foreign particles are sand or fine metals
resulted from grinding of gears and chipping of metals. These tiny unwanted
particles regularly enter the bearings through defective bearing seals and may build
up as time goes by. The existence of these abrasive particles may cause the inner
clearance of bearings to increase or in worst case scenario, create misalignments in
bearings and reduce the bearings life.

2.2.2 Wear – Bruising and Pitting


Similar to abrasive contamination, this type of damage is also caused by the presence
of hard foreign elements in the internal of the bearing. However, in this case, the
foreign particles travel around the bearing with the lubrication flow and create dents
and bruises along the surfaces of the inner and outer races of the bearing. Raised
metal and dents caused by these unwanted particles may cause stress and reduce
bearing life.

2.2.3 Wear - Grooving


This type of damage is also caused by the existence of foreign particles in the bearing,
but it involves particles that are greater in size. As large particles are crammed into
the narrow space in bearings, they tend to cut grooves and cause tremendous heavy
wear. This situation may generate improper rolling contact geometry and reduce the
bearings’ lifetime of operation.

2.2.4 Etching Corrosion


This is one of the most common damage types of bearings. This normally happens
when the bearings are not effectively protected from moisture. As water enters the
bearings, the bearings may face the possibility of corrosion damage.

2.2.5 Inadequate Lubrication


Improper lubrication of bearings can generate a variety of damage conditions, either
in appearance or performance. Damages will occur if there in not enough lubricants

5
in the bearings to minimize the friction between the rolling elements and the contact
surfaces during their operation. Therefore, it is crucial to determine the right amount
of lubricants in the bearings, the viscosity, the type and the grade of the lubricants.
There are four types of damages that can be caused by inadequate lubrication:

- Discolouration
o Caused by the insufficient lubricants in the bearings, which lead to
excessive high temperature of bearings due to the presence of friction.
- Scoring and Peeling
o Caused by inadequate amount of lubricant which can results in
immediate alteration in temperature and operating conditions.
- Excessive roller end heat
o Damages at the ends of the rollers with excessively high temperature
because of improper lubricant
- Total bearing lockup
o Change in the bearing’s initial geometry and all of the elements in the
bearings due to localized high heat
o Some of the examples of the damages are destructions of bearing
cages, skewing of rollers and in worst case scenario, jam in the whole
bearing.

2.2.6 Damaged Bearing Cages


Since bearing cages are made up of soft thin materials, they can be damaged
relatively easier compared to the other components of bearing if they are not handled
well. This type of damage can occur if inappropriate tools are used during the
installation process of bearings.

2.2.7 Burns from Electric Current


This normally happens when electricity flows through a bearing. Whenever the
electricity passes through the inner elements of bearing, sparks will be created,
followed by the development of pits within the bearing. As the pits grow larger,
strong vibration may occur, thus generating loud irritating sound. There are few
factors that causes electrical arch in bearings, namely static electricity caused by
extended time of operation, and also current leakage due to poor cable management.

2.2.8 Fatigue Spalling

6
Chipping or spalling in bearings may be caused by the weakening of bearing
materials. Normally the failures of bearing races and rolling elements start as a small
fracture and gradually become more severe until the particles of the metal eventually
flake away. This may lead to the increase of roughness to the surface of the bearings
races and introduce the presence of loosen metal particles within the bearings.

2.3 Condition Monitoring and Acoustic Emission Response


As indicated by A.B. Kufman in 1975, there is innumerable number of techniques
that have been created to screen the state of machines. Indeed, even with simply the
utilization of sight and sound, we can decide the state of a machine effectively.
Despite the fact that there are numerous systems that can be utilized to screen the
state of a machine, vibration checking and investigation is the most all around
acknowledged and broadly utilized strategy for the reason. This strategy is utilized to
gather the vibration information and capture important frequencies that determine
whether or not the machine is in good condition.

A.B Kufman expressed that condition monitoring of machine has proved to reduce
maintenance costs, improve productivity and increase the availability of the machine.
Thus, great amounts of monitoring instruments have emerged throughout the years.
These instruments have the ability to ceaselessly analyse flaws present in the
machines as well as keeping the records of all the data.

There have been various studies conducted to research the Acoustic Emission (AE)
response generated by faulty bearings. AE is the phenomenon by which transient
elastic wave generation in materials under tension or stress. The study led by T.
Yoshioka and T. Fujiwara in 1984 demonstrated that AE parameters, like ringdown
checks and peak amplitude of signal are capable of detecting faults before they
appear in vibration acceleration range and are able to identify the possible area of
where AE was generated during a loaded ball bearing fatigue test. In 1992, T.
Yoshioka proposed a system to spot where the vibration was originating from, and
two years later, the system was improved by implementing two AE sensors in the
system to determine the difference of arrival times for acoustic emission at the
sensors. Acoustic Emission signal has been shown to distinguish small deformities
such as fine scratch on the inner race of a rolling element bearing at low speed

7
(Smith, 1982). In 1990, a study conducted by N. Tandon and B.C. Nakra showed that
Acoustic Emission parameters for instance, peak amplitude and counts are capable of
detecting defect in radially loaded rolling element bearing at low to medium speed.
The peak amplitude and counts are also utilized to inspect the quality of bearings (V.
Bansal et al, 1990).

