30th International Conference on Lightning Protection - ICLP 2010
(Cagliari, Italy - September 13th -17th, 2010)
ANALYTICAL STUDY ON LIGHTNING OVERVOLTAGES OF RAIL TRACK
AND RAILWAY SIGNALLING EQUIPMENT
2
2
l
l
Hideki Arai , Ikuo Watanabe , Hideki Motoyama and Shigeru Yokoyama
2
I Railway Technical Research Institute (RTRI), Tokyo, Japan
Central Research Institute of Electric Power Industry (CRIEPI), Kanagawa, Japan
e-mail: poo@rtri. or.jp
ABSTRACT
The development of the effective and economical protection
measures is vital for the railway signalling systems because
the lightning damages cause the disruption of the railway
transportation systems. In general, the railway signalling
systems are set up at a wayside and directly connected to the
rails through cables. The lightning surges can invade the
railway signalling systems via the cables and cause damages
to the systems. Accordingly, clarifying the propagation
characteristics of lightning surges along the rails is essential
to develop the lightning protection measures for the railway
signalling systems. We have carried out the field tests to
examine both surge impedance and surge propagation
velocity of the rails. This paper proposes a calculation model
of the surge propagation characteristics along the rails. The
results of surge impedance and surge propagation velocity
calculated by the proposed model almost agree with the
experimental results. Moreover, this paper proposes a
calculation model of the lightning overvoltages on the
railway signalling equipment. The calculation model
consists of the rail model and the equivalent circuit model of
signalling equipment. We indicated the validation of the
calculation model of the railway signalling equipment due to
the comparison with the results of the field test. This model
is applicable to the development of lightning protection
measures for the railway signalling systems.
1
INTRODUCTION
The railway signalling systems have made remarkable
progress in recent years with their components becoming
increasingly compact and multi-functional due to the
adoption of microcomputers and other electronic devices
in wide ranges. However, Iightning damages such as
circuit burnout, system failure or malfunction have
frequently occurred in the railway signalling systems
because of the vulnerability of electronic devices to the
lightning surges. Furthermore, the railway signalling
systems are widely installed at a wayside and constitutes
a network by being connected each other with rails and
cables. Accordingly, there are a lot of parts that lightning
surges can invade easily. In addition, troubles extend to
wide ranges in case of occuring lightning damages on the
railway signalling systems. Therefore, it is required to
build up effective and economical countermeasures for
preventing lightning damages on the railway signalling
systems since suspension and delays of trains due to
lightnings may cause social confusion.
In general, the railway signalling systems are set up at
a wayside and directly connected to the rails through
cables. The rails can be one of the invasion routes of
lightning surges to the railway signalling systems.
Accordingly, clarifying the propagation characteristics of
lightning surges along the rails is essential to develop the
lightning protection measures for the railway signalling
systems.
We carried out field experiments on the following [1].
1.
Measurement of the surge impedance of the rail and
the surge propagation velocity in the rail
2. Measurement of the surge attenuation ratio in the
rail
This paper proposes a calculation model of the surge
propagation characteristics along the rails.
Moreover, this paper proposes a calculation model of
the lightning overvoltages on the railway signalling
equipment.
This paper describes the results of the field tests, the
components of proposal calculation model, and the
comparison between experimental results and calculated
results.
2
OUTLINE OF FIELD TEST SECTION
We selected a non-electrified and single track section
as the test site because noises induced on rails could be
reduced in the measurement.
The line profile of the rail track at the test section is
shown in Fig. 1. A rail track consists of rails, cross ties, a
rail bed, and a track bed. The test section consists of
timber cross ties and a ballast rail bed.
The condition at the test section is shown in Table I.
The ground resistivity in Table I was measured due to the
4 electrode methods of Wenner. The field tests were done
on a [me day when both the rail and rail bed were dry.
1124-1
Receiving end
/ Insulated rail joint
Sending end
Induction Rail (No. 2)
Track bed
Figure I. Line profile of rail track at the test site.
Table
I:
Ground resistivit
534 fl'm
3
3.1
Figure 2. Outline of measuring surge parameters of rail.
Condition of test site.
