1

EMTP Applied to Evaluate Three-Terminal Line

Distance Protection Schemes
K. M. Silva, W. L. A. Neves and B. A. Souza

Abstract— Digital protection schemes have been around for dynamics during transient conditions, revealing malfunctions
decades. Although there are many relay schemes reported in of protective schemes. The EMTP has been used to model
the literature, the protection engineer may gain more insight on and test distance relays in closed-loop simulations [6]–[8], i.e.,
how schemes work properly if well known cases are previously
simulated with the EMTP to evaluate relay algorithms. Here, a simulations in which the relay may interact with the system
very simple case study is presented in which the EMTP is used to network models, tripping breakers in order to switch-off the
evaluate the off-line performance of distance protection schemes faulted portion of the system. However, only the protection of
applied to the three-terminal line of a 230 kV three-bus power two-terminal lines has been evaluated not including distance
network. The digital relays were modeled considering the logic schemes.
of different distance schemes and relay-to-relay communication.
The EMTP is a powerful tool to pinpoint limitations on the Here, a very simple case study is presented in which the
applicability of these distance protection schemes and may help EMTP is used to evaluate the off-line performance of distance
engineers to develop new protection schemes. protection schemes applied to the three-terminal line of a 230
Keywords— Power system protection, three-terminal transmis- kV three-bus power network. The distance protection relay
sion lines, distance schemes, EMTP. schemes are implemented using the MODELS environment
of the Alternative Transient Program (ATP) version of the
EMTP [9], taking into account the logic of different distance
I. I NTRODUCTION schemes and relay-to-relay communication. Some well known

T HERE are technical and financial reasons to avoid the

construction of a full switching electric power station for
some high-voltage transmission line. There are cases in which
aspects about the performance of these schemes and limita-
tions on their applicability are discussed. The EMTP may be
a very useful tool to help engineers to develop new protection
the line must be tapped and divided into separate line seg- schemes.
ments, originating the well known multi-terminal transmission
lines [1]. Among the several configurations, the simplest and II. C ASE S TUDY D ESCRIPTION
most used is the three-terminal line with generation sources The performance of distance protection scheme applied to
behind each terminal. the 230 kV three-bus power network shown in Fig. 1 is
The protection of multi-terminal lines is a challenge to evaluated. The three line segments originated from the buses
engineers, owing to the large number of line configurations A, B and C to the junction point P form a three-terminal
with varying numbers of terminals, line lengths, source and line, named line ABC. The two-terminal line between bus B
load conditions [2], [3]. Both unit-protection and distance and C is named line BC. The power system generation is
schemes may be adapted for use on multi-terminal lines represented by the voltage sources ÊA , ÊB and ÊC behind
protection, but distance schemes are more used [4]. their series impedances ZA , ZB and ZC , respectively. The
The distance schemes may be divided into three groups:
intertripping, permissive and blocking schemes [4]. They are
used to accelerate in-zone fault clearance and/or prevent out-
of-zone tripping, by means of the ON/OFF data exchange
between the protection devices of each line terminal. How-
ever, their application is not straightforward, requiring careful
consideration and schematic checking of all system operating
conditions.
In order to evaluate the performance of protective systems,
the use of Electromagnetic Transients Program (EMTP) has
increased [5]. In fact, EMTP simulations provide a very good
understanding of both relay performance and power system

This work was supported by Brazilian National Research Council (CNPq).

The authors are with Department of Electrical Engineering at Federal
University of Campina Grande, 882 Aprı́gio Veloso Av, Bodocongó, Camp-
ina Grande - PB, CEP:58.109-970, Brazil. E-mail: {kms, waneves, bene-
mar}@ee.ufcg.edu.br.

