Professional Documents
Culture Documents
Dynamic Featuresof Planetary Gear Systemwith Tooth Crackunder Different Sizesand Inclination Angles
Dynamic Featuresof Planetary Gear Systemwith Tooth Crackunder Different Sizesand Inclination Angles
net/publication/275088504
CITATIONS READS
65 1,300
2 authors, including:
Zaigang Chen
Southwest Jiaotong University
99 PUBLICATIONS 2,690 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Zaigang Chen on 26 May 2015.
Keywords: gear tooth crack, dynamic feature, mesh stiffness, planetary gear
1 Introduction because of the changes in the dynamic behavior of the gear sys-
tem, it is possible to detect and extract the feature signals corre-
Planetary gears are widely used in the transmissions of helicop-
sponding to the tooth faults. This is why the investigation on gear
ters, automobiles, aircraft engines, wind turbine transmission,
tooth fault has attracted so much attention. As one of the fre-
heavy machinery, and marine vehicles because of their advantages
quently encountered gear failure modes, crack at the gear tooth
over parallel shaft drives, like compactness, large torque-to-
root is also of concern and a series of research literatures on its
weight ratio, large transmission ratios, reduced noise, and vibra-
propagation have been published by NASA [27–34]. Some other
tions and high efficiency due to relatively smaller and stiffer com-
research work about gear tooth crack has been reviewed and dis-
ponents. That is the reason why studies on planetary gear
cussed in our previous paper [35]. In addition, up to now, only a
transmission systems attracted much attention, particularly in the
few published papers attempted to examine the impact of gear
two most recent decades. Investigations on the vibration- and
tooth faults on the dynamics of a planetary gear system due to its
noise-related dynamic responses caused by the inherited internal
structural complexity.
(mesh stiffness, transmission errors, etc.) and/or external (fluctua-
Kahraman [36] employed a computational model of a planetary
tions of load or speed, etc.) time-varying factors were and will be
gear set to study the influence of surface wear on the dynamic
going on. Previous studies on dynamic analysis of planetary gear
behavior of a typical planetary gear set by combining a wear
systems mainly focused on prediction of natural modes [1–4],
model that defines geometric description of contacting gear tooth
load sharing among planet gears [5–9], effect of time-varying
surfaces having wear and a deformable-body dynamic model of a
mesh stiffness [10–13] and transmission error [7,14,15], spacing
planetary gear set. Sparis [37] investigated the cracks on the
[4,16], mesh phasing [17–19] and backlash related nonlinear dy-
planet carrier plate using experimental data from ground and air-
namics [20–24], etc. Its dynamic performances, especially the
craft tests, and Sunyoung [38] analyzed the cracked failure of the
nonlinear characteristics, were highly concentrated. While in the
planetary carrier in terms of metallurgical and finite element anal-
area of investigations on planetary gear system with tooth fault,
ysis. The dynamic responses of a differential planetary system due
fewer published literature is available.
to the combination of backlash and tooth defects were studied by
Due to the possible influencing factors like inadequate lubrica-
Wu [24], where the interactions between the “chipped” sun gear
tion, inappropriate operating conditions, poor specifications, mate-
tooth and backlash were concerned by combining CAD models
rial defects, and manufacturing or installation problems, gear
with ADAMS to overcome geometric simplifications. Chaari [26]
tooth damage may be in presence with undesirable dynamic
simulated the dynamic responses of a planetary gear failure
behavior changes resulting in noise, acoustic emissions, and unac-
caused by tooth pitting and cracking by changing the phase and
ceptable performance characteristics, including serious reductions
amplitude of the square waveform mesh stiffness. Dhanasekaran
in the durability of the planetary system [25,26]. However, just
[39] analyzed crack failure on the sun gear of a planetary gear in
detail, including micro hardness (HRC), chemical properties, opti-
1
Corresponding author. cal microscopy (Core), and scanning electron microscopy (SEM),
Contributed by the Design Engineering Division of ASME for publication in the
JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received February 5, 2012; final
and drew a conclusion that crack occurred due to fatigue. Based
manuscript received October 2, 2012; published online March 28, 2013. Assoc. on the discussions above, there are few studies on the effect of the
Editor: Philippe Velex. gear tooth crack, as one of the gear tooth failure modes, on the
Crack1 (q0, Wc/W, q2, ac) (0,0,0,0 deg) (0.1,0.5,0,60 deg) (0.2,1,0,60 deg) (0.3,1,0.15,60 deg) (0.3,1,0.3,60 deg)
Crack2 (q0, Wc/W, q2, ac) (0,0,0,0 deg) (0.5,0.5,0,60 deg) (1,1,0,60 deg) (1.5,1,1,60 deg) (1.5,1,1.5,60 deg)
Crack3 (q0, Wc/W, q2, ac) (0,0,0,0 deg) (1,1,0, 0 deg) (1,1,0,20 deg) (1,1,0,40 deg) (1,1,0,60 deg)
Peak
Crf ¼ (24)
RMS
where, RMS, Kur, and Crf are the indicators of root mean square,
kurtosis factor and crest factor, respectively; x is the sampling
data with a length of n data points and x is the average of x. Peak
is the peak level of the raw time series.