SURGE PARAMETERS OF THE RAIL
Measuring methods
The surge impedance and the surge propagation
velocity are important parameters as the surge
�haracteristics of a rail. We investigated the surge
Impedance and the surge propagation velocity between
the rail and the ground due to measure the injection
current to the rail and the induced voltage on the rail
when the steep-front current was injected into the rail.
The outline of measuring method is shown in Fig. 2.
Two measuring raits are insulated from adjacent raits
by inserting insulated rail joints at both ends as shown in
Fig. 2. A steep-front current generated by a pulse
generator (PG) was injected into the sending end of one
side rail (for example the rail No. 1 in Fig. 2). Then, the
injected current waveform (I), voltage waveform of the
rail No. 1 against the ground (Vs), and voltage waveform
of the rail No. 2 against the ground (11,,,) were measured
with an oscilloscope. The self surge impedance of the rail
No. 1 (ZI/) and mutual surge impedance between rails No.
1 and No. 2 (Z2a were calculated from measured
waveforms.
In addition, the above-mentioned measurement was
executed at the case of the open/short between the
receiving end of the injection rail and ground respectively.
�hlS. makes the measurement of round trip propagation
tIme of the surge easy due to clear the difference of the
reflec�ion at the receiving end. The surge propagation
velocIty (ca between the rail No. 1 and the ground was
calculated from the round trip propagation time of the
surge which is obtained with the measurement and the
length of test rail which is known.
The measurement which is similar to description above
was implemented with a steep-front current injected into
the rail No. 2. The self surge impedance of the rail No. 2
(Z22), mutual surge impedance between rails No. 2 and
No. 1 (Z12) and the surge propagation velocity (C2)
between the rail No. 2 and the ground were calculated
from measured waveforms.
3.2
Measured results
Fig. 3 shows the waveforms of the injected current (I),
the voltage of the injection rail against the ground (Vs),
and the voltage of the induction rail against the ground
(Vm) which was measured in the case of a steep-front
current injected into the rail No. 1. Furthennore, the
�oltage waveforms of the injection rail (Vs) and the
mduction rail (Vm) in Fig. 3 is shown at the case that the
receiving end of the injection rail was grounded or non­
grounded respectively.
From Fig. 3, the round trip propagation time of the
surge in the rail, the self surge impedance and the mutual
surge impedance of the rail was calculated by the way
mentioned below.
As shown Fig. 3 (a), we can fmd that the voltage
wavefonns of the injection rail (Vs) begin to divide from
the time ß when the receiving end of the injection rait
was opened or shorted against the ground. When the
receiving end is grounded, negative reflection is shown
when it is non-grounded, the positive reflection is shown :
Namely, the time ß is the time that the reflected surge
traveled back from the receiving end to the sending end
of the r�il. �he time ß is defmed as the round trip
propagatIOn time of the surge in this measurement.
Furthennore, the surge propagation velocity is able to
calculate by the length of the rail.
In addition, we can see the period which the voltage
and the current wavefonns stabilize between to the time ß.
For example, the period is between the time a and the
time ß in Fig. 3 (a). This period is not affected by the
length of the rail because this period is before the round
trip propagation time. The self surge impedance of the
rail is defined those which the voltage of the injection rail
(Vs) exclude by the injection current (I) in this period. In
the same way, the mutual surge impedance of the rail is
defined those which the voltage of the induction rail (Vm)
exclude by the injection current (I) in this period.
The surge impedance matrix of the rail and the surge
propagation velocity between the rail and the ground
calculated by above-mentioned methods are shown in
Table 11.
As shown in Table 11, the surge impedance matrix is
symmetry and the surge propagation velocity against the
ground is same between the rail No. 1 and No. 2. The
measured results are valid in consideration of the
1124-2
geometry of two rails shown in Fig.
u
c:
o
..,
50
"
40
�
20
'2 �
'::::.,'"
ß
11
j
1
1
o
-10
(Vr) was measured at the receiving end.
).-
lI,a open -
30
10
l.
11, a short r--
._..JA..-;-'
-5
o
I
10
15
I
20
�
0.8
�
1.0
0.6
....
�
0
c:
0
0.4
/I
{
1.2
25
30
0.2
0.0
-0.2
�
.
:.5'
Time [l1s]
(a) Voltage waveforms of injection rail and injection current
waveform.
c:
o
..,
"
��
15 -:;,.E
g>"
.l9
"0
>
I
I
50
40
30
Vm
20
10
I
I
I
I
I.