Presented at the International Conference on Power Systems Transients Fig. 1. Diagram of the test power system.
(IPST’07) in Lyon, France on June 4-7, 2007.
 2

series impedance of each power system component is also

presented in Fig. 1. The first ”0” subscript corresponds to a
zero sequence quantity whereas the first ”1” subscript corre-
sponds to a positive sequence quantity. For instance, Z0EA and
Z1EA are the zero and positive sequence impedance of source
ÊA ; and Z0L and Z1L are the zero and positive sequence
impedance in Ω/km of all lines.
The following distance schemes are evaluated: direct un-
derreaching transfer trip (DUTT), permissive underreaching
transfer trip (PUTT), permissive overreaching transfer trip
(POTT), directional comparison blocking (DCB) and direc-
tional comparison unblocking (DCUB) [4]. Some aspects
about the performance of these schemes are discussed, taking
into account the infeed and outfeed current effects, which are
inherent to multi-terminal lines protection and well known to
protection engineers [2]. Fig. 2. Overall block diagram of the simulation using MODELS.
The distance relay algorithms and distance scheme logics
are implemented by means of MODELS environment of the
Alternative Transient Program (ATP) version of EMTP [9]. In A. Instrument Transformers Models
addition, the relay-to-relay communication is also emulated in Both CT and coupling capacitor voltage transformer
MODELS environment. In this way, by the emulation of data (CCVT) models and their parameters were reported in the
exchange between the relays RA1, RB1 and RC1, the breakers reference paper of IEEE Power System Relay Committee
A1, B1 and C1 are tripped properly, switching-off the line [10]. The CT model considers saturation effects of the core
ABC whenever an internal fault is detected. The description including the point by point flux-current curve. In the CCVT
of the EMTP simulation is presented next. model, the ferroresonance suppression circuit is modeled using
a capacitor connected to a non-saturable transformer, in which
III. EMTP S IMULATIONS primary and secondary windings are connected in such a
way that parallel resonance occurs only at the fundamental
The MODELS environment provides the monitoring and frequency.
controllability of the EMTP power systems model. The voltage Auxiliary ideal instrument transformers are used to scale
nodes, current branches and switches status of the power down the CTs and CCVTs outputs to levels suitable to be
system model are the inputs to MODELS. These signals are used by analog-to-digital (A/D) converters. Their secondary
processed and the output signals may interact with the EMTP burden is chosen in a appropriate way to obtain secondary
power system model changing its state by controlling the voltages ranging from -10 to 10 V [8].
switches operation. In this way, the states of the EMTP power
system model may be dynamically changed in response to the
outputs of MODELS environment [9]. B. Breaker Model
The overall block diagram of the EMTP simulation is shown For the sake of simplicity, the non-linear arc dynamics
in Fig. 2. For each time step, four parameters are sent from and losses are ignored in the breaker model. The breaker is
the power system model to the inputs of MODELS: essentially an ideal switch that opens whenever a trip signal is
1) The current in the secondary of the auxiliary current received. However, if reliable arc dynamic models are available
transformers (CTs). these effects may be included in the breaker model [11].
2) The voltage in the secondary of the auxiliary voltage The interrupting time of a circuit breaker used in 230 kV
transformers (VTs). transmission lines lies around 2 cycles [1]. In this way, the
3) The status of the switches which represent the breakers. breaker model was implemented in MODELS to emulate this
4) The ON/OFF data from the relays in remote ends, which interrupting time. This delays the trip signal sent from the
depends on the chosen distance scheme. relay model by 2 cycles and then coordinates the opening of
each breaker pole in such way that a pole opens only when
The relay algorithm evaluates the MODELS inputs and
the current waveform crosses the zero line.
computes two outputs:
1) The decision to trip local breakers.
2) The decision to send ON/OFF data to relays in remote C. Signalling Channel Model
ends. The signalling channel was modeled as a simple delay in
The MODELS outputs depend on the detection of a fault the signals transmitted between the relays in each end of the
by the relay algorithm. If this is the case, by the data protected line.
exchange between the relays in each end of the line, the overall The worst stand-alone channel performances for distance
protection system trips the local breakers in order to switch-off schemes are [12]:
the line. • 40 ms for intertripping schemes (DUTT).
 3

• 20 ms for permissive schemes (PUTT and POTT). from −Vmax to Vmax . The digitized value vd of a voltage v
• 15 ms for blocking schemes (DCB and DCUB). may be computed as:
The chosen values are presented in following sections.  " ¡ ¢#

 v 2b − 1

 RON if v > 0

 Vmax
D. Relay Model
vd = (2)
The main features of the relay model are summarized in 
 · ¸

