Вы находитесь на странице: 1из 16

Journal of http://jvc.sagepub.

com/
Vibration and Control

Experimental investigations to establish correlation between Stribeck curve, specific film thickness and
statistical parameters of vibration and sound signals in a spur gear system
Sang-Kwon Lee and M Amarnath
Journal of Vibration and Control published online 4 August 2014
DOI: 10.1177/1077546314544164
The online version of this article can be found at:
http://jvc.sagepub.com/content/early/2014/08/01/1077546314544164

Published by:
http://www.sagepublications.com

Additional services and information for Journal of Vibration and Control can be found at:
Email Alerts: http://jvc.sagepub.com/cgi/alerts
Subscriptions: http://jvc.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://jvc.sagepub.com/content/early/2014/08/01/1077546314544164.refs.html

>> OnlineFirst Version of Record - Aug 4, 2014


What is This?

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Article

Experimental investigations to establish


correlation between Stribeck curve,
specific film thickness and statistical
parameters of vibration and sound signals
in a spur gear system

Journal of Vibration and Control


115
! The Author(s) 2014
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1077546314544164
jvc.sagepub.com

Sang-Kwon Lee1 and M Amarnath1,2

Abstract
The use of a gear as a power transmitting machine element in engineering applications is quite extensive. Under service
conditions gear teeth contacts have a tendency for deterioration due to contact fatigue, wear, material defects, lubrication failure etc. Gear teeth surface deterioration triggers several types of failures such as abrasive wear, scuffing, micro
pitting and spalling. The abovementioned faults influence the changes in the vibration and sound signals in the gear
system. This paper presents the novel results of experimental investigations carried out to assess the surface fatigue
wear in a spur gear system. Estimation of specific film thickness, measurement of reduction in tooth thickness, visual
examination of wear mechanisms on the gear teeth and their effects on the statistical parameters of vibration and sound
signals are considered in this study. In order to confirm the hypothesis of dependency of surface fatigue wear, specific film
thickness and sound and vibration signals the results of statistical parameters obtained from sound and vibration signals
have been correlated with the Stribeck curve.

Keywords
Lubricant film thickness, sound, tooth thickness, vibration, wear

1. Introduction
Gear drives are one of the very common power transmission systems which appear in various kinds of rotating machinery and control systems. The service life of a
gear system depends upon a variety of factors viz. lubrication, design, contamination, environment etc.
Lubricating oils are used in a gear system to reduce
friction and wear by interposing a lm of material
between the gear teeth. If the lubricant lm at the
gear teeth contacting surfaces collapses, the heat and
wear on the gear teeth increases rapidly thereby causing
various wear mechanism such as micro pitting, abrasive
wear spalling, scoring etc. The presence of severe wear
on the gear teeth reduces stiness in the gear teeth contact and such an operating condition alters the vibration and sound signals of a gear system (Bartz, 1993;
Lee and White, 1998; Hohn and Michaelis, 2003;
Yesilyurt, 2003). In the last two decades several
researchers have published their research articles on

detection and diagnosis of rolling and sliding contact


machine elements such as gears, roller bearings, cams
and journal bearings (Glodez et al. 1997; Castro and
Seabra, 1998; Hoskins et al., 2001; Hohn and
Michaelis, 2003; Yesilyurt, 2003; Yesilyurt et al.,
2003; Amarnath et al. 2012). However, very few
researchers focused on the versatility of combination
of condition monitoring techniques which provide
more reliable fault diagnostic information.

1
Acoustics and Vibration Signal Processing Laboratory, Inha University,
Republic of Korea
2
Machine Dynamics and Tribology Laboratory, Indian Institute of
Information Technology Design and Manufacturing, Jabalpur, India

Received: 25 December 2013; accepted: 6 June 2014


Corresponding author:
Sang-Kwon Lee, Department of Mechanical Engineering, Inha University,
253 Yonghyun-Dong, Nam-Gu, Incheon 402751, Republic of Korea.
Email: sangkwon@inha.ac.kr

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Journal of Vibration and Control

