Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Engineering Failure Analysis 105 (2019) 110–126

Contents lists available at ScienceDirect

Engineering Failure Analysis


journal homepage: www.elsevier.com/locate/engfailanal

Failure analysis of gearbox in CRH high-speed train


T
Yongxu Hua, , Jianhui Lina, Andy Chit Tanb

a
State Key Laboratory of Traction Power, Southwest Jiaotong University, Chengdu 610031, China
b
LKC Faculty of Engineering and Science, Universiti Tunku Abdul Rahman, Jalan Bandar Sungai Long, Cheras, Kajang 43000, Selangor, Malaysia

ARTICLE INFO ABSTRACT

Keywords: In this study, an analysis was conducted to investigate how misaligned cardan shaft could result
Gearbox in failure of gearbox in CRH high speed train. The cardan shaft was used to provide rotating
Bearing failure motion from motor to gearbox. The study began with an analysis of the failure of a serviced
Shaft misalignment gearbox in CRH train. It was revealed that rotation of misaligned cardan shaft was the most likely
Cardan shaft
candidate for gearbox fracture which led to total failure of the train. To explain how misaligned
Universal joint
cardan shaft led to failure, a kinetic analysis of misaligned cardan shaft was first performed by
analytically calculating the force acted on gearbox and followed by an experimental test to va-
lidate the results. The results from kinematic analysis indicated a small offset of cardan shaft led
to huge inertial force and frictional force acting on the universal joints. These forces resulted in
bending load acting on bevel gear shaft and led to vibration. The vibration caused excessive
rubbing of the connection and resulted in looseness of the fit. Eventually the gearbox was frac-
tured by violent collision. An experiment was conducted to validate the failure process. The
experimental results confirmed the rotational misaligned cardan shaft caused impact between
shaft and bearing block. Consequently, this paper proposes a detection method to monitor cardan
shaft used in CHR high speed train. For the first time this work conforms it was essential to
consider transport acceleration when analyzing unbalance force due to misalignment of cardan
shaft.

1. Introduction

A specially designed cardan shaft in CRH high-speed train which connected the bevel gear shaft in gearbox and rotor of motor had
decreased the dynamic force between the wheel set and rail track [1]. The bevel gear shaft connected to the gearbox was used to
transmit torque through a pair of bevel gears. The transmission method required a certain structural stiffness of the gearbox which
was different from other gearboxes used in other kinds of CRH train. The structure of the transmission system was shown in Fig.1.
The gearbox assembly was a key transmission component of a high-speed train traction system and exceeding the permissible
stress limit was a serious problem. Tan [2] tested a kind of gearbox equipped on automobile and analyzed the cracks. The study
explained that imbalance of shaft resulted in alternating load. Fatigue cracks were easy to occur when the alternating load was
applied on the contact surface compressed by the preload, such as bolt and bolt hole. Bearings in gearbox supported shafts carried
huge axial and radical load. Once the bearings were damaged, the rotation of shaft supported by bearings would be unstable. Many
failure cases of bearing had been introduced, totally, fatigue failure and inadequate lubrication were main reasons leading to bearing
failure. H.Ilive [3] introduced a failure analysis of a thrust bearing supported the vertical shaft of a hydraulic turbine and generator.


Corresponding author at: Post Box 302, Jiulidi Campus, Southwest Jiaotong University, No. 111, North 1st Section of Second Ring Road, Jinniu
District, Chengdu City, Sichuan Province, China.
E-mail address: huyongxu51@my.swjtu.edu.cn (Y. Hu).

https://doi.org/10.1016/j.engfailanal.2019.06.099
Received 8 December 2018; Received in revised form 8 April 2019; Accepted 30 June 2019
Available online 02 July 2019
1350-6307/ © 2019 Elsevier Ltd. All rights reserved.
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Nomenclature ω Angular velocity of drive shaft


hconnecting Offset at axle of universal joint and the connecting
mp Mass of element P in cardan shaft shaft at drive side
aCoriolis Coriolis acceleration γ Rotating angle of drive shaft
IVdrive Unbalance value at drive side ldriven Offset at axle of universal joint and the driven
atransportn Transport normal acceleration shaft
IVdriven Unbalance value at driven side BC Line indicating axis of connecting shaft
arelativen Relative normal acceleration lconnecting Offset at axle of universal joint and the connecting
XE, YE, ZE Coordinate value of point E in coordinate system shaft at driven side
arelativeτ Relative tangential acceleration λ Parameter represents the position of point Q on
x .y .z X, Y, Z coordinate of a vector line BC
hdrive Offset at axle of universal joint and the drive shaft

Fig. 1. Structure of transmission system in CRH high speed train.