 (2Vmax − |v|) 2b
Tab. I. Voltages and currents are filtered by analog filters, in 
 RON if v < 0
order to minimize the effect of aliasing as well as to attenuate Vmax
high frequency components. These signals are converted into
discrete forms by means of A/D converter models and the where RON is the rounding operation. In this way, the
voltage and current phasors are estimated by the digital filter. floating-point output representation vf may be computed as:
These phasors are used in the phase comparator model in order

to detect a fault within the relay protective zones. Finally,  Rvd if v > 0
depending on the relay logic, a trip is sent to the local breaker. vf = (3)
The relay logic takes int account the relay settings and phase  ¡ ¢
R vd − 2b+1 if v < 0
comparator outputs. In addition, depending on the distance
scheme, the relay may send a trip, a permissive or a blocking
where R is the A/D resolution which may be computed as:
signal to relays in the remote ends of the line by the signalling
channel model.
Vmax
1) Analog Filter: According to sampling theory, an analog R= (4)
2b − 1
signal must be sampled using a sampling rate at least two times
greater than the maximum frequency of the analog signal.
The chosen sampling rate is 1920 Hz, that corresponds to
Otherwise, it may occur the aliasing effect. In the relay model,
16 samples/cycle for the fundamental frequency of 60Hz.
an analog third-order Butterworth low-pass anti-aliasing filter
is employed, whose transfer function is [13]: 3) Digital Filter: The chosen digital filter applied to phasor
b0 estimation was the cosine filter of one cycle, because it has
H(s) = , (1) been widely used in protective relays due to its inherent
s3 + a2 s2 + a1 s + a0
characteristics such as, rejection of exponentially-decaying dc
where: b0 = 1.6452 · 109 , a0 = 1.6452 · 109 , a1 = 2.7873 · 106 offsets, rejection of all harmonics and good transient response
and a2 = 2.3611 · 103 . [14].
2) A/D Conversion: The A/D converter takes instantaneous 4) Relay Settings: In order to simulate the relay, it is
value of its input and converts it into an n-bit binary num- necessary to set its parameters: the maximum torque angle
ber, by using the sample-and-hold technique and the two’s τ , the impedance reaches and operation time of both zone
complement representation [8]. For instance, suppose an A/D 2 and 3, for ground and phase-phase units of the relay; the
converter with word size of b + 1 bits and full-input ranging value of the zero-sequence current compensation factor K0 ;
the transformer ratios of both CTs and VTs; and the distance
TABLE I scheme. The chosen values are presented in Section IV.
M AIN F EATURES OF THE R ELAY M ODEL .
5) Phase Comparator: The phase angle comparator sub-
Requirements Features module implements the mho autopolarized characteristic, com-
• Butterworth analog filter
paring the angle between (Z Iˆr − V̂r ) and V̂r , where: V̂r and
• A/D converter Iˆr are, respectively, the measured voltage and current; and Z
Components
• Mho autopolarized phase comparator is the impedance reach of the protective zone [15].
• Relay logic depend on chosen distance scheme
• 4 channels of node voltages inputs and 4 6) Relay Logic: The relay logic takes into account the
channels of branch currents inputs
• 3 channels of breaker status contact inputs
phase comparator output, the relay time coordination and the
Interface distance scheme logic. In this way, the relay acts to trip
• 6 channels of pilot signal inputs
• 3 channels of trip signal outputs local breakers and to send ON/OFF data to relays in remote
• 3 channels of pilot signal outputs
terminals of the line, thereby the remote breakers may be
Protection • Phase distance
Functions • Ground distance
tripped.
• Intertripping: DUTT
Distance • Permissive: PUTT and POTT 7) Digital Inputs and Outputs: These modules are respon-
Schemes • Blocking: DCB and DCUB sible to exchange ON/OFF data between the protective relays
• Generation of oscillography files, fault reports in each terminal of the line, depending on the chosen distance
Others and event reports scheme. In addition, they are responsible to get the status of
• Relay settings
the local breaker and to send to it tripping signals.
 4