.cl
-5
0
10
15
20
Surge attenuation ratio
= Vr 12Vs
(I)
Namely, eq. (l) indieates that this surge attenuation
ratio beeomes lower as the attenuation of the surge
voltage eaused by traveling along the rail inereases.
The surge attenuation ratio in the rail was measured as
parameters of the duration of wave front (Tc) and wave
tail (TI) of the lightning surge voltage impressed to the
rail.
Receiving end
Sending end
/ Insulated rail joint
/I shol
rm at ?pen
o
-10
Supposing that the impressed surge voltage reflects
perfeetly on the insulated rail joint at the reeeiving end,
we defined the surge attenuation ratio in the rail as eq. (l).
Induction Rail
f-25
30
Time [l1s]
(b) Voltage waveforms of induction rail.
Figure 3. Measured waveforms.
Table 11: Surge impedance and surge propagation velocity of
rai!.
28
(45 )
28 43
4
(Q)
69. 8 (m/�s)
(Round-trip time: 8. 4 �s )
SURGE ATTENUATION RATIO IN THE
RAIL
4.1
Measuring methods
The surge attenuation caused by traveling along the
rail is important parameters as the surge characteristics of
a rail. We investigated the lightning surge attenuation
ratio due to measure the voltage waveforms between the
rail and the ground at the sending end and the receiving
end respectively when the lightning surge voltage of an
arbitrary waveforrn was impressed to the sending end of
the rail. The outline of measuring method is shown in Fig.
4.
As shown in Fig. 4, insulated rail joints are inserted
into the both ends of the test seetion. The lightning surge
voltage of an arbitrary waveform generated by a PG was
impressed to the sending end of one side rail as shown in
Fig. 4. Then, the voltage waveform of the injection rail
against the ground (Vs) was measured with an
oscilloscope at the sending end. In the same way, the
voltage waveforrn of the injection rail against the ground
Figure
4.2
4.
Outline of measuring surge attenuation ratio in rail.
Measured results
Fig. 5 shows the measured voltage waveforms of the
injeetion rail at the sending end (Vs) and at the reeeiving
end ( V r) when a 1/5 Ils lightning surge voltage was
impressed to the rail. Fig. 5 indieates that the wave erest
deereases and Tc beeomes longer as a lightning surge
voltage travels along the rail.
Fig. 6 shows the measured results of T.-dependenee of
the surge attenuation ratio. From Figs. 6, the surge
attenuation ratio depends on TI and inereases as Tt
beeomes longer, beeause the wave erest is deeided at the
wave tail in the ease of distorting the wave front of the
voltage waveform at the reeeiving end.
As a result, it is clarified that the attenuation of a
lightning surge voltage eaused by traveling along the rail
is extremely large eompared with that of an overhead line.
Rails may have extremely high leakage eonduetanee
against the ground in eomparison with overhead Iines.
90
2
:;.'
2'
v
"
60
�
"
30
(5
>
-30
90
-2
0
2
2
"
4
6
Time
[flS]
2'
v
.....
8
10
60
30
(5
>
-30
If' ......�
-5 0
5
10 15 20 25 30
Time
(a) Sending end.
Figure
:;.'
5.
[flS]
(b) Receiving end.
Measured waveforms.
1124-3
0
�
c
0
:p
'"
c
:l
"
t:i
'"
�
Cf)
:l
(4)
0.30
0.25
0.05
f...-"
_.It
, __
0.15
0.10
where
(S/m).
-'"
0.20
•
-.
0.00
10
100
1000
Duration of wave tail (Tt) of V, [Ilsl
Figure 6. Tt-dependence of surge attenuation ratio.
5
THE RAlL
Distributed parameters of the rail
We studied calculation models used EMTP (Electro­
Magnetic Transients Program) for the surge propagation
characteristics along the rail which can retlect
experimental results, expressed in chapter 4 and 5.
The electric circuit, which is fonned between sending
end and receiving end of rail showing in Fig. 2, can be
considered a two-port circuit composed from the
distributed-parameter line such as rails. We investigated
the frequency-dependent four-terminal parameters
(resistance R, inductance L, conductance G and
capacitance C) of the rails to estimate the distributed
parameter adopted in the calculation model due to
measure the open circuit impedance and the short circuit
impedance by non-grounding or grounding at the
receiving end of rail respectively. In the same way, we
measured the open/short circuit impedance between rails.