According to the analysis, misalignment of pad caused disruption of oil film which resulted in uneven wear of bearing surface. The
misalignment of pad had a disastrous effect on bearing performance. Yu and Yang [4,5] introduced a failure case of a grease-
lubricated roller bearing of an electric motor that drove a supply blower. Reason of the failure was the surface contact fatigue
between the rollers and the raceways. Loss of lubricating capacity and failure of the greases due to excessive temperature was
responsible for the fatigue failure. R.K. Upadhyay [6] explained over load led lubricant squeezed out between the contact area of
rolling elements and raceways, resulting in direct metal-to-metal contact that causes wear.
Misaligned cardan shaft led to additional force acting on the needle bearings and the bearing assemblies in the universal joints
[7]. The dynamics of misaligned shaft and balancing methods were widely studied [8–10], but theoretical dynamics studies of
misaligned cardan shaft were barely known until recently years. Bayrakceken [11] carried out a fracture analysis of a universal joint
yoke and a drive shaft of an automobile power transmission system. The study pointed out failures were occurred as a result of fatigue
process. The fatigue cracks began at the location of the joint yoke corresponded to highest stress points. Feng [12] studied the super-
harmonic resonance of the nonlinear torsional vibration of misaligned rotor system driven by universal joint. The study considered
both natural structure misalignment and actual error misalignment. It was found that jumping phenomenon and dynamic bifurcation
occurred when the rotating angular velocity of driving shaft was half of the natural frequency of the driving system. Masarati [13]
presented a comprehensive formulation of an ideal universal joint and carried out a real time simulation of universal joint. Lu [14]
studied the potential influence and dynamic response of assembly clearance at the neck of cross shaft. The research indicated that a
small clearance can have a big influence on the transient dynamic response. Štefan [15] analyzed the angular accelerations between
the drive shaft, central shaft and output shaft. The study confirmed the presence of angular acceleration in rotational connecting
shaft. The angular acceleration increased if the angle between central shaft and drive or output shaft increased. Desnica [16] analyzed
reliability of agriculture double universal joint shafts. According to their research, misalignment of the input and output shafts led to
the increase of inertial force. The moment of inertia, inadequate lubrication and incorrect use reduced reliability of cardan shaft.
Patel [17] investigated the influence of misalignment on the forcing characteristics of flexible coupling. The results showed that the
presence of misalignment caused additional force acting on the shaft during rotation and the force fluctuation leaded to higher
harmonics.
The aim of the study was to analyze and validate how misaligned cardan shaft could lead to gearbox failure. To achieve the goal,
theoretical force generated by misaligned cardan shaft acting on universal joints had been investigated and then validated by an
experiment. Then effect on bevel gear shaft due to bending load generated by rotational misaligned cardan shaft was analyzed. The
study explained vibration due to bending load was responsible for the failure of gearbox. Finally, this paper proposed an innovative
technique to improve the existing measurement method for testing unbalance of cardan shaft used in CRH high-speed trains.

111
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

2. Gearbox fracture failure analysis

The gearbox in CRH train is supported by the axle of wheel of the train, similar with the supporting structure of a gearbox in a car
which is front engine rear wheel drive. Reaction bar is used to reduce vibration. A bevel gear with transmission shaft is equipped in
the gearbox to transfer the torsion load. The bevel gear shaft is supported by bearing. A flange connecting the bevel gear shaft
utilizing interference fit is connected to the cardan shaft by bolts. An example of a failed gearbox discovered during maintenance is
shown in Fig. 2.
As shown in the above figure, the fractured part of gearbox housing is close to the flange and the gear shaft is extensively
damaged. The impact pits on flange surface revealed that impact of the flange and the gearbox was responsible for the fractured
housing. The fractured part of the gearbox is shown in Fig. 3. The bevel gear shaft is supported by a deep groove ball bearing. To
guarantee the sealing effect of the gearbox, the distance between the flange and gearbox housing was designed close to each other.
Once the gear shaft is out of alignment, the flange would impact with gearbox and led to the part close to the flange to fracture.
Inspection of the bearing showed that the cage and outer ring were fractured and the bearing assembly was seriously distorted. An
analysis confirmed the material of bearing is suitable and the bearing was well lubricated. On further analysis of the cardan shaft, it
was revealed that the connecting shaft was misaligned. The result of examination confirmed that the cross shaft shifted and the
wearing pad in the bearing sleeve was seriously damaged as shown in Figs. 4(a) and 4(b).
A typical cardan shaft consists of a drive shaft, driven shaft, connecting shaft and a pair of universal joints. The length of the
connecting shaft between the universal joints is not fixed and the motion of the shaft can adapt different relative angles between the
drive and driven shafts. Universal joint of cardan shaft contains cross shaft and bearing sleeves. The bearing sleeve can rotate around
cross shaft neck through needle bearings in it. Fig. 4(a) showed irregular pits of about 4 mm at the end face of the cross shaft and at
the bottom of the bearing sleeve. The rim of wear pad in bearing sleeve was completely worn off as shown in Fig. 4(b). The structure
of universal joint and detail structure in bearing sleeve are shown in Fig.5.
As Fig. 5 shows, the wear pad separates housing of bearing sleeve and needle bearings. The pits at bottom of the bearing sleeve
and the excessive wear region of wear pads indicates the cross shaft had shifted and result in impact with the bearing sleeve. Finally,
the shift of cross shaft in the bearing sleeve resulted in misalignment of the cardan shaft. As rotation of misaligned shaft led to large
bending load acting on bevel gear shaft, it resulted in the shaft vibration. The flange connected with the shaft impacted with the
gearbox housing finally resulted in fracture. The excessive load on the flange and the bevel gear shaft was responsible for the fracture
of the gearbox. The load generated by rotation of misaligned cardan shaft would be discussed next.

3. Kinetics of misaligned cardan shaft

3.1. Configuration of misaligned cardan shaft

As the drive and driven shafts are designed parallel with the drive system in CRH train, the motion of a misaligned connecting
shaft would rotate around the axles of the universal joints as shown in Fig. 6. The connecting shaft is misaligned due to the offsets in
universal joints, where the offsets are named hdrive, hconnecting, ldriven, and lconnecting,respectively. Under this situation, when the drive
shaft rotates, the axis of misaligned cardan shaft is no longer static but rotating as well and the length of connecting shaft also
changed periodically. The mentioned offsets are shown in Fig. 6, where symbol O refers to the intersection of drive shaft axle and
universal joint; symbol A refers to the center of drive universal joint; and symbol B refers to the intersection point of universal joint
with connecting shaft at the drive side. Symbol C refers to the center of driven universal joint and symbol D refers to the intersection
point of universal joint and connecting shaft. The driven shaft connects with universal joint at E. The drive shaft and driven shaft are
arranged parallel in the following discussion.

