IET Generation, Transmission & Distribution

Research Article

ISSN 1751-8687
Modelling, simulation, and verification for Received on 11th April 2015
Revised on 12th October 2015
detailed short-circuit analysis of a 1 × 25 kV Accepted on 27th October 2015
doi: 10.1049/iet-gtd.2015.0501
railway traction system www.ietdl.org

Tsai-Hsiang Chen, Rih-Neng Liao ✉

Department of Electrical Engineering, National Taiwan University of Science and Technology, 43, Keelung Road, Section 4,
Taipei 10607, Taiwan
✉ E-mail: [email protected]

Abstract: This study presents a modelling and simulation method for the analysis of short-circuits in 1 × 25 kV railway
traction systems (RTSs). First, the authors developed models of major components and a full-scale model of a RTS.
The component models developed in this study include a specific Le Blanc traction transformer and an overhead
catenary system. The models are represented using corresponding coupling-free equivalent circuits to facilitate
implementation using software packages MATLAB® /Simulink ® . The short-circuit fault simulation of a RTS was
conducted on a personal computer operating with MS Windows 7® . Field testing data was used to verify the
simulation results. The modelling and simulation technology developed in this study provide a valuable reference for
power engineers involved in the planning, design, and operation of a railway traction network.

1 Introduction regards to all major components and the overall RTS, while the
operating conditions of the RTS (both normal and abnormal) must
The electric locomotive is a strong contender for the title of the most
also be considered. In the technical literature, several studies have
efficient and economical motive power for high-density
been performed on: (i) traction system design, components
transportation systems. In the final decades of the twentieth
modelling, and calculation methods [6–11] and (ii) critical and
century, the steam railway system in Taiwan was converted to
short-circuit conditions [12–15]. A detailed short-circuit modelling
electricity. In late 2014, the railway electrification system from
of a RTS can be derived and implemented in the time and frequency
Hualien to Taitung was completed, thereby largely overcoming the
domains. In general, a frequency domain simulation is considered a
traffic problems in Eastern Taiwan. Power to the railway is
feasible solution for the steady-state analysis, and it is typically used
supplied by nearby traction substations fed from a three-phase
for both, system design and harmonic studies. The time domain
69-kV high-voltage grid of the Taiwan Power Company (TPC).
simulation is generally used for transient analysis, in which
Single-phase 1 × 25-kV and autotransformer-fed 2 × 25-kV
information on the waveform characteristics and time evolution are
traction systems are commonly used in modern railway
required. In [16, 17], a time-domain model of a complete 2 × 25-kV
electrification system, including high-speed railways [1, 2]. The
high-speed railway system was proposed using the MATLAB®/
short-circuit fault current in an autotransformer-fed 2 × 25-kV
Simulink® Power System Blockset (PSB) tool, where short-circuit
traction system tends to exceed that in a 1 × 25-kV traction system,
analyses were performed. To reduce the complexity of the
and fault locating is more complicated [3]. In 1 × 25-kV traction
simulations, simplified models were proposed in [14, 18, 19]. In
systems, the large single-phase non-linear traction loads supplied
[14], only three major lines (overhead contact wire, return feeder,
by three-phase power systems also tend to cause the voltage
and rail) in the overhead catenary system (OCS) are considered. In
unbalance, voltage fluctuation, harmonic pollution, and problems
[18, 19], simplified models are suggested to reduce the number of
associated with a poor power quality [4, 5].
phases of the system by grouping the conductors. These models are
Short-circuit analysis is critical in determining the rating of power
suitable for applications where details for each individual line are not
equipment and the implementation of protection devices, as well as
examined. Moreover, some components are mathematically
can also be used to check the voltage profile of a network,
represented by algebraic equations rather than by physical elements
particularly the busses near a fault. Particular attention must be paid to
(R, L, or C); thus, these component models cannot be directly
an accurate modelling of the traction transformers, autotransformers
applied to commercial circuit simulation programs. Although, several
as well as self and mutual impedances of the AC railway overhead
component models have been developed in existing commercial
contact line systems. In short-circuits analysis, the impedance of
software packages, including (i) a specific traction transformer and
system components and the configuration of the railway traction
(ii) a detailed overhead contact lines system (existing in MATLAB®/
system (RTS) are the main factors determining the short-circuit
Simulink® PSB tool; however, their functions are insufficient for
current. This paper examines a single-phase 1 × 25 kV RTS in which
detailed analysis such as asymmetrical structure simulation).
the key factors determining the fault current are as follows:
Therefore, to ensure accurate simulation and easy implementation of
(i) The short-circuit capacity (SCC) at the busbar on the primary RTSs, more detailed models of RTSs need to be developed.
side of the traction transformer; that is, the source transformer. In this study, we developed a system of modelling and simulated for
(ii) The winding connections (Le Blanc) and impedance of the analysis of short-circuits in a full-scale 1 × 25 kV RTS, including
transformers in traction substations. all major components. All models are represented using their
(iii) The self and mutual impedances of overhead contact systems corresponding coupling-free equivalent circuits to facilitate
(including overhead messenger wire, overhead trolley contact wire, implementation within in-house or commercial software packages,
and return feeders) as well as the rails. such as the Electromagnetic Transients Program, the Personal
Computer Simulation Program with Integrated Circuit Emphasis,
Estimating disturbances resulting from the various types of MATLAB®/Simulink®, or self-developed program for short-circuits
short-circuit faults require comprehensive data collection with analysis. We then applied the models to the development of a