2.4 Sources of vibration


Complex vibration frequencies are generated as the components in a rolling element
bearing, namely races, rolling elements and cage interact together. Despite the fact
that bearings now are manufactured with strict quality control using profoundly
précised machines under clean environment, the bearings will still possess
imperfections and produce vibration as they operate. With today’s technology, even
though we are able to reduce the amplitude of surface imperfections of a bearing to a
small as nanometres range, vibrations can still exist in the frequency within the entire
audible range which is 20 Hertz to 20 Kilo Hertz (S.J. Lacey, 2008). The intensity of
bearing vibration relies on several factors, namely the impact strength, the geometry
of the bearing and the exact location where vibration is measured.

2.3.1 Roughness of Surface


Rough surface in a bearing can be one of the main contributors in vibration of
bearings. The asperity or harshness of a rough surface has the tendency of breaking
through the layer of lubricants and create the phenomenon of metal-to-metal contact,
thus forming friction between two interacting surfaces. The pattern of the vibration
may be in random sequence caused by small impulses.

As indicated by S.J. Lacey in 2008, the vibrations frequencies brought in by surface


roughness are generally 60 times more than the bearings’ speed of rotation.
Therefore, it normally shows up as resonances.

The degree of surface asperity is measured using lambda ratio (Λ). It is a proportion
of thickness of lubricant to harshness of surface. Lambda ratio can be written as 𝛬𝛬 =
ℎ(𝜎𝜎𝑏𝑏2 + 𝜎𝜎𝑟𝑟2 )0.5.

Whereby, Λ = degree of asperity interaction


h = thickness of lubricant film
8
𝜎𝜎𝑏𝑏 = ball RMS roughness
𝜎𝜎𝑟𝑟 = raceway RMS roughness

Figure 1: Percent film vs lambda ratio (Λ) (S.J. Lacey, 2008)

Figure 1 above shows the relationship between percent film and the lambda ratio. We
are able to see that lambda ratio increases with percent film. As the lambda ratio is
less than 0, then the bearing is likely to fail prematurely.

2.3.2 Waviness
Waviness in bearing can also be one of the factors that cause vibration and noise in
bearings. According to T. Momono et. al in 1999, waviness in bearing can never be
eliminated even with today’s advanced manufacturing technology, it can only be
reduced to a certain degree. However waviness in bearings is regularly considered
negligible. But as the waviness becomes more severe, the vibration caused by it may
bring in more problems. Waviness in bearing has the tendency of providing
frequencies three hundred times more than rotational speed however it is generally
strongest when it is less than sixty times of that rotational speed.

In 1962, Gustafsson conducted a study to test the effect of bearing waviness on the
vibration of bearings. The surface waviness of inner and outer races of a bearing
were considered and he was able to determine the number of peaks on the wavy
surface that created vibration and its frequency by solving the force balance within
the components of the bearings and compute the vibration caused by the waviness.
The figure below shows how waviness and their peaks in bearing surface can cause

9
vibration in bearings. Gustafsson’s analysis in his study has been confirmed by
numerous researchers.

Figure 2: Vibration caused by the wavy raceway (T. Momono et.al, 1999)

Although the waviness of bearing cannot be eliminated completely, the vibration


caused by this case can only be minimized by reducing the waviness of the surfaces
of bearing races. Regardless, extra attention is needed while mounting the bearing
because any geometry imperfection of shaft or housing where bearing is mounted
may somehow affect the waviness of bearing surfaces.

2.3.3 Discrete Defects of Bearings


Unlike surface waviness and surface roughness which occurred primarily due to
imperfection of manufacturing processes, bearings discrete defects can be caused by
poor handling or poor maintenance of bearings. There are a huge variety of damages
categorized in this mode of defect. They normally include scratches and dents on the
races or the presence of pits and tiny particles in the lubricant of the bearing. As
indicated by Lacey , simple measurements of bearing vibration has been adopted by
various manufacturers to detect the defects of bearings, however these
measurements often vary with the dimension and specification of bearings.

10
Figure 3: Vibration Signal of a healthy bearing (S.J. Lacey, 2008)

Figure 4: Vibration Signal of a Faulty Bearing (S.J. Lacey, 2008)

From figures 3 and 4 above, we can see the comparison of vibration signal between a
healthy and a faulty bearing. It was shown that the fault in the bearing has produced a
significantly high peak in its vibration signal. The RMS in faulty bearing is also
several times greater than in healthy bearing.

2.3.3.1 Raceway Defect


Defects in the raceway will generate impulsive force as a ball goes pass through the
defective zone. The amplitude of the impulses will be highest as the ball is exactly
inside the defective spot, but will reduce as soon as it leaves, therefore causing
amplitude modulation. It does not only increase the amplitude of the ball pass
frequency, but also both side of ball pass frequency as well as illustrated in figure 6
below.

11
Figure 5: The effect of bearing raceway defect on the positions of bearing ball (J. Liu et.al, 2012)

Figure 6: Amplitude modulation around the defective area (S.J. Lacey, 2008)

Although the faults in outer and inner races of a bearing are likely to possess the
same behaviour, the amplitude of spectrum caused by inner race defect may seem to
be much smaller than outer race. It could be caused by the reason that the inner
raceway only enters a load zone one time in one rotation of the bearing, and its signal
has to travel through multiple interfaces such as lubricants, rolling elements and
outer races in order to reach the transducer. This can also be one of the reasons why
faults in inner bearing are more difficult to detect (S.J. Lacey 2008).

