2017 2nd International Conference on System Reliability and Safety
A Study on the Estimation of Bearing Life of Electric Motor Using ISO 281 and
Accelerated Life Test
Youn-Hwan Kim, Hae-Joong Kim, Jae-Won Moon
Rotating Machinery Center
Korea Testing Certification
Gunpo, South Korea
e-mail: younhwan@ktc.re.kr, motor@ktc.re.kr,
moon@ktc.re.kr
Abstract—This paper introduces the results of the study on the
development of accelerated life test methods for the motor used
in machine tools. In recent years, as well as efficiency for
motors, there is a growing need for research on life expectancy
of motors. It is considered impossible to calculate the
acceleration coefficient by increasing the rotational load or
temperature load as the acceleration stress in the motor system
because the temperature of the copper exceeds the wire
thermal class rating. In general, the failure of bearings in a
motor accounts for most of the total failure factor. This paper
describes the development of the procedure for the lifetime
estimation of the 12 kW three-phase squirrel-cage induction
motor. The test to predict bearing life was conducted and the
result was compared with the bearing life expectancy by ISO
281. Finally, a lifetime correction coefficient was calculated.
Keywords-accelerated life test; bearing; life expectancy;
lifetime correction coefficient; motor
I. INTRODUCTION
Machine tool is a machine for shaping or machining
metal or other rigid materials, usually by cutting, boring,
grinding, shearing, or other forms of deformation.
Machine tools are known to be the foundation of all
manufacturing. In particular, the importance of machine
tools in the automotive industry has a great impact on
component productivity and quality. Therefore, there is a
growing demand for durable machine tools in the
manufacturing industry and the importance of the reliability
evaluation is increasing [1].
In a machine tool, a motor converts electrical energy into
mechanical energy. The rotational speed of the motor shaft is
controlled by the inverter so that the machine tool functions.
Spindle system has huge effect on performance of machine
tools and working quality as well as is required of high
reliability. In the machine tool industry, a built-in spindle
with direct drive method is more popular than a conventional
belt driven spindle because of the wide range of torque and
output of direct drive motor.
Recently, studies on the reliability and accelerated life
test of a motor have been carried out. A study is the
calculation of remaining useful life for machine tool [2]. The
load condition according to the actual operation cycle usually
is not taken into consideration in the life test, so that the
actual life may be estimated differently.
978-1-5386-3322-9/17/$31.00 ©2017 IEEE
Youn-Hwan Kim, Sang-Yong Jung
School of Electronic and Electrical Engineering
Sungkyunkwan University
Suwon, South Korea
e-mail: younhwan@ktc.re.kr, syjung@skku.edu
In general motors, failure occur mostly in bearings and
windings. Bearing failure accounts for 37 to 60% of the fault
of the induction motor and windings failure are reported to
account for 16 to 37% of induction motor failure [3].
According to the literature, bearing failures account for more
than 80% of stepper motors [4]. Most of the failures of the
machine tool were found to occur in the motor bearings.
In this study, we analyzed the failure mechanism of a
motor used in a machine tool and proposed a method of the
bearing lifetime estimation.
II. TEST METHOD DESIGN
The motor for the test is shown in Table I below.
Through the Failure Mode and Effect Analysis [FMEA], the
bearings were identified as the most frequent failure and
critical component. As a result of analysis of the operating
environment, the rotation speed of the shaft at the time of
drilling was 6,000 rpm and the axial load was 150 N. The
radial load during face milling was 150 N.
TABLE I. ELECTRIC MOTOR FOR RELIABILTY TEST
Item Information
Motor Type Induction Motor
Output power 12 kW
Number of poles 4P
Voltage 380V
Base speed 6,000 rpm
Axial load 150 N
Radial load 150 N
Bearing arrangement O arrangement
Fig. 1 shows the bearing life prediction method in motor.
The first method is by calculation using bearing dynamic
load rating. This method is a method of bearing lifetime
calculation by ISO 281 (Rolling bearings: Dynamic load
ratings and rating life). It is applicable when the bearing life
prediction is possible. It is necessary to check the load
applied to the bearing in order to predict the life. A simple
method is to use the load distribution equation. It is difficult
to apply when a concentrated load is applied to the bearing.
223
If a concentrated load is expected, the load should be
estimated using the finite element method (FEM).
The second method is an accelerated life test by using of
motor specimens. In this case, since a large number of
samples and test equipment are required, a large amount of
motor capacity causes a lot of cost and long period of time.
Figure 1. Bearing Life Prediction Method in Motor
III. BEARING LIFE CALCULATION BY ISO 281
In 1947, Lundberg and Palmgren proposed a basic
calculation of nominal life of rolling bearings to calculate
If the same two or more ball bearings are used in series
combination, the basic dynamic load rating of a set of
combined bearings can be obtained by squaring the number
of bearings by 0.7. However, this calculation can be applied
when each bearing is subjected to an even load.
Figure 2. Bearing stress distribution
The analysis by FEM was carried out considering the
load applied to the actual machine tool. The load on the
bearing is not analyzed as uniform distribution. The
maximum concentrated load of 0.86 Mpa was confirmed.
Pr is a radial load with a constant direction and load
applied to the bearings that can achieve the same life as in
the actual load case. Pr of the radial bearing subjected to
constant radial load and axial load is obtained by (3).
bearing life. Based on this theory, ISO has established an
international standard for calculating bearing life. Bearing
life calculation theory has recently become more and more
necessary as the life of bearings becomes longer due to the
development of production technology. The revised ISO 281
proposed a method of calculating the nominal rated life as
well as the compensated nominal life. [5-7]. The basic
nominal life (L10) of a radial ball bearing is obtained by (1).
3 of the shaft.
Pr XFr YFa 
X is the equivalent radial load coefficient, and Y is the
equivalent axial load coefficient. The X and Y values were
selected in consideration of the nominal contact angle and O
arrangement conditions of the bearing.
Fr is the radial component of the actual bearing load, and
Fa is the axial component of the actual bearing load. Fr was
calculated by (4) using the weight of the rotor and the length
§C r ·  
L10 ¨ ¸
a b
F 
© Pr ¹ d
Fr FI 
b c  d II
Cr represents the basic dynamic radial load rating and Pr
represents the equivalent radial load. L10 means rating life It is calculated that the front bearing is weaker than the
associated with 90% reliability for bearings manufactured rear bearing in respect of the radial load.
with commonly used high quality material, of good
manufacturing quality, and operating under conventional
operating conditions. Bearings must be made using common
materials. Cr means constant stationary radial load which a
rolling bearing can theoretically endure for a basic nominal
life of 1 million cycles. In the motor, the bearings are
combined into an O arrangement. The number of identical
bearings combined is two. Therefore, the load rating of
single bearing was used for the calculation as (2).