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 1
Fig. 1 Arrangement of traction substations in Eastern Taiwan railway electrification system

short-circuit analysis program based on a bus (node) admittance transformer employs a special winding connection transformer (a
matrix. This makes it possible to derive the fault currents in each Le Blanc transformer) to facilitate the balancing of system voltage.
branch of a RTS as well as the profiles of fault currents along the The primary windings used with the traction transformer are a
route of the railway. It also enables the calculation of voltages in delta connection fed by two 69 kV sub-transmission lines from
overhead messenger wires, overhead trolley contact wires, and discrete TPC bulk power substations. The secondary windings
return feeders along the route of the railway. present a special Le Blanc connection, which provides two 25 kV
electric power with 90° phase displacement for the moving trains.
The traction substations are equipped with protective equipment,
2 Railway electrification system such as circuit breakers, disconnecting switches, and lightning
arresters.
The railway electrification system in Eastern Taiwan includes five
traction substations and four disconnecting switches, as shown in
Fig. 1. A remote operation control centre situated in Hualien is 3 Mathematical model
used to supervise and control moving trains and the entire RTS.
The power-supply zone for each traction substation is The elements essential to a RTS include traction substations with
approximately 40 km, each of which is further separated into two traction transformers and OCS comprising overhead messenger
power-supply sub-zones fed single-phase power from phases M wire, overhead trolley contact wire, return feeder, and rail. The
and T of a Le Blanc transformer in traction substations. This modelling of system components is the key process dictating the
necessitates the inclusion of a neutral zone to isolate out-of-phase success of system simulation. The precision of the simulation
electric power from phases M and T of the Le Blanc transformers. results as well as the convergence characteristics of solution
Traction substations are usually located in the centre of a techniques are determined by the system component models and
power-supply zone. Fig. 2 presents a single-line diagram of a full-scale network model. In this study, we employed a bus
traction substation typical of the Eastern Taiwan railway admittance matrix and coupling-free equivalent circuit for the
electrification system. As shown in Fig. 3, the traction substation development of the proposed models.

Fig. 2 Single-line diagram of typical traction substation in Eastern Taiwan Fig. 3 Le Blanc traction transformer connected using three single-phase
railway electrification system transformers with delta-connected primary windings

IET Gener. Transm. Distrib., pp. 1–12

2 & The Institution of Engineering and Technology 2016
3.1 Le Blanc traction transformer transformer with wye-connected primary windings, as shown in (5).

In the Eastern Taiwan railway electrification system, a special D

YLeBlanc = Y #1 + Y #2 + Y #3
winding connection transformer (a Le Blanc transformer) is used
in traction substations to alleviate system voltage imbalance caused A B C M+ M− T+ T−
by large single-phase traction loads. This approach is generally ⎡ ⎤
A 2y1 −y1 −y1 2y2 −2y2 0 0
preferred over a Scott connection in railway electrification system, ⎢ −y
due to a common core with three-limb, three-phase design is ⎢
B 1 2y1 −y1 −y2 y2 y3 −y3 ⎥
⎥
⎢ ⎥
employed for the construction of a Le Blanc transformer compared ⎢ −y1
C −y1 2y1 −y2 y2 −y3 y3 ⎥ (4)
⎢ ⎥
with two single-phase cores for the Scott transformer. In addition ⎢ ⎥
= M+ ⎢ 2y2 −y2 −y2 y3 −y3 0 0 ⎥
to a somewhat simpler standard core arrangement, the Le Blanc −⎢
⎢ ⎥
transformer is less costly to manufacture because for a given M ⎢ −2y2 y2 y2 −y3 y3 0 0 ⎥⎥
⎢ ⎥
rating, less active materials are required for its construction [20]. T+ ⎣ 0 y3 −y3 0 0 y3 −y3 ⎦
Fig. 3 illustrates the winding connection of a Le Blanc
transformer, in which the primary windings are in delta connection. T− 0 −y3 y3 0 0 −y3 y3
The mathematical model of a Le Blanc transformer can be derived
as follows: A B C M+ M− T+ T−
⎡ ⎤
2 1 1 1 1
A ⎢ − y1 − y1 − y1 y2 −y2 − y3 y
(i) The Le Blanc transformer with delta-connected primary
⎢ 3 3 3 3 3 3 ⎥⎥
windings is portioned into three sub-networks (i.e. sub-networks ⎢ 1 2 1 2 2 ⎥
⎢ y2 − y2 ⎥
#1, #2, and #3) as shown in Fig. 3. The mathematical models of B ⎢ − 3 y1 − y1 − y1 0
3 3
0
3 3 ⎥
⎢ ⎥
the three sub-networks (i.e. three single-phase transformers) can be ⎢ 1 1 2 1 1 ⎥
obtained using the algorithms described in [21]. According to the ⎢ − y1 − y1 − y1 −y2 y2 − y3 y3 ⎥
C ⎢ 3 3 3 3 3 ⎥
structure of the single-phase transformers, the corresponding bus YLeBlanc =
Y ⎢ ⎥
+ ⎢ y2 0 −y2 y3 −y3 0 0 ⎥
admittance matrix can be, respectively, represented using (1)–(3), M ⎢ ⎥
⎢ −y 0 y2 −y3 y3 0 0 ⎥
where yt represents the per unit leakage admittance of the M+ ⎢ ⎥
2
⎢ 1 ⎥
single-phase transformers, obtained in a short-circuit test. ⎢ − y
2 1
y2 − y3 0 −y3 ⎥
T+ ⎢ ⎥
0 y3
⎢ 3 3 3 3 ⎥
⎣ ⎦
T− 1
y
2
− y2
1
y3 0 0 −y3 y3
A B a1 a2 a3 a4 3 3 3 3
⎡ ⎤ (5)
A 2 −2 −b−11 b−1
1 −b−12 b−1
2
B⎢ ⎢ −2 2 b−1
1 −b−11 b−1
2 −b−12
⎥
⎥ where
⎢ ⎥
a1 ⎢ −b−1 b−1 b−2 −b−2 0 0 ⎥
Y #1
= ⎢ 1 1 1 1 ⎥ × yt √ √ √
a2 ⎢
⎢ b1
−1
−b−1 −b−2 b−2 0 0 ⎥ ⎥ 3− 3 3( 3 − 1) 3(3 − 3)
⎢ 1 1 1
⎥ y1 = yt , y2 = yt , y3 = yt
a3 ⎣ −b−12 b−1
2 0 0 b−2
2 −b−22
⎦ 2 4 4
a4 b−1
2 −b−12 0 0 −b−2
2
−2
b2
(1) (iii) According to (4) and (5), the corresponding coupling-free
equivalent circuits of the Le Blanc transformer are presented in