2.3.3.2 Cage Defect


Bearing cage are often low in mass compared with other elements in a bearing, its
deformity does not have a particular frequency and its vibration will likely be
random and hard to predict. However, as the cage starts to wear, its rotation may
12
become less stable and result in the increasing of sidebands of other fundamental
bearing frequencies (S.J. Lacey 2008). Large impact forces will also be generated
between the cage and the balls while accelerating and decelerating as the clearance of
the cage gets bigger.

2.3.3.3 Rolling Element Defect


Damages on the rolling elements of bearing can produce frequencies two times as
quickly as ball spin frequency and also the fundamental train frequency. This is
because the defect on the rolling element hits both inner and outer raceways of the
bearing. However, this can be difficult to detect as the defects in the rolling elements
will not always strike both of the raceways as the rolling elements are able to spin in
various directions.

2.5 Vibration Analysis on Bearings faults


Vibration analysis is one state of the art method for monitoring rolling element
bearings fault by utilizing vibration information such as waveform, phase and
spectral. It allows bearing faults to be detected with cutting edge precision so that
corrective action can be taken before the machine fails catastrophically and causes
costly downtime. Although there are many ways to detect rolling element bearing
faults, vibration analysis is chosen to be the most practical way. It is because it is less
expensive to execute, reliable and significantly easier to implement compared to
other methods (O.G. Gustafsson et al, 1962). Besides that, the different vibration
patterns can be obtained using the vibration analysis, and from the data and
information gathered, we are able to determine the type bearing faults (D. Dryer et al.
1962).

In 1972, P.A.Boto et al. suggested that good rolling element bearings tend to have a
greater amplitude ratio of low to high frequency signals. Defects such as crack on the
bearing race or spalling can cause the ratio to reduce remarkably. Thus, at the early
phase of bearing failure, high vibration frequency can better provide the sign of
bearing condition. The generation impulsive force due to the flawed bearing surface
can result in vibration which consequently can be monitored and to signify the
presence of defect. Although the defects can be tiny in size, the impulsive force may

13
oscillate and further destroy the components of the machine if the issue is not
managed properly.

In 1982, T. Igarashi et al stated that the presence of defects in rolling element


bearings can cause the vibration level to increase in the high frequency range of
spectrum. This is due to the natural frequencies of the bearing being excited by the
impulsive force caused by the defects in the bearing. According to T. Igarashi et al,
the frequency of faulty bearings usually falls in the low frequency region, which is
less than 500Hz, while the resonance frequency of bearing lies around medium to
high frequency range, somewhere around 10 kHz.

The interaction between the local defect on a bearing element and its mating
components creates an impulsive force that lasts for a very short duration. This
impact results in vibration which can be analysed and examined for the presence of
defect. In 1979, K. Nishio et al stated that there are two techniques to investigate the
mechanism of bearing failure. The first technique is by running a brand new healthy
bearing on a rotating shaft of a working machine until it fails. The vibration signal is
monitored and plotted a graph against time whereby any differences in the graph can
be further analysed for the defects (T.Williams et al, 2001). To execute this, the
failure can be sped up by overloading, over speeding or removing of lubricant in the
bearings (B.V.A. Rao et al, 1986). The second technique is by mimicking defects on
the bearing. In this technique, faults are created intentionally on brand new
undamaged bearing components. This can be done through scratching, laser, carving
or simply through mechanical indentation. The vibration signals of the bearing with
simulated defect can be compared with the results produced good bearing (D. Dryer
et al, 1972). By simulating defects on bearing, it is much quicker to analyse bearing
faults rather than having to wait for the bearing to fail by itself. However, special
tools are needed for simulating the defects on bearing and it can be different as real
life failure.

Typically, it is almost impossible to predict how a bearing fails, however, a study


conducted by Brian Graney and Ken Starry published in 2011 showed that majority
of these failures, 43%, are often associated with poor lubrication. Rollers tend to skid
as soon as the smooth rolling motion of the rollers on the raceways is degraded.
Frictions caused by skidding of rollers can be examined with high-frequency natural

14
bearing resonance indicators (HFNBRIs). HFNBRIs can detect frequencies ranging
from 3 kHz to 50 kHz which are can be both sonic (<20 kHz) and ultrasonic (>20
kHz). Archambault, 2009 stated that shock or friction can produce these frequencies
and HFNBRI method is an effective way in identifying these frequencies and show
early indication of bearing faults.

To determine the reasons of HFNBRIs’ reactions, discrete frequency indicators are


used. Normally, Fast Fourier Transform (FFT) velocity spectrum is unable to detect
bearing faults until they get more severe. In spite of that, an acceleration spectrum
which is partitioned into two bands can effectively track the vibration from different
sources. The first one is used to monitor frequencies produced by bearing faults,
while the second one is to inspect the characteristic of the frequencies produced. By
observing each of these bands, analysts are able to distinguish the severity of bearing
faults the rate bearing wears. (Berry & Robinson, 2001)

In 2011, Brian Graney and Ken Starry stated that conditions of bearings diagnosed
by HFNBRI and discrete frequency indicators can be established with FFT and time-
waveform analysis. FFT spectrum analysis comprises of harmonic families, their
individual origins and magnitude, meanwhile time-waveform analysis includes
identification of pattern, the magnitude and crest factor. Crest factor can be
represented as ratio of peak amplitude to root-mean-square the value of waveform.
Crest factor can surpass the value of 5 during the earliest stage of wear and will
eventually be reduced to 2.5 at the later stage.

