48

Chapter 3

Experimentation and Data Acquisition

In order to achieve the objectives set by the present investigation as

mentioned in the Section 2.5, an experimental set-up has been

fabricated by mounting the engine, gearbox, sensing and analyzing

equipment on a steel frame. The schematic diagram of set-up is

shown in Figure 3.1. It is proposed to induce a defect on a particular

tooth of a chosen gear in the gearbox and generate the vibrations. To

sense vibration signal generated by the gearbox, an accelerometer has

been used. In order to process the vibration signal sensed by the

accelerometer an FFT analyzer has been selected. The details of

gearbox, sensing and processing equipment has been presented in the

subsequent sections.

FFT Analyzer
Accelerometer
Gear Box

Faults

Figure 3.1 Schematic Diagram of Set-up

 49

3.1 Gearbox

A standard transmission gearbox of two wheeler with engine was

purchased and modified as test gearbox. It has four gear ratios to

drive the counter shaft at four different speeds as shown in Figure 3.2

by engaging the output shaft with the input shaft with different gear

pairs such as C-D, E-F, G-H, I-J. The gearbox was driven by a 2-

stroke petrol engine and no load was applied on the output shaft of

the gearbox.

Primary
Engine Drive

Clutch
A
Input shaft

E B
C I
Output shaft G

J
D F H

A to J – Gear wheels

Counter shaft sprocket

Number of teeth on E:F=29:60

Figure 3.2 Schematic Diagram of Gearbox

 50

The engine assembly with gearbox was welded to a steel frame

for the convenience of operation is shown in Figure 3.3. Further the

frame was kept on the thermo-coal sheets of 10 mm on the flat floor,

to dampen the vibrations of gearbox, and simultaneously prevents the

transmission of other ground vibrations to the present investigating

gearbox.

Figure 3.3 Fabricated Test Gearbox

Among the four pairs of gears C-D, E-F, G-H, I-J having speed

ratios of 2.97:1, 2.07:1, 1.43:1, 1:1 respectively, the gear pair E-F was

selected for the investigation. The front portion of gearbox was cut

partially as shown in Figure 3.4 for the purpose of inducing defects on

one of the gear tooth. While doing so all the care has been taken to

avoid disintegration of the assembly. Case hardened gear pair with

 51

involute profile having 200 pressure angle with a speed ratio of 2.07:1

and a tooth ratio 29:60 has been selected for experimentation.

Figure 3.4 Gearbox with Cut Portion

3.2 Induced Gear Defects

The vibration excitation in gears is mainly due to errors in the gears.

The source of excitation which causes impacts are manufacturing

errors and advancing local faults. So the vibration signal from the

gearboxes due to impact excitation has non-stationary characteristics.

In some cases an individual gear tooth may be weaker than others on

the same gear due to bending fatigue, shock loading or an internal

void. Most gear engineers consider that a worn gear tooth is one that
 52

has a layer of metal more or less uniformly removed from the surface.

Tooth breakage is a fatigue failure and its occurrence increases when

wear exists. These gear defects of wear and broken tooth alter

significantly the tooth geometry and leads to different degrees of

deflection of a faulty tooth. Both factors of tooth deflection and

geometrical variation contribute to gear vibration.

On gear F, the gear defect of wear was progressively induced

as represented in Figure 3.5 in four stages by grinding 0% to 80% on

one side of the selected tooth of a non faulty gear having 3.5 mm

thickness at the pitch circle circumference as explained in Table 3.1.

The non-faulty gear is represented as Case1, 20% wear as Case2, 40%

wear as Case3, 80% wear as Case4 for ease of referring. In addition to

inducing defect of wear, the present investigation has broke the wear

induced tooth as a gear defect which is referred as Case5.

t/2

Figure 3.5 Cases of Tooth Wear

t- tooth thickness, tv-tooth thickness after wear

 53

Table 3.1 Different Cases

Case No. Induced fault Tooth Thickness

in Gear tooth t- tooth thickness,
tv-tooth thickness after wear
Case1 No fault gear t/2+(tv)/2=3.5 mm

Case2 20% wear tooth t/2+(tx0.8)/2= 3.15 mm

Case3 40% wear tooth t/2+(tx0.6)/2= 2.8 mm

Case4 80% wear tooth t/2+(tx0.2)/2= 2.1 mm

Case5 Broken tooth 0

3.3 Accelerometer

A piezoelectric accelerometer was preferred as it is widely accepted

vibration sensor for condition monitoring of mechanical machinery.

