A real-time DSP based quadrilateral relay for distance protection of 25 kV AC traction overhead equipment

Abstract

This paper presents the design, implementation and testing of a single-phase distance relaying scheme for 25 kV 50 Hz AC traction system using a Texas instruments TMS320C50 digital signal processor (DSP). The three-phase system with substations, track section with rectifier-fed DC locomotives and a detailed traction load are modeled using power system block set (PSB)/SIMULINK software package. The model has been used to study the effect of loading and fault conditions in 25 kV AC traction. The relay characteristic proposed is a combination of two quadrilaterals in the X-R plane in which resistance and reactance reaches are independently controllable. The algorithms, hardware and software are also briefly described.

Full text

Page 1

2004 tntemational Conference on
Power System Technology - POWERCON 2004
Singapore, 21-24 November 2004
A Real-time DSP based quadrilateral relay for distance
protection of 25 kV AC traction overhead equipment
U. J. Shenoy, Member, IEEE, K. G. Sheshadri, K. Parthasarathy, Senior Member, IEEE,
H. P. Khincha, Senior Member, IEEE, D. Thukaram, Senior Member, IEEE
Abstruct-This paper presents the design, implementation and
testing of a single-phase distance relaying scheme for 25 kV 50
Hz AC traction system using a Texas Instruments TMS320C50
digital signal processor @ S P ) , The three-phase system with
substations, track section with rectifier-fed DC locomotives and a
detailed traction load are modeled using Power System Block set
(PSB) I SlMULlNK software package. The model has been used
to study the effect of loading and fault conditions in 25 kV AC
traction. The relay characteristic proposed is a combination of
two quadrilaterals in the X-R plane in which resistance and
reactance reaches are independentiy controllable. The
algorithms, hardware and software are also briefly described.
Inda Terms-Traction system, PSB/SIMULINK, Wrong
phase coupling, TMS32OC50 DSP, Quadrlateral relay
characteristic
I. INTRODUCTION
he function of an AC traction system is to deliver power
T to the locomotives as efficiently and economically as
possible. Problems involved in providing protection to
traction systems are different from those faced in protecting
other transmission lines or distribution systems workmg at the
same voltage level. This is due to the continuous movement of
locomotive load, change in the length of the line during
operation, nature of loading, voltage drop due to the flow of
the lagging reactive current in inductive components of the
overhead system and the high levels of harmonic distortion
[I]. The situation is firther aggravated due to the use of DC
series motors in electric locomotives, which d r a w large
current on starting. It may happen at times that several
locomotives run in the same section of the overhead
equipment (OHE), leading to large increase in load. The
impedance seen by the relay on such heavy loads may be even
smaller than that on distant earth faults. Fig. 1 shows the
typical feeding arrangement of a 25 kV electrified railway
system The load current drawn by locomotives is rich in large
odd harmonic components [Z].
substations are fed from different phases of the three-phase
supply in rotation having a phase difference of 120'. The
supply to the OHE can be switched ONiOFF through
intemptors. Normally power supply from the traction
substation extends upto the sectioning post (SP) on either side
of the substation, but in case of an emergency necessitating
total shut down of the substation, it can be extended upto the
failed substation by closing the bridging intemptors at the
The adjacent traction
U.J.Shenoy, corresponding author (e-mail: u j s@ee . iisc. ernet. in) is
Senior Scientific Officer, K.G. Sheshadri (e-mail: kgsheshadri@yahw.com)
is thc Project Assistant, H.P. Khincha (e-mail: hpk@ee.iisc.emet,in) and D.
Thukaram (e-mail: dtram@ee.iisc.emet.in) are Professor; in the Electrical
Engg. Department, Indian Institute of Science, Bangalore-560 012.
K . Panhsaraihy (e-mail: prdc@vsnl.com) is with the Power Research &
Development Company, Bangalore.
Fig. I. Typical feeding arrangement of 25 kV traction system of Indian
Railways
two SPs. Fault on the OHE can be of two types (i) Earth faults
(ii) Phase-to-phase faults. The second fault can occur by
accidental closure of the bridging interruptor at the SP during
normal feeding condition or by a short circuit at the insulated
overlap opposite a traction substation at times of emergency
feed conditions. This is termed as Wrong phase coupling
(WPC) fault. Under emergency feed conditions, however the
zone would extend upto the next traction substation, which is
double the normal zone and the relay should provide
protection upto the end of next section.
The harmonic currents drawn by the dc motor locomotives
degrade the power quality of the traction supply [3]. The
excessive voltage drop due to the flow of lagging reactive
current makes the performance of the system even worse.
Voltage regulation with shunt compensation
overcoming these drawbacks. Static VAR Compensators
(SVCs), Thyristor controlled reactors (TCRs) and Thyristor
Switched Capacitors (TSCs) can be used to provide such
compensation. However, TCRs are expensive and require
additional filters against the harmonic pollution they add into
the system in addition to the harmonic load current. TSCs are
cheaper devices and do not produce as much harmonic
pollution as TCRs. They can provide step changes of the
compensation levels from a shunt compensator.
This paper presents the
implementation and testing of a quadrilateral characteristic
single-phase digital distance relay for 25 kV AC traction
applications. A Texas Instruments TMS320C50 digital
processor (DSP) has been employed to support the high-speed
numeric processing capabilities required for high-speed
transmission line protection.
allows
modeling, simulation,
11. RAILWAY TRACTION SYSTEM MODEL
In order to investigate the performance of faults and
loading conditions, the OHE of a typical 25 kV traction
0-7803861 0-8104/$20.00 Q 2004 IEEE
1339