Baydar and Ball (2003) demonstrated the results of


fault diagnosis experiments conducted on two stage helical gearbox. Authors have considered sound and vibration signals to detect local faults in the helical gear
tooth. Sound and vibration signals acquired from the
gearbox were processed using Morlet wavelet transform. Amplitude and phase maps obtained from wavelet analysis provided a good visual inspection tool to
detect gear faults in the early stage. Raja Hamzah and
Mba (2009) conducted experimental investigations to
nd the inuence of speed and load in generating acoustic emission (AE) signals for the helical and spur gears
operating under dierent load and speed conditions.
Specic lm thickness analysis was also considered in
conjunction with AE signal analysis. Results showed
that the percentage reduction in specic lm thickness
values causes an increase in root mean square (RMS)
values of acoustic signals in a nine-fold and four-fold
percentage change. This information highlighted exact
lubrication condition in the geared system.
Peng et al. (2005) carried out a series of experimental
investigations on worm and worm wheel gear systems
to obtain the correlation between vibration and wear
debris analyses. The data obtained from the wear particle morphology analysis was compared with vibration
spectrum analysis to quantify the eectiveness of these
condition monitoring techniques. Wear debris analysis
revealed the wear modes of the gear wheel whereas the
vibration signature in fast Fourier transform (FFT)
provided quick reliable information on faulty operating
conditions of the bearing.
Ebersbach et al. (2006) highlighted the eectiveness
of combining vibration and wear debris analysis methods to assess surface fatigue wear in spur gears. Wear
particle morphology analysis provided evidence of fatigue and sliding wear modes on gear teeth surfaces. Fast
Fourier transform analysis of vibration signals showed
gear eccentricity and misalignment faults on the amplitudes of gear mesh frequency harmonics. Manoj et al.
(2008) carried out fatigue test experiments on heat treated rollers of rolling contact fatigue test rig. Sound and
vibration signal analysis techniques were used to detect
faults on rollers operating under dierent loads and
speeds. Hoskins et al. (2001) performed fault detection
investigations on spur gears made of various polymeric
materials. A sound signal monitoring method was used
to diagnose faults on the gear teeth of a back-to-back
power recirculation type gearbox. Tribological parameters such as tooth surface roughness, wear propagation and lubricant temperature were also measured to
provide additional diagnostic information. Results
obtained from the frequency spectra of sound signals
revealed erroneous operating conditions in the gearbox
thereby suggesting a potential use of noise frequency
analysis method for fault detection.

Tekiner and Yeslyurt (2004) have explored the best


suitable machining conditions using statistical values of
sound signals. In this work the authors have highlighted the inuence of sound signals to analyze the
machining process parameters viz. ank wear, workpiece surface roughness, chip morphology and builtup edge formation. Kurtosis parameter values were
not considered due to their susceptibility to high frequency acoustic signals obtained during machining
process.
Li and Machefske (2006) carried out experiments to
diagnose defects in an induction motor using vibration,
sound and stator current monitoring methods. A
broken bar fault and a combination of bearing faults
i.e. inner race, outer race and ball defects were articially introduced in a variable speed three phase induction motor. Experimental results showed that the
stators current monitoring method was sensitive to
the broken rotor bar faults while the sound and vibration signal monitoring and analysis methods were
found to be very sensitive to detect bearing faults.
Amarnath et al. (2012) carried out experiments to
asses distributed faults in a spur gear system. In this
work, fatigue test experiments were carried out on
power recirculation type back-to-back spur gearbox.
The results provide a good understanding of the
dependent roles of gearbox operating conditions and
vibration parameters as measures for eective assessment of wear in spur gears. Sound signal analysis methods were implemented by Amarnath and Praveen
Krishna (2012) to detect simulated faults in a helical
gear system. A series of trials were made in their
work to extract reliable diagnostic information from
healthy and faulty helical gears, empirical mode decomposition (EMD)-based statistical parameters were used
to extract fault related features in the helical gear
system.
In the present work, the experiments were conducted
on the spur gearbox to provide a verication of the
dependency of wear propagation, reduction in minimum lubricant lm thickness, specic lm thickness,
tooth thickness reduction and statistical parameters of
vibration and sound signals to assess the severity of
surface fatigue wear in a spur gear system.

2. Lubrication, wear mechanism,


vibration and noise levels
in spur gear system
Sucient lubrication is an important factor for the durability of a gear transmission system. Gear failures viz.
micro pitting, scung, spalling, scoring etc. depend
strongly on lubrication conditions in gear teeth contacts (Bartz, 1993; Yesilyurt, 2003). Lubrication in rolling/sliding contacts such as gears, rolling element

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

bearings, journal bearings and cams can occur in three


regimes: (i) Hydrodynamic lubrication (HL) (ii)
Elastohydrodynamic lubrication (EHL) (iii) Boundary
lubrication (BL). The characteristic of each lubrication
regime is summarized in the Stribeck curve as shown in
Figure 1.
Transition from one lubrication regime to the other
may occur due to run-in wear or operating conditions
(Castro and Seabra, 1998; Jacobson, 2003; Serrato
et al., 2007; Halme and Andersson, 2009). The minimum oil lm thickness (hmin) and specic lm thickness (l) parameters are calculated according to the
Dowson and Higginson equation (Dowson and
Higginson, 1977; Bartz, 1993).
hmin 

1:60:6 o u0:7 E0 0:03 0:44


w0:13
l

hmin
R

The nomenclature details are included in


Appendix A. l values can be used in the Stribeck
curve to understand operating conditions in the spur
gear system. Reduction in l causes wear propagation
on gear tooth surfaces. Wear propagation on gear tooth
surfaces results in tooth stiness reduction which causes
an increase in the vibration and noise levels in the gear
transmission system (Halme and Andersson, 2009;
Amarnath et al., 2012).
In the case of HL, the contacting surfaces are supported by the hydrodynamically pressurized lubricant.
The specic lm thickness is greater than the combined
surface roughness (l  3). The vibration signals generated in this regime show characteristics of rotational
vibration, which can be considered as base line vibration
signals. Elastohydrodynamic lubrication refers to the
type of lubrication where the pressure between contacting surfaces is very high so that the surfaces deform elastically to an amount comparable to the oil lm thickness.
In the EHL regime, the specic lm thickness is found to

Figure 1. Gear teeth contacts in Stirbeck diagram as a function of specific film thickness .