B C b1 b2
⎡ ⎤
B 1 −1 −(2b2 )−1 (2b2 )−1
C⎢ ⎢ −1 1 (2b2 )−1 −(2b2 )−1 ⎥
⎥
Y #2 = ⎢ ⎥ × yt
b1 ⎣ −(2b2 )−1 (2b2 )−1 (2b2 )−2 −(2b2 )−2 ⎦
b2 (2b2 )−1 −(2b2 )−1 −(2b2 )−2 (2b2 )−2
(2)

C A c1 c2 c3 c4
⎡ ⎤
C 2 −2 −b−11 b−1
1 −b−12 b−1
2
A⎢ ⎢ −2 2 b−1
1 −b−11 b−1
2 −b−1 ⎥
2 ⎥
⎢ ⎥
c ⎢ −b−1 b−1 b−2 −b−2 0 0 ⎥
Y #3 = 1⎢ 1 1 1 1 ⎥ × yt
c2 ⎢
⎢ b1
−1
−b−11 −b−21 b−2
1 0 0 ⎥ ⎥
⎢ ⎥
c3 ⎣ −b−12 b−1
2 0 0 b−2
2 −b−22
⎦
c4 b−1
2 −b−12 0 0 −b−2
2 b−2
2
(3)

(ii) The bus admittance matrix of a Le Blanc transformer with

delta-connected primary windings can then be obtained by
combining the bus admittance matrices of the three single-phase
transformers. In addition, the internal nodes (i.e. a3, a4, b2, c2, and Fig. 4 Coupling-free equivalent circuit of Le Blanc transformer
c4) can be removed using Kron’s law, as shown in (4). Following a Delta-connected primary windings
the same procedure, we can obtain a model of the Le Blanc b wye-connected primary windings

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 3
 Fig. 4a (with delta-connected primary windings) and Fig. 4b (with
wye-connected primary windings).

3.2 Overhead catenary system

Depending on traffic density, a railway electrification system may

consist of one, two, three, or four tracks. The OCS for Eastern
Taiwan railway electrification system consists of overhead
messenger wire, overhead contact wire, rails, return feeder 1 and
return feeder 2, as shown in Fig. 5, in which the overhead contact
system includes an overhead messenger wire, overhead trolley
contact wire and two return feeders. The circuit used for the return
of traction power to traction substations includes running rails and
return feeders.
The mathematical model of OCS was derived as follows:

(i) In general, an OCS has inherent geometrical asymmetrical

structure and unbalanced current carried in different lines;
therefore, its detailed model should consider the arrangement of
the overhead contact lines, taking into account the spacing
between overhead lines and running rails, and the physical
position for each individual overhead line. Carson proposed a
method to determine the self and mutual impedances for
Fig. 5 Geometric structure of AC railway traction pole used in Eastern overhead multi-conductor circuits considering the effect of the
Taiwan railway electrification system earth [22]. Further, the equations derived by Carson can be
applied to underground cables and railway OCSs. Carson used

Fig. 6 Schematic representation of self and mutual impedances of OCS in the Eastern Taiwan railway electrification system

Fig. 7 Coupling-free equivalent circuit of an OCS

IET Gener. Transm. Distrib., pp. 1–12

4 & The Institution of Engineering and Technology 2016
the method of images based on the assumption that the earth an (7) are used to find the self and mutual impedances,
equipotential plane, and therefore, the electric flux lines are respectively. The corresponding primitive impedance matrix
forced to cut the surface of the earth orthogonally. Hence, the [zpr] that considers the self and mutual impedances of the
earth effect can be represented by placing an oppositely charged overhead messenger wire, overhead contact wire, running rails,
conductor (image conductor) below the surface of the earth by a and two return feeders are given as
distance equal to that of the overhead conductor above the earth
for all conductors in the OCS. Carson’s original equations are ⎡ ⎤
zM zMC zMR zMF1 zMF2
given in [22]. Because Carson’s original equations are somewhat
complex, some approximations are made to simplify their  ⎢ z
⎢ MC zC zCR zCF1 zCF2 ⎥ ⎥
calculations. In [23], Carson’s modified equations are developed, zpr = ⎢
⎢ zMR zCR zR zRF1 zRF2 ⎥ ⎥ (9)
⎣ zMF zCF1 zRF1 zF1 zF1 F2 ⎦
and these equations are applied for calculating the primitive self 1
and mutual impedances of the overhead lines in an OCS in this zMF2 zCF2 zRF2 zF1 F2 zF2
study. It was assumed that: (i) the frequency of the railway
electrification system is 60 Hz, and (ii) the earth resistivity is
100 Ω-metre. A set of modified Carson’s equations are given in
(6) and (7). (iii) The branch-bus incident matrix C for Fig. 6 is as follows

1
ẑii = ri + 0.09327 + j0.12134 ln + 7.95153 (6) bus i bus j
GMRi
    M C R F1 F2 m c r f1 f2
ẑij = 0.09327 + j0.12134 ln
1
+ 7.95153 (7) ⎡ ⎤
Dij #1 1 0 0 0 0 −1 0 0 0 0
#2 ⎢
⎢0 1 0 0 0 0 −1 0 0 0 ⎥ ⎥ (10)
⎢ ⎥
where ẑii : self-impedance of conductor i in Ω/mile; ẑij : C = brach #3 ⎢
⎢0 0 1 0 0 0 0 −1 0 0 ⎥ ⎥
mutual-impedance between conductors i and j in Ω/mile; ri: ⎢ ⎥
#4 ⎣ 0 0 0 1 0 0 0 0 −1 0 ⎦
resistance of conductor i in Ω/mile; GMRi: geometric mean
radius of conductors i in feet; and Dij: distance between #5 0 0 0 0 1 0 0 0 0 −1
conductors i and j in feet.
(ii) Fig. 6 presents an equivalent schematic circuit of an OCS
including running rails. All overhead lines are modelled as a π (iv) Using (9) and (10), the bus admittance matrix of railway OCS
model. However, all shunt capacitances of the overhead lines are was obtained as (11). (see (11))
disregarded because the power-supply zones of phases T and M of
the traction transformer from traction substations are quite short,
usually less than 20 km.

