Fault diagnosis in gears using acoustic sensor

Chapter 1

1.INTRODUCTION

In the industry today, in order to meet consumer needs, continuity of

the production and safety of the processes are of crucial importance. Continuous Fault
Condition monitoring (FCM) can reduce downtime and the total cost of products. In
spite of the fact that the initial capital investment cost is high for setting the FCM
system, monitoring the health of the structures has a great importance, and FCM
requirements in the industry are increasing day by day. Providing a method of early
detection of structural, mechanical, or electrical problems allows operators to predict
where faults occurred and suggest an approximate time that the system will break
down. Unexpected machine failure causes both hitches of the production schedule and
increases in cost with financial loss. For this reason, Predictive Maintenance methods
are very important. One of them is the Acoustic Emission (AE) technique. It requires
a contactless measurement, which is an advantage for mechanical systems and their
parts. Any abnormalities or defects in the machinery or equipment must be detected
and analyzed at an early stage to avoid major problems. Therefore, FCM of rotating
machinery, such as at gears and bearings, has a crucial role in the industry, as it keeps
the system in a healthy condition for maximum productivity, while detecting and
diagnosing faults at an early stage. As a result, it is possible to prevent serious
problems, damages, and more cost. It is difficult to diagnose possible gear faults such
as micro cracks and pitting because of their minimal effect on the system, and these
kinds of faults can lead to a rapid destruction of the teeth in meshing gears. Micro-
cracks and pitting in gear teeth can be a catastrophic problem resulting in tooth
breakage while the system is running. Nowadays, fault diagnosis techniques for
rotating machinery, such as vibration monitoring and temperature detection analysis,
need a measurement technique with contact. Although vibration monitoring
techniques are quite common for detecting and diagnosing faulty conditions of
rotating machinery, the AE technique provides early detection of faults. In the last
two decades, a new method for early fault diagnosis, the AE technique, has been
researched and gained increasing attention. AE is defined as a matter that results in
the generation of structure- or fluid-borne waves due to the rapid release of energy

RIT, Islampur Page1

 Fault diagnosis in gears using acoustic sensor

from localized sources within or on the surface of a well-documented in recent

research, and investigators have observed some advantages over classical monitoring
techniques.

AE was also found to be more sensitive to surface damage than other fault
condition monitoring techniques. In spite of the common use of vibration monitoring
method in rotating machinery, up to now, classical vibration monitoring analysis
techniques have not been enough for detecting incipient failures such as pitting,
micro-cracks, and surface wear in gears. Machine conditions and faults such as gear
failures, unbalance, shaft misalignment, and bearing defects can be monitored by
using acoustic emission signals at machines operating conditions. These signals can
be used not only for growing failures, but also for incipient failures of the rotating
machineries, to reduce or eliminate catastrophic damages. In this study, the AE
technique was used to analyze and evaluate the gear defects on spur gears.

1.1 Problem Statement

Unexpected machine failure causes both hitches of the production schedule

and increases in cost with financial loss. For this reason, Predictive
Maintenance methods are very important. One of them is the Acoustic
Emission (AE) technique. The purpose of this mini project is to detecting and
diagnosing faults at an early stage. As a result, it is possible to prevent serious
problems, damages, and more cost.

1.2 Objective

• Test Carried out for early detection of fault in rotating machine elements.
• To reduce catastrophic damages.
• To ensure continuity of the production by analysing system.
• To ensure safety of the processes.

RIT, Islampur Page2

 Fault diagnosis in gears using acoustic sensor

1.3 Theory

• Fast Fourier Transform:

FFT analyzer is the fast Fourier Transform itself. The fast Fourier Transform,
FFT uses the same basic principles as the Fourier transform, developed by Joseph
Fourier (1768 - 1830) in which one value in, say, the continuous time domain is
converted into the continuous frequency domain, including both magnitude and phase
information. However to capture a waveform digitally, this must be achieved using
discrete values, both in terms of the values of samples taken, and the time intervals at
which they are taken. As the time domain waveform is taken at time intervals, it is not
possible for the data to be converted into the frequency domain using the standard
Fourier transform. Instead a variant of the Fourier transform known as the Discrete
Fourier Transform, DFT must be used.