When the sine wave voltage of 1-100 kHz was
impressed from the sending end of the rail against
between one side rail and the ground or between rails, the
voltage of the rail, the current through the rail, and those
phase were measured. The measurement executed at the
same test section shown in Fig. 2.
Four-terminal parameters (R, L, G and C) of the
distributed line can be derived from eq. (2)-(5). The
open/short circuit impedance obtained with above­
mentioned measurement is inputted in eq. (2) and (3).
(2)
where Zk is the characteristic impedance of rail (Q), Zo is
the open circuit impedance (Q) and Zs is the short circuit
impedance (Q).
r=tanh-
l
�ZoIZsll
Y
is the admittance
R=ZcosBz
L =ZsinBz 127if
G =YcosBy
C =YsinByI27if
(5)
where R is the resistance (n/m), L is the inductance
(H/m), G is the conductance (S/m), C is the capacitance
(F/m), ez is the phase of the impedance (rad), er is the
phase of the admittance (rad) andfis the frequency (Hz).
The calculated results of four-terminal parameters (R,
G and C) of between the rail and the ground and
between rails due to measure the open circuit impedance
and the short circuit impedance by non-grounding or
grounding at the receiving end of rail is shown in Fig. 7,
respectively.
L,
CALCULATION MODEL FOR SURGE
PROPAGATION CHARACTERISTICS OF
5.1
Z is the impedance (n/m) and
(3)
where r is the propagation constant (I/rn) and I is the
length of rail (m).
1000
EE
-'" -'"
"- "u.
I
s,3
...J
ü
E
-'" E
-'"
"- "S!!!.
c::
Cl
100
10
R
=-z.
1
...0.1
I
G
0.01
100
10
Frequency [kHzl
(a) Between rail and ground.
1000
EE
-'" -'"
"- "u.
I
E
...J
�
::t
�
ü
E
-'" E
-'"
"- "S!!!.
c::
Cl
b'
100
10
..-
lt�
L
I I
C"
0.1
G
0.01
10
100
Frequency [kHzl
(b) Between rails.
Figure
7.
Distributed parameters of rail.
As mentioned at section 4.2, rails are considered as a
distributed line with extremely high leakage admittance,
so we applied G and C to the calculation model.
Furthermore, G and C where it exists between the rail
and the ground and between rails shown in Fig. 7 are
something for the kind of circuit which is shown in Fig. 8.
We can calculate the true value of Ge and Ce which exist
between the rail and the ground from the measured value
of Ge meas and Ce meas which show in Fig. 7 due to eq. (6).
_
_
In the same way, the true value of Gs and Cs which exist
1124-4
between rails can be derived from the measured value of
Gs_meas and Cs_meas which show in Fig. 7 due to eq. (6).
2Ge meas · G s /IIeas
4G s _/lleas - Ge _ m eas
length of 293 m. This reason is that when number of
partitions is too few, the admittance component of the rail
is added as concentrated constant against 1 place.
Otherwise, when number of partitions is too multi, the
model becomes troublesome.
2
Gs =
Ce =
4G s meas - 2Ge /IIeas · G s meas
4G s - /lleas - Ge - meas
2Ce /IIeas . C s meas
RailNo.
(6)
G,: 44.3 I1S/rn
C: 0.03 nF/rn
4C s meas - 2Ce /IIeas , C s meas
4C s - /IIeas - Ce - meas
where Ge is the conductance between the rail and the
ground (S/m), Gs is the conductance between rails (S/m),
Ge meas is the measured value of conductance between the
rail and the ground (S/m), Gs_meas is the measured value
of conductance between rails (S/m), Ce is the capacitance
between the rail and the ground (F/m), Cs is the
capacitance between rails (F/m), Ce_meas is the measured
value of capacitance between the rail and the ground
(F/m) and Cs meas is the measured value of capacitance
between rails (F/m).
G,_"oe",: G, II (G,+G,)
G,_"".,: G, II (Gi2)
MeasUling between rails
Ce_",,",: C, II (C,+C,)
C_"".,: C II (Ci2)
Figure 8. Admittance between rail and ground and between rails.
