A Comprehensive Condition Monitoring of Modern Railway
S.L. 101, K.K. Lee', K.Y. Lee2, H.Y. Tam1, W.H. Chungt, S.Y. Liul, C.M. Yip1, T.K. Ho1
'Dept. of Electrical Engineering, Hong Kong Polytechnic University, Hong Kong
2Kowloon-Canton Railway Corporation, Kowloon, Hong Kong
Keywords: Condition Monitoring, FBG Sensors
Abstract
The demand for high quality rail services in the twenty-
first century has put an ever increasing demand on all rail
operators. In order to meet the expectation of their patrons,
the maintenance regime of railway systems has to be
tightened up, the track conditions have to be well looked
after, the rolling stock must be designed to withstand
heavy duty. In short, in an ideal world where resources are
unlimited, one needs to implement a very rigorous
inspection regime in order to take care of the modem
needs of a railway system [1]. If cost were not an issue, the
maintenance engineers could inspect the train body by the
most up-to-date techniques such as ultra-sound
examination, x-ray inspection, magnetic particle inspection,
etc. on a regular basis. However it is inconceivable to have
such a perfect maintenance regime in any commercial
railway. Likewise, it is impossible to have a perfect rolling
stock which can weather all the heavy duties experienced
in a modem railway. Hence it is essential that some
condition monitoring schemes are devised to pick up
potential defects which could manifest into safety hazards.
This paper introduces an innovative condition monitoring
system for track profile and, together with an instrumented
car to carry out surveillance of the track, will provide a
comprehensive railway condition monitoring system
which is free from the usual difficulty of electromagnetic
compatibility issues in a typical railway environment [2].
1 Detection System
There are many aspects in trains that needs condition
monitoring. The paper describes two essential aspects, one
on the rails and one on the trains. Both systems to be
described monitor the strains using optical sensors. It is
well known that the use of conventional strain gauges can
be highly corrupted by noise. However, optical sensors are
immune to electromagnetic fields. Furthermore, many
optical sensors can be fabricated onto one single optical
fibre and this is different from conventional sensors which
require two independent wires and a pre-amplifier for each
sensor [3]. In other words, if one needs to monitor twenty
points on the track (Fig. 1), one needs only one single
optical fibre linking all twenty sensors. The same
argument is applicable to the optical sensors installed on
the train body (Fig. 2). For conventional strain gauges one
would require forty wires and twenty pre-amplifiers if 20
sensors are installed. Notwithstanding the fact that the
optical fibre and the sensors are as thin as human hair
125
Authorized licensed use limited to: IEEE Xplore. Downloaded on February 26, 2009 at 04:30 from IEEE Xplore. Restrictions apply.
(excluding the mechanical reinforcement, of course), one
can therefore easily understand why the authors believe
that the future railway sensors should be optical rather than
electrical.
2 Measurements using Optical Sensors
A Fibre Bragg Grating (FBG) sensor is an optical device
that measures strains by means of detecting changes in the
wavelength of light. A FBG sensor can be fabricated by
exposing the sensors to ultra-violet light through a face
mask as shown in Fig. 3.
UV beams
U HEEHE Phase
Mask
111111 Optic Fibre
Fig. 3 Fabrication of Optical Sensors using Ultra-violet lights and masks
The FBG sensor consists of a short length of periodic
refractive-index changes inside an optical fibre which is as
thin as human hair. A light source is used to pass a band of
light spectrum into the sending end of the optical fibre.
Without the FBG sensor, the light passes through the
optical fibre un-obstructed. When there is a FBG sensor, a
narrowband of wavelength of the light spectrum is
reflected back to the sending end and these reflected
wavelengths are analyzed by an optical interrogator. The
light reflected back has a spectrum that characterizes the
pitch (e.g. the separation between two periodic marks) of
the periodic refractive-index variation. The pitch is
changed when the FBG is subjected to strain (i.e. the pitch
becomes longer, when the FBG sensor is being extended,
and becomes shorter when the FBG sensor is being
compressed), resulting in a corresponding change in the
reflected wavelength. As the parameter of measurement is
the wavelength of light, the process is immune to
electromagnetic interference and hence is intrinsically
more stable than any electrical monitoring system in an
electromagnetically noisy environment which is typical in
an electrified railway.
Fig. 4 shows a typical pickup from a sensor installed on
the track with the passage of a 12-cars train. It is noted that
there are 4 wheels in one car and hence the first 4 peaks on
the left-hand side correspond to the first car. It can be seen investigation revealed that there were roundness problems
that there are more vibrations in Car 9 (when counting in the wheels of Car 9 and this will be described in more
from the left hand side) and such observation guides the details in subsequent sections of this paper.
engineer to pay attention to the cars. Subsequent

Seve al t B(.W Sensort w tahric ated onto one single c iitical fibrce

--\<04 SlD0 >Xit:.l0

4..saums.a..a----a..u
r
2 a a a I
I U I
_ S I an I
,_ I__ I I . I . I I .