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Journal of Vibration and Control

be 1.4 < l < 3, which indicates mixed wear lubrication


and some wear on contacting surfaces is predicted
(Jacobson, 2003; Halme and Andersson, 2009;
Amarnath et al. 2012). Wear mechanism such as pitting,
mild wear and scung may be observed in an EHL
regime, propagation of these defects in machine elements
leads to an increase in the vibration levels and periodic
impacts in the vibration signals. On the other hand, the
boundary lubrication regime is shown at the extreme left
of the graph (Figure 1), this regime indicates the presence
of asperity contacts between the gear teeth contacting
surfaces. As a result of closer contact of asperity, wear
on the gear tooth surfaces is to be expected by the tribo
chemical or mechanical wear. Metal asperity contacts
under boundary lubrication act as a source of vibration
and noise. Machine elements operating under this lubrication regime show an increase in vibration and noise
levels along with random impacts (Cameron, 1981; Sato
and Takanashi, 1982; Bartz, 1993).

The crest factor (C) (Yesilyurt et al., 2003) is dened


as the maximum positive peak value of the signal x
divided by RMS and is given by

3. Vibration and sound signal analysis


using statistical parameters

This parameter is designed to increase in the presence of heavy uniform wear on gear teeth surfaces. On
the other hand, spectrum analysis of sound and vibration signals can be used in conjunction with these statistical parameters to detect wear severity on the gear
tooth surfaces.

A number of vibration and sound-based signal analyses


methods have been developed to detect local/distributed faults in geared systems. Application procedures
and the success of these techniques have been discussed
in several papers (Baydar and Ball, 2003; Yesilyurt
et al., 2003; Li and Mechefske, 2006; Parey and
Tandon, 2007; Amarnath and Praveen Krishna, 2012;
Ma et al., 2012). Gear faults such as pitting, scung
and severe wear on gear teeth can be evaluated by
employing the statistical indicators of the sound and
vibration signals.
The RMS of the signal (Yesilyurt, 1997; Lee and
Lee, 2014) is dened as the square root of the average
of the sum of the squares of the signal samples and is
given by
v
u N
N
u1 X
1X
RMS t
xn  x 2 x
xn 3
N n1
N n1
where N is the number of samples, x(n) is the amplitude
of the signal of the nth point and x is the mean value of
all the amplitudes.
The kurtosis is the fourth normalized moment of a
given signal and provides a measure of the peakedness
of the signal, i.e. the number and amplitude of peaks
present in the signal, this parameter is quite useful in
detection of faults in gears and bearings (Yesilyurt,
1997).
Kurtosis

N
1X
xn  x 4
h P
i2
N n1 1 N
 2
n1 xn  x
N

maxx
RMS

C is a normalized measurement of the amplitude of


the signal and is designed to increase in the presence of
a small number of high amplitude peaks, such as those
caused by some types of local tooth damage such as
pitting, scoring, spalling etc.
The energy ratio (ER) (Yesilyurt, 1997) is dened as
the ratio of the RMS of the dierence signal d and the
RMS of the signal containing only the regular meshing
components, yd, and given by
ER

RMSd
RMSyd

4. Experiments
4.1. Experimental setup, sensors and equipments
Figure 2 shows the experimental setup used to conduct
fatigue test experiments on the spur gear pair. It consists of 0.75 kW direct current (DC) motor and a single
stage spur gearbox with a pair of gears mounted on two
parallel shafts, the driving pinion had 24 teeth and the
driven gear had 25 teeth. The pinion was considered for
fault assessment.Table 1 gives the detailed specications of the gearbox and test conditions. The DC
motor rotation was controlled by a variable speed controller which was used to drive the input shaft. The
eddy current magnetic brake was used to provide
torque load to the gearbox, this device has a maximum
torque capacity of 12 Nm.
The vibration and sound signals were acquired using
a Bruel & Kjaer 4506 triaxial accelerometer and Bruel
& Kjaer 4189 microphone respectively. The accelerometer was mounted on the input shaft bearing. A commercial data acquisition system LMS SCADASIII was
used to acquire the sound and vibration signals, these
signals were sampled at 8.2 kHz, acquired simultaneously and stored in a personal computer for post processing. Each time domain signal consisting of the
average of 16 time history plots, which seemed to be
sucient to reduce the noise in vibration and sound

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

Figure 2. Schematic representation of the experimental setup installed in anechoic chamber.

Table 1. Gear test rig specifications.


Parameter

Pinion

Number of teeth
24
Deport angle
0.00
Pressure angle
20
Height
6.53
Module
3
Face width
30
Pitch diameter
72.72
Diameter of base
68.34
Diameter of head
78.78
Pinion speed
450 rpm
Nonimal load
44 N
Torque on pinion shaft
012 Nm
Material properties of gears
SM45C
Brinell hardness number hardness (HRB) 167
Poissons ratio
0.3
Youngs modulus
2  105 N/mm2

Gear
25
0.00
20
6.53
3
30
75.75
71.18
81.81

signals. Vibration signals in the Z - direction were considered to assess the wear propagation on gear teeth.
The microphone was used to acquire sound signals
from the gearbox which was installed in an acoustically
rigid anechoic chamber. The microphone was kept at a
distance of 5.5 cm to 6 cm in the vicinity of input shaft.
This position was considered after many trials (near
eld condition): this procedure was successfully used
in our previous work (Amarnath and Praveen
Krishna, 2012). The walls, ceiling and oors were
glued with 8 inch acoustic wedges. The photograph of
the experimental setup with data acquisition system
and sensors is shown in Figure 3.