With the statistic indicators, the relative severity of the gear
fault can be assessed by,
Xi X0
RXi ¼ 100% (25)
X0
where RXi is the indicator to present the severity of gear fault rela-
tive to the healthy gear case. X stands for the statistic indicators
defined in Eqs. (22)–(24). Subscript i refers to the gear tooth crack
Fig. 7 Ring-planet mesh stiffness (Krp) cases listed in Table 3, namely, i ¼ As, Bs, Cs, Ds, Hs or i ¼ Ap,
Bp, Cp, Dp, Hp, and subscript 0 means that X is the indicator corre-
Table 3 Natural frequency of the planetary gear train (unit: Hz) sponding to the healthy gear case.
Fig. 9 Radial vibration of planet gear No.1 for TRCS: (Bs), (Cs), (Ds) and (Hs) are correspond-
ing to the TRC2 listed in Table 1
Fig. 10 Trajectory of the sun gear center for TRCS: (Bs), (Cs), (Ds) and (Hs) are
corresponding to the TRC2 listed in Table 1
The amplitude and frequency of the impulsive signal in the trajec- observe the impulsive signals as displayed in Figs. 9(Bs) and
tory of the sun gear center is directly determined by the size and 11(Bp). The defect passing frequencies can be calculated, accord-
location (on the sun or the planet gear) of the tooth crack in ing to Eqs. (6)–(7), as 0.0214 s and 0.0571 s corresponding to the
engagement. Difference between the trajectory curves of the sun conditions of tooth crack on the sun gear and on the planet gear,
gear center for the cracks on the sun and the planet gears is a no- respectively, which coincides exactly with the results in Figs. 9
ticeable observation. These numerical symbols in Fig. 10(Hs) and 11 obtained by numerical simulation.
ranging from 1 to 14 and that in Fig. 12(Hp) from 1 to 7 are used However, an obvious discrepancy between the vibration wave-
to indicate the contact number when the cracked tooth is in mesh forms is observed for the two cases. For the case of a crack seeded
and they are determined by the ratio of the number of teeth of the in a tooth of the sun gear, the shape of the impulsive signal varies
sun, ring, and planet gears. When the gear tooth crack lies at the with contacting of the cracked tooth with different planets. Just like
early stage, namely the crack size is very small, it is difficult to what is shown in Fig. 9(Hs), the radial vibration of the planet No.1
Fig. 12 Trajectory of the sun gear center for TRCP: (Bp), (Cp), (Dp) and (Hp) are
corresponding to the TRC2 listed in Table 1
appears to be different when the faulty tooth contacts with different addition, the statistical indicators (root mean square (Rms), kurto-
planets. While for the case of a crack planted in the planet No.1 sis, and crest factor) defined in Eqs. (22)–(25) are applied to detect
gear, the profile of the impulsive vibration signal keeps the same the presence of the gear tooth crack and shown in Figs. 13–14.