As the DFT uses discrete samples for the time domain waveform, this
reflects into the frequency domain and results in the frequency domain being split into
discrete frequency components of bins. The number of frequency bins over a
frequency band is the frequency resolution. To achieve greater resolution, a greater
number of bins is needed, and hence in the time domain a large number of samples is
required. As can be imagined, this results in a much greater level of computation, and
therefore methods of reducing the amount of computation required is needed to ensure
that the results are displayed in a timely fashion, although with today's vastly
increased level of processing power, this is less of a problem. To ease the processing
required, a Fast Fourier Transform, FFT is used. This requires that the time domain
waveform has a the number of samples equal to a number which is an integral power
of two.

RIT, Islampur Page3

 Fault diagnosis in gears using acoustic sensor

• FFT spectrum analyzer:

The block diagram and topology of an FFT analyzer are different to that of the
more usual super heterodyne or sweep spectrum analyzer. In particular circuitry is
required to enable the digital to analogue conversion to be made, and then for
processing the signal as a Fast Fourier Transform.

The FFT spectrum analyzer can be considered to comprise of a number of different

blocks:

Fig1:FFT Spectrum Analyzer Block Diagram

• Analogue front end attenuators / gain: The test instrument requires

attenuators of gain stages to ensure that the signal is at the right level for the
analogue to digital conversion. If the signal level is too high, then clipping and
distortion will occur, too low and the resolution of the ADC and noise become
a problems. Matching the signal level to the ADC range ensures the optimum
performance and maximises the resolution of the ADC.
• Analogue low pass anti-aliasing filter: The signal is passed through an anti-
aliasing filter. This is required because the rate at which points are taken by
the sampling system within the FFT analyzer is particularly important. The
waveform must be sampled at a sufficiently high rate. According to the
Nyquist theorem a signal must be sampled at a rate equal to twice that of the
highest frequency, and also any component whose frequency is higher than the
Nyquist rate will appear in the measurement as a lower frequency component -

RIT, Islampur Page4

 Fault diagnosis in gears using acoustic sensor

a factor known as "aliasing". This results from the where the actual values of
the higher rate fall when the samples are taken. To avoid aliasing a low pass
filter is placed ahead of the sampler to remove any unwanted high frequency
elements. This filter must have a cut-off frequency which is less than half the
sampling rate, although typically to provide some margin, the low pass filter
cut-off frequency is at highest 2.5 times less than the sampling rate of the
analyzer. In turn this determines the maximum frequency of operation of the
overall FFT spectrum analyzer.
• Sampling and analogue to digital conversion: In order to perform the
analogue to digital conversion, two elements are required. The first is a
sampler which takes samples at discrete time intervals - the sampling rate. The
importance of this rate has been discussed above. The samples are then passed
to an analogue to digital converter which produces the digital format for the
samples that is required for the FFT analysis.
• FFT analyzer: With the data from the sampler, which is in the time domain,
this is then converted into the frequency domain by the FFT analyzer. This is
then able to further process the data using digital signal processing techniques
to analyze the data in the format required.
• Display: With the power of processing it is possible to present the
information for display in a variety of ways. Today's displays are very flexible
and enable the information to be presented in formats that are easy to
comprehend and reveal a variety of facets of the signal. The display elements
of the FFT spectrum analyzer are therefore very important so that the
information captured and processed can be suitably presented for the user.
• In recent years a form of spectrum analyzer, termed a real-time spectrum
analyzer, RSA has grown in popularity.
• These real-time spectrum analyzers are particularly useful in looking at
waveforms where changes may be seen, and need to be captured. Often
spectrum analyzers that take time to process the waveforms may miss spurious
signals and these can be particularly important when testing for compliance
and out-of-band signals.
• As the name implies, real-time spectrum analyzers operate in real time.