Cr i0.7 u C 1
Figure 3. Conceptual diagram for load distribution calculation

224
 Fa is calculated assuming that the axial load of the
machine tool is evenly distributed in each of the two bearings.
As a result, the L10 was analyzed as 797,711 hours. Table II
is the value of each parameter used to calculate the bearing
life.
TABLE II. THE VALUES OF BEARING LIFE CALCULATION

Symbol Description Value

L10 Basic nominal life 797,711 hours

Basic dynamic
Cr 18,844 N
radial load rating
Pr Equivalent radial load 285 N
Load rating of single Figure 4. Motor life test equipment
Ci 11,600 N
bearing
Equivalent radial load Vibration sensors were installed to observe the defect
X 0.72
coefficient frequency of the bearing. When the outer ring is fixed and
Equivalent axial load the inner ring rotates together with the rotating shaft, if there
Y 2.39
coefficient is no slip between the raceway wheel and rolling elements
Radial component of and there is no deformation, the fundamentals of bearing
Fr 272 N
the actual bearing load synchronous signal averaging [8-9] are based on the
Fa
Axial component of
37.5 N
commonly utilized bearing kinematics (fault frequencies)
the actual bearing load equation as in (5).

In Table II, the L10 lifetime needs to be adjusted. This is

N D
because the load on the bearing is not a uniform load and the fi f (1 b cosD )
life of the bearing may be less than the expected value 2 r Dp

depending on the degree of environmental pollution. N D
Therefore, it is necessary to apply the life adjustment factor. fo f (1 b cosD )
2 r Dp
To find the life adjustment factor, bearing life were
recalculated by selecting axial load and radial load as the 1 D
fc f (1 b cosD )
acceleration stress because the experiment of 797,711 hours 2r Dp
is practically impossible. In order to calculate the axial load
and the radial load for the acceleration stress, the limit of where fi and fo are the inner race and outer race fault
bearing life equation must be confirmed. When a large load frequencies; fc and fr are the cage and shaft rotational
is applied, deformation may occur at the contact area frequencies, respectively. N is the number of balls; Db and Dp
between the ball and the raceway surface. In general, the are the ball and pitch diameters. Parameter ¢ is the contact
bearing lifetime calculation can be applied in case that Pr is angle in degrees between the raceway and balls.
less than 0.5 of Cr. The accelerated life test was carried out at
TABLE III. THE VALUES OF BEARING FAULT FREQUENCIES
15 times the actual load considering the test time.
Symbol Description Value
IV. EXPERIMENT AND EVALUATION
Inner race fault
For the accelerated life test, the test equipment was fi
frequency
1216.4 Hz
constructed as shown in the Fig. 3. Axial and radial load Outer race fault
devices were made to be able to apply loads with screws, and fo 1283.6 Hz
frequency
loads were measured by using the load cells. The load was fc Cage fault frequency 51.3 Hz
designed to be able to be applied and measured up to 3000N.
The test motor is the same as the drive motor. The drive
motor was also installed for the test because when the The fault occurred after 360 hours at 15 times load. This
bearing failure occurs, the torque sensor between the two is less than 1,067 hours of bearing life calculation at 15 times
motors detects the increased torque and stops the drive of the load. Therefore, the lifetime correction coefficient is
inverter. The torque before the failure is detected to be less estimated to be 0.337. It was regarded as a fault that the
than 1 Nm, and when the torque exceeds 6 Nm due to the inverter was stopped because the torque exceeding 6 Nm was
bearing failure, it is judged as a failure and the operation is detected. Failure frequency had been observed until the fault
stopped. with envelope analysis, and the vibration acceleration value
of the cage fault frequency was increased more than 10 times.

