LIGHTNING INTERACTION WITH THE SWEDISH RAILWAY NETWORK

Dr N Theethayi; Z Mazloom; P-A Lindeberg Dr T Schütte

Prof R Thottappillil
Uppsala University
BOX 534
SE-751 21 Uppsala
Sweden
Nelson.Theethayi@Angstrom.uu.se ,
Ziya.Mazloom@Angstrom.uu.se ,
Rajeev.Thottappillil@Angstrom.uu.se
Banverket HK/BJ
Jussi Björlings väg 2
SE-781 85 Borlänge
Sweden
Per-Anders.Lindeberg@Banverket.se
Rejlers
Ängsgärdsgatan 13
SE-721 30 Västerås
Sweden
Thorsten.Schutte@Rejlers.se

KEYWORDS: Transmission line, lightning attachment, crosstalk, transients.

ABSTRACT
In the beginning, the design of signal and control network of railways in early times was not always done in
accordance with the strict rules of Electromagnetic Compatibility (EMC). This created problems due to
Electromagnetic Interference (EMI) when old electro-mechanical signalling, control and communication
systems were replaced by modern sensitive electronic circuits. Modern developments in the railway systems
have made the system more vulnerable to lightning transients, a natural source of EMI, because the overall
network is not designed to reduce the lightning surges to the low levels tolerable to the electronic systems
widely being introduced in the railway system. Railway networks are extensive and modernization of
signal/control/communication systems is carried out only in stages at different times, it is not unusual to find
transient related problems in a section of the network on account of inadequate lightning protection design or
EMC design of the existing network in which a new module (equipment or system) is introduced. One
wonders if these problems act as a brake on the upgrading plans of the railways. Also the new railway
systems which incorporate advanced signal/control/communication systems sometimes suffers from the
effects of lightning transients because many of the standards and guidelines used in the design dates back to
the age of electromechanical devices and hence do not include robust transient protection design. There were
many attempts to address EMC issues in railways, but usually these attempts were confined to solving
immediate problems of EMI due to introduction of new locomotive drives and EMI due to sparks at the
pantograph. A comprehensive review of railway system from the angle of lightning protection and EMC is
not yet carried out. This work discusses the EMC problems of large distributed systems, with particular
reference to lightning interaction with Swedish rail network.

INTRODUCTION
Statistics from Banverket (Swedish National Rail Administration) show that the delay caused in the regular
running of trains during thunderstorms is several hundreds of hours (Thottappillil et al, 2001). The
equipments that were damaged in the railway network were track circuit systems that include rectifier and
relays, wayside telephone systems and cables, booster transformers, communication equipments, signal
circuits, etc. Uppsala University and the Banverket have agreed upon to work together in an effort to come
up with efficient schemes of lightning protection to minimize delays and to ensure normal operations of
trains. Most organizations dealing with railway transportation technology has practical experience with
lightning protection and EMC. They try to consolidate this experience and share it with others in the
organization in the form of internal standards and guidelines. Often these documents are designed for very
specific systems and not well suited when there are system changes. If there is a simulation software for
engineers at railway industry that can simulate the important parts of the railway system from the lightning
protection and EMC point of view, and if the internal standards and guidelines are coupled to this software,
then such a software can cope up with the changes in the system more effectively. Well, then the heart of
such software would be a set of simulation programs that use various electromagnetic techniques, which
would be capable of simulating the effects of the interference sources on different parts of the railway
network. The aim of this paper is to present the concepts behind the development of such simulation
programs and modelling of lightning interaction with railway network based on theoretical studies, giving
also importance to relevant experimental investigations for future model validation and to evaluate the failure
modes/vulnerability of various systems used in the railway network. From the perspective of lightning
protection, this work therefore should be able to give engineers an opportunity to ascertain the following:

• Whether the installed protective device provide sufficient level of protection as the lightning induced
voltage appearing at the terminals where the protection is installed is known a priori.
• Using the simulation programs it is possible to trace whether the installed protection will hamper the
normal operation of the other interconnected systems, perhaps due to different ground potentials and
possible low impedance paths in the event of lightning transient diversions.
• Clear guidelines on how and where to provide the protection, like from the traction power side or from
the track side, or both as in the event of lightning strikes both the tracks and overhead traction and
auxiliary power wires (if present) will be raised to considerable over voltages, whose magnitudes
depends on the regional ground conditions, grounding system, induced/coupled lightning
electromagnetic fields, non-linear phenomena like insulator flashover and soil ionization, etc.
Bearing in mind that the first step in any lightning protection study is to identify the way in which the
lightning transients enter into different parts of the systems. The devices mentioned in earlier paragraphs that
get damaged during the thunderstorms are associated with the tracks and overhead conductor systems. Both
the tracks and overhead conductors can be thought of as Multi-conductor Transmission Lines (MTL) above
ground. All the transmission lines are taken to be infinite in either direction as they are several hundreds of
kilometres long. This confines our problem to be a two-dimensional one. Our first task will be to find out
how a lightning stroke interacts with this overhead traction system. A photograph of the railway line near
Gistad (Sweden) is shown in Figure 1. This shows the complexity of the MTL. To start with only a single-
track system is considered here. However, an analysis with the double track system would be similar as
described here but with more conductors.