when the cracked tooth is getting into mesh with the sun gear each Seen from the figures, amplitudes of all the three statistical indica-
time. The reason for the discrepancy between the shapes of the im- tors ascend gradually with the growth of gear tooth crack and the
pulsive signals in the two cases can be attributed to that whether the inclination angle. When the crack is at its initial stage, such as the
mesh position when the cracked tooth in mesh relative to the place crack case Bs and Bp, it is hard to detect its presence due to the little
(say the center of the planet gear No.1) where the data are collected change of these statistical features. Another phenomenon can be
varies each time or not. These findings can be the theoretical observed that the crest factor (Crf) is more sensitive to the propaga-
approaches to distinguish the faults on sun gear or on planet gears. tion of gear tooth crack than the other two indicators while the indi-
For the case of Crack3, namely the effect of the inclination cator; Rms has the lower sensitivity. This is because the crest factor
angle of crack, it has a similar influence on the mesh stiffness is very sensitive to the impulsive signals, while for the root mean
which is shown in Fig. 5, and the similar results can be obtained square, it represents the average level of the vibration energy which
as discussed above. As such, the results will not be plotted in time is insensitive to impulsive signals. This is especially for the detec-
domain and frequency domain in this study except for the statisti- tion of the incipient tooth crack shown Fig. 13(a) which exhibits a
cal indicators discussed subsequently. good performance of the crest factor in detecting the crack at early
Statistical features are the commonly used indicators in mechani- stage. It coincides with the state in Ref. [54] that crest factor could
cal fault detection to measure the vibration severity [35,51–54]. In indicate the damage in an early stage. What is shown in Fig. 13(b)
Fig. 14 Effect of crack inclination angle on the statistic (for crack cases in Crack3)
demonstrates that only a relatively large tooth crack could result in root of the planet gear, all the amplitudes of the sidebands
an obvious increase in the amplitudes of the indicators of kurtosis increase with the propagation of the crack and they are shown in
and Rms. Based on the foregoing discussion, the indicator of crest Fig. 16. The interval between the adjacent sidebands is equal to
factor is recommended for the gear tooth crack detection, especially 17.3 Hz, which is the defect frequency calculated by Eq. (7).
for the incipient ones. While for the case of the crack on the sun gear, as displayed in
Influence of the gear tooth crack on the frequency structure is Fig. 15, the amplitudes of the low frequencies (<1800 Hz) have a
also investigated in this study. For visual convenience, only the drastic increase in the presence of gear tooth crack. Furthermore,
crack cases As, Bs in Crack1 and Ds in Crack2 for the tooth root the amplitudes of some local frequency components do not
crack happened to the sun gear and Ap, Bp in Crack1 and Dp in increase with the growth of tooth crack, such as the frequency
Crack2 for the tooth root crack seeded in the planet gear are ana- range of 5300–5580 Hz in Fig. 15(a)). A noticeable observation
lyzed in this paper, which stand for the healthy condition, incipi- can be found that the mesh frequency and its harmonics are not
ent crack and moderate crack, respectively. The frequency modulated by the fault passing frequency of 46.67 Hz obtained
spectrums of the radial vibrations of the planet gear No.1 are from Eq. (6), but by the equivalent rotational frequency of the sun
shown in Figs. 15–16. In Fig. 15 (16), the dotted solid curve gear fe ¼ fs–fc ¼ 11.67 Hz (see Fig. 15(b) and 15(c)). For the low
stands for the crack case As (Ap) indicating the healthy condition, frequency range about lower than 1800 Hz, the sidebands corre-
the thick solid curve for the crack case Bs (Bp) in crack 1 and the sponding to the frequencies f ¼ fm 6 n fe (n ¼ 2, 4, 6) are sup-
slender solid curve for the crack case Ds (Dp) in crack 2. pressed (see Fig. 15(b)), where fm refers to the mesh frequency
When the planetary gear train is in a healthy status, the mesh and its harmonics. The appearance of such a phenomenon is due
frequency and its harmonic components dominate the vibrations to the varying path from the impulsive exciting source caused by
which are shown by the dotted solid curves and there is no side- the tooth crack to the radial direction of the planet No.1 gear
band available. When a tooth crack appears, there are a group of where the vibration data are acquired. Another suggestion from
sidebands around the mesh frequency and its harmonic compo- Figs. 15–16 is that the amplitude of those sidebands within the
nents whose amplitudes are hardly affected by the tooth crack. low frequency range, say lower than the fourth mesh harmonic, is
The same phenomenon, that is to say the amplitudes of the side- prone to be affected by the crack, and those sidebands are more
band frequency components rather than the mesh harmonics, are suitable for the fault diagnosis analysis.