Fig. 2. Fracture failure of gearbox.

112
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 3. Structure of driven shaft assembly in fractured gearbox.

Fig. 4. (a) Damaged cross shaft and bearing bushing. (b) Abnormal and normal wear pad.

Fig. 5. (a) Structure of universal joint.

3.2. Kinetic analysis of misaligned cardan shaft

To derive the kinematic equations of connecting shaft, a coordinate system was built. The axes are plotted in red lines as shown in
Fig. 6. The origin of the coordinate is set at O, the X-axis is along the drive shaft axis, Y-axis is parallel with the direction of initial
direction of line AO, Z-axis is parallel with the direction of X-axis and Y-axis. When the drive shaft rotates with a certain mis-
alignment, it contained two parts of motion, one is the relative motion and the other is transport motion. As shown in Fig. 7, the
motion mode could be illustrated as a random point on cardan shaft named Q which rotates about two lines, one is line FG and the

113
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 6. Magnified view of misaligned universal joint.

Fig. 7. Schematic diagram of motion modes and acceleration components.

other is line GG'. Point F is the intersection of line AD and line FQ, which is parallel with line AB. Point G is the intersection of line
OE and line FG which is parallel with line AO. Then GG' is the line crosses point G and parallel with the axis of drive shaft, referred to
as X-axis coordinate. According to kinematics, point Q in rotation contains transport motion and relative motion, therefore accel-
eration of point Q includes transport acceleration, relative acceleration in normal direction, relative acceleration in tangential di-
rection and Coriolis acceleration as shown in Fig. 7. Point G' was the intersection of line GG' and a line crosses at point Q and is
perpendicular with line GG'. The direction of normal transport acceleration is along line QG'. The relative acceleration has two
components, normal relative acceleration which is along line QF and the other is tangential relative acceleration which is perpen-
dicular with plane GFQ. The direction of Coriolis acceleration is parallel with line OA. The expression of acceleration at point Q is:
n n
aQ = atransport + arelative + arelative + aCoriolis (1)

The expressions of acceleration components in Eq. (1) are illustrated in Appendix A. According to the deduced equations with
respected to accelerations, different relative positions between drive shaft and driven shaft lead to different effects. The relative
positions are described through relative angle between line OE and X-axis. The definition of the angle is shown as below:

Angle = arctan( YE2 + ZE2 /XE ) (2)

To discuss effects of different relative angles on accelerations, conditions of rotational speed 3300 RPM and offsets

114
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

hdrive = hconnecting = 0.1 mm, ldriven = lconnecting = −0.15 mm are chosen. The length of shaft is set 1.725 m which was same as the
length of cardan shaft equipped in CRH train. The point Q selected is 0.1725 m away from location B on line C. The angles discussed
are listed in Table 1.
The values in Table 1 represent the relative angles corresponding to different relative angular positions. The accelerations of
different relative angles are shown in Fig.8 using different line symbols listed in Table 1.
As shown in figure, different kinds of accelerations are marked with different colors. The transport acceleration remains almost
unchanged under different relative angles and distributed in a circle with different rotating positions. This is because the direction of
transport acceleration is stable and is always perpendicular with line GG', besides, the variation of distance between point Q and G' is
small. As a result, transport acceleration is insensitive to relative angle, but the other acceleration components showed certain
peculiar motions and sensitive to the relative angles. The amplitude of acceleration components under different angles between drive
shaft axis and driven shaft axis are shown in Fig. 9.
As shown in Fig. 9, excepted transport acceleration, all the other accelerations decreased to zero when the angle between axis of
drive shaft and axis of driven shaft was zero which indicate the axis of drive and driven shaft are coincidental (angle = 0). From the
above plots the values of transport acceleration (Y-axis) are much larger than the other kinds of acceleration. The reason for this
phenomenon is because the rotating distance of transport motion is much larger than the rotating distances in other motions. At the
same time, the speed of transport motion is higher than the other kind of motions. Take for example; under conditions of rotational
speed 3300 RPM and offsets hdrive = hconnecting = 0.1 mm, ldriven = lconnecting = −0.15 mm, when the relative angle of drive and
driven shafts was 4.4°, the root mean square (RMS) values of the four acceleration components at point Q are listed in Table 2.
The RMS values in Table 2 indicated transport acceleration has the largest magnitude (12.66 m/s^2) and is considered as primary
component when compared with the other accelerations. Consequently, it is also the main component which caused gearbox failure.
The results also showed that the other acceleration components are relatively small, between 0.033 and 0.49 m/s^2. The root mean
value of transport acceleration occupies about 95% of the overall acceleration. Hence, understanding the effect of accelerations on
cardan shaft is important in understanding the status of cardan shaft.