2.6 Bearing Fault Frequency


Bearing fault frequencies are harmonics which occurs simultaneously with consistent
repeating impacts between the races and rollers of bearings. There will be no
significant peaks on these frequencies if the surfaces of a bearing are flawless. On the
other hand, if there is any imperfection on the surfaces of a bearing such as pits or
dents, then high amplitude will be expected due to collisions within the internals of
the bearing. The collision will occur repeatedly according to the rotating speed of
bearing.

15
These frequencies will be generated while bearings operate, and they vary with the
geometry of bearings. According to a journal by Baldor Dodge in 2007, there are
four types of frequencies, each of them is related to a particular part of a rolling
element bearing, namely:

2.5.1 Ball Pass Frequency Inner Race (BPFI)


The reoccurrences of rollers that go through a particular point in the inner race of a
bearing. In other words, in one rotation of inner race, how many times will a specific
point in inner race passes through rollers.

𝑁𝑁𝑏𝑏 𝐵𝐵𝑑𝑑
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑜𝑜𝑜𝑜 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 , 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 + 𝑐𝑐𝑜𝑜𝑜𝑜𝑜𝑜)
2 𝑃𝑃𝑑𝑑

2.5.2 Ball Pass Frequency Outer Race (BPFO)


The frequency of specific point in outer race that passes through rollers in one single
revolution of inner race.

𝑁𝑁𝑏𝑏 𝐵𝐵𝑑𝑑
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 , 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 − 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐)
2 𝑃𝑃𝑑𝑑

2.5.3 Fundamental Train Frequency (FTF)


Also called cage frequency, it is the rotational speed of bearing cage and ball
assembly. FTF signifies the presence of redundant looseness in the bearing or crack
in the cage.

𝑓𝑓𝑠𝑠 𝐵𝐵𝑑𝑑
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 , 𝐹𝐹𝐹𝐹𝐹𝐹 = (1 − 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐)
2𝑑𝑑 𝑃𝑃𝑑𝑑

2.5.4 Ball Spin Frequency (BSF)


The number of spins a single roller will make as the inner ring turn for one round.
This is often duplicated for two times, this is because if there is a flaw on the ball of
the bearing, then the flawed area of the ball will touch both internal and external race
of the bearing.

𝑃𝑃𝑑𝑑 𝐵𝐵𝑑𝑑2
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓, 𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 − 𝑐𝑐𝑐𝑐𝑐𝑐 2 𝛷𝛷)
2𝐵𝐵𝑑𝑑 𝑃𝑃𝑑𝑑2

𝑃𝑃𝑑𝑑 = 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝ℎ 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

𝐵𝐵𝑑𝑑 = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

16
𝑁𝑁𝑏𝑏 = 𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛𝑛 𝑜𝑜𝑜𝑜 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

ø = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎

2.7 Previous Research


Vibration analysis is often regarded as one of the most reliable methods of
identifying problems or flaws that occur within rolling element bearing. Different
methods of vibration analysis have been developed over the years to further improve
the reliability of the analysis. Plenty of researchers have hypothetically and
theoretically conducted studies and experiments on the development of faults in
bearings and the methods of locating the faults in bearings.

In 1999, Dr Alexej. V. Barkov expressed that the faults detection techniques in


rolling element bearing by making use of spectral analysis on high frequency
vibration envelope started in the mid 1970’s. By that time, the algorithms for
diagnosis and faults prediction of rolling element bearings were just recently created.
These algorithms have permitted the identification of various kinds of damages and
their depths from one single evaluation. Many years later, the algorithms were then
computerized in 1990, and the computerized investigation systems for bearings
quickly became popularised started to be used in several industries since.

In the year 2002, Nikolaou N.G., et al came out with a technique to present bearing
faults using wavelet packets (WPT). In this method, a time-frequency decomposition
of vibration signals collected from experiment and simulation is analysed, and only
those signal components that are having important information are chosen for further
investigation with the use of PCMIA DAQ Card-1200 data acquisition card.
According to Nikolaou N.G., instead of using other methods, filters or WPT will
provide the added benefits in terms of flexibility and efficient computational
implementation.

In the journal published by V.K.Rai and A.R. Mohanty in 2007, a discussion was
made regarding the utilization of Fourier Transform (FFT) of intrinsic mode
functions in Hilbert-Huang Transform (HHT) to find bearing faults. Conventional
data acquisition card was used for measuring vibration data and they were able to
calculate characteristic defect frequencies (CDFs) analytically. By comparing the

17
readings for brand new flawless bearings and the flawed bearings with inner and
outer races faults, HHT technique was able to provide multiple resolutions in distinct
frequency scales and take the variation of frequencies into consideration. Through
this comparative analysis, V.K.Rai and A.R. Mohanty have demonstrated the
existence of amplitude regulations and were able to get hold of the frequencies of
defect accurately.

Tuncay Karacay et al. in 2009 conducted an experiment by installing two brand new
rolling element bearings onto a test rig and the bearings let to operate nonstop
throughout their entire lifespan under consistent rotating velocity and load. The
development of defects was measured regularly at 15 minutes interval and the
parameters that were collected are the vibrations’ peak-to-peak amplitude and its root
mean square (RMS). Meanwhile, the bearings’ crest factor and kurtosis number were
also determined to predict the condition of the bearings. From the experiment, they
have discovered that the first defect formed in the bearings is at the inner race caused
by the slight defect found in bearing balls. As experiment went on, the defects on
both inner race and balls developed, and finally followed by the outer race.