Generally in this type of accelerometers the magnitude of output

voltage signal is low and rapidly dissipates. Further current and

voltage are substantially out of phase. To adjust these conditions,

preamplifiers are required. By building integration networks into the

preamplifier, either velocity or displacement can be measured from the

accelerometer signal. In view of this, an Integrated Circuit Preamplifier

(ICP) type PCB Piezotronics make model 601A01 accelerometer was

selected which is shown in Figure 3.6 has a built in amplifier powered

by a DC polarization of the signal. Hence no extra amplifier and wiring

was needed for amplification purpose.

The weight of the accelerometer was 80 gm which was less than

1/10th of the test gearbox weight as per requirement. The sensitivity of

 54

the accelerometer was 0.2 mV/(m/s) or 10.2 mV/(m/s2) which was

quite good for our application. It has a measurement range ±24745

m/s or ±490 m/s2 which was less than our operating range. It is

having and a frequency range of 0.03-10000 Hz which can cover all

the part frequencies of test gearbox. Its non-linearity was ±1% which

gives accurate readings for 99% of its range.

Figure 3.6 Accelerometer

The output of the accelerometer signal is in terms of

displacement, velocity and acceleration of the vibrating body.

Generally displacement measurements give large value outputs at low

frequencies, acceleration measurements give large value outputs at

high frequencies and velocity measurements give reasonably large

values at low and high frequencies. So velocity is selected as vibration

parameter for measurement purpose using the accelerometer.

 55

3.4 Fast Fourier Transform (FFT) Analyzer

The accelerometer was connected to PHOTON a Portable Dynamic

Signal Analyzer manufactured by LDS-DACTRON to acquire the

vibration signals which is shown in Figure 3.7. It is USB powered,

portable having a weight of 227 gm, easy to install and fast to

connect. It has the facility to act as a signal source and also performs

processing such as FFT, independent of PC. It can obtain time domain

as well as frequency domain waveforms. The output from Photon was

given to the computer which can store the time domain waveform. No

separate power supply is required as PHOTON is powered via the host

PC's USB port. Its real-time rate as an FFT Analyzer is 42 kHz in each

channel.

Output
Channel

Input
Channels

Figure 3.7 FFT Analyzer

High quality design of the PHOTON gives exceptional accuracy

and fidelity for all acquired or generated signals. Its 110 dB dynamic
 56

range allows resolution of signals that differ in amplitude by a ratio of

over 300,000 to 1. All inputs have both analog and digital filters

providing complete protection for aliasing and ensuring full data

integrity. This high quality design together with the 24-bit resolution

provides high accuracy for all measurements.

3.5 Mounting of Sensing and Measuring Equipment

The sensing and analyzing equipment which are described above were

mounted on casing of gearbox in the radial direction using epoxy resin

adhesive as shown in Figure 3.8.

Figure 3.8 Accelerometer Mounted on Casing

Care has been taken such that the mounted accelerometer has

become an integral part of test gearbox. In order to achieve high

 57

signal-to-electrical noise ratio the accelerometer output was connected

to the FFT analyzer input through a BNC cable.

The power to the FFT analyzer was supplied through USB input,

whereas power to accelerometer was supplied through the analyzer.

The output of the FFT analyzer was connected to computer to store

data which is shown in Figure 3.9. Using RT-Pro software the windows

were drawn, required time domain or frequency domain scale was

selected and vibration data was collected.