4. Dynamic analysis of misaligned cardan shaft

According to acceleration of all points on line BC which represented connecting shaft, the resultant force acting on the universal
joint could be determined. Considering the cardan shaft in CRH train consisted of two parts which allowed for relative movement of
the two shafts as shown in Fig. 10, total length of shaft was L and first part of connecting shaft was L'. As such, not only inertial force
was generated but also the frictional force acting on the universal joints should be considered.
The relative motion of two shafts resulted in frictional force acting on universal joints is due to length variation and changes of
direction of periodic force. A simplified mechanical model of generated inertial force and frictional force acting on universal joint is
shown in Fig. 11. As shown in the figure, two parts of shaft are connected at location H, the distance between location B and H is L'
and the distance between locations B and C is L, the length of overlap between the two parts is L". The contact force between them is
signified as Fcontact. The connecting shaft is equivalent to two cylinders, a solid shaft and the other a hollow shaft. Assuming the
radius of the solid shaft is R1 and radius of the hollow shaft is R1 to R2. Symbol P in the figure referred to a point of shaft in plane ∏
which crosses point Q and perpendicular with line BC. Considering the connecting shaft to be composed of micro mass elements and
defining material density of the shaft as ρ. The height of the element length of micro arc is rdθ. Length of the element is Ldλ; width of
the element is dr. So that mass of the element at point P is:
mP = rLd drd (3)
λ is a variable which varies from 0 to 1, representing different point on axis of connecting shaft. When λ = 0, point Q represents
location B, when λ = 1, point Q represents location C. The symbol dθ referred to angle of element in circle.
In rotation, element P rotates around line BC in plane ∏. According to kinetics, acceleration of point P is:
aP = aQ + aPQ (4)
The expression aPQ is acceleration of point P relative to point Q. Expression of aPQ is illustrated in Appendix B. According to
D'Alembert principle [18], the inertial force of element P is:
FP = mP aP (5)
Defining the unit vector along with line BC.

lBC
eBC =
| lBC | (6)

Table 1
Relative angle between driven shaft and drive shaft.
Line style o – .

Angle(degree) 0 2.2 4.4

115
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

(a)Transport accelera!on (b)Normal rela!ve accelera!on

(c)Tangen!al rela!ve accelera!on (d) Coriolis accelera!on


Fig. 8. Acceleration components under different relative angle.

Fig. 9. Amplitude of acceleration components against different angle.

Table 2
Root mean square values of acceleration component.
Acceleration name Value

Transport acceleration 12.66 m/s^2


Relative tangential acceleration 0.033 m/s^2
Relative normal acceleration 0.49 m/s^2
Coriolis acceleration 0.037 m/s^2

116
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 10. Structure of a cardan shaft.

Fig. 11. Mechanical model of misaligned shaft.

According to geometrical position of misaligned connecting shaft, direction of line BC is given by:

lBC
= (XE (lconnecting hconnecting ) lAB xYE + (ldriven hdrive ) cos ( ) (lconnecting hconnecting ) lAB yZE + (ldriven hdrive)
sin ( ) (lconnecting hconnecting ) lAB z ) (7)
The projection of force FP in the direction of line BC is:

FP //BC = (FP eBC ) eBC (8)


Defining a force F′ which is FP − FP//BC. It can be proved F′ is perpendicular to eBC which is unit vector of line BC as follow:

F eBC = (FP (FP eBC ) eBC ) eBC (9)

= FP eBC ((FP eBC ) eBC ) eBC

= FP eBC (FP eBC )( eBC eBC ) = FP eBC (FP eBC )| eBC|2

As the vector eBC is a unit vector, so | eBC|2 = 1, as a result:

F eBC = 0 (10)
So force F′ is perpendicular to line BC, force F′ is given by another name FP⊥BC.
FP BC = FP FP //BC (11)
So that force FP could be decomposed into components along with line BC and perpendicular to line BC. According to the
principle of mechanical equilibrium [19], the expression of the reaction force perpendicular to BC at location B and C are as follows:

117
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

FB BC = (1 ) FP
FC BC = FP (12)
The total reaction forces perpendicular with line BC at locations B and C are obtained by integration as follows:
L
L R1 2 1 R2 2
FB BC = (1 ) FP BC d drd + (1 ) FP BC d drd
0 0 0 L L R1 0
L
L
L R1 2 1 R2 2
FC BC = FP BC d drd + FP BC d drd
0 0 0 L L R1 0
L (13)
According to mechanics of internal force of shaft [20], the contact force between the two parts of the shaft are determined as
follow:
L / L R1 2
Fcontact = (1 ) FP BC d drd FB BC
0 0 0 (14)
Assuming the friction coefficient to be f, then frictional force is given by:
Ffrictional = f Fcontact (15)
Since the connection shaft consists of two parts, the inertial and frictional forces along BC should be considered separately. Hence,
location B bears the force along line BC at the first part of connecting shaft, and location C bears the force along line BC at the second
part of connecting shaft. The resultant reaction force along line BC at location B and C are determined by the principle of mechanical
equilibrium [19].
L / L R1 2
FB //BC = FP //BC d drd Ffrictional
0 0 0
1 R2 2
FC //BC = FP //BC d drd Ffrictional
L L R1 0
L (16)
Total reacting force at locations B and C are:

FB = FB BC + FB //BC
FC = FC BC + FC //BC (17)
Generally, the resultant force acting on the universal joint is the sum of inertial force and frictional force. According to the
parameters of transmission system of CRH train, the mass, length, and rotational speed of cardan shaft are listed in Table 3.
According to the parameters of cardan shaft listed in Table 3, inertial and frictional force of misaligned cardan shaft with different
offsets are listed in Table 4.
As listed in the table, an offset larger than 0.1 mm would lead to relatively huge inertial force and frictional force. The inertial
force at the driven side is greater because of the bigger offsets at driven side. The offsets shown in Table 4 increased linearly, offsets in
the second, third and fourth rows are 2, 3 and 4 times of the length of offsets listed in the first row. The ratio between the result of
each row and the result of the first row is listed in Table 5.
Results show that the ratio of calculated inertial force in the 2nd,3rd and 4th rows are 2,3 and 4 which is same with the increasing
ratio of the offsets length. The ratio of calculated frictional force in the 2nd,3rd and 4th row is lower than the increasing ratio of
offsets length. The inertial force is about 10 to 15 times larger than the frictional force. To determine the misaligned degree of the
shaft, a parameter named unbalance value is utilized. Unbalance value indicates the resultant force acting on the supporting device of
the rotating shaft divided by the square of angular speed. The detected unbalance results of the cardan shaft are listed in Table 6. The
results were obtained at a rotation speed of 1500 RPM, it can be seen the unbalance at the drive side and driven side are much larger
than the allowable unbalance of drive side and driven side which are 512 g.cm and 384 g.cm (refer to the G100 unbalance re-
quirement standard of cardan shaft of GEWES co., LTD.). The testing results indicate that there will be more than the allowable