Page 2

system of the Indian Railways has been considered. The
Power System Block set (PSB) of MATLABISIMULMK is a
modern design tool used to build the simulation models for
electric power system as well as its interactions with other
systems [4]-[6f. The basic function blocks of the individual
subsystems are developed initially and are interconnected to
form the hIl system model. Each system element is modeled
based on its specifications [7j.
A. Three-phase AC suppry system
A three phase 220 kV, 50 Hz AC supply system with the
220 kV single circuit transmission line has been modeled as
shown in Fig. 2. The power received from the supply
authority grid network is transmitted to the railway's own
kansmission lines by a series of transformer and line
sectioning facilities. The substations have been modeled as
subsystems. A bridging interruptor modeled as a switch
connected between Substation 1 (Subl) and Substation 2
(Sub2) facilitates the simulation of WPC faults.
a
m ku
ON state resistance & , , = lmil, Forward voltage = 0.8 V,
Snubber resistance = 100 IZ as shown in Fig. 4. The upper
and lower half-bridge converters convert AC voltage to a
controlled DC voltage.
2m-w
-g.c-
flrzz=
Fig. 3. Model of Substation I With 25 kV 40 km traction feeder and loads
AC voltage from the 25 kV feeder is reduced to the
required voltage of the power converters. Each thyristor-diode
bridge is fed from a 25 kVI 2 X 400 V three winding single
phase transformer having 8% impedance and saturable
characteristics. The thyristor converters are used with delayed
firing to control the current in lower speed ranges, but for
most of the time, the converters operate without any firing
delay and speed increase is achieved by field weakening.
When two or more locomotives are running together hauling a
single train, they are assumed to have identical firing angles.
The DC machine motor model in PSBISIMULINK
implements
a
separately
electromechanical torque developed T, is proportional to the
armature current I,.
The parameters chosen are moment of
inertia J=O.1 kg.m2, Initial speed = 10 rads, Armature
res,stance ~,=0.06 Q, Amatwe inductance L, = 0.0012 H.
*
excited machine.
The
Rr*(
h4m+a
Fig. 2. Model ofthree-phase supply grid with substations
E. Substation and Track section model
Fig. 3 shows thf: modeling of Substation 1. The modeling of
Substation 2 is identical to that of Substation I. The 25 kV
supply for traction system is drawn through a single phase
step down transformer. This is modeled as a 25 MVA, 220
kV/25kV, two winding single phase transformer with
impedance of 12% at 25 MVA base. The average length of the
catenary to be protected during normal feed conditions is 40
km. This feeder is modeled as ten 4 km pi sections, each
having a longitudinal impedance of 0.169+]0.432 n/lan at 50
Hz and shunt capacitance of 0.01 1 f l k m [SI. This facilitates
the simulation of earth faults f r o m 10% to 90% of the line. In
their simplest configuration, the TSCs are constituted of a
capacitor bank, where each capacitor may be connected to the
system through a thyristor switch and a damping reactor to
limit rate of cum-nt. The TSC is modeled appropriately by
choosing reactor and capacitor values tuned to a particular
frequency (i.e. the third harmonic) and can reduce the
harmonic pollution.
C. Locomotive model
The locomotives are assumed to be of the conventional
thyristor type with a total locomotive rating of 2.5 MW (rated
at 25 kV). They are modeled as two half-controlled thyristor-
diode bridge rectifiers with each rectifier having parameters of
d .
Fig. 4. Model of a single 2.5 MW locomotive
111. SIMULATION RESULTS AND ANALYSIS
In order to investigate the effects of the faults and loading
conditions in the traction system, a number of cases have been
studied with and without TSC.
1340