RIT, Islampur Page5

 Fault diagnosis in gears using acoustic sensor

Chapter 2

2. LITERATURE REVIEW

Tan and Mba [1] investigated the relations between the surface roughness,
contact resistance, and AE levels of spur gears under partial elasto-
hydrodynamic lubrication. These observations showed that rotation speed was
more efficient than loading conditions on AE levels. They emphasized the
surface roughness, rotation speed, and oil temperature for AE measurements.
Energy levels of the acquired signal in the analyzed time-domain and the
Energy Index (EI) for statistical measurements were noted.

Al-Balushi and Samanta’s [2] They studied one helicopter gearbox by using
the AE technique, and comparisons between AE and vibration measurements
were implemented. They indicated that AE was more sensitive on early fault
diagnosis than the vibration method. In their work, spur gears were used, and
relations between oil film thickness, temperature, and AE activity were
analyzed. It was shown that AE activity was affected by surface asperity
contacts more than loading condition.

Belsak and Flasker [3] investigated the crack at the tooth root of the gear,
which was a common fault for gearboxes, and they used time- and frequency-
domain for damaged and defect-free conditions on crack propagation.
investigated the crack size and sideband occurrence by comparing defect-free
conditions by using short time frequency spectrum analysis. Fault diagnosis on
a single-stage gearbox test rig with only spur gears, similar to vehicle
gearboxes, was studied by using FFT and RMS for obtaining the fault
characteristics of the gear. It was proved that the early fault detection on crack
propagation was possible by AE.

RIT, Islampur Page6

 Fault diagnosis in gears using acoustic sensor

Combet, et al.[4] studied two different waveform methods to obtain local and
instantaneous averages of the signals. Local faults were recognized, and they
found a diagnosing method dependent on the angle at a special frequency
range. AE results for spur gears were more sensitive than those for helical
gears. Surface roughness and contact length was also important for the acoustic
characteristics of gears.

Husin, et al.[5 ] investigated the online fault condition monitoring techniques

and noted that AE was a complimentary tool for proactive maintenance.

Loutas, et al.[6] They employed some experimental research on fault diagnosis

by using the AE technique. They generated some artificial defects and
obtained satisfactory results from AE measurements.

2.1 Literature Gap

Most of the researchers describe the relationship among surface roughness,
contact resistance, acoustic emmision technique and vibration methods. In the
literature review show that importance of acoustic emmision over conventional
vibration methods. In this project the experimental work in laboratory using
gear box test system by focusing on the fault diagnosis of gears by using
acoustic emission technique will carried out.

RIT, Islampur Page7

 Fault diagnosis in gears using acoustic sensor

Chapter 3

3 PROCESS PLAN AND ACTION PLAN

3.1 The process below illustrate the planning for completion of the project. we
should complete the expected work in the following manner.

Phase-1

This phase consists reading of the literature of present theories and practices regarding
Fault diagnosis by referring reputed journals from reputed publishers like IEEE,
Elsevier, Science Direct, ASME and previously published master and Doctoral thesis.
This will help to build theoretical background regarding fault analysis.

Phase-2

In this phase, Design and development of gear box test rig is carried out. selection of
various parameters like motor rating, three-phase frequency inverter, Delta VFD-E,
were carried out.

Phase-3

In this phase, Manufacturing and fabrication of all parts are carried out. assembly of
all parts are made ready for test.

Phase-4

In this phase, experimental measurements, AE data was recorded after the following
processes: (i) the gearbox system was run for about 2 hours to satisfy a working
condition stability of the system, and (ii) an acoustic emission sensor was located to
collect the signal for data processing. and the output of the signals sent to the
frequency analyzer to get the character of the noise. For all that, the test environment
and compatibility of the devices were checked.

RIT, Islampur Page8

 Fault diagnosis in gears using acoustic sensor

3.2 Expected Outcome

Following are the expected outcomes:

1. To get the proper working results of acoustic sensors in gearbox without any
external exigencies and difficulties.
2. To get the proper results and expected graphs for acoustic emission sensors.
3. Due to gearbox vibrations there should not be damage of acoustic sensors.
4. Acoustic sensors should provide proper results for gears even working under
lubrication.