Figure 1. Railway lines near Gistad, which shows the complexity of MTLs

The following material to be presented is a concise summary of the Ph.D. thesis “Electromagnetic
Interference in Distributed Outdoor Electrical Systems, with an Emphasis on Lightning Interaction with
Electrified Railway Network”, by Nelson J. Theethayi (Theethayi, 2005). Note that a German version of
thesis summary is presented in (Lindeberg et al, 2007).

THE SWEDISH RAILWAY SYSTEM

The voltage used in the Swedish railway traction power is 15 kV at 16.7 Hz. This voltage is obtained from
substation transformers that are operating at even higher voltages. Auxiliary power in this system works at 11
kV or 22 kV at 50 Hz. All other equipments connected to this system work at lower voltages, mostly at few
tens or less than 10 V. The track circuit works on DC. Booster transformers are used to guide the traction
return current away from the rails.

There are some studies conducted on electrified railways, but these are mainly concentrated to the low
(Ferrari & Pozzobon, 1998) and medium (harmonic) frequency problems (Zynovchenko et al, 2006), which
do not pose any direct threat to the functionality of the railway system. The higher frequency problems which
may damage the components and equipments must be given more attention in future studies. The EMC
problems addressed have mostly been confined to EMI problems due to introduction of new locomotive
drives and due to sparks at the pantograph (Equiluz et al , Proceedings of the IEEE International Symposium
on Industrial Electronics, 2000); (Equiluz et al, IEE Conference Publication, 2000). The biggest high
frequency threat today is lightning. There are some studies carried out on this topic, but these have been
mainly statistical (Biesenack et al, 2006). Therefore, after considering the previous studies, lightning
interaction with electrical systems will be investigated.

Single Track Overhead Traction System in Swedish Railways

Before any simulation regarding the lightning interaction with any small or large system can be done
knowledge about the system geometry is needed, and later the parameters needed for running the simulations
have to be defined.

The conductor layout in a standard single track overhead traction system in the Sweden railways may look as
illustrated in Figure 2.

Figure 2. Layout of standard single track railways in Sweden

A complete system layout including operating voltages, communication cables, booster transformers, track
circuits and relays used in the Swedish railways can be found in (Stevenson, 1988); (Lagerstam, 2001);
(Bülund et al, 2004); (Schütte & Thiede, 2000).
COUPLING OF LIGHTNING
There are two main ways for the lightning to couple into electrical systems, either by direct hit or by indirect
hit. The meaning of direct hit is obvious; the system is physically hit by the lightning. Indirect hit means that
the system is disturbed, damaged or by any other mean affected by a lightning strike close by. It is evident
that a direct hit is, in most cases, more devastating than an indirect hit. But an indirect hit may also be very
severe. The damage may also be latent, leading to failures later on. The direct and indirect hits are visualized
in Figure 3a and Figure 3b, respectively.

Figure 3. a) Direct lightning hit, b) Indirect lightning hit

The coupling path of disturbance, between the emitter and receptor, is called as crosstalk. Crosstalk normally
happens through the EM coupling in the near field and occurs due to the inductive and capacitive coupling.
The emitter may be one of the conductors of a transmission line which is hit by a lightning. The receptor may
be another conductor in the transmission line which will be influenced by the current flow in the emitter. If
the distance between emitter and receptor is very small, the receptor is said to be in the near field of the
disturbance (Paul, 1994).

The effect of the lightning does not only depend on how the object is hit by the lightning. It also depends on
the object and lightning characteristics. The parameters affecting the effects of lightning are current and
electromagnetic fields created as lightning strikes. The most important parameters of lightning current
causing damages are the peak current, maximum rate of current change, the integral of the current over time
(charge) and integral of square of current over time (action integral). The parameters of most importance
among the field parameters are the peak electric field and the maximum change of electric or magnetic field
per unit time.

The stroke current in the return stroke phase of the lightning is the most important part of indirect strikes.
The first return strokes and subsequent strokes in a lightning have different properties. The first return strokes
are rising slowly and have their main frequency contents in the range from a few tens of kHz to a few
hundreds of kHz. Subsequent strokes are rising faster with frequency contents up to a few MHz (Rakov &
Uman, 2003); (Cooray, 2003).