4.2. Operating conditions


Fatigue test experiments were carried out on the spur
gear system over 1000 hours (27  106 cycles), the gearbox was allowed to run-in wear for a period of 30
hours. After run-in wear test gears were cleaned and
mounted again. The gearbox was operated at a speed
of 450 rpm, a load of 12 Nm was given to the gear pair
using an eddy current brake. The sound and vibration

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Journal of Vibration and Control

Figure 3. View of the test rig showing sensors and equipments.

signals were measured at regular intervals. SAE 40


lubricant was used in gear fatigue test experiments,
lubricant temperature was monitored using a k-type
thermocouple (0900 C), this probe was placed in the
gear mesh zone. The temperature values were measured
after every 30 minutes. Oil viscosity values corresponding to the measured temperatures were used for computing the lubricant lm thickness for dierent
temperatures. During the experiments, the gearbox
operation was paused to allow internal inspections of
the gear anks. After every 200 hours the surface
roughness and tooth thickness measurements were carried out on the gear teeth anks. A Hommel-Etamic
(w5) roughness measurement instrument was used to
measure the roughness on the gear tooth ank and
tooth thickness measurement was performed using a
digital vernier caliper. A USB microscope was used to
analyze wear mechanisms on the gear tooth surfaces.
Figure 4. Oil temperature versus time from the start of the
test at 0 Nm, 6 Nm and 12 Nm gearbox input torque.

5. Results and discussions


5.1. Lubricant film thickness and specific
lubricant film thickness
Lubricant lm thickness between the gear teeth contact
is of signicant importance to maintain eective operating conditions. To analyze the increase in temperature and corresponding decrease in minimum lm
thickness, the experiments were carried out for 8
hours under three load conditions i.e 0 Nm, 6 Nm and
12 Nm.
Figure 4 shows the oil bath temperature versus the
operating time at 0 Nm, 6 Nm and 12 Nm input torques, stabilization in temperature is observed after
6 hours. In general, stabilization in the lubricant temperature occurs due to the equalization of temperature

from the generation and dissipation of heat in the


system. An increase in lubricant temperature causes a
decrease in viscosity These viscosity values along with
velocity, load and pressure viscosity coecient values
are used to calculate oil lm thickness as shown in
equation (1). A gradual decrease in hmin values
obtained for three load conditions (0, 6, 12 Nm) are
shown in Figure 5. Further, the gear fatigue test experiment was continued by considering the highest load i.e.
12 Nm over a period of 1000 hours.
At the initial stage, after 200 hours of the test period,
due to cyclic fatigue load and variations in lubricant
properties (oxidation, nitration, sulphation etc.) the distributed faults such as mild wear and pitting have

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

Figure 5. Lubricant film thickness versus time from the start of


the test at 0 Nm, 6 Nm and 12 Nm gearbox input torque.

Figure 6. Combined surface roughness versus time at 12 Nm


gearbox input torque.

propagated on the gear tooth surfaces. The combined


surface roughness values on the gear teeth anks are
increased under these operating conditions. Figure 6
depicts the increase in combined surface roughness
values as a function of time. The specic lm thickness
l is calculated by the ratio of minimum lubricant lm
thickness to combined surface roughness, which represents the most critical tribological contacts in determining contact fatigue life of a gear system.
Figures 7(a)(e) show specic lm thickness l estimated during 0200, 400, 600, 800 and 1000 hours of
operation obtained at 12 Nm gearbox input load
condition.
Figure 7(a) shows the specic lm thickness values
obtained during 0200 hours, the l values obtained
during this operating period correspond to the
hydrodynamic lubrication i.e. l  3, about 8 hours
observation is shown in the Figure 7(a). However
approximately equal l values were obtained throughout 200 hours. After 200 hours the experiment was
paused, the propagation of surface wear on the test
pinion was examined using a USB port microscope,
the tooth thickness and teeth surface roughness values
were measured using a digital vernier calliper and
perthometer.
Further, due to long running hours, surface fatigue
wear on the gear teeth causes an increase in the combined roughness values of the contacting surfaces.
Specic lm thickness values obtained after every 200
hours show a decrease in trend due to an increase in the
combined surface roughness values. Figure 7(b) shows
specic lm thickness values obtained between 200400
hours of operation, these values correspond to the EHL
regime (1.4 < l < 3). Under this regime the initiation of

scung damage was observed on the gear tooth surfaces. The EHL regime provides a protecting layer
between the gear tooth surfaces. The EHL lm formation is decreased at high temperature and lower viscosities resulted in scung wear propagation on gear tooth
surfaces An increase in the wear has an eect on the
roughness of the gear teeth anks (Hohn and
Michaelis, 2004; Rak Sari et al., 2007; Hohn et al.,
2009) . A gradual increase in the gear tooth surface
roughness values, the EHL and boundary lubrication
regimes are observed during 400 h and 600 h of operation as shown in Figure 7(c). Due to the prolonged
operating time, surface wear propagates thereby causing the transition of the lubrication regime i.e. the gear
system enters into a boundary lubrication regime after
800 hours of operation as shown in Figure 7(e). Wear
mechanisms and their corresponding eects on vibration and sound signals of the gear system are discussed
in the next sections.