IV. S ETTINGS C ONSIDERATIONS In order to analyze the effect of infeed currents, assume that
the line BC of the Fig. 1 is out of service. Consider that a three-
The relays RA1, RB1 and RC1 (Fig. 1) of the line ABC
phase fault with incidence angle of 30o and fault resistance of
were set as follow:
1 Ω occurs 40 km from the junction point P toward bus B.
• Zone 1 is required for the DUTT and PUTT schemes
According to the relays settings aforementioned, it is expected
and may be used in POTT, DCB and DCUB schemes that all relays see this fault within their zone 1 tripping all
to improve performance. Its reach was set to cover 85 breakers simultaneously. However, due to infeed currents, the
percent of the actual positive sequence line impedance relays RA1 and RC1 underreach the fault and see it within
to the nearest remote terminal, in order to avoid relay their zone 2 as shown in Fig. 3, where the dynamic locus of
overreach under all operating conditions [2]. the apparent impedance seen from each relay unit is plotted.
• In POTT and DCUB schemes the zone 2 was set to
cover 125 percent of the larger positive sequence apparent
impedance, in order to prevent all expected infeed current
distribution [2]. In this paper, the zone 2 of DUTT,
PUTT and DCB schemes was set the same way, and
it was considered that the relay never operate on load
impedance.
• The reverse-looking zone 3 in DCB scheme was set to be
greater than the zone 2 reaches of the remote terminals.
In fact, it was set to cover 25 percent of the difference
between the larger apparent impedance and the actual
positive sequence impedance of the line to the furthest
terminal, but in the reverse direction.
• The K0 factor was computed taking into account the
larger positive and zero sequence apparent impedances, (a) (b)
in order to prevent the effects of infeed currents [3].
The relays RB2 and RC2 (Fig. 1) of the line BC were set
as follow:
• The zone 1 and zone 2 distance functions were set as
85 and 125 percent of the actual positive sequence line
impedance, respectively.
• The K0 factor was computed taking account the actual
zero and positive sequence line impedances.
The zone 2 operation time of all relays were set to 150
ms, whereas the zone 3 operation time of the relays RA1,
RB1 and RC1 were set to 400 ms. In addition, the maximum
torque angle of all relays was chosen to be 60o , in order to
increase the fault resistance coverage.
In DCB scheme, the short time lag (STL) was chosen to be (c) (d)
20 ms, in order to accelerate in-zone 2 fault clearance in case
of no blocking signal is received [4].

V. S IMULATION R ESULTS
On protection of three-terminal lines, the distance schemes
performance is affected by the junction point location and
current distribution for line faults under all operating condi-
tions. The well known effects of infeed and outfeed currents
in distance schemes performances are discussed next.

A. The Infeed Effect

Infeed describes a condition in which fault current flows into
(e) (f)
the faulted line from all line terminals. As a consequence, the
distance relay may ”see” an apparent impedances greater than
Fig. 3. Apparent impedance plotting considering infeed current distribution:
the actual positive sequence line impedance from its location (a) phase-phase units of the relay RA1; (b) phase-ground units of the relay
to the point of fault. In other words, the relay may underreach RA1; (c) phase-phase units of the relay RB1; (d) phase-ground units of the
the fault due to infeed currents. relay RB1; (e) phase-phase units of the relay RC1; (f) phase-ground units of
the relay RC1.
 5

For the relay RC1, the actual positive sequence impedance the junction point P. According to the relays settings, only the
from its location to the point of fault is 47.6∠82.8o Ω. relay RB1 would de expected to see the fault within its zone 1,
However, the impedance seen by its all units are nearly meanwhile the relays RA1 and RC1 would see it within their
76.5∠88.0o Ω. zone 2. However, the part of the fault current coming from bus
The total fault clearing time without considering any dis- A has two pathways: one from bus A to point P and to the
tance scheme is 198.5 ms. Distance schemes are needed to fault location; the other one from bus A to point P, then to bus
allow high speed line relaying. B and finally to the fault location. Thus, the relay RC1 sees
Tab. II is a summary of the distance scheme performance, the fault within its reverse-looking zone 3 until the breaker B1
considering the worst stand-alone channel performances (Sec- opens. Then, the current direction seen by the relay RC1 is
tion III-C). DCUB scheme is the best and the DUTT scheme reversed and it now sees the fault within its zone 2 (Fig. 4).
the worst. The permissive schemes PUTT and POTT have In other words, the fault will be cleared by the sequential trip
intermediate performances between intertripping and blocking signals from the relays RB1, RA1 and RC1. As a consequence,
schemes. the total fault clearing time without considering any distance
Tab III is the summary of distance schemes performances, scheme is 245.8 ms.
considering the signalling delay of 10 ms for relay-to-relay
signalling channel. Comparing to Tab. II, all schemes have
better performances, except the DCB, which presents the same
performance in both cases. It is also observed that DUTT and
PUTT schemes have the same performance, with total fault
clearing time of 64.07 ms. Whereas, the POTT and DCUB
schemes totally clear the fault in 58.82 ms, but DCUB would
be chosen since it is more reliable than POTT.
TABLE II
D ISTANCE SCHEMES PERFORMANCES FOR INFEED CURRENTS ,
CONSIDERING DIFFERENT SIGNALLING DELAYS .