The primitive equations that relate the node voltages and

branch currents of overhead lines in an OCS are shown in a
compact form as
⎡ ⎤ ⎡ ⎤ ⎡ ⎤
VM Vm zM zMC zMR zMF1 zMF2
⎢ VC ⎥ ⎢ Vc ⎥ ⎢ zMC zC zCR zCF1 zCF2 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ VR ⎥ = ⎢ Vr ⎥ + l × ⎢ zMR zCR zR zRF1 zRF2 ⎥
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ VF ⎦ ⎣ V f ⎦ ⎣ zMF zCF1 zRF1 zF1 zF1 F2 ⎦
1 1 1
VF2 V f2 zMF2 zCF2 zRF2 zF1 F2 zF2
⎡ ⎤
IM
⎢ IC ⎥
⎢ ⎥
×⎢⎢ IR ⎥
⎥ (8)
⎣ IF ⎦
1
IF2
Fig. 8 One-line diagram of the Eastern Taiwan railway electrification
where l is the overhead line length of an OCS. Further, (6) and system at Guanshan traction substation

⎡ ⎤
yM yMC yMR yMF1 yMF2 −yMm −yMc −yMr −yM f1 −yM f2
⎢ ⎥
⎢ yMC yC yCR yCF1 yCF2 −yCm −yCc −yCr −yC f1 −yC f2 ⎥
⎢ ⎥
⎢ yMR yCR yR yRF1 yRF2 −yRm −yRc −yRr −yR f1 −yR f2 ⎥
⎢ ⎥
⎢ ⎥
⎢ yMF1 yCF1 yRF1 yF1 yF1 F2 −yF1 m −yF1 c −yF1 r −yF1 f1 −yF1 f2 ⎥
⎢ ⎥
−1 ⎢ y yCF2 yRF2 yF1 F2 yF2 −yF2 m −yF2 c −yF2 r −yF2 f1 −yF2 f2 ⎥
⎢ MF2 ⎥
Y RES = C zpr C = ⎢
t
⎥ (11)
⎢ −yMm −yCm −yRm −yF1 m −yF2 m ym ymc ymr ym f1 ym f2 ⎥
⎢ ⎥
⎢ −y −yCc −yRc −yF1 c −yF2 c ymc yc ycr yc f1 yc f2 ⎥
⎢ Mc ⎥
⎢ ⎥
⎢ −yMr −yCr −yRr −yF1 r −yF2 r ymr ycr yr yr f1 yr f2 ⎥
⎢ ⎥
⎢ −y −yC f1 −yR f1 −yF1 f1 −yF2 f1 y f1 f2 ⎥
⎣ M f1 ym f1 yc f1 yr f1 y f1 ⎦
−yM f2 −yC f2 −yR f2 −yF1 f2 −yF2 f2 ym f2 yc f2 yr f2 y f1 f2 y f2

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 5
Fig. 9 Block diagrams created using MATLAB®/Simulink®
a Full-scale model of the Eastern Taiwan railway electrification system at a typical traction substation area
b Coupling-free equivalent circuit of a Le Blanc traction transformer
c Coupling-free equivalent circuit of an OCS

(v) According to (11), the corresponding coupling-free equivalent power sources is represented by a Thevenin’s equivalent
circuit of the OCS is shown in Fig. 7. circuit [24].

4 Implementation procedure

3.3 Utility power source Short-circuit field testing in a high-voltage network can be highly
dangerous and time-consuming; therefore, only limited testing
The impedance of the power-supply source should be taken into cases are subsidised. In contrast, a precise component model and a
account for fault analysis because of its sizable effect on the full-scale RTS model can be used to obtain nearly the same results
fault current. The equivalent impedance of the source can be as those obtained from the testing field. An appropriate computer
obtained by converting the SCC on the source side of the program not only helps engineers select, design, and operate a
traction transformer. In this paper, the circuit model of a utility RTS, but also provides a useful tool for determining the correct

IET Gener. Transm. Distrib., pp. 1–12

6 & The Institution of Engineering and Technology 2016
Fig. 9 (Continued).

short-circuit currents for various fault scenarios. A commercial simulation results and confusing conclusions, and may lead to
simulation software package called MATLAB®/Simulink®, which program divergence. All required parameters of the major
has powerful simulation ability and a user-friendly graphical components in the target RTS must be collected.
interface, was adopted to simulate the RTS of the Eastern Taiwan (iii) Building a full-scale model of the target RTS According to the
railway electrification system. All major components need to be structure of the target RTS, a full-scale system model can be built by
properly modelled by their corresponding equivalent circuits combining the major component models, such as Le Blanc traction
before the RTS can be simulated using MATLAB®/Simulink®. transformer and OCS.
The Le Blanc traction transformer and OCS can be represented
explicitly in the MATLAB®/Simulink®. Hence, the OCS can be The one-line diagram of the Eastern Taiwan railway electrification
analysed in more detail. The steps for modelling the RTS of the system at Guanshan substation is shown in Fig. 8 and its
Eastern Taiwan railway electrification system are as follows: corresponding full-scale network model implemented using
MATLAB®/Simulink® is illustrated in Fig. 9a. The branches and
nodes of the major components, such as the Le Blanc traction
(i) Forming equivalent circuit models for all major components in transformer and OCS are explicitly represented as shown in
the target RTS In MATLAB®/Simulink®, the coupling-free Figs. 9b–c.
equivalent circuits can be represented by simple elements such as
resistance, inductance, capacitance, and voltage/current sources,
that is, coupling-free equivalent circuits can be modified and
modularised to meet the requirements of different simulation 5 Field testing and simulation results
objectives, and then be implemented with MATLAB®/Simulink®.
Modularisation is useful for MATLAB®/Simulink® functions 5.1 Field testing
where users can construct their own component models and
full-scale system models. A short-circuit fault was field tested within the power-supply zone of
(ii) Setting necessary parameters for all system components the Guanshan traction substation [25]. A solid short-circuit fault was
Unsuitable or incorrect parameters may result in incorrect introduced between the overhead contact line and the running rail.