5.2
RailNo. 2
Ce: 0.44 nF/rn
4C s /IIeas - Ce - meas
2
Cs =
I
G,: 166.7),S/rn
Constitution of calculated model
The calculation model of the surge propagation
characteristics along the rails is constituted of the line
model insulated from the ground, which is ep line model
in EMTP, and the admittance component against the
ground.
The rail used for the field test, which is called the 50N
rail, has the property that equivalent cylindrical body
radius is 93.9 mm, specific resistance is 20.3xl0·s n'm,
relative magnetic penneability is 70. These electric
qualities above were made to retlect ep line model. In
addition, the ep line model at 30 kHz was applied.
Furthennore, Ge, Ce, Gs and Cs at 30 kHz derived from
the measured value ( Ge mea" Ce mea" Gs-meas and Cs_meas)
and eq. (6) were appliedto the model.
The calculation model of the surge propagation
characteristics along the rail is shown in Fig. 9. As shown
in Fig. 9, the model forms 8 divisions of the rail with
CP line model
(36.625 midivision)
Figure 9. Rail model for surge propagations.
5.3
Surge analysis by calculated model
We simulated the current and voltage characteristics of
the rail when the steep-front current was injected into the
sending end of the rail by using the calculation model
shown in Fig. 9.
The simulation result which reproduces the experiment
which is expressed in chapter 3 is shown in Fig. 10.
Furthermore, at the time of analysis, the ground
resistivity is 534 n'm, same as the condition of the field
test section.
When Fig. 10 which is the analytical result and Fig. 3
which is the experimental result is compared, the
analytical result by the surge propagation model of the
rail has become smaller than the experimental result
about the voltages occurred between the rail and the
ground with the steep-front current impression. While,
the voltage waveform of analytical result almost agrees
with the experimental result.
In addition, the surge impedance calculated by the rail
model for surge propagation analysis is smaller than that
of experimental result shown in Table 11. On the other
hand, the surge propagation velocity of analytical result
calculated on Fig. 10 is about 70 m/Ils. It is similar to the
experimental result shown in Table 11.
From this, we can estimate that the admittance
component between the rail and the ground which is
applied to the surge propagation model of the rail is
larger than the admittance component with the rail laid on
the field.
Next, we simulated the surge attenuation caused by
traveling along the rail when a lightning surge voltage
was impressed to the rail by using the calculation model
shown in Fig. 9.
The simulation result which reproduces the experiment
1124-5
which is expressed in chapter 4 is shown in Fig. 11. A
lightning surge waveform impressed to the rail in the
simulation is 1/5 �s, similar to the experiment.
When Fig. 11 which is the analytical result and Fig. 5
which is the experimental result is compared, the surge
attenuation and the strange stain which accompany the
propagation of the surge along the rail can be confrrmed
on the surge propagation model of the rail, similar to the
experimental result. But, the analytical result with the
model, the attenuation has become large. In regard to this,
as above-mentioned, we can estimate that the admittance
component between the rail and the ground which is
applied to the surge propagation model of the rail is
larger than the admittance component with the rail laid on
the field.
The results of surge impedance and surge propagation
velocity calculated by the proposed model almost agree
with the experimental results. Moreover, the proposed
model can calculate the surge attenuation along the rails.
But we need to discuss the admittance component applied
to the surge propagation model of the rail.
.�
60
50
o
40
'ß
. � '>
30
. :: '---'
20
10
o
-10
I,
I
I
c
I "L--i ---.-L-
I
I
I
v: at open
I--
_
/1
V. at short
-5
o
5
10
15
Time
[Ils]
20
25
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
�
.....
...
�"
c
()
0
c
...,
()
Q)
g
30
(a) Voltage waveforms of injection rail and injection current
waveform.
.�
60
50
40
'ß
'>
-e
30
c
�
� ':,.� 20
Vm at short
�
l,
10
o I--I-- VI" a open
"0
-10
>
-5
0
5
10
15
20
90
2:
�
"bIl
.s
"0
>
90
'\
60
r--.
30
2:
::,;-
.....
0
il'o
.s
"0
>
-30
60
30
0
/r-- ....
- 30
-2 0
2
4
6
8 10
-5 0 5 10 1520 2530
Time [Ils]
nme [fls]
(a) Sending end.
(b) Receiving end.