Sweden is, due to its geographical location and the less number of lightning days in a year, not a lightning
dense country. But lightning interference is a bigger concern in the Sweden compared to some other places
due to the following reasons (Biesenack et al, 2006):

• Number of could to ground (CG) lightings are higher than that in tropical areas as the clouds are closer to
ground.
• The railways in Sweden are vast, stretched across the entire country.
• High ground resistivity will increase the voltages and electric potentials induced.
• High number of single track rails, which are more responsive to interference.
• The Swedish railways are spread out; there are in most cases no alternative routes for rerouting the traffic
if something would happen to the railway operation.
Lightning Attachment to Railway Overhead Conductors
A leader mechanism of breakdown has been observed in the breakdown of long air gaps stressed with
impulse voltages (Golde, 1977); (Baldo, 1999). This motivated researchers to extrapolate the data from long
air gap experiments to lightning protection (Baldo, 1999). Their aim was to arrive at a suitable model that
relates charge on the stepped leader with the attractive effect of the grounded structure. Efforts in this
direction have led to the so-called Electro-geometric model (EGM) (Golde, 1977); (Horvath, 1991). EGM is
based on the principle that the lightning strike point is determined only during the final jump (at the striking
distance) (Horvath, 1991). The electric field produced by the charge of the leader channel initiates connecting
leader (upward leader) as shown in Figure 3. The final path is thus dependent upon this moment.

Figure 4. Striking distance for a given conductor held by tower of a given height

The position where the lower end of the downward leader (stepped leader) is at this moment is called the
orientation point because the path of the lightning is believed to turn towards the point of the strike. The gap
between the orientation point and the earth/earthed structures is called the orientation distance or the striking
distance and it is determined by the field strength at the earth. The striking distance rs is shown in Figure 4.
Imagine now that a sphere of radius equal to the striking distance is rolled in the outward directions and
around the conductor surfaces of the system shown in Figure 2 of the single-track system. Then the exposure
surface or the collection surface through which the descending stepped leader can attach to the overhead
conductor system (Horvath, 1991) is obtained.
 Figure 5. Collection surface offered by a single-track railway system to a descending stepped leader with
prospective return stroke amplitude of 5 kA and 100 kA

Figure 5 (left windows) shows the possible collection surface offered by the conductors for two different
values of the prospective return stroke currents 5 kA and 100 kA respectively using modified EGM. It can be
seen that the strokes can only attach to the conductors R7, R8 and R9 of the system and that conductors R7,
R8 and R9 shelter all the other conductors, acting as shielding wires. Any stroke descending outside this
region of the collection surface of R7, R8 and R9 will terminate on the ground and will result in an indirect
strike, more of which will be discussed later. Note that the striking distance and the lateral distance are the
same in the case of modified EGM. However they have different definitions (Rizk, 1990). Lateral or
attractive distance is the maximum horizontal distance from the structure location at which an incepted
upward leader from the tip of the structure is capable of intercepting the stroke of a given prospective current
(Rizk, 1990). (Rizk, 1990) mentions that it is not necessary that all the incepted upward leaders for a given
prospective current will intercept the stroke. To verify whether the above statement that R7, R8 and R9 are
the only possible candidates that can be struck by lightning, simulations were also carried out using the
recent models for lightning attachment proposed by (Rizk, 1990). Rizk's model is based on three important
criteria namely upward leader inception criterion, upward leader propagation criterion and final jump
criterion (Rizk, 1990), which all were considered in the simulations. Upward leader inception criterion of
Rizk's model considers both the height and the radius of the conductors, derivation of which can be found in
(Rizk, 1990). Still it was seen that the strokes mainly attach to the conductors R7, R8 and R9 of the system.

WAVE PROPAGATION IN MULTICONDUCTOR TRANSMISSION LINES ABOVE GROUND

In the case of MTL systems above ground the reference conductor is the ground that carries or returns the
currents back to the source. The problem one could face here is what would happen to the voltages and
currents propagating on the MTL system if the reference ground is real earth and thus not perfect? We shall
try to answer this question in this section. More emphasis is given for direct numerical time domain solutions
of the telegrapher's equations. This is necessary because later while calculating the induced voltages,
continuous monitoring of the voltages and currents on the MTL system of the Swedish railway system is
needed, for modeling the non-linear phenomena, like insulator flashover and soil ionization at the pole
footings.

Telegrapher’s Equations
The voltage and current wave propagation in frequency domain in MTL is well represented by the two set of
equations, described for finitely conducting ground as shown in (1). They are popularly known as
Telegrapher’s Equations (Carson, 1926); (Wedepohl & Wilcox, 1973).

dV ( x, jω )
+ jωLe ⋅ I ( x, jω ) + Z g ⋅ I ( x, jω ) = 0 (1a)
dx
dI ( x, jω )
+ jω C e ⋅ V ( x , jω ) = 0
dx (1b)
The per unit length parameter Le and Ce matrix calculations under perfect ground conditions are calculated
from the image theory (Paul, 1994); (Paul, 1992); (Tesche et al, 1997); (Degauque & Hamelin, 1993). The
term Zg is called the ground impedance due to the frequency dependant penetration of the electromagnetic
fields into the ground (Tesche et al, 1997); (Rachidi et al, 2003); (Rachidi et al, 1996). Various expressions
for the ground impedance have been derived in the literature based on the low frequency and high frequency
approximations (Carson, 1926); (Semlyen, 1981). In time domain the expressions (1) can be written as (2).