5.2. Microscopic observations of damaged gear


Prolonged running time and the transition of lubrication regimes trigger several wear mechanisms on the
gear tooth surfaces such as pitting, scung, scoring
and spalling (Bartz, 1993; Glodez et al., 1997;
Amarnath et al., 2012). Figure 8 depicts wear propagation on dierent sections of gear tooth surfaces. Figure
8(a) shows the appearance of pitting damage observed
after 200 hours. Pitting is considered to be negligible if
pits are small, such a pitting mechanism has no eect
on operational characteristics, however large size pits
may spread on the tooth surface and the other teeth to
destroy gear teeth completely (Bartz, 1993; Glodez

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Journal of Vibration and Control

Figure 7. Specific film thickness versus time at 12 Nm gearbox input torque (a) 0200 hours (b) 400 hours (c) 600 hours (d) 800
hours (e) 1000 hours.

et al., 1997; Hohn and Michaelis, 2004). Figures 8(b)


and (c) show an increase in pit size and pitting extending on the other teeth, these faults were observed after
400 hours. EHL and boundary lubrication regimes
have been observed between 300600 hours of
operation.Microscopic inspection of the gear teeth surfaces after 800 hours showed scung wear as depicted
in Figure 8(d), a combination of sliding and rolling
action on the gear tooth tip resulted in the scung
wear mechanism progressing from the tooth tip

towards the tooth root. Further, a scoring defect also


appeared on the gear tooth ank as shown in Figure
8(e). This wear mechanism was observed after 1000
hours, a characteristic wear pattern which developed
in the form of patches concentrated in the dedendum
region of the gear teeth.
The above discussed wear mechanisms cause loss of
material from the gear teeth surface thereby resulting in
a reduction in tooth thickness. Table 2 gives average
values of the tooth ank thickness measurement carried

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

Figure 8. Surface fatigue wear on the pinion teeth surfaces obtained at 12 Nm gearbox input torque.

Table 2. Tooth thickness measurement and % reduction in


tooth flank thickness.

Operating
time

Measured tooth
flank thickness (mm)
(three teeth
separated by 120 )

Average
tooth
thickness
(mm)

Reduction
in
thickness
(%)

0 hours
200 hours
400 hours
600 hours
800 hours
1000 hours

5.62
5.51
5.44
5.34
5.21
5.18

5.58
5.47
5.45
5.41
5.32
5.22

0
1.9
2.3
3.04
4.65
6.41

5.59
5.48
5.49
5.51
5.25
5.22

5.55
5.42
5.43
5.38
5.53
5.28

out on the three gear teeth separated by 120 . Figure 9


shows the percentage increase in the reduction in tooth
thickness with respect to operating time. Transition in
lubrication regimes, wear propagation, the gear teeth
surface and tooth thickness loss showed a signicant
eect on vibration and noise levels in the gear system.

5.3. Correlation of the vibration and sound signal


analysis with stribeck curve
The surface fatigue failure is a root cause for the gear
teeth failure modes viz. pitting, scoring and scung.
These wear modes propagate with an increase in running time thereby resulting in time varying mesh stiness and gear transmission errors, which cause
considerable changes in sound and vibration levels of
the gearbox (Yesilyurt, 1997; Amarnath et al., 2012).
Fault detection experiments using the vibration and
sound signal monitoring and analysis methods were
successfully conducted by many authors (Baydar and

Figure 9. Reduction in tooth flank thickness versus operating


time.

Ball, 2003; Yesilyurt et al., 2003; Li and Mechefske,


2006; Parey and Tandon, 2007; Amarnath and
Krishna, 2012). Simulated faults detection in induction
motors, bearings, gears were considered in these works.
In order to conrm the existence of the correlation
between the specic lm thickness and damage severity
on the gear teeth, the statistical parameters of sound
and vibration signals are plotted against the specic
lm thickness l. Comparisons of l versus the statistical
parameters viz. RMS, crest factor, energy ratio and
kurtosis values of sound and vibration signals are
made with the Stribeck curve.
Figure 10 shows the RMS values of vibration signals
versus l. At the initial stage of the experiment i.e.
during 0200 hours the hydrodynamic lubrication
regime was observed in the operating condition of the