Table 1 Parameters of the overhead contact line and running rail of the target OCS
Measuring Measurement data Calculated data
point
AC voltage, AC current, cos θ Active power, P Reactive power, Q Resistance, R Reactance, X Impedance, Z Angle,
V A (kW) (kvar) (Ω) (Ω) (Ω) θ

Dongzhu 305.40 33.97 0.2554 2.64 10.03 2.29 8.69 8.990 75.2
Luye 305.40 41.30 0.2870 3.62 12.07 2.05 7.10 7.394 73.9

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 7
 (i) Ensure that there is no train in the power-supply zone and the
OCS is de-energised and temporarily grounded so that there is no
residual voltage on the target overhead contact lines.
(ii) Short-circuit the overhead contact line and the rail at the fault
points, that are located at the Dongzhu (Case 1) and Luye (Case 2)
train stations.
(iii) Measure the parameters of the target overhead contact lines and
calculate their corresponding impedance and resistance. The
overhead contact line system is a complicated system; the
magnitude, phase angle of voltage and current, and the active and
reactive powers are measured while a proper current is injected
between the overhead contact line and the rail at the traction
substation. Next, the equivalent impedance of the overhead contact
line and running rail of the target power-supply zone is calculated.
Table 1 summarises the measured and calculated results. This table
indicates that the system equivalent resistance at the Luye train
station is less compared with that at the Dongzhu’s train station.
Fig. 10 Schematic connection diagram for short-circuit current (iv) Using the calculated data in Table 1, along with the given
measurement traction transformer impedance and system equivalent impedance
at the primary side of the traction transformer, calculate the
short-circuit current before the real short-circuit is performed. The
calculated fault currents were applied to the settings of the fault
To reveal the variations in the short-circuit current along the current recording instruments so that the waveform of the
power-supply zone, the ends of the power-supply zone – phases M short-circuit current can be exactly recorded, and unnecessary
and T of the transformer from the Guanshan traction substation – repetitive tests can be avoided.
were selected as fault points that are located at the Dongzhu (Case (v) Reenergise the target overhead contact line before the
1) and Luye (Case 2) train stations, respectively (see Fig. 1). short-circuit test are conducted. More functions of the related relay
The short-circuit fault testing procedure is as follows: and vacuum circuit breaker (VCB) should be endorsed.

Fig. 11 Field testing result of short-circuit fault

a (ISC-case1) at Dongzhu train station
b (ISC-case2) at Luye train station

IET Gener. Transm. Distrib., pp. 1–12

8 & The Institution of Engineering and Technology 2016
Table 2 Parameters used in railway electrification system output current of the current transformer (CT) to voltage signals.
The CT was installed on the secondary side of the Le Blanc
Conductors Resistance, Ω/mile GMR, ft. X, m Y, m
traction transformer; therefore, the transient fault current could be
messenger wire 0.19490 7.2033 3.15 5.95
recorded using this transient recorder. Fig. 10 illustrates the
contact wire 0.16355 12.000 2.95 4.75 connection for short-circuit current measurement wherein the
rails 0.04076 45.160 2.62 0.00 overcurrent relay (51) with automatic action capability is
return feeder 1 0.27210 9.5557 −0.20 6.05 automatically triggered to interrupt the VCB to isolate the faulted
return feeder 2 0.27210 9.5557 −0.20 5.30
zone if a short-circuit fault occurs.
The transient recorders record the current waveform when a
short-circuit fault occurred, whereupon a VCB caused an
(vi) Perform short-circuit tests, record the transient fault currents, interruption after approximately four cycles as shown in Fig. 11.
and analyse the fault current waveform for comparison with Fig. 11a presents the results of field testing of a short-circuit fault
simulation results. on the phase M at Dongzhu train station. Fig. 11b presents the
results of field testing of a short-circuit fault on the phase T, at
Luye train station.
In most cases, short-circuit current would include components of
The short-circuit test at high voltage system is not only dangerous symmetric AC current as well as asymmetrical DC. The
but also highly expensive. The equipment, overhead contact line asymmetrical DC component of the fault current lasts only a few
circuit, and all the connections should be checked in detail cycles, as shown in Figs. 11a and b. However, the asymmetrical
visually and using instrument inspection equipment before the DC component of the fault current mainly depends on momentary
short-circuit test can be carried out. In addition, the testing voltage, system X/R ratio, and the timing of the short-circuit fault.
procedure should be strictly followed to ensure safety. Moreover, Thus, the maximum direct short-circuit current was difficult to
the route of the overhead catenary line system should be guarded, obtain by controlling the voltage injection time.
especially at railway crossings. All personnel on duty at fault Furthermore, the field testing results in Figs. 11a and b provide a
points, measuring locations, traction substations, and railway valuable reference for use in verifying the precision of simulation
crossings, should keep close contact using telephones or mobile results. For example, each peak value of the measured fault current
phones. These requirements confirm that the short-circuit test is waveform can be used to calculate the time constant and X/R of
undoubtedly time-consuming and expensive. Therefore, in this the fault point, whereas the recorded amplitude is used to calculate
study, only two of these tests were completed owing to time and the AC fault current. All of the fault currents are in root mean
cost constraints. square (rms) values; therefore, momentary fault current can be
To measure the fault current waveform during the short-circuit obtained by multiplying the recorded peak voltage value
test, the current-to-voltage carbon stick was used to convert the (representing the short-circuit fault current) by the turn ratio