Table 3
Basic parameters of cardan shaft in CRH train.
Mass Length of shaft Length of 1st part Length of 2nd part

95Kg 1.725 1.1 m 0.9 m


Radius of 1st part Radius of 2nd part Relative angle Rotational speed
0.05 m 0.066 m 4.4° 3300RPM

118
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Table 4
Root mean square values of inertial force and frictional force.
hdrive hconnecting ldriven lconnecting RMSinertial_B RMSinertial_C RMSfrictional

0.1 mm 0.1 mm −0.15 mm −0.15 mm 138.2 N 534.5 N 63.5 N


0.2 mm 0.2 mm −0.3 mm −0.3 mm 276.5 N 1069 N 98.4 N
0.3 mm 0.3 mm −0.45 mm −0.45 mm 414.7 N 1603.5 N 138.2 N
0.4 mm 0.4 mm −0.6 mm −0.6 mm 553.0 N 2138.0 N 179.7 N

Table 5
Ratio between calculated results.
Row number Ratio of RMSinertial_B Ratio of RMSinertial_C Ratio of RMSfrictional

2 2 2 1.55
3 3 3 2.18
4 4 4 2.83

Table 6
Detected unbalance results of the cardan shaft.
Unbalance at drive side Unbalance at driven side

1734.8 g.cm 4690.3 g.cm

magnitude of inertia forces acting on the supporting system of cardan shaft. The bevel gear shaft connected to the cardan shaft will
vibrate violently because of the bending load caused by the inertial force.
According to the method of balancing test, this requires the axes of drive shaft and driven shaft to be coincidental. Utilizing Eqs. 3
to 17, the relationship between unbalance and offsets at universal joints were derived as below.
m
IVdrive = (lconnecting + 2hconnecting ) 2 + (ldriven + 2hdrive )2
6 (18)

m
IVdriven = (2lconnecting + hconnecting )2 + (2ldriven + hdrive ) 2
6 (19)

Eq. (18) and Eq. (19) revealed that different offsets at universal joints produced different results. The relative angle between the
drive shaft and driven shaft affects the relative acceleration and Coriolis acceleration which in turn led to different inertial force and
frictional force. The offsets derived from unbalance test through Eq. (18) and Eq. (19) are shown in Fig. 12. The positive values of
offsets in the figure mean the direction of the offsets is along with the positive direction of the coordinate axis, and vice versa. As the
figure shows, according to the range of all axes in the figure, the possible offsets with respect to the result in unbalance test dis-
tributed from about −2 mm to 2 mm.
Take one of the results in Fig. 12, under the condition listed in Table 3, the spatial distributions of the inertial force and frictional
force were shown in Fig. 13.
As shown in Fig. 13, the direction of frictional force is almost in line with the direction of driven shaft and the direction of inertial
force was approximately perpendicular to the direction of driven shaft. As such the inertial force was responsible for creating the
bending load on the flange and bevel gear shaft in gearbox This confirms our hypothesis that the cause of gearbox failure is due to the

Fig. 12. Offsets derived from unbalance test result.

119
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 13. Distributions of Inertial force and frictional force in space.

bending load.
To analyze the force generated by different offsets, analytical results with offsets values sampled at 0.01 mm interval were applied
in the analysis. At a rotational speed of 3300 rpm, with the sampled offsets, angular velocity and different sample points were applied
to Eqs. 3 to 17. The maximum and minimum amplitude of force generated by the misaligned cardan shaft are calculated and shown in
Fig. 14 and Fig. 15.
The X-axis in the figures represent rotation angle of cardan shaft relative to initial position. As shown in the above figures, the
waveforms of inertial force and frictional force are similar to a sine wave. Inertial force wave has twice the frequency of frictional
force wave. The amplitude range of maximum inertial force is larger than the minimum inertial force. The amplitude fluctuation
range of minimum inertial force has some overlaps of amplitude fluctuation range of maximum inertial force. The amplitude range of
maximum frictional force is similar to the minimum frictional force and there is no overlap between the two frictional forces. It has to
be noted that the amplitude differences of inertial force were smaller than the frictional force, with the amplitude of inertial force
varied from 5600 N to 5640 N, while the amplitude of frictional force varied from 400 N to 650 N.
The bending load acting on bevel gear shaft and flange led to vibration [21,22,23]. As the distance between the flange and
gearbox housing was designed close to each other, the vibration of bevel gear shaft is transmitted to the flange which results in impact
with gearbox housing. Finally, the violent impact leads to the fracture gearbox housing of. The failure process due to the bending load
could be summarized in follow steps:

• Rotation of misaligned cardan shaft generates bending load acting on bevel gear shaft in gearbox of CRH train.
• The bevel gear shaft which is supported by deep groove ball bearing vibrates violently due to the huge bending load.
• As gearbox housing was designed very close to the flange, the flange impacted with gearbox housing due to vibration of bevel gear
shaft.
• The violent collision between flange and gearbox result in fracture of gearbox housing and loosen the interference fit between
flange and bevel gear shaft.