Meanwhile in the same year, Kankar P.K. et al. have discussed about using response
surface method to diagnose of bearing faults. Unlike what Tuncay Karacay et al.
have mentioned earlier, dispersed damage like waviness of bearing races and internal
radial clearance are put into consideration. According to them, these factors are
important for designing reliable bearings with high capability. They have utilized
Hertzian theory for elastic deformation elastic contact deformation theory and have
drawn to a conclusion that nonlinear dynamic responses are the main reason behind
the excessively huge inner radial clearance of bearings and are often linked to
distributed bearings defects. On top of that, their findings also showed that even a
small variation in ball waviness and internal radial clearance may cause significant
behavioural changes of the bearing systems.

18
3 Methodology

3.1 Methodology Description

Bearings are abundant in many types of machines, especially the ones that exhibit
rotational movements. Bearings reduce friction as component in a machine rotates,
thus providing smoother and quiet spin. Indirectly, bearings also prevent heat
generation caused by friction. However, bearings may be damaged at some point of
the time. Several factors, such as misalignment, rust, excessive loading or the
presence of foreign particles can cause bearings to be faulty. When bearings are
damaged, rotation will be rough, thus creating vibration and unpleasing sound. This
condition will certainly be very unfavourable for machineries as vibration can causes
looseness in other parts of machineries and in a worst case scenario, the entire
machinery may face a catastrophic failure.

Bearings are usually located very deep inside a machinery, thus making it to be very
inaccessible, especially when one has to take apart many different machine
components first, in order to reach the bearing. To make the matter worse, bearings
are normally not designed to be taken apart, therefore, visual inspection method of
diagnosing bearing condition can be literally impractical. One alternative way for
condition monitoring of bearings is through their vibrations while they are spinning.
From the vibration signal of the bearings, it is possible to detect the mode of bearing
failure.

In this research project, vibration analysis was conducted to detect bearing faults
using FFT analysis. Artificial defects were done to bearings at different locations to
mimic real life bearing conditions. Conditions that were created were healthy bearing
(brand new out of the box), defect on the inner race, defect on the outer race and
defect on the ball. For the test, a variable speed single rotor test rig was used as the
apparatus. As the bearings were loaded onto the test rig and set to rotate at a constant
speed of 17Hz, the FFT analyzer was used to collect the vibration signals of the
rotating bearings. The FFT analyser then displayed the vibration signals into the form
of graph, acceleration (m/s2) vs frequency (Hz), this was when the analyzation took
place. From the vibration graph, we were able to determine the location of defects in
bearings. This was done by checking whether there were any amplitude spikes at

19
specific characteristic defect frequencies. This was based on the journal by Baldor
Dodge stated earlier in the literature review section, where it indicated that each
component in the bearing will have their own specific frequencies. Ball spin
frequency of inner race (BPFI) for the inner race, ball spin frequency of outer race
(BPFO) for outer race, Ball Spin Frequency (BSF) and Fundamental Train
Frequency (FTF) for the cage. At any location of the bearing where defect was
present, amplitude spike would show up at their respective frequencies.
Characteristic defect frequencies were obtained through a series of calculation
according to the bearings’ dimension and the rotational speed.

20
3.2 Flow Chart

Selection of Bearings

Creation of Artificial Faults


onto the Bearings
Calculation of
• Healthy characteristic defect
• Cracked frequencies based on the
o Inner race specification of the
o Outer race chosen bearing
o Ball o BSF
o BPFO
o BPFI
Load the bearings o FTF
onto the test rig

Run the bearing


with a constant
speed of 17Hz

Take readings by Analysis of spectrum


using 4CH 25KHZ produced by FFT analyser
sound vibration
Acoustic FFT
analyser

Justification of bearing defects


through the amplitude at
characteristic defect frequencies

Conclusion

21
3.3 Apparatus and Materials

3.3.1 Bruel & Kjaer Lan I/F 3560C 4CH 25kHZ Sound Vibration Acoustic FFT
Analyzer
Bruel and Kjaer FFT analyser is a versatile, noise and vibration analysis system. It
contains input and output channels for microphone and accelerometers and is capable
of performing real-time measurement for signal and system analysis. The FFT
analyser collects vibration signals through accelerometer and displays them in a form
of readable graphs or spectrums. This apparatus is connected to a desktop computer
and can only be accessed through its software, called “PULSE”.

Figure 7: Bruel & Kjaer FFT Analyzer

3.3.2 Bruel & Kjaer Accelerometer (Type 4508 B 001)


This small accelerometer is designed for vibration measurement. While performing
measurements, this device has to be mounted onto the surface of the parts that needs
to be measured. While being very light in weight, it has a high sensitivity of 10mV/g.

Figure 8: Bruel & Kjaer Accelerometer

22
3.3.3 Bruel & Kjaer AO-0038-D-030 Cable
This cable acts as a connector that links the accelerometer and FFT analyser together.
This cable is designed to eliminate mechanically induced noise, thus giving a more
accurate result.

Figure 9: Cable connector cable

3.3.4 Single motor test rig (Gunt TM170 Balancing Apparatus)


The will be the platform where the bearing will be installed to. It is essentially a
balancing apparatus, however it can easily be converted to a simple single rotor test
rig by removing all the imbalance masses. The rig consists of a variable speed
controller, a motor and a long shaft. The shaft is belt driven by the motor. It also has
an LCD screen located at the front of the console that displays the real-time rotation
speed (RPM) of the shaft. The rotation speed of shaft can be tuned using the speed
control knob, located right next to the screen. It is capable of running up to 60Hz.