Figure 11. Simulation results about surge attenuation along the
rail.
6
CALCULATION MODEL OF THE
LIGHTNING OVERVOLTAGES ON THE
RAILWAY SIGNALLING EQUIPMENT
6.1
Field test for validation of the model
We propose the calculation model of the lightning
overvoltages on the railway signalling equipment. The
calculation model consists of the rail model above
mentioned and the equivalent circuit model of signalling
equipment.
As shown in Fig. 12, we temporarily installed an actual
railway level crossing equipment representing typical
examples of wayside electronic signalling equipment for
field lightning surge tests, and measured the lightning
overvoltages on the level crossing equipment in case of a
potential rise of the rail for the validation of the
calculation model of the lightning overvoltages on the
level crossing equipment [1]. A potential rise of the rail
was caused by injecting a 11100 �s lightning surge
current of 3 A generated with an IG.
o
c
!
CLLrrent injection line
{
Time
25
30
[Ils]
(b) Voltage waveforms of induction rail.
Figure 10. Simulation result of the surge propagation
characteristics.
'--_--:-_--"'-' Contral cable to level erossing
Potential referenee line
-=
Remote potential electrode
Figure 12. Outline offield test.
1124-6
6.2
propagation velocity between the rail and the
ground is about 70 m/Ils. In general, the surge
impedance of the rail is much small and the surge
propagation velocity of the rail is much slow
compared to the over-head conductors.
Comparison of the field test results and the
calculation results
Fig. 13 shows the comparison of the calculated results
and the experimental results regarding the lightning
overvoltages in case of the lightning surge current
injection into the rait. From Fig. 13, the proposal model is
almost correct because of agreement with the
experimental results.
The calculation model of the lightning overvoltages on
railway signalling equipment is applicable to the
development of lightning proteetion measures for the
railway signalling systems.
100
�
�.
,"
o
-20
�
'\
I
I
I
I
I
I
I---
j
Experimental waveform
�
I
�
-5
I
>- Calculated waveform
o
I
10
I
15
I
20
25
30
Time [Ilsl
(a) Voltage wavefonns of terminal 'S-' connected to injection
rail side.
c
100
'" "
re,
80
c
0
ß
"0
�
b1)
.�
e
0;
>
..!1
�
c
"> ....0 "E
0
' :;
c � �
"
�
CL
b1)
-'"
E
:J
b1)
"1ij
c
.�
.1l
60
40
n .....
20
I
I
I
I
IL L
I
-5
0
5
10
15
20
The attenuation of the surge propagating along the
rail is large compared to the over-head conductors.
The rail is the conductor which possesses the high
admittance components against the ground.
3.
We proposed a calculation model of the surge
propagation characteristics along the rails. The
results of surge impedance and surge propagation
velocity ca1culated by the proposed model almost
agree with the experimental results. Moreover, the
proposed model can ca1culate the surge attenuation
along the rails.
4.
We propose the ca1culation model of the lightning
overvoltages on the railway signalling equipment.
The ca1culation model consists of the rail model
above mentioned and the equivalent circuit model of
signalling equipment. The ca1culation model is
applicable to the development of lightning
protection measures for the railway signalling
systems.
8
[1]
Experimental waveform
\Calculated waveform
0
-20
I
I
2.
_
REFERENCES
H. Arai, H. Matsubara, K. Miyajima, S. Yokoyama and K.
Sato
:
"Experimental
Study
of
Surge
Propagation
Characteristics of Rail and Lightning Overvoltages on
25
Level
Crossing",
IEEJ
Trans.
PE,
Vo1.l23,
No.ll,
pp.1307-1312, November 2003
30
(b) Voltage waveforms of terminal 'S+' connected to induction
rail side.
Figure 13. Comparison of the calculated waveform and the
experimental waveform regarding the lightning
overvoltages.
7
CONCLUSIONS
This paper describes the experimental results
conceming to the surge propagation characteristics of the
rail and the calculation model for surge analysis which
can reflect the experimental result. Moreover, this paper
proposes a Iightning surge calculation model of the level
crossing equipment representing typical examples of
wayside electronic railway signalling equipment.
The main conclusions are summarized as folIows:
1.
The self surge impedance between the rail and the
ground is about 50 n, the mutual surge impedance
between the rails is about 30 n, and the surge
1124-7