3.3 Available Facilities

1. Central Workshop
2. Digital Library
3. Acoustic sensors

3.4 Work Plan

Jan. 2019 Feb. 2019 Mar. 2019 Apr. 2019

Month
Work
Activity
Selection of
Topic

Literature
Survey

Concept Design

Manufacturing
of experimental
setup and
analyse

Report
Writing and
Submission

RIT, Islampur Page9

 Fault diagnosis in gears using acoustic sensor

Chapter 4

4. ACOUSTIC EMMISSION ANALYSIS

Acoustic emission is defined as the generation of transient elastic waves by
the rapid release of energy within or on the surface of a material. According to Kaiser,
crystalline solids could propagate sound under a mechanical load. Kaiser used high
frequency sensors and electrical amplifiers to hear sounds in a wide range of materials
under loading conditions. AE sensors are one of the most important tools for
obtaining acoustic sound, and they are usually made of piezocrystals. AE sensors
detect mechanical movements or stress waves, and convert them into an electrical
signal. The main advantage of AE is its high sensitivity. The AE method is the most
sensitive monitoring technique for early fault detection in rotating machine elements,
especially in today’s industry. AE sensors can detect sound beyond the human hearing
frequency range, between 100 kHz to 1 MHz AE detection may not be enough for
fault condition monitoring itself. The important thing with the application of the AE
technique is the attenuation of the signal, and the AE sensor has to be closet source of
the noise generation. Therefore, in this study, the AE sensor was placed close to the
source of the noise generation. Signals obtained from the AE sensor are transferred to
multi-channel pulse analyzer and data acquisition system, passing through the pre-
amplifier, which has a range 40to 60db.The traditional techniques for damage
detection are based on some statistical parameters of the vibration or acoustic energy.

For defect-free conditions of gears, experiment indicated that the regular

meshing component of the signal is dominated by the mesh frequency and its
harmonics. It is also observed that the sidebands about the meshing components at the
rotational frequency of the gear. Signal processing techniques (such as time and
frequency analysis) and some statistical-mathematical parameters (such as RMS, CF,
and average fault) were used to evaluate the faulty conditions of a material. Sound
pressure generally can be expressed as the effective sound pressure for a given period
of time.

RIT, Islampur Page10

 Fault diagnosis in gears using acoustic sensor

Chapter 5

5. DESIGN OF EXPERIMENTAL SETUP

5.1 Proposed Model

The model is designed as per the requirement of test. the gear specifications
along with their alignment with bearings are designed by using creo parametric
software. whole arrangement of gears and bearings are placed over the bed to avoid
vibrations due to rotating parts. motor is also aligned with one shaft , where we want
to provide input and proper meshing of gears are done as shown in picture1.

Picture 1:Model for experimental measurement.

5.2 Gear specification

SPUR GEAR SPECIFICATION

Parameters Pinion Driven Pinion Driven
No. of Teeth 67 67
Module 2 2
Shaft Angle 90° 90°
Addendum 3mm 3mm
Dedendum 3.5mm 3.5mm
Face width 12 mm 12 mm
Pitch Circle Diameter 114mm 114mm
Table 1: Gear Specification

RIT, Islampur Page11

 Fault diagnosis in gears using acoustic sensor

Chapter 6

6. EXPERIMENTAL SETUP

In order to observe and analyze AE in gears, a gearbox test rig was designed.
Experiments were performed on a single stage gearbox test rig

1. The gearbox consists of various gears containing first setup of healthy

gears combinations and another setup of faulty gears combinations.

2. The gears are driven by an electric motor for torque energy.

3. An Acoustic emission sensors or FFT and preamplifier also a digital

display with wavelength change detection.

4. Variation in the signals is analyzed in comparison with healthy gear pair

and unhealthy gear pair.

5. The collected data is being used to predict perspective faults in the

gearbox, with respect to condition monitoring of Gear Box.

Picture2:Gear Box Test Setup

RIT, Islampur Page12

 Fault diagnosis in gears using acoustic sensor

In order to observe and analyze AE in gears, a gearbox test rig was designed.
Experiments were performed on a single stage gearbox test rig that is, as a vehicles
transmission system. A variable speed electric motor was used to test gears. A three-
phase frequency inverter, Delta VFD-E, was used for the case of different working
speeds. A pinion gear has 67 teeth and the driven gear has also 67 teeth. the meshing
frequencies for both pinion and driven gears are 447 Hz, 558 Hz and 670 Hz for 400
rpm, 500 rpm, and 600 rpm rotation speeds respectively. In the experimental
measurements, AE data was recorded after the following processes: (i) the gearbox
system was run for about 2 hours to satisfy a working condition stability of the
system, and (ii) an acoustic emission sensor was located to collect the signal for data
processing.