Figure 10: Single rotor test rig (www.gunt.de)

23
3.3.5 NTN 127 Bearings
Few bearings of this type will be used in this research report. The dimension or
specifications of the bearings given are as follow:

Inner/Bore Diameter 7mm


Outer Diameter 22mm
Bearing Pitch 22mm
Pitch Diameter 14.7mm
Number of Balls 7
Ball Diameter 3.969mm
Table 1: Dimension of 627 Bearing

Figure 11: NTN 627 Bearing

24
3.4 Experimental Setup

3.4.1 Bearing Preparation


Different modes of defects were created artificially to the bearings to simulate
different real bearing conditions:

i. Healthy Bearing (Figure 11)


ii. Inner race Crack (Figure 12)

Figure 12: Defect on bearing inner race

iii. Outer Race Crack (Figure 13)

Figure 13: Defect on bearing outer race

iv. Damage on Bearing ball (Figure 14)

Figure 14: Spall on bearing ball

25
3.4.2 Experimental Platform Configuration
There are two supports for the shaft on the test rig as shown in Figure 15. The shaft is
held in place with two bearings. For the experiment, bearing on one end of the shaft
was replaced with test bearings created earlier while another end was left unaltered.
As the bearings are loaded to a test rig, they were set to rotate at a constant speed of
17HZ. At the same time, an accelerometer is used to pick up the vibration signals of
the test bearings, in this case, it was situated right on top of the test bearing as shown
in Figure 16 for better capturing of vibration signal. The signals are then sent to FFT
analyzer, where vibration spectrums for each of the bearings are produced for
analyzation.

Figure 15: Test rig configuration

Figure 16: Location of Accelerometer

3.5 Characteristic Defect Frequencies


Characteristic defect frequencies are the frequencies that are closely related to the
defects found on bearings. The parameters used are based on the dimensions of

26
bearings, namely the number of balls, pitch diameter of bearings, diameter of bearing
balls and the contact angle.

Calculations were done to determine characteristic defect frequency for NTN 627
bearings. Based on the specifications and dimensions of the bearings shown in
section 3.3.5, the defect frequencies are shown in Table 2 below. Theoretically, as
the bearing was loaded onto the test rig, FFT analyzer will produce a vibration
spectrum. Any faults within the bearing either at the outer ring, inner ring or the ball,
will cause a rise of amplitude at their respective frequencies.

NTN 627 Bearing

Speed of Shaft 17HZ

Ball Spin Frequency


𝑃𝑃𝑑𝑑 𝐵𝐵𝑑𝑑2 29.2HZ
𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 − 𝑐𝑐𝑐𝑐𝑐𝑐 2 𝛷𝛷)
2𝐵𝐵𝑑𝑑 𝑃𝑃𝑑𝑑2

Ball Pass Frequency of Outer Ring


𝑁𝑁𝑏𝑏 𝐵𝐵𝑑𝑑 43.4HZ
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 − 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐)
2 𝑃𝑃𝑑𝑑

Ball Pass Frequency of Inner Ring


𝑁𝑁𝑏𝑏 𝐵𝐵𝑑𝑑 75.6HZ
𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑓𝑓𝑠𝑠 (1 + 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐)
2 𝑃𝑃𝑑𝑑

Fundamental Train Frequency

𝑓𝑓𝑠𝑠 𝐵𝐵𝑑𝑑 6.2HZ


𝐹𝐹𝐹𝐹𝐹𝐹 = (1 − 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐)
2𝑑𝑑 𝑃𝑃𝑑𝑑

Table 2: Characteristic Defect Frequency of 627 Bearing

Where,

Pd = pitch diameter of bearing = 14.7mm

Bd = diameter of bearing ball = 3.969mm

Nb = number of balls = 7

ø = contact angle = 0˚

Fs = inner race speed = 17Hz

27
4 Results and Discussion

4.1 Healthy Bearing

Figure 17: Vibration Spectrum of Healthy Bearing

Figure 17 above shows the vibration spectrum of healthy bearing produced by FFT
analyzer. The graph was set to be in acceleration root mean square (RMS) versus
frequency in HZ. From the figure above, a peak could be observed at 17HZ as the
bearing is set to run at the constant speed of 17HZ. There were some vibrations in the
vibration spectrum, however there was no real peak at characteristic defect
frequencies. Those spikes appeared to be in random pattern and they are non-
harmonic. Those vibration bumps were caused by noise signals, which were not
related to the condition of the bearing. This could be due to the misalignment of
bearings or the vibrations that came from other external sources, such as the motor
from the test rig itself or the surrounding. The accelerometer used could be very
sensitive that it could pick up signals not only from the test bearing but also from
other sources. The purpose of the test was to test the condition of bearing, thus
vibrations that were caused by external factors could be ignored. From the vibration
spectrum, it can be concluded that this is a healthy bearing with no defects.

28
4.2 Bearing with Cracked Inner Race

Figure 18: Vibration Spectrum of Bearing with Cracked Inner Race

Figure 18 shows the vibration of bearing with cracked inner race. From the vibration
spectrum, it can be observed that the vibration is greater than healthy bearing. A peak
can be observed at 17HZ as the bearing was rotating at 17HZ. The crack in inner race
gave an amplitude rise of 0.03m/s2 at ball pass frequency of inner race (BPFI)
calculated earlier which is at 75.6HZ. The harmonic frequency, at 151HZ also
showed a spike. The vibration was repeating, therefore it was concluded that the
crack in bearing’s inner race was causing this type of vibration characteristic. Any
other peak amplitudes besides the characteristic defect frequency and its harmonics,
were irregular and non-repeating, they may be caused by other factors that are not
related to the bearing and its defect.