5. Experimental tests with simulated misalignment

An experiment was conducted to verify the looseness of interference fit caused by the rotation of misaligned cardan shaft. A
specially designed cardan shaft was used in the test and the experimental set-up was shown in Fig.16. The equipment used in the
experiment consisted of a misaligned cardan shaft, two bearing blocks, two bearings, two force sensors, a group of belt pulleys and a

Fig. 14. Inertial force of maximum and minimum amplitude.

120
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 15. Frictional force of maximum and minimum amplitude.

Fig. 16. Picture of the experimental set-up.

test bed. The drive shaft was driven by an AC motor using belt drive and this represents the motor side in CRH train. A cardan shaft
with flexible length connects the drive shaft and driven shaft. The rotation of driven shaft was caused by the rotation of cardan shaft
and this represented the gearbox side in CRH train. The drive shaft and driven shaft were supported by bearings with interference fit.
In order to simulate the drive system in CRH train, bearing block at drive side was designed higher than bearing block at driven side.
The drive shaft and driven shaft were parallel to each other. Two force sensors were installed under the driven shaft block and were
used to acquire the force signals acting on the block using IMC signal acquisition device. Basic parameters of experimental cardan
shaft in experiment are listed in Table 7.
As explained in Section 2.2, cardan shaft was misaligned because of the cross shaft shifted in bearing sleeve. In order to imitate
misalignment of cardan shaft used in CRH train, a specially designed universal joint as shown in Fig. 17 was used in the tests.
Components of the universal joint included a cross shaft assembly and connecting plates. The connecting plates were used as shaft
yokes and bearing sleeves to connect cross shaft with different parts of cardan shaft. As explained in Fig. 6, the misalignment was due
to shift of cross shaft in bearing sleeve, so that the circular concave slot in plates were designed different depth. Analysis of similarity
principle between these universal joints were illustrated in Fig. 18:
The reason leading to misalignment of connecting shaft was the serious wear of the wear pad in the bearing sleeve, resulted in the
center of cross shaft shifted position in bearing sleeve. This resulted in the axis of connecting shaft and the axis of cross shaft shifted
from the rotating center. To simulate this phenomenon, the depths of concave slot in movable plates were designed differently. As
shown in Fig,18, depth of concave slot mounting cross shaft in above plate is designed less than the concave slot in lower plate. In this
way, the center distance of cross shaft shifted down with a distance relative and the connecting shaft which simulate misaligned
similar to the actual misaligned cardan shaft. The offsets leading to minimum amplitude of inertial force were chosen in experiment.
The chosen offsets are listed in Table 8.
Fig. 19 show the force signals acquired from the experiment and by numerical simulations based on the above developed
techniques. The red curve referred to force data acquired from the experiment which represents the force acting on the universal
joints, while the blue curve refer to simulation results with misalignment.

Table 7
Basic parameters of experimental cardan shaft.
Mass Length of shaft Length of 1st part Length of 2nd part

12Kg 0.65 m 0.33 m 0.35 m


Radius of 1st part Radius of 2nd part Relative angle Rotational speed
0.024 0.03 4.4° 420RPM

121
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 17. Structure of universal joint used in the experiment.

Fig. 18. The actual cross shaft leading to misalignment and experimental cross shaft simulation.

Table 8
Offsets in experiment.
Offsets hdrive hconnecting ldriven lconnecting

Length 0.32 mm 1.5 mm 0.56 mm -2.2 mm

Fig. 19. Comparison between theoretical simulations and experimental results.

As shown in Fig. 19, the waveform of experimental force and simulated force are very similar to each other. The force generated
by the manufactured cardan shaft varied from about 200 N to 235 N which is almost similar with the simulated force. The waveform
and amplitude of the experiment validated the accuracy of the kinetics simulation model and dynamic analysis of misaligned cardan
shaft developed in this paper.
As gearbox of the CRH train was hard to reconstruct, bearing, bearing block and driven shaft without flange were used to verify

122
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

effect of misaligned cardan shaft leading to failure in experiment. The principle of experimental structure is similar with gearbox
assembly in the following respects:

• The failure mode of actual cardan shaft and experimental cardan shaft was the same.
• The driven shaft was supported by a deep groove ball bearing which is similar with the bevel gear shaft supported by a deep
groove ball bearing in CRH train gearbox.
• The driven shaft in the experiment was designed an appropriate diameter so that it was close to the housing of bearing block. This
design imitates the small distance between the flange and gearbox housing.

Inspection of housing of bearing blocks after the experiment are shown in Fig. 20.
The wear tracks on bearing block are shown in Fig. 20. The experiment confirmed that the bending load generated by misaligned
cardan shaft led to vibration of the driven shaft and resulted in direct contact and wear of driven shaft and bearing housing block. The
experiment also verified that rotation of misaligned cardan shaft affected the interference fit. In the experiment, the inner race of
bearing connecting with driven shaft could be treated as the flange connected with bevel gear shaft in gearbox because each of them
connected with the shaft through interference fit. So that the effect of bending load on interference fit between inner race and driven
shaft could reflect the effect on interference fit between flange and bevel gear shaft to some extent. The inspection of interference fit
between the driven shaft and bearing at beginning of the experiment is shown in Fig.21(a) and the distance measurement is shown in
Fig.21(b).
As shown in Fig.21(a), at the start of the experiment, the shaft shoulder of driven shaft is close to bearing inner race, and the
distance between shaft block and edge of driven shaft is about 15 mm. Then, the driven shaft was found slipping out from the bearing
and the distance measurement is shown in Fig. 22.
The distance is measured about 20 mm, it shows the driven shaft has begun to slip out from the bearing due to alternating load
generated by the rotation of misaligned cardan shaft. Through opening shaft block and measuring the distance after the experiment,
the interference fit was proved to have been affected. The inspection of interference fit and distance measurement are shown in
Fig. 23(a) and (b), respectively.
The final measured distance as shown in Fig.23(b) reached 23 mm and the driven shaft is found almost completely slip out from
the bearing. The results of experiment proved that under alternating load of rotating misaligned cardan shaft, interference fit fails and
the shaft would slip out from the mounting base.