Distance Fault Clearing Time (ms) (a) (b)

Scheme Bus A Bus B Bus C

DUTT 94.54 56.72 94.54

PUTT 71.43 56.72 73.53
POTT 66.17 56.72 67.22
DCB 67.23 56.72 66.17
DCUB 64.07 56.72 64.07

TABLE III
D ISTANCE SCHEMES PERFORMANCES FOR INFEED CURRENTS ,
CONSIDERING THE SAME SIGNALLING DELAYS .

Distance Fault Clearing Time (ms)

Scheme Bus A Bus B Bus C
(c) (d)
DUTT 64.07 56.72 64.07
PUTT 64.07 56.72 64.07
POTT 58.82 56.72 58.82
DCB 67.23 56.72 66.17
DCUB 58.82 56.72 58.82

B. The Outfeed Effect

Multiterminal lines create the possibility of a current outfeed
condition. Current outfeed occurs when, due to system sources,
loads, and impedance conditions, current flows out from one
or more line terminals during a fault. As a result, distance and
directional relays may be affected, causing either a delay or a
sequential operation. (e) (f)
Assume the system shown in Fig. 1 with the line BC in Fig. 4. Apparent impedance plotting considering outfeed current distribution:
operation and with the source at bus C out of service. Assume (a) phase-phase units of the relay RA1; (b) phase-ground units of the relay
RA1; (c) phase-phase units of the relay RB1; (d) phase-ground units of the
also that a three-phase fault with incidence angle of 30o and relay RB1; (e) phase-phase units of the relay RC1; (f) phase-ground units of
fault resistance of 1 Ω occurs 5 km from the bus B towards the relay RC1.
 6

The outfeed current also reduces the apparent impedance VII. ACKNOWLEDGMENTS
seen by the relay. For example, the impedance seen by the The authors would like to thank the reviewers for their
ZAB unit of the relay RA1, with all terminal closed, is invaluable contributions to improve the paper.
66.6∠67.0o Ω, but the actual positive sequence impedance
from the relay location to the point of fault is 83.3∠82.8o Ω. R EFERENCES
In Tab. IV, it is summarized the distance scheme per-
[1] P. M. Anderson, Power System Protection. Piscataway, New Jersey,
formance for outfeed current distributions, considering the USA: IEEE Press Series on Power Engineering, 1999.
worst stand-alone channel performances. Differently from the [2] Protection Aspects of Multiterminal Lines, IEEE PES/PSRC Special
infeed current situation, the DUTT scheme presented the best Publication 79TH0056-2-PWR, 1979.
[3] G. E. Alexander and J. G. Andrichak, “Application of phase and ground
performance, because the relay RB1 sees the fault within its distance relays to three-terminal lines,” GE Protection and Control, Tech.
zone 1 and quickly trip the local breaker B1 and send a transfer Rep., Markham, Ontario, Canada, 1996.
trip to the remote terminals. The other schemes are delayed [4] Network Protection and Automation Guide, 1st ed., Areva T&D, Paris,
France, 2002.
due to sequential tripping, mainly the DCB scheme, where [5] Tutorial on Electromagnetic Transient Program Applications to Power
relays are blocked until breaker B1 opens. System Protection, IEEE PES/PSRC Special Publication, 2000.
The schemes performances considering the signalling delay [6] C.-H. Kim, M.-H. Lee, R. K. Aggarwal, and A. T. Johns, “Educational
use of emtp models for the study of a distance relaying algorithm for
of 10 ms between all relays are summarized in Tab. V. It was protecting transmission lines,” IEEE Transactions on Power Systems,
observed that all schemes improve their performance except vol. 15, no. 1, pp. 9–15, Feb 2000.
DCB, which presents the same performance in both cases. The [7] J. Y. Heo, C. H. Kim, K. H. So, and N. O. Park, “Realization of distance
relay algorithm using emtp models,” in International Conference on
DUTT presents the best performance again, with total fault Power Systems Transients, New Orleans, USA, Sept 2003.
clearing time of 55.67 ms. Once more, the POTT and DCUB [8] S. G. A. Perez, “Modeling relays for power system protection
schemes showed the same performance. studies,” Ph.D. dissertation, University of Saskatchewan, Saskatoon,
Saskatchewan, Canada, July 2006.
[9] L. Dube, Models in ATP: Language Manual, 1996.
TABLE IV [10] EMTP Reference Models for Transmission Line Relay Testing, IEEE
D ISTANCE SCHEMES PERFORMANCES FOR OUTFEED CURRENTS , PES/PSRC Special Publication, 2004.
[11] V. Paniraj and A. G. Phadke, “Modelling of circuit breakers in the emtp,”
CONSIDERING DIFFERENT SIGNALLING DELAYS .
IEEE Transactions on Power Systems, vol. 3, no. 2, pp. 799–805, 1999.
[12] Teleprotection Equipment of Power Systems, IEC60834-1, 1999.
Distance Fault Clearing Time (ms) [13] A. V. Oppenheim and R. W. Schafer, Discrete-Time Signal Processing.
Scheme Bus A Bus B Bus C New Jersey, USA: Prentice-Hall, 1989.
[14] E. O. Schweitzer and D. Hou, “Filtering for protective relays,”
DUTT 86.13 46.21 79.83 Schweitzer Engineering Laboratories, Inc., Pullman, Washington USA,
PUTT 67.22 46.21 96.64 Tech. Rep., 1993.
POTT 114.49 46.21 94.54 [15] E. O. Schweitzer and J. Roberts, “Distance relay element design,”
DCB 97.69 46.21 122.90 Schweitzer Engineering Laboratories, Inc., Pullman, Washington USA,
DCUB 108.19 46.21 94.54 Tech. Rep., 1993.