A tool for acoustic measuring microphone with a pre-amplifier, were used to obtain
and analyze the results. An analysis program was able to evaluate the time- and
frequency-domain analysis of the obtained signals from the gearbox test rig. The
analyzer is able to give necessary information about AE signals by measuring Sound
Pressure Level (SPL), which is a logarithmic measure of the effective sound pressure
of a sound relative to a reference value. It is measured in decibels (dB) above a
standard reference level. The commonly used ”zero” reference sound pressure in air is
20 Pa. Sound pressure produces a signal in the microphone of the sound level meter,
and the output of the signal is sent to the frequency analyzer to get the character of the
noise. For all that, before all of the measurements, the test environment and
compatibility of the devices were checked.

The AE signal was amplified and filtered using 20/40/60 dB, AE

PREAMP/100 1200 KHz with a band pass filter (BPF). The PAC AE5A amplifier is a
high performance AE system that amplifies and filters received AE signals from the
preamplifier. Both gears having module of 3 mm. Detailed information for the gears
are given in Table 1. A sliding mechanism was added to the system for setting the
different distances between the shaft centers. The test rig used to measure acoustic
signals is shown in Picture 2.

RIT, Islampur Page13

 Fault diagnosis in gears using acoustic sensor

Picture 3:Well Condition Gear

Picture 4:Faulty Gear

Those test will carried out by using well conditioned gears as shown in
picture 3at various rpm stated as 400,500 and 600 rpm. similarly the same process
repeated with same rpm 400,500 and 600 rpm for faulty gear as shown in picture 4.
Test will carried out. Result obtained from both process will compared and we will
get intimation for conditions of gears. Data available by this two steps explained in
next chapter.

RIT, Islampur Page14

 Fault diagnosis in gears using acoustic sensor

Chapter 7

7.RESULT AND DISCUSSION

Signal processing techniques (such as time and frequency analysis) and
some statistical-mathematical parameters (such as RMS, CF, and average fault) were
used to evaluate the faulty conditions of a material. Sound pressure generally can be
expressed as the effective sound pressure for a given period of time. The effective
value of sound pressure is the RMS value of the instantaneous sound pressure taken at
a point over a period of time as

where P(t) is instantaneous sound pressure, and T is the average time interval. The CF
is defined as the maximum peak value of the sound pressure signal x, over a period of
time (T) divided by the RMS value of sound pressure, and is expressed

where PRMS is the RMS value of the sound pressure, and x peak is the maximum
peak value of the signal. CF values are usable to understand the presence of a small
amount of maximum local defects over a signal.
AE peaks were due to the cracked tooth root that appeared in the meshing
harmonics of the gear, and sidebands were also observed near the maximum fault
frequency. As shown in fig. 2 and fig 3, at 400 rpm, the crest factor for the condition
is 22.07. When the gears are in an undamaged state, the crest factor values were
relatively lower than those of gears in a faulty state because of the short peaks of the
undamaged state of the gear. As seen in the results, very high CF values resulted from
high peaks. Results for 500 rpm rotation speed are shown in fig 4 and fig 5. The
obtained SPL of the gear was higher than that of 400 rpm running speed’s results.
Sidebands observed near the meshing harmonics related to gear tooth cracks were
observed. CF were obtained as 23.01. fig 6 and fig 7 gives the results of a 600 rpm
rotation speed. As outlined in this figure, SPL of the gear at 600 rpm was higher than
that of 400 and 500 rpm where CF value is 24.4.