6. Conclusion

An extensive investigation was conducted to understand how misaligned cardan shaft could lead to failure of gearbox deployed in
high speed train. Theoretical kinematic analysis was first employed to obtain the kinetics and dynamics of misaligned cardan shaft.
The inertial force and frictional force generated by misaligned cardan shaft were performed using numerical simulations. This was
followed by an experiment investigation to validate the derived theory of misaligned cardan shaft and the effect of misalignment on
interference fit. This paper can be concluded as follows:

(1). The offsets in universal joints led to misalignment of cardan shaft when in rotation and generated the main vibration mode.
Various acceleration components were generated due to misaligned rotating shaft and the transport acceleration was found to be
the dominant component and is of primary concern as it had the largest magnitude compared with other acceleration compo-
nents.
(2). The relative angle between the drive shaft and driven shaft affected the amplitude of the generated acceleration.
(3). The resultant force acting on the universal joints contained inertial force and frictional force. Their amplitudes were related to

Fig. 20. (a) Upper part of the bearing housing. (b) Lower part of the bearing housing.

123
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Fig. 21. (a) Examination of initial interference fit. (b) Measurement of initial distance.

Fig. 22. Measurement of distance at the second hour of experiment.

Fig. 23. (a) Examination of interference fit at end of test. (b) Measurement of final distance.

the amounts of offsets. Inertial Force led to huge bending load on flange and bevel gear shaft.
(4). The bending load acting on bevel gear shaft caused vibration. As distance between flange and gearbox housing was small,
vibration led to impact between flange and gearbox. The impact resulted in violent collision is responsible for enlarging the
interference fit of the shaft and fracture of gearbox housing

Acknowledgement

The authors acknowledge the support of Sichuan Province Science and Technology Support Program, China (Grant
ID:2016JY0047 and 2017JY0127) to conduct this research and their financial contribution has led to this comprehensive study.

124
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

Appendix A. Equations of acceleration aQ

Parameter ϕ was the angle between vector lAB and X-axis, the expression was:

(ZE cos ( ) YE sin ( )) cos ( ) + XE sin ( )


= arccos
(ZE cos ( ) YE sin ( ))2 + XE2 (A-1)
With offsets hconnecting and lconnecting, the connecting shaft must rotate around axle of universal joint with a small angle and the
expression of the small angle, named η was:

lconnecting hconnecting
= arccos
XE2 + (YE sin ( ) ZE cos ( ))2 2
(A-2)

The expression of vector lAB was:

(ZE cos ( ) YE sin ( )) cos ( ) + XE sin( )


lAB = (ZE cos ( ) YE sin ( )) sin ( ) sin ( ) + XE sin ( ) cos( )
(ZE cos ( ) YE sin ( )) cos ( ) sin ( ) XE cos ( ) cos( ) (A-3)

The vector lAB varies at different rotating angle γ, the directions of relative accelerations were related to direction of vector lAB .
The expressions of accelerations in Eq. 1 were listed below:
0
1 1
lAB . y lconnecting 1 hconnecting ldriven 1 hdrive cos ( )
n 2
atransport =
1 1
lAB . z lconnecting 1 hconnecting ldriven 1 hdrive sin ( )
(A-4)

lAB . x
2
n d
arelative = ( (lconnecting hconnecting ) + hconnecting ) lAB . y
dt
lAB . z (A-5)

lAB . y sin ( ) lAB . z cos( )


d2
arelative = ( (lconnecting hconnecting ) + hconnecting ) lAB . x sin ( )
dt 2
lAB . x cos ( ) (A-6)

0
d
aCoriolis = 2 ( (lconnecting hconnecting ) + hconnecting ) lAB . x cos ( )
dt
lAB . x sin ( ) (A-7)
λ was a variable varies from 0 to 1, representing different point on axis of connecting shaft. According to the derived equations,
relative acceleration and Coriolis acceleration were only affected by offsets at axle of universal joints connecting with the connecting
shaft. The direction of transport acceleration was perpendicular with the drive shaft.

A.1. Equations of acceleration aPQ

According to reference [24], rotating speed of connecting shaft around its axis which was BC in Fig. 12 was:
cos ( )
connecting =
1 sin ( )2sin ( ) 2 (B-1)
Angular acceleration was:

2 cos ( ) sin ( )2sin (2 )


connecting =
(1 sin ( )2sin ( )2)2 (B-2)
Symbol β was the angle between direction vector of drive shaft and direction vector of line BC.

lBC . x
= arccos
| lBC| (B-3)

125
Y. Hu, et al. Engineering Failure Analysis 105 (2019) 110–126

The acceleration aPQ was sum of tangential and normal acceleration. Defining unit vector of lAB was eAB . Vector of line PQ was:

lPQ = r lAB

cos ( ) + (1 cos ( ))( eAB x ) 2 sin ( ) eAB z + (1 cos ( )) eAB x sin ( ) eAB y + (1 cos ( ))
eAB y eAB x eAB z

sin ( ) eAB z + (1 cos ( )) eAB x eAB y cos ( ) + (1 cos ( ))( eAB y ) 2 sin ( ) eAB x +
(1 cos ( )) eAB y eAB z

sin ( ) eAB y + (1 cos ( )) eAB z eAB x sin ( ) eAB x + (1 cos ( )) eAB z cos ( ) + (1 cos ( ))
eAB y ( eAB z )2
(B-4)
Expression of tangential and normal acceleration were:

lBC × lPQ
aPQ = connecting r
| lBC × lPQ| (B-5)

n 2 lPQ
aPQ = connecting r
| lPQ| (B-6)
Expression of acceleration aPQ was:
n
aPQ = aPQ + aPQ (B-7)