58 Sf s

7M
II Is1 1 1I
m
I1IF
0 m
I I I I I I I I I I I I I I
m

I I I I I I I I I
A\ typical Sensor

x-vithont any plh'\N'sical

p rotect ion

Fig. 1 Optical Sensors mounted on the rail track

........................... . . . ...

FBG Sensor-s mioun-ted on the

undlerfranie equipmients to
detect high str-ess points

D " VI|I',pr1J fI
1 q| §|§ § g } ?
11 .t'I.I
tj
. I.
^ ~~ ~ ~ ~ u t21
,-Mm.

; Strain readings of one of the

FBG Senso-rs
Fig. 2 Optical Sensors mounted on trains

126

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 26, 2009 at 04:30 from IEEE Xplore. Restrictions apply.
 Strain Signals of a noisy tirain in Car 9 (counted from left-hand side)

1
ltulid;
jl 1 21 H f~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~-4i4
Si1
A j 1544M

cz
, ..
Strain Signals of a normal and healthN train
a .r

,AO,

t t , '^
4g:3Ee
t.,7:"t.:144
Figt. 4Straispickeupby1p
tica'l4en sorsm
4 4 o4u.nted
t
e, v.
1 ote rack
w,.
1i ;
zt t:
..4 . 114
0 i
t1he pasag
'4

of1a
5

car
144 tra

Fig. 4 Strains picked up by optical sensors mounted on the track with the passage of a 12 car train

FL TWO *okph. ISMI Mwarl 2006 Eqs*poewt case Z1 FL TWO a Ospn¶. IWb March 2006

IT!
~ ~ ~ ~ ~ ~ ~. . . . . . ,K .w I:o ....:-,
>l.:.v:.:.: .. ~.r::.v:::>:. :.::::.:...

Fig. 5 Strain on the metal case which is about 18 mm (i.e. 22 mm- 4 Fig. 7 Strain on the metal case which is very near thie welded bracket
mm) from the welded bracket

FL TWO okrpt 150f March 2006 Eqiupwert case 22

Fig. 6 Strain on the metal case which is about 9
mm - 5 mm) from the welded bracket
Apart from being immune from electromagnetic
interference, these sensors could be made very small and
hence a string of these sensors could be connected together
in a string to measure how the strain patterns are varying
in the vicinity of cracks or points suspected to be suffering
from high strains. Figs. 5 to 7 show the strain
measurement of 3 FBG sensors in the vicinity of a weld
which is suspected to be suffering from high strains in a
typical train. All these 3 FBG sensors are connected in
series in one single optical fibre and many measurements
have indeed been carried out. The length of each sensor in
this 3-senor pack is around 4 mm and each sensor is
separated from each other by another 5 mm. In other
mm (i.e. 22 mm - 4 words, the total length of the three sensors is (4+5+4+5+4)
mm or 22 mm.

127

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 26, 2009 at 04:30 from IEEE Xplore. Restrictions apply.
3 Applications The local vibrations such as those in Cars 4 and 10
(counting from the left) of Fig. 8 have been found to be
3.1 Monitoring of Imperfections in Train Wheels higher than those of the other cars in the same train.
With the installation of the sensors on the tracks, a wealth It can be seen from Fig. 9 that the vibration index for most
of investigations could be carried out readily. For example, cars are below 2. However the vibration index for Cars 4
from the strain measurements at the track it was confirmed and 10 are higher than 4.
that there were some noisy trains. It was suspected that
these noisy trains might have relatively imperfect wheels. The wheel out-of-roundness of the noisy cars (car No.4 &
Hence a series of tests were carried out. The first test is to 10) were measured and the results are shown in Table. 1.
identify a noisy train (a train producing strains as shown Table. I Wheel out-of-roundness measurement for the noisy cars under
similar to the one in Fig. 4) using a FBG sensor installed investigation.
on the rail track as described in the previous section.
Wheel out of roundness (mm)
The principle of wheel imperfection detection by FBG Car no.4 Left Right
strain sensors installed at track is based on the fact that Axle
wheel defects such as flange pits, wheel flats and 1 0.4 0.5
particularly out-of-round wheels which are also known as 2 0.4 0.6
3 0.4 0.4
polygonal wheels, will exert periodic impact force on the 4 0.4 0.4
track. In this work, it was founded that an imperfect wheel Wheel out of roundness (mm)
will produce an uneven strain impulse on the track. In Car no. 10 Left Right
contrary, a newly turned wheel will produce a symmetrical Axle
strain impulse. In another approach, one attempts to relate 1 0.8 1.00
2 1.00 0.75
the inter-axial vibrations as shown in Fig. 8 with wheel 3 0.7 0.7
out-of-roundness. In order to compare parameters from 4 0.8 1.2
two different systems, the inter-axial vibrations were
quantified by a vibration index which is obtained The wheels on car No.10 were turned right after the out-
arithmetically by considering the train speed, vibration of-roundness measurement while the wheels on car no.3
frequency and magnitude as a whole. Fig. 9 shows the were kept unturned as a control. The wheels out-of-
vibration indexes obtained from a noisy train. roundness after tuming were between 0.05 to 0.07mm.
Strain Pattern for a Noisv Car The strain pattern and vibration indexes of the serviced
train are shown in Fig. 11.
0.1
Noise From Fig. 11 it can be observed that the vibration at car
no. 10 was eliminated after wheel turning while the
vibration at car no. 4 persisted. By comparing Fig.9 with
Fig. 11, it can be seen that the vibration index of Car No. 4
(unturned wheels) is around 4.4 while the index of Car No.
10 (turned wheels) was greatly reduced from 4.7 to 1.8.
6 1 4 1 6.
-4 2.
2l
This shows that the vibration index, which is deduced
from the FBG strain sensor measurement results, is an
Fig. 8 Strains picked up on the track due to the passage of a noisy Train effective means to distinguish out-of-round wheels from
good wheels.
V1ibatloll IiOCdex
After W87heel Turning
j~~ '227 'kph