Figure 3.9 Gear Pair with Data Acquisition

The ICP accelerometer has a low frequency roll-off due to the

amplifier itself, and this is at 1 Hz. When the ICP accelerometer is

 58

connected to the power source it takes few seconds for the amplifier to

stabilize. Any data collected during this time may be contaminated by

a slowly varying voltage ramp. To avoid this, a small time delay of 5 to

10 seconds was incorporated in the data collectors. A low pass filter of

3 kHz was used to overcome the aliasing effect. Hanning is the most

commonly used window function for random signals because it

provides good frequency resolution and leakage protection with fair

amplitude accuracy. So it was employed to average the time signal

and to overcome the leakage problems.

3.6 Methodology

The engine was started using the kick rod and motion was

transmitted to the gearbox. Using the gear shift lever the motion was

further transmitted to the driven shaft through a gear pair E-F (2nd

gear). The fuel at constant rate was supplied continuously and no load

was applied on the driven shaft. The system was run till it got

stabilized.

Initially the experiments were conducted on gear pair E-F

without inducing any defect on gearwheel F (Case1) and by

transmitting motion to the driven shaft. The vibrations generated due

to transmission of motion were sensed by the mounted accelerometer

and the output of accelerometer was accessed and stored in the

computer through FFT using BNC cables. Using RT-Pro software

which was installed in the computer velocity signal in time domain

scale was selected. From RT-Pro window using the start icon, data
 59

acquisition was commenced and signal was observed sometime till it

got stabilized and that part of stabilized signal for one revolution of

gear was stored as sample vibration signature for this particular case

of no induced defect. The same procedure was repeated number of

times and 80 samples of vibration signatures were collected for Case1.

Among them three sample signatures are presented in Figure

3.10 for Case1, where Figure 3.10(a) shows sample no.69 which has a

maximum amplitude of vibration 0.0052 mm/s, Figure 3.10(b) shows

sample no.23 which has a maximum amplitude of vibration 0.0038

mm/s and Figure 3.10(c) shows sample no.8 which has a maximum

amplitude of vibration 0.0031 mm/s. As it was established earlier that

the maximum amplitude of vibration signature reflects the intensity of

the defect it was recorded. The present investigation has carefully

found the maximum amplitude of vibration in all the 80 samples

collected and tabulated in Table 3.2. These amplitudes for Case1 (No

fault) were in the range of 0.001-0.005 mm/s.

 60

mm/s

a) Case1 (No Fault gears) – Sample 69

mm/s

b) Case1 (No Fault gears) – Sample 23

mm/s

c) Case1 (No Fault gears) – Sample 8

Figure 3.10 Snapshots of Vibration Measurement-No Fault Case

 61

Then experiments were repeated after inducing 20% wear as

described in Section 3.2 on selected tooth of gearwheel F (Case2) and

by transmitting motion to the driven shaft. The vibration signatures

were collected for 80 samples after the signal got stabilized. Among

them three sample signatures are presented in Figure 3.11 (a, b and c)

for sample no.02, 18 and 74 respectively. From the Figure 3.11 (a, b

and c) it may be noted that the maximum amplitude of vibration are

0.0089 mm/s, 0.0088 mm/s and 0.0073 mm/s respectively.

The maximum values of vibration signatures of 80 collected

samples of Case2 have been tabulated in Table 3.2. It may be

observed from the table that these amplitudes for Case2 (20% wear)

are ranging between 0.002 mm/s to 0.018 mm/s.

 62

mm/s

a) Case2 (20% wear) – Sample 2

mm/s

b)Case2 (20% wear) – Sample18

mm/s

c) Case2 (20% wear) – Sample 74

Figure 3.11 Snapshots of Vibration Measurement-20% Wear

 63

Similarly the experiments were carried out with 40% & 80%

wear induced gears by transmitting the motion to the driven shaft.

The vibration signal collected for Case3 and Case4 were presented in

Figure 3.12 & 3.13 respectively and tabulated in Table 3.2. Further it

may be noted from the table that the maximum amplitude of vibration

for Case3 is in the range of 0.06 mm/s to 0.09 mm/s and for Case4 in

the range of 0.07 mm/s to 0.09mm/s.