∂V ( z, t ) ∂ I ( z, t ) ∂I ( z ,τ )
t
+ Le + ∫ ζ (t − τ ) =0 (2a)
∂z ∂z 0
∂τ
∂ I ( z, t ) ∂ V ( z, t )
+ Ce ⋅ =0
∂z ∂z (2b)
Z g ( jω) ⎤
In the above equations ζ (t ) = F −1 ⎡⎢ ⎥ is known as the transient ground impedance in time domain. One
⎣ jω ⎦
may wonder why there isn’t any ground admittance or transient ground admittance in the discussion or in the
above equations. The transient ground admittance term in the current wave equation (2b) is not used because
of the fact that, among the fields that penetrate the lossy ground, the magnetic field penetrates more than the
electric field. Thus the contribution of ground admittance to the overall shunt admittance for the overhead or
above ground lines is negligible. Further, in (Chen & Damrau, 1989) it is mentioned that the ground
admittance becomes important only when the wires are touching the ground. It has been shown in (Rachidi et
al, 1996); (Rachidi et al, 1999) that for wires above the ground the overall admittance including ground is
approximately equal to the shunt admittance due to external line capacitance. Since the magnetic field
penetrates the earth more than the electric field, ground impedance becomes very important and cannot be
neglected. Soon we will demonstrate, with an example, to see the error one would expect by neglecting the
ground admittance.

We need the ground impedance from both Carson’s expressions (low frequency) and Semlyen’s expressions
(high frequency), which will be combined to use in the transition region. As Carson’s expressions are low
frequency approximations in time domain, its contribution will be in late time (LT), while as Semlyen’s
approximation is a very high frequency approximation, its contribution will be in early time (ET).
Researchers have contributed to the development of LT and ET transient ground impedance expressions. The
transient ground impedance expression is given in (Rachidi et al, 2003); (Araneo & Celozzi, 2001);
(Theethayi et al, in press) and its implementation in FDTD method can found in (Araneo & Celozzi, 2001).
The best available and simple method to solve the telegrapher’s equations is to use the finite difference time
domain technique. It is well known as FDTD method. Later several researchers developed it to solve the
transmission line type problems, the simplicity being that the problem we have is one dimensional
propagation and all the main parameters with regard to field coupling between wires are inherent in the
inductance, capacitance and ground impedance and admittance matrix. Thus we have space and time
variables connected to voltages and currents along the line in the differential equations. The FDTD method
splits (2) using central difference approximations. The solution to the problem is achieved by means of leap-
frog scheme (Paul, 1994); (Tesche et al, 1997). The stability of FDTD method depends on the time and space
discritisations and it must satisfy the Courant conditions Δx ≥ v p (Paul, 1994); (Tesche et al, 1997), where
Δt
vp is the maximum phase velocity of the system.

Detailed analysis on the influence of ground on the induced voltage and currents can be found in the papers
(Araneo & Celozzi, 2001); (Theethayi et al, in press), where the influences of ground conductivity, conductor
heights, terminal load on the lines, etc. are studied. Using the FDTD method we solve a simple three
conductor problem both the conductors are 10 m in height and separated by 1 m and are 1 km long. The ends
of the lines are terminated in the self surge impedance of the line. A lightning current source having a double
exponential wave shape with 11 kA peak and 0.1 μs rise time with half-peak time 70 μs (typical subsequent
return stroke) is placed at the middle of the line which is called the emitter line and the induced voltages at
three points on the other line, center of the line, half way to the load and at the load point of the line, are
calculated using the FDTD method is shown in Figure 6.

Figure 6. Induced voltages on the receptor line for direct strikes on the emitter line

Indirect Strikes Pulse Propagation on MTL

There are many models available in the literature for field to wire coupling type problems. The most popular
and simplest model is the Agrawal et al. model (Agrawal et al, 1980). Nucci et al. (Nucci & Rachidi, 1995);
(Rubenstein et al, 1989) have compared the various coupling models, namely Taylor et al. model (Taylor et
al, 1965) and the Rachidi model (Rachidi, 1993), through simulations for the induced voltage on a single
power line from a nearby return stroke and concluded that all the models agree very well. We will use the
Agrawal et al. model (Agrawal et al, 1980) for the discussions here.
The associated telegrapher’s equations in time domain based on the scattered voltage is given by (3),

dVi S ( x, t ) dI ( x, t )
t
∂I ( x,τ )
+ Le ⋅ + ∫ [ζ (t − τ )]⋅ dτ = E xi ( x, hi , t ) (3a)
dx dt 0
∂τ

dI ( x, t ) dVi S ( x, t )
+ Ce ⋅ =0
dx dt (3b)
The total voltage on the line and the scattered voltage are related as shown in (4).