0.70-
0 L-

CIA

1"
0- 1 1 P 1 1l,r1i
11 I1 21 .1.. - 2
2 A4 5 7 1I 11 12 1
,i 6 7..:
9 9 |0 II 1 '1.;'? I)8i 18'g 20 2'1 2-' 2 91 X
Tiff On)

Fig. 9 Vibration index of the 12 cars

Fig. 10 The strain readings after wheel turning

128

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 26, 2009 at 04:30 from IEEE Xplore. Restrictions apply.
 "V1Ibration index
0-.
6 -lo
Left sole baw FFT-
-290

5- -400

-60 0
4-
-80 0-

-100 0-

2-
-120,09

-140 0-
-150.0-
.....

18.(H).) d30. 31.2

Fnsueny
~~~~~~~~~~~~~~~.....................
Nurik 1 2 3 4 5 6 7 0 9 10 1.1 12 SuiLt Fig. 14 Strain picked up on the sole bar with old suspension system
FCir
Fig. 11 Vibration index of the train with the wheels car 3 turned

For a typically healthy train the strain measurements

picked up and the corresponding vibration index are
shown in Fig. 12 and 13, respectively.
Normal Strain Pattern

07~~~~~~~~~~~~~~~~~~~~-

.En'1~ ~ ~ ~ ~ i 'i l'" ei' gi l..

0.3-
Fnqomncy Mz)
027
Fig. 15 Strain picked up on the sole bar with new suspension system
01-

00-

2 1 2c232
4 Conclusion
Fig. 12 Strains picked up on the track due to the passage of a normal
Train A very powerful condition monitoring system using FBG
sensors to monitor the conditions of rail tracks and train
Vibration Index borne equipment have been developed and tested on a
b -
heavily used mainline railway.
5.-
5 Acknowledgement
4

The authors wish to express their grateful to KCRC for

supporting the investigations and allowing the authors to
2- publish the work as described in this paper.

6 Reference :
0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
-I 1 2 -. ;4
0 ------I t
Snnti Fr 7 8 9 10 11i Ntth
[1] Weston, P.F., Goodman, C.J., Li, P., Goodall, R.M.,
5

Fig. 13 Vibration index of a healthy train
3.2 Monitoring of Strain on Train Borne Equipment
The strains of the various equipment on another train were
measured with i) the old suspension system and ii) a
totally new suspension system. The spectral density of the
strains measured as given in Figs. 14 and 15.
It can be seen from these two figures that there were
improvements after changing the suspension system and
hence a very thorough investigation with new components
being replaced one at a time was carried out. Very useful
conclusions could be drawn from the measurement data
although the authors cannot reproduce such results here
due to commercial considerations.
Ling, C.S., Roberts, C., Track and Vehicle Condition
Monitoring During Normal Operation Using
Reduced Sensor Sets, HKIE Transactions, Special
Issue on Railway Development in the 21' Century,
March, 2006, pp. 47-53.
[2] Lee, K.Y., Lee, K.K., Ho, S.L., Exploration of Using
FBG Sensor for Axle Counter in Railway
Engineering, WSEAS Transactions on Systems,
Issue 6, Vol. 3, August, 2004, pp. 2440-2447.
[3] Lee, K.K., Ho, S.L., Unconventional Method of
Train Detection Using Fibre Optic Sensors, HKIE
Transactions, Special Issue on Railway Development
in the 21st Century, March, 2006, pp. 16-21.

129

Authorized licensed use limited to: IEEE Xplore. Downloaded on February 26, 2009 at 04:30 from IEEE Xplore. Restrictions apply.