Fig. 12 Simulation results of short-circuit fault

a (ISC-case1) at Dongzhu train station
b (ISC-case2) at Luye train station

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 9
settings of the carbon stick and CT, which is then divided . On the Table 3 Comparison of short-circuit rms currents obtained through
basis of the short-circuit fault current waveform shown in field testing and simulation
Figs. 11a and b, their corresponding actual rms short-circuit fault Field testing Fault distance from Fault rms current, A
currents can be calculated is as following: point substation, km
Field Simulation
testing
(i) ISC-case1 = (50 mV/DIV) × (1.33 DIV) × (0.707) × (0.02 A/1
mV) × (1000/1 A) = 940.45 A, for fault at Dongzhu train station. Dongzhu 19.935 940.5 948
(ii) ISC-case2 = (50 mV/DIV) × (1.48 DIV) × (0.707) × (0.02 A/1 Luye 16.377 1046 1026
mV) × (1000/1 A) = 1046 A, for fault at Luye train station.
(iv) Ground resistivity was assumed to be 100 Ω-m. The railway
poles are grounded with multiple grounding branches used to
5.2 Parameters prevent harmonic current flowing through the rail from interfering
with nearby communication lines.
The parameters used in the simulation are listed as follows: (v) The parameters used for the AC railway line (see Fig. 5) of the
Eastern Taiwan of railway electrification system are listed in Table 2
(i) The equivalent impedance of the utility power source is using a Cartesian coordinate system [26].
calculated according to a SCC of 364.5 MVA and rated voltage of
69 kV, as provided by the TPC. This voltage source was
considered ideal and the phase-to-phase voltages are specified. 5.3 Simulation
(ii) The parameters used in the Le Blanc traction transformer are as
follows: primary windings in delta connection, rated capacity 25 Short-circuit faults were simulated using the full-scale model of the
MVA, 69/25 kV, the leakage impedance of 10.56%. railway electrification system established in the previous section.
(iii) The distance between two railway poles is 50 m. The distance Simulation enabled the control of the phase angle of the pre-fault
between Dongzhu train station and Guanshan substation is 19.935 voltage. Figs. 12a and b presents the simulation results of
km. The distance between Luye train station and Guanshan short-circuit current waveforms. As shown in Table 3, a
substation is 16.377 km. comparison of results from computer simulation and field testing

Table 4 Short-circuit currents and voltages under various fault conditions

System operating mode| RSC, Ω d, km Branch current, A Voltage, kV

ISC IM IC IR IF VTr

Tr.#1: on Tr.#2: off 0 20 963.7∠97.9° 443.8∠98.3° 519.9∠97.6° 520.6∠−87.1° 447.4∠−76.3° 11.98

18 1011.7∠97.5° 465.9∠97.9° 545.8∠97.2° 546.5∠−87.5° 469.7∠−76.6° 11.31
16 1064.7∠97.1° 490.3∠97.5° 574.4∠96.8° 575.1∠−87.9° 494.3∠−77.1° 10.58
12 1188.9∠96.1° 547.5∠96.5° 641.4∠95.8° 642.2∠−88.9° 552.0∠−78.1° 8.864
8 1345.4∠94.8° 619.6∠95.2° 725.8∠94.5° 726.8∠−90.2° 624.6∠−79.4° 6.687
4 1548.1∠93.1° 712.9∠93.5° 835.2∠92.8° 836.3∠−91.9° 718.7∠−81.1° 3.847
2 1673.5∠92.1° 770.7∠92.5° 902.9∠91.8° 904.0∠−92.9° 776.9∠−82.1° 2.080
1 20 955.2∠101.0° 439.9∠101.4° 515.3∠100.7° 516.0∠−84.0° 443.4∠−73.2° 12.30
18 1002.6∠100.8° 461.7∠101.2° 540.9∠100.4° 541.6∠−84.2° 465.5∠−73.4° 11.68
16 1055.0∠100.5° 485.8∠100.9° 569.1∠100.2° 569.9∠−84.5° 489.8∠−73.7° 10.99
12 1177.9∠99.9° 542.5∠100.3° 635.5∠99.6° 636.3∠−85.1° 546.8∠−74.3° 9.372
8 1333.1∠99.1° 613.9∠99.5° 719.2∠98.8° 720.1∠−85.9° 618.9∠−75.1° 7.365
4 1534.9∠98.1° 706.9∠98.5° 828.1∠97.8° 829.2∠−86.9° 712.6∠−76.1° 4.874
2 1660.4∠97.5° 764.6∠97.9° 895.8∠97.1° 896.9∠−87.5° 770.8∠−76.7° 3.521
5 20 898.9∠112.5° 413.9∠112.9° 484.9∠112.2° 485.6∠−72.5° 417.3∠−61.7° 14.27
18 940.6∠112.8° 433.2∠113.2° 507.4∠112.5° 508.1∠−72.2° 436.7∠−61.4° 13.89
16 986.3∠113.2° 454.2∠113.6° 532.1∠112.8° 532.8∠−71.8° 457.9∠−61.0° 13.49
12 1092.4∠114.0° 503.1∠114.4° 589.4∠113.6° 590.1∠−71.0° 507.2∠−60.2° 12.65
8 1223.8∠115.0° 563.6∠115.4° 660.2∠114.6° 661.1∠−70.0° 568.2∠−59.2° 11.85
4 1390.5∠116.3° 640.3∠116.7° 750.2∠115.9° 751.1∠−68.7° 645.5∠−57.9° 11.27
2 1491.7∠117.0° 687.0∠117.4° 804.8∠116.7° 805.8∠−68.0° 692.5∠−57.2° 11.19
Tr.#1: on Tr.#2: on 0 20 1037.9∠98.3° 478.0∠98.7° 560.0∠98.0° 560.7∠−86.7° 481.8∠−75.9° 12.89
18 1093.9∠97.9° 503.8∠98.3° 590.2∠97.6° 590.9∠−87.1° 507.8∠−76.3° 12.23
16 1156.1∠97.4° 532.4∠97.9° 623.7∠97.1° 623.7∠−87.6° 536.7∠−76.7° 11.49
12 1304.2∠96.4° 600.6∠96.8° 703.6∠96.0° 704.5∠−88.6° 605.5∠−77.8° 9.724
8 1495.1∠95.0° 688.5∠95.4° 806.6∠94.6° 807.6∠−90.0° 694.1∠−79.2° 7.431
4 1749.8∠93.1° 805.8∠93.5° 944.0∠92.8° 945.2∠−91.9° 812.4∠−81.1° 4.349
2 1911.7∠91.9° 880.4∠92.3° 1031.4∠91.6° 1032.7∠−93.1° 887.5∠−82.3° 2.376
1 20 1027.5∠101.6° 473.2∠102.0° 554.4∠101.3° 555.1∠−83.4° 477.0∠−72.6° 13.24
18 1082.7∠101.4° 498.6∠101.8° 584.1∠101.0° 584.8∠−83.6° 502.6∠−72.8° 12.61
16 1144.0∠101.1° 526.9∠101.5° 617.2∠100.8° 618.0∠−83.9° 531.1∠−73.1° 11.91
12 1290.2∠100.5° 594.2∠100.9° 696.1∠100.2° 697.0∠−84.5° 599.0∠−43.7° 10.27
8 1479.0∠99.8° 681.1∠100.4° 797.9∠99.4° 799.0∠−85.2° 686.7∠−74.4° 8.171
4 1732.1∠98.7° 797.7∠99.1° 934.5∠98.4° 935.7∠−86.3° 804.1∠−75.4° 5.500
2 1893.8∠98.1° 872.1∠98.5° 1021.7∠97.7° 1023.0∠−86.9° 879.2∠−76.1° 4.016
5 20 959.2∠113.9° 441.7∠114.3° 517.5∠113.5° 518.2∠−71.1° 445.3∠−60.3° 15.23
18 1006.7∠114.3° 463.6∠114.7° 543.1∠113.9° 543.8∠−70.7° 467.4∠−59.9° 14.86
16 1059.1∠114.7° 487.7∠115.1° 571.4∠114.4° 572.1∠−70.3° 491.7∠−59.5° 14.48
12 1181.8∠115.8° 544.2∠116.2° 637.6∠115.4° 638.4∠−69.2° 548.7∠−58.4° 13.69
8 1336.0∠117.1° 615.3∠117.5° 720.8∠116.8° 721.7∠−67.9° 620.3∠−57.1° 12.94
4 1535.4∠118.8° 707.1∠119.2° 828.3∠118.5° 829.4∠−66.2° 712.8∠−55.4° 12.45
2 1658.4∠119.9° 763.7∠120.3° 894.7∠119.6° 895.8∠−65.1° 769.9∠−54.3° 12.44