Page 3

A. Simulation waveforms
Figs. 5-7 show some representative simulation waveforms
of the feeder voltage and feeder current. The commutation of
the locomotive loads generates severe distortion on the track
voltage. When four or more locomotives are running, the
distortion becomes even worse and also the voltage becomes
too low. Initially earth faults were simulated at different
points on the traction feeder ('from 10% to 90% of the line)
with the bridging intemptor open and fault block (modeled as
a switch in series with resistance RF and external timer
control) connected. The SIMULINWPSB also facilitates the
timing of the fault by varying the timer parameters, fault
resistance and location of the fault, The bridging intemptor
was then closed with the earth fault block removed and the
WPC simulation studies were camed out. In this case, the
waveforms of the feeder voltage, feeder current and load
current were monitored at both the leading and lagging end
substations as shown in Fig. 6 and Fig. 7.
WPC maxlnmnrssistance rcnch
WC
minimum reactancc mach X,,
WPC minimum rcactance ma& X ,
IV. RELAY CHARACTENSTIC
Quadrilateral characteristics have proved very versatile in
protecting railway overhead lines. It provides higher resistive
coverage than mho characteristic [9]. They permit each relay
to protect longer sections of the line, while avoiding the
traction load. Heavy traction loading can lead to load
encroachment problems. In the proposed digtal distance
relay, the quadrilateral characteristic is as shown in Fig. 8.
The detection of the fault using the above logic applies to
Zone-1 protection. The relay reach settings and other
parameters have been chosen as given in Table I.
The traction OHE is subjected to frequent earth faults
caused by failure of insulation, or by the OHE snapping and
touching the earth. These faults are cleared by the feeder
circuit breaker. The relay characteristic under normal feed
with earth fault conditions is indicated by the first quadrant in
the relay characteristic.
Due to overlap of the earth fault and WPC characteristics,
some times the earth fault relay operates on WPC fault too, A
tripping decision based only on angle is not sufficient enough
to detect these fault conditions accurately. The impedance
seen by the relay on WPC fault at the substation w i t h lagging
voltage always lies in the second quadrant of the relay
characteristic while that for earth fault lies in the first
quadrant. These two faults can be discriminated by having
two relay characteristics as shown in Fig. 8.
20 /180 n
5mn
30 # d fl
rim-
LEADING END
in -
".IIIC. lodk.kd . n .*r -
-"It .I t-a.l&
Fig. 6. Waveform for Case with 3 locos in Subl, No lwo i n Sub2
(Leading end), WPC fault at P O . 14 s
Fig. 7. Waveforms for Case with 3 locos in Subl, No loco in Sub2
(Lagging end), WPC fault at t=O. 14 s
Fig. 8. Quadrilateral relay characteristic for traction feeder
TABLE r
TYPICAL RELAY CHARACTERISTIC PARAMETERS
I
FaramcteT
ICated Voltage (AC)
I
value s t t t h g
110 v
Rated current
I
I
5 A
Setti= angle 8, (Zm~c-1, Fault)
mSctti
a .
WPC rcttinp mglc 0 (Zone-2)
65 degrees
30 d c E t s
64.35 degTecs
I
e ~ ~ ~ y - 1
RELAY REACH SETTINGS
Fonvard Rosistancc rcach R,
Forward renctmce reach
Reverse redstancc reach RB
Rcvcrre reactancc rcachXB
WPC ninhnum resirtarrc reach R,, I
20 190 5 - l
1
6 m n
8/180n
1341