V i ( x, t ) ≈ − E zi ( x,0, t ) ⋅ h (4a)
V ( x, t ) = V S ( x , t ) + V i ( x , t ) (4b)
V S (0, t ) = − Z S ⋅ I (0, t ) − V i (0, t ) (4c)

V S ( L, t ) = − Z L ⋅ I ( L , t ) − V i ( L , t ) (4d)

In (4) Ez is the vertical component of electric field due to the lightning return stroke. In (3) Ex is the
horizontal component of electric field along the line at line height. Zs and Zl are the line termination
impedance. More details on the lightning induced voltage on indirect strikes can be found in (Rachidi et al,
1999). Note that the horizontal component of the lightning return stroke electric fields incident on the line is
dependant on the finite ground conductivity and can be calculated using the Cooray-Rubinstein
approximation (Cooray, 1993).

Current Pulse Propagation in Underground Wires and Coupling Through Cable Shields - A Transmission
Line Analysis
In (Theethayi, 2005) and (Thottappillil & Theethayi, 2005); (Theethayi & Thottappillil, 2005) simplified
transmission line models for the pulse propagation in buried insulated wires and cables are developed. For
the buried cables the transmission line equations are given by (5) for bare wires and (6) for insulated wires,
corresponding to Figure 7. These can be solved in time domain using the FDTD methods.

For bare wire

dV ( x, jω ) ⎫
= − Z gb ⋅ I ( x, jω )⎪
dx ⎪ (5)
⎬
dI ( x, jω )
= −Ygb ⋅ V ( x, jω ) ⎪
dx ⎪⎭
For insulated wire
dV ( x, jω )
= −( jω ⋅ L + Z gi ) ⋅ I ( x, jω )
⎫
⎪
dx ⎪ (6a)
⎬
dI ( x, jω ) ⎛ C ⋅ Y ⎞
= − jω ⋅ ⎜ ⎟ ⋅ V ( x, jω )⎪
gi

dx ⎜ jω ⋅ C + Y ⎟ ⎪
⎝ gi ⎠ ⎭
 μ0 ⎛ b ⎞ ⎫
L= ln⎜ ⎟ ⎪
2 ⋅π ⎝ a ⎠ ⎪⎪
2 ⋅ π ⋅ ε in ⎬ (6b)
C= ⎪
⎛b⎞ ⎪
ln⎜ ⎟
⎝a⎠ ⎪⎭

Figure 7: Variables used in equations (5) and (6)

An empirical expression for ground impedance that is computationally efficient and valid up to 10 MHz can
be found in (Theethayi, 2005).

Once the currents due to lightning on the cable shields are calculated the currents induced into the inner
conductors can be easily calculated using the concept of coupling through cable shields as explained in
(Horvath, 1991). More details can be found in the thesis (Theethayi, 2005).

SIMULATIONS OF LIGHTNING EFFECTS

Through simulating the lightning and the effects of lightning on electrical equipments and components in the
vicinity of the lightning, we can estimate and in many cases even find a solution to reduce or, in best cases,
nullify noise and damages. Many complex components and equipments cannot be simulated; these have to be
tested experimentally in order to find their voltage and current thresholds and their effect on surge pulse
propagations. In order to later tailor a protection system.

A booster transformer model has been derived (Theethayi, 2005), and simulated (Mazloom et al, 2007) on a
two conductor line similar to the catenary and return wires in the Swedish railway system. In this paper the
influence of the booster transformer on the surge propagation on an infinitely and finitely conducting ground
are shown and compared.

Complexity Associated With Simulation of Railway Above Ground Conductor Systems

Most of the studies in the recent past focused on the power line lightning interaction (Rachidi et al, 1996);
(Rachidi et al, 1999); (Nucci & Rachidi, 1995); (Rachidi et al, 1997) and the configurations of the MTL were
such that conductors were located at similar heights (typically at 10 m above ground) and with symmetrical
spacing. The Swedish railway traction system has MTLs that are spread out from heights of 10 m to as low as
0.5 m above ground and has many conductors located in an unsymmetrical manner as seen in Figure 2. In a
multi-track railway system, there are more conductors and the conductors are positioned in a more
unsymmetrical manner.