29
4.3 Bearing with Cracked Outer Race

Figure 19: Vibration Spectrum of Bearing with Cracked Outer Race

Figure 19 shows the spectrum by bearing with damaged outer race. There was an
amplitude of vibration at 17HZ as this was the speed that the bearing was spinning.
Vibration can be seen at 43.3HZ with an amplitude of 0.045m/s2 as shown in the
figure above, which was essentially the ball pass frequency of outer race (BPFO),
indicating that there was a defect on the outer race of the bearing. Harmonics were
also present at 86.6HZ, 130HZ and 173.2HZ respectively. The vibration amplitude at
BPFO is relatively higher than the vibration at BPFI shown in Figure 18. As like
what S.J. Lacey has mentioned in 2008, this might be due to the location of outer
race, which is nearer to the accelerometer. For inner race, the vibration signal had to
travel through multiple layers of bearing components such as lubricants, rolling
elements and outer race, in order to reach the accelerometer, dampening of signal
might happen during the process.

30
4.4 Bearing with Damaged Ball

Figure 20: Vibration Spectrum of Bearing with Damaged Ball

Figure 20 shows the vibration of bearing with defected ball. Similarly, there was an
amplitude rise at 17HZ because of the constant speed that the bearing was rotating.
Vibrations could be seen clearly in Figure 20. Peak amplitude at 0.032m/s2 can be
observed at ball spin frequency (BSF), at 28.9HZ, along with its harmonic due to the
defect at the ball.

31
5. Conclusion

In this paper, the vibration characteristics of different types of bearings are being
compared. FFT analyzer was used to generate the vibration spectrum for bearings
with different conditions, namely healthy bearing, bearing with defected ball, bearing
with cracked inner race and bearing with cracked outer race. From the results
obtained, the vibration of healthy bearing was clearly lower than the defected ones.
For bearings with defects, peak amplitude will show up at the characteristic defect
frequencies, which is calculated according to the dimensions of bearing. FFT
analysis has proven to be an effective way to diagnose rolling element bearing faults
without having to disassemble it. It can be concluded that different types of bearing
defects will affect the vibration characteristic in a certain way:

• Defected outer race - rise in amplitude at ball pass frequency of outer race
(BPFO)
• Defected inner race – rise in amplitude at ball pass frequency of inner race
(BPFI)
• Defected ball – rise in amplitude at ball spin frequency (BSF)

With the successful implementation of this project, the key deliverables that we have
achieved are:

• Determined the types of rolling element bearings faults without actually


disassembling the bearings.
• Without taking apart the bearings, we were able to know whether the bearings
need replacement.
• Identified the developing problems in rolling element bearings earlier and
minimized the risk of machine damages due to faulty bearing.

32
6. Recommendations for Future Work

From the results obtained, noise is present in every situation. Noise is random and
very unpredictable. Although not significant, these random signals can spike to a
higher amplitude in a way that it can even causes confusion while analysing the
vibration graph. Noise cannot be totally eliminated, as the accelerometer is very
sensitive, not only to bearing’s vibration, but also to the test rig and other sources
that cause vibration. However, it can be reduced to some extend through the
following recommendations:

• Isolate the test rig from any other moving mechanisms


• Install dampening material under the feet of test rig to minimize vibration
signal from the surrounding

There are some other recommendations that can be done to further study bearing
faults using vibration analysis. These will benefit us in understanding deeper the
relationship between bearing defects and its vibration characteristics. They are:

• Aside of localized defect, we can create defect around the circumference of


inner and outer ring to mimic a severely worn out bearing
• Instead of creating defect in a form of pit, we can create a hump to the races
of the bearings and test its vibration characteristics

The research is deemed a success, thereby, defects in any other bearing types can
also be tested with the same methodology. Nevertheless, some other techniques
involving simulation technologies such as the utilization of Finite Element Analysis
(FEA) or Matlab can be developed for vibration analysis to save time and effort.