RIT, Islampur Page15

 Fault diagnosis in gears using acoustic sensor

Autospectrum(Mic) (Real) \ FFT Cursor values

[dB/20u Pa] X: 2.350k Hz
Y: 72.239 dB/20u P

80

70

60

50

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz]

Figure 2: Faulty gear under 400 rpm

Autospectrum(Mic) (Real) \ FFT Cursor values

[dB/20u Pa] X: 700.000 Hz
Y: 77.327 dB/20u Pa

80

70

60

50

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz]

Figure 3: Good Gear under 400 rpm

RIT, Islampur Page16

 Fault diagnosis in gears using acoustic sensor

Autospectrum(Mic) (Real) \ FFT Cursor values

[dB/20u Pa] X: 550.000 Hz
Y: 71.811 dB/20u P

80

70

60

50

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz]

Figure 4:Good Gear under 500 rpm

Autospectrum(Mic) (Real) \ FFT Cursor values

[dB/20u Pa] X: 550.000 Hz
Y: 76.912 dB/20u P

80

70

60

50

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz]

Figure 5: Faulty gear under 500 rpm

RIT, Islampur Page17

 Fault diagnosis in gears using acoustic sensor

Autospectrum(Mic)1 (Real) \ FFT Cursor values

[dB/20u Pa] X: 5.125k Hz
Y: 63.832 dB/20u P
80

70

60

50

40

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz] 

Figure 6:Good Gear under 600 rpm

Autospectrum(Mic) (Real) \ FFT Cursor values

[dB/20u Pa] X: 5.125k Hz
Y: 68.504 dB/20u P

80

70

60

50

0 1k 2k 3k 4k 5k 6k 7k 8k 9k 10k
[Hz]

Figure 7:Faulty gear under 600 rpm

RIT, Islampur Page18

 Fault diagnosis in gears using acoustic sensor

Chapter 8

8.CONCLUSION

In this study, AE results obtained from the gearbox test system were
analysed under different running speeds for spur gears. AE data was taken in time
domain and converted to frequency domain by using FFT in order to clearly
understand the occurrence of the fault location and its source. Measurements were
implemented by using an acoustic sensor near the meshing points of the gears without
any contact to the test system. The artificially created gear damages, such as pitting
and cracks . were apparent in the AE graphs. At the meshing frequencies, AE peaks
indicate the normal condition of the gear. AE peaks observed near the meshing
harmonics, which are called sidebands, indicate the gear tooth damage, such as cracks
and pitting faults.CF was measured and indicated the damage with its high value in all
cases. CF shows the damage severity .The increasing peak level/RMS ratio indicates a
bigger CF value, which shows the severity of the local fault. Corresponding to the
running speed of the gear, in general cases, an increasing speed increases the SPL of
the signal. Gear defects, loading, and running speed increasingly affect the maximum
and average values of the signal. All these statistical measurements confirm the
existence of a fault in the gear system by comparing their reference value.

RIT, Islampur Page19

 Fault diagnosis in gears using acoustic sensor

Chapter 9

9.REFERENCES

[1] Loutas, et all, 2006, "Diagnosis of artificial gear defects on single stage gearbox
using acoustic emission" , Advanced Material Research, 13–14, 415–420.

[2] Husin, S., Mba, D., and Raja Hamzah, 2010, "R. I. Viability of application of the
acoustic emission (AE) technology for the process and management of maintenance
in industries: defect detection, on-line condition monitoring, diagnostic and
prognostic tool", Proceedings of the International Multi Conference of Engineers and
Computer Scientists, 3, 978–988.

[3] Al-Balushi, K. R. and Samanta, B. ,2002, " Gear fault diagnosis using energy-
based features of acoustic emission signals", Proceedings of the Institution of
Mechanical Engineers, PartI: Journal of Systems and Control Engineering, 216–249.

[4] Baydar, N. and Ball, A.,2001, " A comparative study of acoustic emission and
vibration signals in detection of gear failures using Wigner Ville distribution
", Mechanical Systems and Signal Processing, 15 (6), 1091–1107.

[5] Combet, F., Gelman, L., and La Payne, G.,2012, " Novel detection of local tooth
damage in gears by the wavelet bicoherence ", Mechanical Systems and Signal
Processing, 26, 218–228.

RIT, Islampur Page20

 Fault diagnosis in gears using acoustic sensor

RIT, Islampur Page21