The lightning interaction to the railway system studied here is a bit complex and different compared to power
line conductor system owing to many reasons, but the main reasons based on (Theethayi et al, 2005) are:

• There are nine conductors in the system (Figure 8), which are located at different heights. Hence their
ground impedances, as discussed in the previous chapter, required for the wave propagation analysis
are unique and have to be determined independently. Whereas in power line conductors located at a
height of 10 m and close to each other, the self and mutual ground impedances are comparable. More
details on the magnitudes and shapes of ground impedance curves for various conductor heights can be
found in (Bülund et al, 2004).
• Insulators on the pole (supporting the conductors) are of different types and in particular have different
impulse withstand levels, because each conductor has different operating voltages. In the case of power
lines, as they have common operating voltages, similar insulators are used. This causes flashover
behavior to be more unique at every pole location.
• The pole footing at every pole suffers the soil ionization phenomenon (Technical Council of the IEEE
Power Engineer Society, 1999) and hence they additionally modify the pole footing resistance. This
poses an additional termination on one of the tracks at every pole (one of the tracks is grounded at
every pole) location. While the other track is floating without any termination.

Some important conclusions for a direct strike of 31 kA to the auxiliary conductors of the single-track
railway system are presented by (Theethayi et al, 2005), who attempted to include all the above complexities.
Details of all modeling strategies can be found in (Theethayi et al, 2005).

Figure 8. Insulators, connections and impulse withstand voltages in the single track railway system

EXPERIMENTS
From the simulations investigated and presented by (Theethayi et al, 2005) it is certain that there will be large
common mode and differential mode voltages that would be appearing along and across the auxiliary power
conductors and the rails/tracks. Therefore it becomes necessary to identify the failure modes of various
equipments connected along the above ground lines and tracks. Here we discuss the failure modes of two
important devices connected to the tracks; the relay unit and the rectifier unit. Their connections to the tracks
are shown in Figure 9. Since the relay and rectifier sections are connected in parallel across the tracks, the
best way to test the system would be to test the two sections independently and identify the most vulnerable
components in each section that may fail due to surges.

Figure 9. Connections between the relay and rectifier unit to the tracks

The rectifier section, which has a power supply section, consisting of over-voltage protection, the rectifier
unit, a choke coil and an adjustable resistor (0-6 Ω), has been tested. The unit was subjected to lightning type
of surges in four different configurations: a) Common mode from trackside, b) Differential mode from track
side, c) Common mode from auxiliary power supply side and d) Differential mode from auxiliary power
supply side. The schematic of the test configurations are shown in Figure 10.

Figure 10. Schematics of the rectifier test configurations

The relay withstand tests were carried out in two stages: one is the injection to the I-Rail with respect to the
S-rail as reference and the second is injection to the S-rail with I rail as reference. This is because in the event
of lightning strikes the S-rail and I-rail could be at different potentials (Theethayi et al, 2005). The
connection for the I-Rail injection is shown in Figure 11. These experiments are documented in (Theethayi,
2005).
 Figure 11. Connections to the relay unit as current is injected into the I-Rail

One main concern for engineers designing lightning protection devices and systems is to find the ways
lightning transients enter or couple into the system. The most effective way to heed to this concern is to
conduct experiments at the place of interest.

In summer 2003 Uppsala University, in association with Banverket, conducted experiments to measure the
lightning transients entering a technical house (Theethayi, 2005); (Theethayi et al, 2007). This experiment
was conducted in Tierp train station, about 145 km north of Stockholm. In this paper lightning flashes in the
vicinity of Tierp (up to 50 km away from the station), detected with the LLP system (Sørensen, 1995)
(lightning location system used in Sweden) were correlated with measured transients inside the technical
house.

The measurement setup recorded totally 4885 data files, of which 174 had perfect correlation with lightning
strikes. The other measured transients depended on transients entering into the house due to locomotive/train
movements. These could not be correlated to lightning strikes and the transient shape looked different from
the wave shape of lightning transients. Four lightning strike locations and the corresponding induced
transients entering the technical house are shown in Figure 12.
 Figure 12. Lightning strike locations and corresponding transients entering the technical house in Tierp,
with date and GPS times marked

CONCLUSIONS
The effect of lightning cannot be neglected in the design and modification of electrical systems in the future.
Diversion of the current from direct lightning strikes is not enough to protect a system from lightning
interference. Systems designed incorrectly, from EMC point of view, must be examined and appropriate
measures should be taken to make these systems compatible with electromagnetic interference.

FUTURE WORK
In order to verify models proposed, experiments have to be conducted. For pulse propagation in underground
cables an experiment was planned in an abandoned train station in Söderhamn, 250 km north of Stockholm,
disconnected from the main railway network with all signaling and communication equipments still
connected. This experiment would be very informative since it is be a real case study and it would have show
the behavior of none ideal conductors used for a long time in real railway systems. Due to some technical
problems the experiment was postponed to the summer of 2007.
One of the key components in the railway system is the rail, which in the current and frequency range of
interest has strongly nonlinear properties (Trueblood & Wascheck, 1934); (Hill & Carpenter, 1991). Thus an
experimental investigation on the rail impedance has started.

At Uppsala University an EMC software is being developed for the Swedish railway administration,
Banverket. The idea is that the software later will aid the railway engineers to design a, from EMC
perspective, safe system.