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

10

Journal of Vibration and Control

Figure 10. Root mean square of vibration signals versus  at


12 Nm gearbox input torque.

gear system. A sucient lubricant lm is present


between the gear teeth contacts, hence large values of
l and lower RMS values of the vibration signals are
obtained as shown in Figure 10. Further, after 200
hours, an increase in the combined surface roughness
values R caused a reduction in the specic lm thickness, hence partial metal-to-metal contacts occurred in
the EHL regime tended to increase in RMS values of
the vibration signals, as depicted in the Figure 10. On
the other hand, increase in surface fatigue wear on gear
teeth causes increase in the combined roughness value
(R) of gear teeth contacts which results decrease in l
values. Hence, the transition take place to boundary
lubrication regime at l < 1.4, a signicant increase in
RMS values of vibration levels has been observed in the
boundary lubrication regime as depicted in Figure 10.
The crest factor and energy ratio values are also computed from the vibration signals, an approximately
similar increase in trends is observed from these
values as shown in Figures 11 and 12 respectively.
The crest factor and energy ratio values showed a
gradual increase in trend in HL, EHL and BL regimes.
Further, abovementioned statistical parameters are also
extracted from the sound signals. Figure 13 shows
RMS values of sound signals versus l. The transition
of lubrication regime from HL to EHL and BL resulted
in gradual increase in crest factor and energy ratio
parameters of sound signals as shown in Figures 14
and 15 respectively. These sound signal parameters,
along with RMS values of vibration signals, showed
some correlation to the Stribeck curve shown in
Figure 1. In the hydrodynamic lubrication regime a
thick lubricant lm is present between gear teeth contacts, hence coecient of friction m and statistical parameters of sound and vibration signals showed lower
values. An increase in the co-ecient of friction values

Figure 11. Crest factor of vibration signals versus  at 12 Nm


gearbox input torque.

Figure 12. Energy ratio of vibration signals versus  at 12 Nm


gearbox torque.

caused a decrease in the l values, hence the transition


occurred in the lubrication regimes of a gear system
resulted in an increase in sound and vibration levels.
Figure 16 depicts kurtosis values of both sound and
vibration signals as a function of operating time. This
parameter showed an uneven trend, hence these values
are not suitable to discriminate healthy and faulty operating conditions of the gearbox. This shortcoming of
the kurtosis parameter is due to its high susceptibility to
the spurious eect of noise and high frequency vibration and sound signals. Based on the fact that, in some
cases, the adverse eect of spurious vibrations on the
values of kurtosis is more than the benet gained from
the higher sensitivity of kurtosis to detect incipient
faults in rotating machine elements. Some case studies

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

11

Figure 13. Root mean square of sound signals versus  at


12 Nm gearbox input torque.

Figure 15. Energy ratio of sound signals versus  at 12 Nm


gearbox input torque.

Figure 14. Crest factor of sound signals versus  at 12 Nm


gearbox input torque.

Figure 16. Kurtosis values of sound and vibration signals


versus time at 12 Nm gearbox input torque.

have been discussed in the authors previous publications (Amarnath and Praveen Krishna 2012, 2013) and
interested readers can refer to these research articles.
The vibration spectrum encompasses abundant information related to the wear propagation on gear teeth
and the corresponding eects on dynamical characteristics of the gear meshing process.
Spectrum analysis of vibration and sound signals
can be used to locate and obtain the severity of fault
growth on the gear teeth (Yesilyurt 1997; Yesilyurt
et al., 2003; Serrato et al., 2007; Amarnath et al.,
2012; Amarnath and Praveen Krishna, 2013). In the
present experiment, the pinion speed was set to

450 rpm giving a fundamental mesh frequency of


180 Hz. Figure 17 shows the frequency spectra which
consist of fundamental gear mesh frequency (1 fm) and
its higher harmonics (2 fm, 3 fm and 4 fm) around 360 z,
540 Hz and 720 Hz respectively. The structural resonance of the gearbox system used for the test occurred at
the frequency of 450 Hz.
The surface fatigue wear on the gear teeth was the
result of the transition in lubrication regimes that had
been manifested in the fundamental gear mesh frequency component and its higher harmonics. Figure
17(a)(e) show vibration spectra obtained over a
period of 1000 hours at regular intervals of 250

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

12

Journal of Vibration and Control

Figure 17. Fast Fourier transform spectra of vibration signals at 12 Nm gearbox input torque (a) 0 h (b) 250 h (c) 500 h (d) 750 h
(e) 1000 h.

hours. From the Figure 17(a), lower amplitudes are


seen on gear mesh frequency harmonics after a run-in
wear test under healthy operating conditions. Micro
pitting damage observed on the gear teeth after 250
hours has no signicant eect on gear mesh frequency
harmonics. Figure 17(b) shows a small reduction in
amplitudes at gear mesh frequency components. At
the initial stage, wear propagation in gear pair causes
counter-institutive characteristics i.e. small deviations
in the involute tooth form. Gear vibration levels and
the corresponding strongest gear mesh frequency components in the frequency spectrum can be reduced due
to convex tooth prole modication (Yesilyurt, 1997;
Yesilyurt et al., 2003). Apparently, in the present work,
wear propagation on the gear teeth resulted in a 6.41 %
reduction in tooth thickness, hence material removal on
the gear teeth caused convex prole modication. The

minor modications in the convex tooth prole resulted


in a decrease in amplitudes at gear mesh frequency harmonics as shown in Figure 17(b). Further, after 400
hours the surface fatigue wear which propagates on
the other gear teeth results in a signicant increase in
the overall vibration levels in the gearbox. Figure 17(c)
(e) show frequency spectra of gearbox vibration signals
obtained after 500, 750 and 1000 hours respectively.
Wear mechanisms viz. macro pitting, scung, scoring
faults were observed on gear teeth as shown in Figure 8.
A gradual increase in amplitudes of gear mesh frequencies along with a signicant increase in sidebands
of the gear mesh frequency harmonics are attributed to
the deterioration of gear teeth. Frequency domain plots
of microphone outputs obtained from healthy and
worn gears are shown in Figure 18(a)(e). Gearbox
sound signals were acquired in an anechoic chamber