References

[1] S. Leva, A.P. Morando, P. Colombaioni, Dynamic Analysis of a High-Speed Train, 57 (2008), pp. 107–119 1.
[2] Xiu-Qing Tan, L.I. Shun-Ming, Cracked Diagnosis and Analysis of Automotive Gear-Box, Equi- Pment Manufacturing Technology, (2007).
[3] H. Iliev, Failure analysis of hydro-generator thrust bearing, Wear 225–229 (PART II) (1999) 913–917.
[4] Z.Q. Yu, Z.G. Yang, Failure analysis of fatigue fracture on the outer ring cof a cylindrical roller bearing in an air blower motor, J. Fail. Anal. Prev. 12 (4) (2012)
427–437.
[5] Z.Q. Yu, Z.G. Yang, Fatigue failure analysis of a grease-lubricated roller bearing from an electric motor, J. Fail. Anal. Prev. 11 (2) (2011) 158–166.
[6] R.K. Upadhyay, L.A. Kumaraswamidhas, M.S. Azam, Rolling element bearing failure analysis: a case study, Case Stud. Eng. Fail. Anal. 1 (1) (2013) 15–17.
[7] N. Sad, A. Asonja, T. Serbian, D.M. Tehni, G. Milanovac, Reliability Model of Bearing Assembly on an Agricultural Cardan Shaft, No. January (2015).
[8] W. Zheng, C. Rui, J. Yang, P. Liu, Research on dynamic balancing simulation of rotary shaft based on ADAMS, IOP Conf. Ser. Mater. Sci. Eng. 307 (1) (2018).
[9] M. Lal, R. Tiwari, Experimental identification of shaft misalignment in a turbo-generator system, Sadhana Acad. Proc. Eng. Sci. 43 (5) (2018) 1–20.
[10] K. Huang, G. Chen, Q. Chen, Z.F. Zhang, Dynamic balance analysis of automobile drive shaft based on Solidworks, Appl. Mech. Mater. 215–216 (2012) 986–991.
[11] H. Bayrakceken, S. Tasgetiren, I. Yavuz, Two cases of failure in the power transmission system on vehicles: a universal joint yoke and a drive shaft, Eng. Fail.
Anal. 14 (4) (2007) 716–724.
[12] C.L. Feng, Y.Y. Zhu, D.S. Wang, Super-harmonic resonance analysis on torsional vibration of misaligned rotor driven by universal joint, Appl. Mech. Mater.
26–28 (50875259) (2010) 1226–1231.
[13] P. Masarati, M. Morandini, An ideal homokinetic joint formulation for general-purpose multibody real-time simulation, Multibody Syst. Dyn. 20 (3) (2008)
251–270.
[14] J.W. Lu, G.C. Wang, H. Chen, A.F. Vakakis, L.A. Bergman, Dynamic analysis of cross shaft type universal joint with clearance, J. Mech. Sci. Technol. 27 (11)
(2013) 3201–3205.
[15] K. Simulation, O. F. the, and C. Shaft, Faculty of Materials Science and Technology in trnava Kinematics Simulation of the Cardan Shaft for Investigation of the
Cardan Error in Catia v5, 24 (2016), pp. 159–166 39.
[16] A. Asonja, E. Desnica, Reliability of agriculture universal joint shafts based on temperature measuring in universal joint bearing assemblies, Span. J. Agric. Res.
13 (1) (2015) 1–8.
[17] T.H. Patel, A.K. Darpe, Experimental investigations on vibration response of misaligned rotors, Mech. Syst. Signal Process. 23 (7) (2009) 2236–2252.
[18] M. Izadi, S. Amit, Rigid Body Pose Estimation based on the Lagrange–d'Alembert Principle, 71 (2016).
[19] G. Gallavotti, On the mechanical equilibrium equations, Nuovo Cim. 57 B (1) (1968) 208–211.
[20] N.K. Baishev, V.N. Kononov, M.N. Tolstyakova, Using the program ‘matLab’ to solve the problems of mechanics: determining the calculated values of internal
forces in a statically indeterminate continuous beam from the actions of constant stationary and temporary Mobile loads by the matrix method, IOP Conf. Ser.
Mater. Sci. Eng. 463 (4) (2018).
[21] N. Driot, J. Perret-Liaudet, Variability of modal behavior in terms of critical speeds of a gear pair due to manufacturing errors and shaft misalignments, J. Sound
Vib. 292 (3–5) (2006) 824–843.
[22] N. Wang, D. Jiang, Vibration response characteristics of a dual-rotor with unbalance-misalignment coupling faults: theoretical analysis and experimental study,
Mech. Mach. Theory 125 (2018) 207–219.
[23] S. Ganeriwala, S. Patel, H. Hartung, The Truth behind Misalignment Vibration Spectra of Rotating Machinery #349 (Session 32), (1999).
[24] F. Schmelz, C.H.-C. Seherr-Thoss, E. Aucktor, Designing Joints and Driveshafts BT - Universal Joints and Driveshafts: Analysis, Design, Applications Eds.
Springer Berlin Heidelberg, Berlin, Heidelberg, 1992, pp. 79–169.

126

You might also like