ACKNOWLEDGMENTS
The financial support of Banverket (contact persons: Ulf Hellström and Roger Byström) and Bombardier
Transportation (contact persons: Michael Zitnik and Georg Bohlin) are greatly acknowledged.

REFERENCES
Agrawal AK, Price HJ and Gurbaxani S (1980), “Transient response of a multiconductor transmission line
excited by a nonuniform electromagnetic field”, IEEE Transactions on Electromagnetic Compatibility, Vol.
22, No. 2, pp. 119-129.
Araneo R and Celozzi S (2001), “Direct time domain analysis of transmission lines above a lossy ground”,
IEE Proceedings on Science and Measurement Technology, Vol. 148, No. 2, pp. 73-79.
Baldo G (1999), “Lightning Protection and the Physics of Discharge”, Proceedings of 11th International
Symposium on High Voltage Engineering, London, Conference Publication No. 467, Paper No. 2.169.S0.
Biesenack H, Dölling A and Schmieder A (2006), “Schadensrisiken bei Blitzeinschlägen in Oberleitungen”,
Elektrische Bahnen, No. 4, pp. 182-189.
Bülund A, Hellström U and Varju G (2004), “Changing from booster transformer system to autotransformer
system in the Kiruna-Råtsi-Svappavaara line in Sweden”, EMC York International Conference and
Exhibition, EMC York, (Presentations for download).
Carson JR (1926), “Wave propagation on overhead wires with ground return”, Bell Systems Technical
Journal, Vol. 5, pp. 539-556.
Chen KC and Damrau KM (1989), “Accuracy of approximate transmission line formulas for overhead
wires”, IEEE Transactions on Electromagnetic Compatibility, Vol. 31, No. 4, pp.396-397.
Cooray V (1993), “Some consideration on the “Cooray-Rubenstein” formulation used in deriving the
horizontal electric field of lightning return strokes over finitely conducting ground”, IEEE Transactions on
Electromagnetic Compatibility, Vol. 35, No. 1, pp. 75-86.
Cooray V (Editor) (2003), ”The Lightning Flash”, IEE Power Series; 34, Published by the IEE, London.
Degauque P. and Hamelin J. (Editors) (1993), “Electromagnetic Compatibility”, Oxford University Press.
Equiluz RP, David MP, Roboam X and Fornel BD (2000), “Dead time effect in railway traction system”,
Proceedings of the IEEE International Symposium on Industrial Electronics, Puebla, Vol. 1, pp. 151-156.
Equiluz RP, David MP, Roboam X and Fornel BD (2000), “Pantograph detachment perturbation on a railway
traction system”, 8th International Conference on Power Electronics and Variable Speed Drives, London,
IEE Conference Publication, No. 475, pp. 437-442.
Ferrari P and Pozzobon P (1998), ”Railway lines models for impedance evaluation”, 8th International
IEEE/PES and NTUA Joint Conference on Harmonics and Quality of Power, ICHQP 98, Athens, pp. 641-
646.
Golde RH (1977), “Lightning”, Academic Press, London.
Hill RJ and Carpenter DC (1991), “Determination of rail internal impedance for electric railway traction
system simulation”, IEE Proceedings-B, Vol. 138, No.6, pp. 311 – 321.
Horvath T (1991), “Computation of Lightning Protection”, Research Studies Press Limited, John Wiley and
Sons Incorporation.
Lagestam L (2001), “Kabelsystem – Huvud-/mellanortskabel för Banverkets telenät, Elektriska kvalitetskrav
på par- och fyrskruvkabel”, BVF 518.1101, BVF 518.1002, Banverket.
Lindeberg PA, Mazloom Z, Theethayi N, Thottappillil R and Schütte T (2007), “Blitzeinwirkungen auf
Oberleitungs- und Signalanlagen in Schweden“, Elektrische Bahnen, No. 1-2, pp. 67-80.
Mazloom Z, Theethayi N and Thottappillil R (2007), ”Influence of Discrete Series Devices on Crosstalk
Phenomena in Multiconductor Transmission Lines”, to be on the Proceedings of the 18th International Zurich
Symposium on Electromagnetic Compatibility, Munich.
Nucci CA and Rachidi F (1995), “On the contribution of the Electromagnetic Field Components in Field-to-
Transmission Line Interaction”, IEEE Transactions on Electromagnetic Compatibility, Vol. 37, No. 4, pp.
505-508.
Paul CR (1992), “Introduction to Electromagnetic Compatibility”, John Wiley and Sons Incorporation.
Paul CR (1994), “Analysis of Multiconductor Transmission Lines”, John Wiley and Sons Incorporation.
Rachidi F (1993), “Formulation of the Field-to-Transmission Line Coupling Equations in Terms of Magnetic
Excitation Field”, IEEE Transactions on Electromagnetic Compatibility, Vol. 35, No. 3, pp. 404-407.