Notes: The variables are described as follows; d: distance of fault location away from traction substation; RSC: fault resistance; ISC: short-circuit fault
current; IM: fault current in overhead messenger wire; IC: fault current in overhead contact wire; IF: the sum of fault currents in the two return feeders; and
VTr: voltage at the traction station outgoing busbar

IET Gener. Transm. Distrib., pp. 1–12

10 & The Institution of Engineering and Technology 2016
revealed a difference of between 0.79% and 1.91%. The factors conditions. Finally, the simulation and field test results was
contributing to this variation are as follows: compared. The comparison indicated that the proposed models can
precisely retrieve the characteristics of a 1 × 25 kV railway
(i) Visual error: This is the primary source of error due to the fact electrification system for short-circuit analysis. Based on the above
that even slight visual errors can be magnified when a short-circuit facts, the proposed models and solution approach could provide a
fault of up to 1000 A appears on the screen of the transient recorder. valuable reference for power engineers involved in the planning,
(ii) Instrumentation error: The instruments used for testing are design, and operation of a railway traction network.
likely to introduce measurement error. In addition, the proposed models and solution approach can be
(iii) The validity of the components and full-scale models, the used to perform more simulations, and therefore, help detect the
technical solutions that are selected, and the assumptions made for features and limitations of the target system that cannot be done
the simulation can all exert a profound effect on the results. only through field tests. At present, the proposed models are
being used to optimise and upgrade the Taiwan railway’s
electrification system. Future works will aim at power flow study,
Table 3 confirms that the proposed models can precisely retrieve the harmonic analysis, voltage drop calculation, substation grounding
characteristics of a 1 × 25 kV railway electrification system for system analysis, and protective relay coordination study.
short-circuit analysis. In addition, the proposed models and solution
approach can be used to make more detailed simulations, and
therefore, can detect the features and limitations of the target system
that cannot be done by only the field tests. The following detailed
7 Acknowledgment
short-circuit analysis results were obtained by the proposed approach.
The author thank the CECI Engineering Consultants, Inc., Taiwan
for providing the measurement data.
(i) For the short-circuit fault occurred at Dongzhu train station
(Case 1), the detailed steady-state short-circuit currents flowing
through the individual contact lines in the OCS are identified as
follows: ISC-case1 = IM + IC = –(IR + IF) = 965.21∠97.92° A, IM = 8 References
444.5∠98.32° A, IC = 520.73∠97.57° A, IR = 521.4∠−87.08° A,
and IF = IF1 + IF2 = 448.1∠−76.26° A (the sum of the currents 1 Courtois, C.: ‘Why the 2 × 25 kV alternative? [autotransformer traction supply]’.
flowing in the two return feeders). The steady-state voltage of the IEE Colloquium on 50 kV Autotransformer Traction Supply System – the
French Experience, November 1993, pp. 1–4
traction station outgoing busbar can also be identified, that is 2 Roussel, H.: ‘Power supply for the Atlantic TGV high speed line’. Int. Conf. on
11.96∠172.18° kV with a voltage deviation of 52.2%. Main Line Railway Electrification, Set. 1989, pp. 388–392
(ii) For the short-circuit fault that occurred at the Luye train station (Case 3 Millard, A., Taylor, I.A., Weller, G.C.: ‘AC electrified railways – protection and
2), the detailed steady-state short-circuit currents flowing through the distance to fault measurement’. Int. Conf. on Electric Railways in United
Europe, March 1995, pp. 73–77
individual contact lines in the OCS are as follows: ISC-case2 = 4 Bailey, D.I., Jones, K.M.: ‘The supply of large rail traction loads from small
1054.28∠7.19° A, IM = 485.52∠7.60° A, IC = 568.78∠6.85° A, IR = isolated power systems’, IET Conference Publication No. 210, Sources Eff.
569.51∠−177.87° A, and IF = 489.45∠−166.99° A. The steady-state Power Syst. Disturb., 1982, pp. 260–266
voltage of traction station outlet end is 10.73∠81.45° kV with a 5 Ross, B.A.: ‘Meeting tomorrow’s railroad power requirements’, IEEE Trans.
Power Appar. Syst., 1971, PAS-90, (2), pp. 393–400
voltage deviation of 57.08%. 6 Hill, R.J., Brillante, S., Leonard, P.J.: ‘Railway track transmission line parameters
from finite element field modelling: series impedance’, IEE Proc. – Electr. Power
Moreover, two fault impedances (0, 1, and 5 Ω), seven fault Appl., 1999, 146, (6), pp. 647–660
7 Hill, R.J., Brillante, S., Leonard, P.J.: ‘Railway track transmission line parameters
locations (2, 4, 8, 12, 16, 18, and 20 km from traction substation) from finite element field modelling: shunt admittance’, IEE Proc. – Electr. Power
and two system operating conditions (Tr #1: on, Tr #2: off; Tr #1: Appl., 2000, 147, (3), pp. 227–238
on, Tr #2: on) were simulated for considering the effect of fault 8 Mingli, W., Chengshan, X., Fan, Y., et al.: ‘Performance and mathematical model