TABLE V
D ISTANCE SCHEMES PERFORMANCES FOR OUTFEED CURRENTS , Kleber Melo e Silva was born in João Pessoa, PB, Brazil, 1980. He received
CONSIDERING THE SAME SIGNALLING DELAYS . his B.Sc. and M.Sc. in Electrical Engineering from Federal University of
Campina Grande, PB, Brazil, in 2004 and 2005, respectively. He is currently
Distance Fault Clearing Time (ms) a PhD student at the same university. His research interest are electromagnetic
Scheme Bus A Bus B Bus C transients, power quality, fault diagnosis and power system protection.

DUTT 55.67 46.21 50.42

PUTT 54.42 46.21 95.59
POTT 102.94 46.21 96.64
Washington Luiz Araújo Neves is an Associate Professor in the Department
DCB 97.69 46.21 122.90 of Electrical Engineering at UFCG, Campina Grande, Brazil,. He received
DCUB 102.94 46.21 96.64 the B.Sc. and M.Sc. degrees in electrical engineering from UFPB, Brazil, in
1979 and 1982, respectively, and the Ph.D. degree from UBC, Vancouver,
Canada, in 1995. From 1982 to 1985 he was with FEJ, Joinville, Brazil. He
was a Visiting Researcher with the University of Alberta, Edmonton, Canada,
VI. C ONCLUSIONS from September 2004 to August 2005, and with UBC, Vancouver, Canada,
from September to December 2005. His research interests are electromagnetic
This paper presented the use of a distance relay EMTP transients in power systems and power quality.
model to evaluate the performance of distance schemes in
three-terminal line protection. Although the case study pre-
sented here is an engineering application of known aspects,
Benemar Alencar de Souza was born in Crato, CE, Brazil, 1954. He received
the obtained results encourages engineers to use any EMTP his B.Sc., M.Sc. and Ph.D. in Electrical Engineering from Federal University
version to evaluate protection schemes prior to putting relays of Paraı́ba, PB, Brazil, in 1977, 1981 and 1995, respectively. He works
in service, pinpointing limitations on the applicability of these currently as a professor at the Department of Electrical Engineering of Federal
University of Campina Grande, PB, Brazil. His research activities are mainly
schemes. This may help engineers to develop new protection focused on optimization methods applied to power systems, electromagnetic
schemes. transients, power quality and fault diagnostic.