225
Therefore, it was predicted that there was a failure in the 1
cage.
Figure 5. Bearing envelope analysis
After the failure, the motor was disassembled and
observed. The failure of the bearings appeared as
discoloration and cage destruction. Discolored ball tracks
and balls are symptoms of lubricant failure. Excessive wear
of cage follow, resulting in overheating and subsequent cage
crack. It is predicted because the continuous lubrication of
the bearings is impossible.
Figure 6. Bearing Faults: Discoloration and Cage Crack
The lifetime correction factor measured by the test can be
applied to the lifetime at a load of 150N. Finally, the lifetime
correction factor was applied to the ISO 281 calculation
method and the lifetime of 26,894 hours was calculated.
V. STATISTICAL ANALYSIS
Statistical analysis of the results of the accelerated life
test is required for life prediction. In general, Weibull
distribution is used as the life distribution of mechanical
parts. The distribution function F(t) of the Weibull
distribution can be expressed as (6).
t
( )E
F (t) 1 e K
ln(ln( )) E lnt  E lnK 
1 F (t)
ƀ is a shape parameter that determines the shape of the
distribution and is described by the failure physics. η is a
measure of durability that can vary depending on the
material.
VI. CONCLUSION
The life expectancy of bearings used in motors was
estimated using the ISO 281 standard. Considering the load
acting on the actual motor shaft, the test time was shortened
by applying the load for acceleration life. Most first failures
in machine tools tend to appear in bearings. Considering
such assumptions, bearing life is considered to be the life of
the motor. The predicted life span can be used to determine
when to replace the motor or bearing. If more motor and
bearing specimens are tested and statistical analysis is
applied, it will be possible to verify the life calculation
method of this study.
ACKNOWLEDGMENT
This work was supported by the Ministry of Trade,
Industry and Energy, Korea, supervised by the Korea
Institute of Energy Technology Evaluation and Planning
(KETEP) (No. 20172010105710)
REFERENCES
[1] T. H. Lee, “A study on the failure and life assessment of high speed
spindle,” J.Korean Soc. Precis. Eng., vol. 31, no.1, pp.67-73, Jan.
2014.
[2] X. Du, “Remaining useful life assessment of machine tools based on
AHP method and euclid approach degree,” Intern. Confer. Syst.
Reliab. and Scie. pp. 46-52, Nov. 2016.
[3] F. J. T. E. Ferreria, “Reliability and operation of high-efficiency
induction motors,” IEEE Trans. Ind. Appl., vol.52, no. 6, pp.4628-
4637, Nov. 2016.
[4] A. D. Nisio, “Methodology for implementing accelerated life stress
on stepper motors,” 12th IMEKO TC 10 Workshop on Techical
Diagnostics, Jun. 2013.
[5] L. J Liang, “Study on bearing life calculation for wind turbine
gearbox,” J.Korean Soc of Manu. Proc. Eng., 13(5), pp.21-27 Oct.
2014.
[6] S. P. Lee. “Bearing life calculation by using the ISO281:2017,” KSPE
2012 Spring conference, pp. 903-904.
[7] ISO 281:2007 Rolling bearings - Dynamic load ratings and rating life.
 [8]
It can be transformed as (7) to draw a straight line on
Weibull probability paper. [9]
W. Wang, “Size estimation for naturally occurring bearing faults
using synchronous averaging of vibration signals,” ASME J. Vib.
Acoust. vol. 138, pp. 051015/1-10, Oct. 2016.
D.F. Shi, “Defect detection for bearings using envelope spectra of
wavelet transform,” ASME J. Vib. Acoust. vol. 126, pp. 567-573, Oct.
2014.

226