33
References

1) Lacey, S.J 2008, An Overview of Bearing Vibration Analysis, Maintenance &


Asset Management Journal, vol. 23, Maintenanceonline, Surrey, pp.32-42.
2) Dong, W 2009, ‘Rolling element bearing fault detection using an improved
combination of Hilbert and Wavelet transforms’, Journal of Mechanical Science
and Technology, vol. 23, no. 2009, pp. 3292-3301.
3) Howard, I.M 1994, A Review of Rolling Element Bearing Vibration “Detection,
Diagnosis and Prognosis”, Aeronautical and Maritime Research Laboratory
Airframes and engines Division, Melbourne.
4) Nikalaou, N.G & Antoniadis, I.A. 2002, Rolling element fault diagnosis using
wavelet packets, NDT&E International, Elsevier B.V, Vol. 35, pp. 197-205.
5) Alexej, B & Natalia, B 1999, Diagnostics of Gearing and Geared Couplings
Using Envelope Spectrum Methods, VibroAcoustical Systems and Technologies
Inc., Saint-Petersburg, Russia.
6) Rai, V.K. & Mohanty, A.R. 2007, Bearing Faults Diagnosis using FFT of
Intrinsic mode functions in Hilbert-Huang Transform, Mechanical Systems and
Signal Processing, vol.21, pp. 2607-2615.
7) Tuncay, K & Nizami, A 2009, Experimental Diagnostics of Ball Bearings using
Statistical and Spectral Methods, Tribology International , vol. 42, pp. 836 – 843.
8) Kankar, P.K., Harsha, S.P., Pradeep, K & Satish, S 2009, Fault Diagnosis of a
Rotor Bearing System Using Response Surface Method, Journal of Mechanics
A/Solids, Vol. 28, PP. 841-857.
9) NTN Americas n.d., Bearing Frequencies, NTN Americas, Mount Prospect,
booklet.
10) SKF 2009, Self-aligning ball bearings 1205 ENT9, SKF, viewed 6 October 2016,
< http://www.skf.com/au/products/bearings-units-housings/ball-bearings/self-
aligning-ball-bearings/self-aligning-ball-
bearings/index.html?designation=1205%20ETN9>.
11) Ravindra, T, Nilesh, K & Shyam, P 2013, Vibration Analysis of Ball Bearing,
International Journal of Science and Research (IJSR), vol. 4, no. 5, pp. 2655-
2665.
12) Kaufman, A.B. 1984, Measure machinery vibration - it can help you anticipate
and prevent failure, Ird Mechanalysis, pp. 59-62.
13) 'Bearing fault frequencies, what they mean and how to utilize them' 2007, Baldor
Dodge, p. 1-11, viewed 09 October 2016,
<http://www.baldor.com/mvc/DownloadCenter/Files/WP0019>.
14) Graney, B & Starry, K 2011, "Rolling Element Bearing Analysis", Materials
Evaluation, vol. 70, no. 1, pp. 78-85. Dryer, D & Steward, R 1978, Detection of
rolling element bearing damage by ststistical vibration analysis, J Mech Design,
TransASME, vol. 100, pp. 299-235.
15) Smith, J 1982, Vibration monitoring of bearings at low speed, Tribol Int., p. 139-
144.

34
16) Tandon, N & Nakra, B 1990, Defect detection in rolling element bearings by
acoustic emission method, J. Acoustic Emission, vol. 9, pp. 25-88.
17) Gustafsson, O & Tallaron, T 1962, Detection of damage in assembled rolling
element bearing, American Society of Lubricant Engineer, vol. 5, pp. 197-209.
18) Yoshioka, T & Fujiwara, T 1982, A new acoustic emission source locating
system for the study of rolling contact fatigue, Wear, vol. 81, pp 183-186.
19) Yoshioka, T & Fujiwara, T 1984, Application of acoustic emission techique to
detection of rolling element bearing failure, Acoustic emission monitoring and
analysis in manufacturing, ASME, pp. 55-75.
20) Bansal, V Gupta, B Prakash, A & Eshwar, Y 1990, Quality in inspection of
rolling element bearing using emission technique, J. Acoustic Emission, vol. 9,pp
142-146.
21) Boto, P & Fernlund, I 1972, Shock Pulse measurement of bearings, Wear, vol.19,
pp.367-371
22) Igarashi, T & Hamada, H 1982, Studies on the vibration and Sound of defective
rolling Bearings, Bulletin of JSME, vol.25, pp. 994-997.
23) Nishio, K Hoshiya, S & Miyachi, T 1979, An Investigation of the early detection
of defects in ball bearings by the vibration monitoring, ASME Paper 79-DET-45
ASME, New York.
24) Williams, T, Ribadeneria, X & Billingtons, S 2001, Rolling Element bearing
diagnostics is tun-to-failure lifetime testing, Mechanical System and Signaling
Processing, vol.15, pp.979-993.
25) Rao, B Swarnamani, S & Varghese, G 1986, Studies on a test rig to check
defective and spurious ball and roller bearings, Proceedings of the national
conference on Industrial Tribology, Bombay, India, pp 1.1-1.10.
26) Tatsunobu, M & Banda ,N 1999, Sound and Vibration in Rolling Bearings,
Motion & Control, vol.9, pp.29-37.
27) Jing, L Yimin, S Teik C, L 2012, Vibration analysis of ball bearings with a
localized defect applying piecewise response function, Mechanism and Machine
Theory, Elsevier, vol. 56, pp. 156-169.
28) 2009, Timken Bearing Damage Analysis with Lubrication Reference Guide, 1st
ed, Timken, Ohio, pp. 7-25, viewed 17 November 2016,
<http://www.bell.si/pdfji/1.%20LE%C5%BDAJI/3.%20TIMKEN/ORODJA/BD
ABrochure5892.pdf>

35
Gantt Chart
No Task Week
FYRP1 FYRP2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 Topic Registration
2 Literature Review
3 Research and Discussion
4 Research Plan
5 Methodology
6 Design Testing Method
7 Materials Selection
8 Progress Report Preparation
9 Oral Presentation Preparation
10 Oral Presentation
11 Acquiring of Materials
11 Set-up Testing
12 Conduct Experiment
13 Data Analysis
14 Discussion of the Result
15 Research Paper Submission
15 Final Report Documentation
16 Final Presentation Preparation
17 Final Presentation
18 Ongoing Assessment
FYRP 2 Assessment 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1 A1_Research Paper
2 A2_Presentation
3 A3_Workbook Assessment
4 A4_Final Report

36

Вам также может понравиться