Rachidi F, Loyka SL, Nucci CA and Ianoz M (2003), “A new expression for ground transient resistance
matrix elements of Multiconductor overhead transmission lines”, Journal of Electric Power System Research,
Vol. 65, No. 1, pp.41-46.
Rachidi F, Nucci CA and Ianoz M (1999), “Transient analysis of multiconductor lines above lossy ground”,
IEEE Transactions on Power Delivery, Vol. 14, No. 1, pp. 294-302.
Rachidi F, Nucci CA, Ianoz M and Mazzetti M (1997), “Response of multiconductor power lines to nearby
lightning return stroke electromagnetic fields”, IEEE Transactions on Power Delivery, Vol. 12, No. 3, pp.
1404-1411.
Rachidi F, Nucci CA, Ianoz M and Mazzetti M (1996), “Influence of a lossy ground on lightning induced
voltages on overhead lines”, IEEE Transactions on Electromagnetic Compatibility, Vol. 38, No. 3, pp. 250-
264.
Rakov VA and Uman MA (2003), “Lightning physics and effects”, Cambridge University press
Rizk F (1990), “Modeling of transmission line exposure to lightning strokes”, IEEE Transactions on Power
Delivery, Vol. 5, pp. 1983-1997.
Rubenstein M, Tzeng AY, Uman MA, Medelius PJ and Thomson EM (1989), “An experimental test of a
theory of lightning-induced voltages on an overhead wire”, IEEE Transactions on Electromagnetic
Compatibility, Vol. 31, No. 4, pp. 376-383.
Schütte T and Thiede J (2000), “Kombinierte Streckenspeisung mit Auto- und Saugtransformatoren”,
Elektrische Bahnen Vol. 98, No. 7, pp. 249 -253.
Semlyen A (1981), “Ground Return Parameters of Transmission Lines an Asymptotic Analysis for Very
High Frequencies”, IEEE Transactions on Power Apparatus and Systems, Vol. 100, No.3, pp. 1031-1038.
Stevenson WD (1988), “Elements of Power system analysis”, McGraw-Hill Book Company.
Sørensen T (1995), "Lightning Registration Systems: Analysis, Optimization and Utilization", Publication
Number 9505, Electric Power Engineering Department, Technical University of Denmark, DK-2800 Lyngby.
Taylor CD, Satterwhite RS and Harrison CW (1965), “The response of a terminated two-wire transmission
line excited by a nonuniform electromagnetic field”, IEEE Transactions on Antennas and Propagation, Vol.
13, No.6, pp. 987-989.
Technical Council of the IEEE Power Engineer Society (1999), “IEEE guide for the application of insulation
coordination”, IEEE Std. 1313.2.
Tesche FM, Ianoz MV and Karlsson T (1997), “EMC Analysis Methods and Computational Models”, John
Wiley and Sons Incorporation.
Theethayi N (2005), “Electromagnetic Interference in Distributed Outdoor Electrical Systems, with an
Emphasis on Lightning Interaction with Electrified Railway Network”, Uppsala University, ISBN 91-554-
6301-0.
Theethayi N and Thottappillil R (2005), “Pulse propagation in underground wires – A transmission line
analysis”, Proceedings of RVK05conference, Linköping, Sweden.
Theethayi N, Liu Y, Montaño R, Thottappillil R, Zitnik M, Cooray V and Scuka V (2005), “A theoretical
study on the consequence of a direct lightning strike to electrified railway system in Sweden”, Journal of
Electric Power Systems Research, Vol. 74, No. 2, pp. 267-280.
Theethayi N, Thottappillil R, Liu Y and Montaño R (in press), “Importance of Ground Impedance in the
Study of Parameters That Influence Crosstalk in Multiconductor Transmission Lines”, Journal of Electric
Power Systems Research, In press and accepted for publication.
Theethayi N, Thottappillil R, Yirdaw T, Liu Y, Götschl T and Montaño R (2007), “Experimental
Investigation of Lightning Transients Entering a Swedish Railway Facility”, IEEE Transactions on Power
Delivery, Vol. 22, No. 1, pp. 354-363.
Thottappillil R and Theethayi N (2005), “Lightning strikes to tall towers, with implications to
electromagnetic interference”, Proceedings of RVK05 conference, Linköping, Sweden.
Thottappillil R, Theethayi N and Zitnik M (2001), “Lightning protection of the Swedish railway network,
final technical report”, Division for Electricity and Lightning Research, Department of Engineering Sciences,
Uppsala University, Banverket project 2001-000282, 2001-03.
Trueblood HM and Wascheck G (1934), “Investigation of Rail Impedance”, AIEE Transactions, Vol. 53, pp.
1771-1780.
Zynovchenko A, Xie J, Jank S and Klier F (2006), “Impedance of contact lines and propagation of current
harmonics”, Elektrische Bahnen, No. 5, pp. 222-227.