Page 4

A. Calculufion of Impedance
A Full cycle Fourier relaying algorithm with 12-sample
data window has been used to extract the fundamental
component from the voltage and current samples. The
window is progressively advanced by one sample as new
samples of voltages and currents become available. With
K=12 sampledcycle, the sine and cosine components of the
incoming voltage and current signals (for fundamental
frequency f=50 Hz) are determined by the expressions (1) and
(2) given below.
and X,j
2 K
EZr,FasB
2 K
K n=l
K -I
zsin =- cznsin8 ; I " -
(2)
where V, and I, are incoming samples of feeder voltage and
current. The sine and cosine values are stored in the form of a
look-up-table. The values of resistance Kd
are then determined by the equations given in (3) and (4).
D.
. Ices
D-Xca, = %os 'Isin -Icm .vsin
where D = I& +I&
The 'D' terms have been cross-multiplied in equations (3) and
(4) to prevent the need for any digital division algorithm,
which could increase the processing burden and delay on the
DSP processor [IO]. In the logics implemented for the
quadrilateral characteristic, the calculated values of resistance
and reactance k,~
and q,,
are compared with the reach
settings.
and reactance XCd
Rcol = V , , . Isin i V , ,
B. Relay Logic
The proposed relay characteristic can be realized by the
following equations given in (5), (6) and (7):
For earth fault detection (Zone 1)
X g < X,,l < L'pl AND
RB f Xcal. Cot 6'1 . : Real < RF + Xc,l. mt @ ,
-
Zone1: 0 < X,,l < x F 2 A N D
+ xcal .cote2 < R , , ~ < o
OR
Zone2: x F 2 < X,,i < x F 3 A N D
+ xCal
Additional logic for discriminating between the earth faults
and W C
faults has been given in the expression (8)
RcaI
AND
(3)
(4)
1
(5 1
(4)
R~
.cote3 < R , , ~ < o
(7)
Iload < IWPC
Ifairit
(8)
V. RELAY DESIGN
A. Relay Hardware
Fig. 9 shows the hardware set-up for implementation of the
proposed relay consisting of a PC based Waveform simulator,
data acquisition system, DSP processor and the host PC
system.
PC based Waveform Simulator system
The Waveform playback simulator system consists of a
Digital-to-Analog converter PAC) card that is interfaced to a
personal computer (PC) system. The data files containing the
samples of the feeder voltage and current obtained from
PSB/SIMULINK based simulation studies of traction system
are reproduced on real-time basis using the DAC card. A
sampling frequency of 60 times the power system frequency
(3000Hz for a 50 Hz system) has been used to generate the
data for voltage and current signals. The proposed
implementation scheme uses a Digital-to-Analog converter
card that supports two B u r r Brown DAC4815 ICs. Each of the
DAC4815 consists of four identical DAC modules having
double buffering capability and a voltage output range of +/-
10 V or +/- 5 V 1 6 1 .
Data Acsuisition Svstem
This interface hardware consists of eight identical Analog-
to-Digital (ADC) channels. All the eight channels are
connected through 8-channel multiplexer. The Burr Brom
ADS7804 ADC has been used in the data acquisition system.
The ADS7804 ADC is of successive approximation type with
a 12-bit resolution, a maximum conversion time of 6 ps and
has the capability of latching the converter output until it is
read [ 1 11. To achieve simultaneous sampling of all the voltage
and current signals, eight S/Hs (LF398- Burr Brown) are used
by the ADC. The S/H delay is about 10 p. The DG508 is an
S-channel single ended CMOS analog multiplexer which
connects the output to one of the eight analog inputs
depending on the state of a 3-bit binary address that is
software controlled. It has a fast access time of 0.2 ps and fast
settling time of 0.6 p. The ADC is interfaced to the Texas
lnstruments TMS320C50 DSP processor by the Intel
Programmable Peripheral interface 8255 chip on the data
acquisition board. The desired control signals for ADC
channel selection, start-of-conversion, end-of-conversion and
read data lines has been programmed through Port-B and
Port-C (PC3 and PC7 respectively) of 8255. The bi-
directional control signals and data lines between DSP board
and data acquisition interface hardware are buffered to ensure
minimum loading of the processor.
DSP Hardware
This consists of the Texas Instruments TMS320C50 DSP
board interfaced to a front-end MS-DOS PC system, which
provides the user interface facility to the DSP board. The DSP
processor board implements the relaying scheme by
processing the acquired signals obtained from the data
acquisition interface board. The TMS320C50 DSP has the
following important features such as 40 ns single-cycle fixed-
point instruction execution
multiply/accumulate (MAC) instructions, 9K X 16-bit single
cycle on chip prograddata RAM and 16 bit programmable
timer [ 121.
time, single cycle
B. ReEay Sofhuare
The software used in the implementation comprises of
Waveform simulator software for signal generation and the
relaying software. The user interface software, written in C,
runs on the personal computer PC-I shown in Fig. 9 to which
1342