impedance, fault location, and system configuration on the fault of three-phase three-winding transformer used in 2 × 25 kV electric railway’, IEE
Proc. – Electr. Power Appl., 2006, 153, (2), pp. 271–281
currents and voltages. The simulation results are summarised in 9 Mariscotti, A., Pozzobon, P.: ‘Determination of the electrical parameters of railway
Table 4. The short-circuit currents provide significant information traction lines: calculation, measurements, and reference data’, IEEE Trans. Power
for the selection of protective relays and circuit breakers. Deliv., 2004, 19, (4), pp. 1538–1546
Accordingly, the proposed approach provides a valuable reference 10 Mariscotti, A., Pozzobon, P.: ‘Resistence and internal inductance of traction rails at
power frequency: a survey’, IEEE Trans. Veh. Technol., 2004, 53, (4),
for power engineers involved in the planning, design, and pp. 1069–1075
operation of a railway traction network. 11 Kneschke, T.A., Hong, J., Natarajan, R., et al.: ‘Impedance calculations for
SEPTA’s rail power distribution system’. Proc. 1995 IEEE/ASME Joint Railroad
Conf., April 1995, pp. 79–85
12 Natarajan, R., Kneschke, T.A., Naqvi, W., et al.: ‘Short circuit currents of the
6 Conclusions SEPTA traction power distribution system’, IEEE Ind. Appl. Mag., 1997, 3, (6),
pp. 52–59
13 Cella, R., Giangaspero, G., Mariscotti, A., et al.: ‘Measurements of AT electric
This study developed mathematical models of the major components railway system currents at power supply frequency and validation of a
as well as a full-scale representation of a 1 × 25 kV railway multiconductor transmission line model’, IEEE Trans. Power Deliv., 2006, 21,
electrification system for analysis of short-circuits. The proposed (3), pp. 1721–1726
14 Chen, T.H., Hsu, Y.F.: ‘Systematized short-circuit analysis of a 2 × 25 kV electric
models of a Le Blanc transformer and a RTS are represented using traction network’, Electr. Power Syst. Res., 1998, 47, (2), pp. 133–142
their corresponding coupling-free equivalent circuits and 15 Battistelli, L., Pagano, M., Proto, D.: ‘Short circuit modelling and simulation of 2 ×
implemented using the commercial software package, MATLAB®/ 25 kV high speed railways’. 2008. AICMS 08. Second Asia Int. Conf. on
Simulink®. The models developed in this paper could be applied Modelling and Simulation, May 2008, pp. 702–707
16 Battistelli, L., Pagano, M., Proto, D.: ‘2 × 25-kV 50 Hz high-speed traction power
to the simulation of short-circuit faults occurring in any location of system: short-circuit modeling’, IEEE Trans. Power Deliv., 2011, 26, (3),
a railway electrification system as a means of avoiding the dangers pp. 1459–1466
and costs associated with field testing. 17 Amendola, A., Battistelli, L., Candurro, L., et al.: ‘Steady state and transient short
The Eastern Taiwan railway electrification system was introduced circuit analysis of 2 × 25 kV high speed railways’. Proc. 8th World Congress on
Railway Research, May 2008
and used as a demonstration system for short-circuit analysis. The 18 Lee, H., Lee, C., Jang, G., et al.: ‘Harmonic analysis of the korean high-speed
full-scale model of the demonstration system was created by railway using the eight-port representation model’, IEEE Trans. Power Deliv.,
combining the equivalent circuit models for its major components 2006, 21, (2), pp. 979–986
such as the Le Blanc traction transformer model and the OCS 19 Mariscotti, A., Pozzobon, P., Vanti, M.: ‘Simplified modelling of a 2 × 25-kV AT
railway system for the solution of a low frequency and large-scale problems’, IEEE
model. The coupling-free equivalent circuit of the full-scale model Trans. Power Deliv., 2007, 22, (1), pp. 296–301
was implemented using MATLAB®/Simulink®. Next, a series of 20 Stigant, S.A., Franklin, A.C.: ‘The J & P transformer book: a practical technology
digital simulations were performed for various short-circuit of the power transformer’ (London, Boston, 1973), pp. 150–205

IET Gener. Transm. Distrib., pp. 1–12

& The Institution of Engineering and Technology 2016 11
21 Chen, T.H., Huang, M.Y.: ‘Network modelling of 24-pulse rectifier transformers 24 Chen, T.H., Kuo, H.Y.: ‘Network modelling of traction substation transformers for
for rigorous simulation of rail transit power systems’, Electr. Power Syst. Res., studying unbalance effects’, IEE Proc.-Gener. Transm. Distrib., 1995, 142, (2), pp.
1999, 50, (1), pp. 23–33 103–108
22 Carson, J.R.: ‘Wave propagation in overhead wires with ground return’, Bell Syst. 25 Tatung Company: ‘East railway improvement project – power supply system: static
Tech. J., 1926, 5, (4), pp. 539–554 test report’ (Railway Reconstruction Bureau Ministry of Transportation and
23 Kersting, W.H., Green, R.K.: ‘The application of Carson’s equation to the Communications, 2011), pp. 221–224
steady-state analysis of distribution feeders’. Power Systems Conf. and 26 Chang, S.H.: ‘Analysis of voltage unbalance in the electric railway depot using
Exposition, March 2011, pp. 1–6 two-port network model’, Korean Inst. Electr. Eng., 2001, 50, pp. 248–256

IET Gener. Transm. Distrib., pp. 1–12

12 & The Institution of Engineering and Technology 2016