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

13

Figure 18. Fast Fourier transform spectra of sound signals at 12 Nm gearbox input torque (a) 0 h (b) 250 h (c) 500 h (d) 750 h
(e) 1000 h.

to reduce intrusive background noise. Figure 18(a)


shows the sound spectrum obtained after a run-in
wear test under healthy operating conditions. First
and second gear mesh frequency components, along
with the other sound sources such as a driving motor
and eddy current brake system, can be observed in this
Figure 18(a). Further, the wear propagation on gear
tooth surfaces after 250 hours causes tiny variations
in the sound signals, hence there is an increase in gear
mesh frequency amplitudes as shown in Figure 18(b).
Figures 18(c)(e) show spectra obtained after 500, 750
and 1000 hours respectively, an increase in gear mesh
frequency harmonics are seen at 180, 360, 540 and
720 Hz. However, spectra obtained after 750 and 1000
hours show a remarkable increase in amplitudes at

second and third gear mesh frequency components as


indicative of wear severity.

6. Summary and conclusions


The hypothesis of wear propagation on the spur gear
teeth and its corresponding eects on the sound and
vibration levels in the spur gear system have been investigated in this study. Wear assessment in the spur gear
system is carried out by monitoring and analysis of
parameters such as lm thickness, specic lm thickness, visiual examination of wear mechanisms on gear
teeth, statistical features of sound and vibration signals.
The following conclusions were drawn from the experimental observations.

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:39am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

14

Journal of Vibration and Control

1. Increase in prolonged operating time over 1000


hours triggered surface fatigue wear which resulted
in an increase in the combined surface roughness of
the gear teeth thereby decreasing specic lm thickness values. The specic lm thickness estimation
provides an alternative method to predict wear
severity on the gear teeth.
2. Visual examination of the worn gear teeth revealed
the details of the wear mechanisms such as pitting,
scung, and scoring which occurred on the tooth
tip, ank and root surfaces.
3. Statistical features viz. RMS, crest factor and energy
ratio values of the sound and vibration signals
showed an overall increase in trend as a function
of the transition in the lubrication regimes.
4. The Stribeck curve, which explains the lubrication
regimes in machine elements, has been considered
in this study. In the Stribeck curve, the co-ecient
of friction varies as a function of the change in lubrication regimes. A similar observation was made in
trend plots of specic lm thickness l versus statistical parameters of sound and vibration signals.
5. Polynomial curve tting plots of RMS values of the
vibration signals and RMS, crest factor, energy ratio
values of the sound signals showed approximately a
similar variation of l in the Stribek curve, which
established the correlation between specic lm
thickness, surface fatigue wear and statistical parameters of sound and vibration signals.
6. Kurtosis values of both sound and vibration signals
fail to provide reliable diagnostic information due to
their susceptibility to the spurious eect of noise.
7. Fast Fourier transform analysis of the vibration and
sound signals showed an increase in amplitudes at
the fundamental gear mesh frequency and its higher
harmonics, which clearly indicated the severity of
surface fatigue wear on the gear tooth surfaces.
8. The wear severity on the gear teeth was also evaluated by considering thickness loss on gear teeth surfaces. About 6.04% tooth loss was obtained over a
period of 1000 hours.
Funding
This work was supported by the Mid-career Researcher
Program through an NRF grant funded by the MEST
(grant number 2010-0014260) and by an Inha University
research grant.

Nomenclature
o

u
E0

absolute viscosity
pressure viscosity coefficient
peripheral velocity
combined
Youngs
modulus
1 E2
E0 E 12E
2 E 1 2
2

E1 , E2
1, 2

d0,1 , d0,2

w
F
FU
db
M
b
l
R
Ra

modulus of elasticity for gear and pinion


respectively
Poissons ration for gear and pinion
materials
effective radius of curvature
 of the two
2
1
1
surfaces 1 Sin

d0,1
d0,2
diameters of gear and pinion respectively
mesh angle
normal tooth force per unit width F =b
FU
normal tooth force F cos
2M
peripheral force db
pitch circle diameter
torque
tooth width
specific film thickness
combined surface roughness
arithmetic average of surface roughness