 64

mm/s

a) Case3 (40% Wear) – Sample 62

mm/s

b) Case3 (40% Wear) – Sample 18

mm/s
mm/s

c) Case3 (40% Wear) – Sample 37

Figure 3.12 Snapshots of Vibration Measurement-40% Wear

 65

mm/s

a) Case4 (80% Wear) – Sample 15

mm/s

b) Case4 (80% Wear) – Sample 9

mm/s

c) Case4 (80% Wear) – Sample 29

Figure 3.13 Snapshots of Vibration Measurement-80% Wear

 66

Then experiments were conducted with broken tooth gear

(Case5) and by transmitting motion to the driven shaft. As the peaks

of vibration due to broken tooth could not be observed in the velocity

window, the acceleration window is taken for experimentation and

found that the peaks of vibration due to induced broken tooth are

visible. The vibration signatures were collected for 80 samples for the

same fault of broken tooth. Among them three sample signatures are

presented in Figure 3.14(a, b and c) for sample no. 15, 34 and 57

which have maximum amplitudes of 22.5mm/s2 (0.219 mm/s), 21

mm/s2 (0.19 mm/s) and 20mm/s2 (0.2 mm/s) respectively. Similarly

the maximum values of vibration signatures have been tabulated in

Table 3.2 for the 80 samples of Case5.

As the objective of this investigation is to prove the success of

Fault Diagnosis system, more effort has been put on overall concept of

integration than in executing the minute details of processes and

standard methods.
 67

mm/s2

a) Case5 (Broken tooth) – Sample 15

mm/s2

b) Case5 (Broken tooth) – Sample 34

mm/s2

c) Case5 (Broken tooth) - Sample 57

Figure 3.14 Snapshots of Vibration Measurement-Broken tooth

 68

Table 3.2 Vibration Amplitudes for Different Faults (mm/s)

Case1 Case2 Case3 Case4 Case5

Sample
No Fault 20% Wear 40% Wear 80% Wear Broken Tooth
1 0.003 0.0172 0.062 0.071 0.200554
2 0.0012 0.0089 0.077 0.08 0.343807
3 0.0052 0.0022 0.065 0.07 0.133703
4 0.0048 0.0012 0.082 0.076 0.362907
5 0.0044 0.0131 0.089 0.071 0.296056
6 0.0024 0.004 0.073 0.078 0.171903
7 0.004 0.0054 0.089 0.09 0.191004
8 0.0031 0.0116 0.074 0.07 0.085952
9 0.0037 0.0131 0.089 0.083 0.334257
10 0.0011 0.0067 0.063 0.078 0.181454
11 0.0041 0.0024 0.083 0.08 0.085952
12 0.0025 0.0089 0.078 0.073 0.343807
13 0.0022 0.0148 0.082 0.076 0.353357
14 0.0014 0.0054 0.085 0.077 0.343807
15 0.0045 0.0148 0.079 0.089 0.219654
16 0.0037 0.0062 0.078 0.084 0.219654
17 0.0044 0.0014 0.077 0.084 0.162353
18 0.0046 0.0088 0.067 0.089 0.191004
19 0.002 0.0133 0.077 0.085 0.095502
20 0.0011 0.013 0.069 0.086 0.296056
21 0.004 0.0066 0.081 0.071 0.181454
22 0.0041 0.0076 0.075 0.084 0.085952
23 0.0038 0.0162 0.087 0.084 0.382007
24 0.0021 0.0029 0.078 0.088 0.210104
25 0.0035 0.0162 0.069 0.085 0.382007
26 0.001 0.002 0.081 0.082 0.210104
27 0.0051 0.0178 0.09 0.087 0.124152
28 0.0036 0.0164 0.088 0.084 0.124152
29 0.0044 0.0067 0.081 0.076 0.200554
30 0.0043 0.0162 0.07 0.08 0.267405
31 0.0041 0.0149 0.069 0.089 0.267405
32 0.0037 0.0156 0.075 0.084 0.286506
33 0.001 0.0062 0.074 0.088 0.305606
34 0.0045 0.0054 0.065 0.087 0.191004
35 0.0042 0.0178 0.063 0.076 0.238755
36 0.0033 0.0017 0.068 0.085 0.257855
37 0.0018 0.0141 0.071 0.074 0.267405
38 0.0015 0.0133 0.082 0.08 0.114602
39 0.004 0.0043 0.088 0.078 0.353357
40 0.0026 0.0014 0.079 0.073 0.248305
41 0.0042 0.0117 0.084 0.08 0.267405
42 0.0039 0.0059 0.066 0.082 0.152803
 69