much more sensitive to the crack propagation, has also been found Some noticeable results have been observed in Fig. 15(b) that
in Ref. [35] for the dynamic features of a spur gear pair in the some sidebands around the first several mesh harmonics are sup-
presence of a tooth root crack. When a crack seeded in the tooth pressed. In order to verify if it also happens to other planetary
Fig. 16 Frequency spectrum of radial vibration of planet No.1 when a crack seeded in the planet gear
gear sets, the same simulation and analysis are also implemented its dynamic responses. Tooth cracks with different sizes and incli-
with the planetary gear system with different number of planets, nation angles are seeded in one tooth of the sun gear and planet
say, 3, 4, 5, and 6 planets. To meet the requirement for the assem- gear, respectively. For more practical simulation, the angular-
bly of planetary gear with equally spaced 3 and 6 planets, the position-dependent mesh stiffness is utilized in this study. The
number of teeth of the planet gear and the ring gear are changed simulated results show that in the presence of the gear tooth crack,
to be 21 and 72, respectively. Moreover, other parameters are the severity of the vibration of the planetary gear train is increased
assumed to be the same as in Table 2. This little change in the based on the statistical analysis and the sidebands around mesh
number of teeth has a negligible effect on the mesh frequency and frequency and its harmonics appear in the frequency spectrum.
rotational frequency of the planetary gear set. The frequency spec- When a tooth crack is seeded in one tooth of the sun gear, the im-
trums around the fundamental mesh frequency with different pulsive vibrations of a planet gear caused by the cracked tooth in
number of planets are shown in Fig. 17. It can be seen that there mesh with the different planets appear to be different from each
are also some sidebands suppressed which vary with the number other, which makes the mesh frequency and its harmonics modu-
of planets. An interesting conclusion about which sidebands will lated not by the so-called defect passing frequency, but by the
be suppressed or not can be obtained that these sidebands with equivalent rotational frequency of the sun gear. In the low fre-
f ¼ fm 6 n fe will not be suppressed, where fm, fe are the mesh quency range, some frequency components are suppressed, and
frequency and equivalent rotational frequency of the sun gear and their relationship with the number of the planets is obtained; this
the index n should meet the condition in Eq. (26), is useful for the selection of the monitored sidebands. Simultane-
ously, observing the first order sidebands beside the fundamental
ni þ niþ1 ¼ i N; i ¼ 1; 2; :::; Zs 1; n1 ¼ 1 (26) mesh frequency is suggested for the detection of the localized
gear tooth faults in the planetary gear set due to its presence de-
where, Zs is the number of the teeth of the sun gear. spite of the number of planets. In contrast, the vibration in the
This phenomenon also results from the modulation due to the time domain and the frequency distribution appear to be much
relative motion between mesh position of the cracked tooth and more regular when a tooth crack happens on the planet gear
the place where the vibration data are collected. This finding ena- where the modulating frequency is exactly equal to the predefined
bles the selection of the proper sidebands for monitoring the se- defect passing frequency. The above-mentioned discrepancies
verity of the tooth crack on the sun gear, which can be extended between the two cases, namely a crack seeded in the sun gear and
to other localized gear tooth faults in the planetary gear set. in the planet gear, will be expected to be applied to fix where the
Another valuable observation is concerned with the first order failure happens. Comparisons are made among three different sta-
sidebands around the fundamental mesh frequency which always tistical indicators and the results reflect that the crest factor is
appear in the spectrum in spite of the number of the planets. This more sensitive in detecting the gear tooth crack in the planetary
is why the sideband level factor [54] was defined to detect the gear train.
gear tooth damage based on the first order sidebands.
Acknowledgment
5 Conclusions The authors are grateful for the financial support provided by the
A coupled planetary gear dynamic model with the tooth crack National Natural Science Key Foundation of China under Contract
model is developed to investigate the effect of gear tooth crack on No. 51035008 and the China Scholarship Council (CSC).
DownloadedViewFrom:
publicationhttp://vibrationacoustics.asmedigitalcollection.asme.org/
stats on 04/09/2013 Terms of Use: http://asme.org/terms