Page 5

the 12-bit DAC4815 DAC card is interfaced. The user
interface software reads the data generation file, which
consists of voltages and currents stored in the form of samples
and the DAC card converts them into a continuous waveform.
The relaying sofrware is written entirely in the Assembly
language of the TMS320C50 processor. The relaying software
comprises of data acquisition software, data processing
software and application software.
The data acquisition software controls the operation of the
data acquisition system. The entire data acquisition procedure
has been implemented in real-time on Interrupt basis. A 16-bit
programmable timer provided in the TMS320C50 board is
used to control the sampling of the input signals. The
appropriate count value for the required sampling time is
loaded into the timer counter that can be used to periodically
generate CPU interrupts. The timer operation is controlled by
the Timer control register (TCR). Two circular buffers are
maintained with appropriate pointers using the auxiliary
registers of TMS320C50 to store the incoming samples. The
data processing sohare performs the necessary task of
extracting the fundamental components of the voltage and
current signals. It implements a 12-sample windowing
technique of the Fourier algorithm. The TMS320C50 uses a
16 X 16 bit hardware multiplier that is capable of computing a
signed or unsigned 32-bit product in a single machine cycle.
The sine and cosine values, stored in the look-up-table are
multiplied with incoming signals. The typical MPYiMAC
instructions of the DSP perform several operations in a single
instruction cycle, thus minimizing the computation timings.
The appIication software implements the relayng algorithms
to compute the impedance values (k,,,
quadrilateral relay characteristic has been realized in sohare
using the suitable relay logic. The relay reach settings and
setting angle have been appropriately chosen for a typical 25
kV, 50 Hz single phase AC traction overhead equipment
(OHE).
and Xed. The
U
jl,Kmxwy
a-
AMbghpulS*latem
> ................................................................
.
h M S l 6 a r I R N
W W M * W
i.
iO@bW
:
1 - - - - . -
pc-2
!
..-.. i
Fig. 9. Block diagram of the hardware set-up for relay characteristic
implementation
VI. RESULTS AND DlSCUSSION
The performance of the relay has been evaluated using the
data simulated from MATLABISJMULINKIPSB based
studies. The earth fault studies have been canied out for
various locations along the traction feeder for different
timing of the faults. The performance of the relay for various
harmonics in the feeder voltage and current has also been
analysed using the Fourier program. For each such case the
phase of the feeder voltage and current in both substations and
also the impedance seen by the relay at both substation has
been tabulated. Table 11 shows the typical simulated cases.
The load distributions have been coded as a number, with
each case representing the number of fully loaded locomotives
at the related loading points in Fig. 3 both in Substation 1 and
Substation 2. For example, a load pattern of 1110 for
Substation 1 means that there are 3 locomotives connected to
the track section, one at 12 km, the second locomotive at 24
km and the third locomotive at 32 km (as 40 km is modeled as
10 sections). The values of impedance for the simulated cases
have been indicated in Table 111.
TABLE I1
LOAD PATTERN AND FAULT CONDITIONS FOR THE SIMULATED
CASES
I cas I
Snbarbn 1
i
Snbstatirm 2
TABLE I I I
TYPICAL VALUES OF IMPEDANCE FOR THE SLMMULATED CASES
Case
Substatim 1
Snbstatirm 2
The feeder voltages and currents have been reproduced
using the DAC card after suitable signal conditioning and
scaling. The simulated signals are used to evaluate the relay
characteristic. For testing, WPC faults, feeder voltage and
currents of the lagging end substation have been fed, In this
case the relay detects the fault in second quadrant of the
characteristic. From Table I1 and Table 111, it is observed that
earth fault is detected in first quadrant of relay characteristic.
For WPC fault, the relay at the lagging end detects the fault as
seen in the second quadrant of the relay characteristic. For
1343