43 0.0018 0.0137 0.065 0.081 0.257855

44 0.0048 0.0065 0.067 0.082 0.114602
45 0.0021 0.0105 0.084 0.073 0.114602
46 0.0028 0.0137 0.085 0.087 0.133703
47 0.0051 0.0151 0.079 0.073 0.162353
48 0.0045 0.0015 0.088 0.072 0.133703
49 0.0029 0.0014 0.065 0.073 0.124152
50 0.0044 0.0169 0.087 0.086 0.095502
51 0.0034 0.002 0.079 0.071 0.382007
52 0.0028 0.0085 0.087 0.084 0.257855
53 0.0038 0.0075 0.069 0.086 0.191004
54 0.0018 0.0063 0.089 0.089 0.362907
55 0.0028 0.0139 0.086 0.085 0.382007
56 0.0032 0.0167 0.074 0.077 0.229204
57 0.0022 0.002 0.073 0.071 0.200554
58 0.0027 0.0038 0.07 0.07 0.362907
59 0.0043 0.004 0.082 0.076 0.171903
60 0.0036 0.0165 0.068 0.084 0.267405
61 0.0035 0.009 0.075 0.071 0.334257
62 0.0038 0.0092 0.072 0.079 0.076401
63 0.0027 0.0045 0.074 0.07 0.238755
64 0.0024 0.0165 0.082 0.076 0.200554
65 0.0031 0.016 0.09 0.077 0.382007
66 0.0046 0.0046 0.088 0.077 0.162353
67 0.0024 0.015 0.07 0.083 0.143253
68 0.0032 0.0021 0.075 0.072 0.124152
69 0.0052 0.0052 0.09 0.073 0.200554
70 0.0036 0.009 0.084 0.071 0.353357
71 0.0025 0.0127 0.083 0.076 0.267405
72 0.0039 0.0132 0.061 0.074 0.305606
73 0.0049 0.0047 0.081 0.077 0.334257
74 0.0029 0.0073 0.066 0.073 0.200554
75 0.0013 0.0044 0.063 0.076 0.219654
76 0.0028 0.007 0.078 0.077 0.324706
77 0.002 0.0054 0.073 0.086 0.276955
78 0.0032 0.013 0.081 0.079 0.267405
79 0.0028 0.0095 0.063 0.088 0.105052
80 0.0017 0.0156 0.082 0.087 0.248305

For fault diagnosis of gears based on maximum vibration amplitudes

collected by the present investigation, an analysis has been carried

out and presented below.

 70

3.7 Analysis of Vibration Amplitudes

As a first step of analysis, the vibration amplitude data pertaining to

gearbox with and without faults, is plotted and presented in Figure

3.15. It depicts maximum vibration amplitudes of all trials for all the

cases studied by the present investigation. On careful observation of

the above mentioned figure and the maximum vibration values

presented in Table 3.2, it can be pointed out that the data pertaining

to different cases of no fault (Case1), different degrees of wear (Case2,

Case3, Case4), broken tooth (Case5) is overlapping with each other, to

a certain degree.

Figure 3.15 Vibration Amplitudes for Different Cases

Though there is some sort of trend in the values of maximum

vibration, still it is ambiguous to state the trend due to overlapping of

 71

data and it is not possible to categorically conclude the condition of

the gear based on this data only.

Hence it is required to extract the features based on this

experimental data in order to probe further for fault diagnosis, which

is presented in the next chapter.