References
Amarnath M and Praveen Krishna IR (2012) Empirical mode
decomposition of acoustic signals for diagnosis of faults in
gears and rolling element bearings. IET Science,
Measurement and Technology 6(4): 279287.
Amarnath M and Praveen Krishna IR (2013) Detection and
diagnosis of surface wear failure in a spur gear system
using EEMD based vibration signal analysis. Tribology
International 61: 224234.
Amarnath M, Sujatha C and Swarnamani S (2012)
Experimental investigations of surface wear assessment
of spur gear teeth. Journal of Vibration and Control 18:
10091024.
Bartz JW (1993) Lubrication of Gearing. London: Mechanical
Engineering Publications Ltd.
Baydar N and Ball A (2003) Detection and diagnosis of gear
failure via vibration and acoustic signals using wavelet
transform. Mechanical Systems and Signal Processing
17(4): 787804.
Cameron A (1981) Basic Lubrication Theory, 3 rd edn.
New Delhi: Wiley Eastern Ltd.
Castro J and Seabra J (1998) Scuffing and lubricant
film breakdown in FZG gears Part 2. New PV scuffing
criteria, lubricant and temperature dependent. Wear 215:
114122.
Dowson D and Higginson GR (1977) Elasto Hydrodynamic
Lubrication. New York: Pergamon Press Ltd.
Ebersbach S, Peng Z and Kessissoglou NJ (2006) The investigation of the condition and faults of a spur gearbox using
vibration and wear debris analysis techniques. Wear
260(12): 1624.
Glodez S, Winter H and Stuwe HP (1997) A fracture mechanics model for the wear of gear flanks by pitting. Wear
208(12): 177183.
Halme IJ and Andersson P (2009) Rolling contact fatigue and
wear fundamentals for rolling bearing diagnostics state
of the art. In: Proceedings of the Institution of Mechanical
Engineers: Part J. Engineering Tribology vol. 225,
pp. 377393.

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

XML Template (2014)


[24.7.201411:40am]
//blrnas3/cenpro/ApplicationFiles/Journals/SAGE/3B2/JVCJ/Vol00000/140170/APPFile/SG-JVCJ140170.3d

(JVC)

[115]
[PREPRINTER stage]

Lee and Amarnath

15

Hohn B-R and Michaelis K (2004) Influence of oil temperature on gear failures. Tribology International 37(2):
103109.
Hohn BR, Michaelis K and Otto HP (2009) Minimized gear
lubrication by a minimum oil/air flow rate. Wear 266(34):
461 467.
Hoskins TJ, Dearn KD, Kukureka SN and Walton D (2001)
Acoustic noise from polymer gears: A tribological investigation. Materials & Design 32(6): 35093515.
Jacobson B (2003) The Stribeck memorial lecture. Tribology
International 36: 781789.
Lee SK and White PR (1998) The enhancement of impulsive
noise and vibration signals for fault detection in rotating
and reciprocating machinery. Journal of Sound and
Vibration 217(3): 485505.
Lee SM and Lee SK (2014) Objective evaluation of human
perception of automotive sound based on physiological
signal of human brain. International Journal of
Automotive Technology 15(2): 283290.
Li W and Mechefske CK (2006) Detection of induction motor
faults: a comparison of stator current, vibration and
acoustic methods. Journal of Vibration and Control 12(2):
165188.
Ma R, Chen Y and Cao O (2012) Research on dynamics and
fault mechanism of spur gear pair with spalling defect.
Journal of Sound and Vibration 331(9): 20972109.
Manoj V, Manohar Shenoy K and Gopinath K (2008)
Developmental studies on rolling contact fatigue test rig.
Wear 264(78): 708718.
Peng Z, Kessissoglou NJ and Cox M (2005) A study of the
effect of contaminant particles in lubricants using wear
debris and vibration condition monitoring techniques.
Wear 258: 16511662.

Parey A and Tandon N (2007) Impact velocity modeling and


signal processing of spur gear vibration for estimation of
defect size. Mechanical Systems and Signal Processing 21:
243.
Rafik Sari M, Haiahem A and Flamand L (2007) Effect of
lubricant contamination on gear wear. Tribology Letters
27(1): 119126.
Raja Hamzah RI and Mba D (2009) The influence of operating condition on acoustic emission (AE) generation during
meshing of helical and spur gear. Tribology International
42(1): 314.
Sato M and Takanashi S (1982) Thermo-elstohydrodynamic
lubrication of an involute gear. Tribology International
15(1): 2330.
Serrato R, Maru MM and Padovese LR (2007) Effects of
lubricant viscosity grade on mechanical vibration of
roller bearings. Tribology International 40: 1270 1275.
Tekner Z and Yesilyurt S (2004) Investigation of the cutting
parameters depending on process sound during turning of
AISI 304 austenitic stainless steel. Materials and Design 25:
507513.
Yesilyurt I (1997) Gearbox fault detection and severity assessment using vibration analysis, PhD Thesis, University of
Manchester, UK.
Yesilyurt I (2003) The application of the conditional moments
analysis to gearbox fault detection - a comparative study
using the spectrogram and scalogram. NDT & E
International 37(4): 309320.
Yesilyurt I, Fengshou G and Ball AD (2003) Gear tooth stiffness measurement using modal analysis and its use in wear
fault severity assessment of spur gears. NDT International
36: 357372.

Downloaded from jvc.sagepub.com at Indian Inst of IT, Design on October 9, 2014

Вам также может понравиться