DOI: 10.48011/sbse.v1i1.

2212

A Performance Analysis of Transmission

Line Differential Protection
Igor F. Prado ∗ Rodrigo P. Medeiros ∗∗ Flávio B. Costa ∗∗∗
Kleber M. Silva∗∗∗∗
∗
Department of Exact and Technological Sciences, State University of
Santa Cruz, BA, Brazil (e-mail: [email protected]).
∗∗
Multidisciplinary Center of Caraúbas, Federal Rural University of
the Semi-Arid, RN, Brazil (e-mail: [email protected])
∗∗∗
School of Science and Tecnology, Federal University of Rio Grande
do Norte, RN, Brazil (e-mail:[email protected])
∗∗∗∗
University of Brası́lia, DF, Brazil (e-mail: [email protected])

Abstract: This work accomplishes a comparative analysis between two methodologies of trans-
mission line protections available in commercial relays: percentage and alpha-plane differential
protections. In order to verify the operation limits of these protections, several internal and
external faults were simulated considering variations of the electrical parameters of fault types,
system loading, fault impedance and CT saturation on transmission lines. The fault detection
time is strongly influenced by the fault impedance variation, the greater the fault resistance,
the longer the time for its detection. Comparing the overall performance of the protections, the
alpha plane method showed more accurate and faster responses, since the sequence units are
not influenced by the fault impedance.

Keywords: Transmission Line Protection, Percentage Differential Protection, Alpha-Plane

Differential Protection.

1. INTRODUCTION more complex communication system, which increases the

implementation costs. (Ziegler, 2012)
An electric power system is characterized by three ma- With the advent of new digital signal processing techniques
jor subsystems: generation, transmission and distribution. and robust communication systems, the differential pro-
Transmission lines are important components of the power tection has become a promising alternative for the trans-
system since they connect the generation system to the mission line protection. Among the differential protection
distribution system. schemes applied in transmission lines, the most traditional
is the percentage differential, based on the implementation
According to brazilian National Electric System Operator
of electromagnetic relays (Kasztenny et al., 2014). How-
(2019), 70% of power system faults occurs on transmission
ever, the alpha-plane differential protection, which takes
lines due to their length, which makes them more suscep-
into account the ratio between the current phasors that
tible to weather conditions and vandalism. Therefore, the
enter and leave a transmission line by using a complex
protection of transmission lines is of utmost importance,
plane, has been commercially applied in the last years
since it prevents the spread of the fault to the electrical
(SEL, 2011). This technique was designed to present a
system by isolating the smallest region with the shortest
more reliable operation when compared to the percentage
possible time, as well as reduces equipment damage and
one, since it provides information about the module and
ensures system stability, taking into account the fault
the phase of the local and remote currents.
currents can be several times greater than the steady-state
currents and may cause thermal and electrical damage to This paper accomplishes a performance analysis between
the equipment. In this way, a fast and reliable transmission the transmission line differential protection schemes: per-
line protection operation is required in order to prevent centage and alpha-plane. For this purpose, challenging
the emergence of faults, which would yield economic losses cases of faults in transmission lines are simulated by vary-
and major power delivery outages. (Paithankar and Bhide, ing the type of fault, system loading, fault resistance, and
2011) current transformer (CT) saturation. The main objective
of this study is to compare the two methods and to verify
The differential protection is an unitary protection, i.e., it
the behavior in critical situations and operations. The
protects the whole line, but it does not protect adjacent
performance of the alpha-plane protection method was
equipment. This protection presents several advantages
slightly higher than the percentage differential protection
over the distance protection, such as better performance
mainly due to the sequence units.
in transmission lines with series compensation and no de-
pendence of the voltage measurement in most applications
(Roberts et al., 2001). However, this technique demands a
 2. THE TRANSMISSION LINE DIFFERENTIAL
PROTECTION PHILOSOPHY

The transmission line differential protection (87L) presents

a high degree of selectivity, being capable to ensure the
operation in all the extension of the transmission line. The
differential protection is based on the comparison of the
currents that flow through the terminals of the protected
element, providing reliable discrimination between internal Figure 2. Differential relay: (a) External fault; (b) Internal
and external faults. The protection zone is delimited by the fault.
connection of the CTs. During an internal fault condition,
for instance, the differential relay sends the trip command (IˆR ) present the same magnitude and opposite directions,
for circuit breakers in order to isolate the equipment from and no current flows in the operating coil, desensitizing the
the electrical system. protection (Fig. 2(a)). However, if a fault occurs inside of
the protection zone, there is a resulting current flux in the
In transmission lines, CTs are usually too far apart. There- operative coil, which enables the protection operation. The
fore, a communication system with data synchronization operation and restriction currents (Iop and Ires ) are given
is required to ensure the differential protection operates by Ziegler (2012):
correctly with the current samples in a same time basis.
The synchronization can be performed by means of a
communication channel as described in Mills (1991), or by Iop = |IˆL + IˆR |, (1)
means of the use of external time references, such as the
Global Positioning System (GPS) Hall et al. (2003). In this
work, phase (87LA, 87LB and 87LC), negative sequence |IˆL − IˆR |
Ires = , (2)
(87LQ) and zero sequence (87LN) units were implemented. 2
where IˆL and IˆR are the local and remote terminal cur-
Fig. 1 depicts all the executed steps by the differential pro- rents, respectively. The protection operates when:
tection scheme, per sampling k. From the acquisition of the
signal, performed by the CTs, until the tripping command,
the following subroutines are executed: the preprocessing Iop > SLP ∗ Ires + k0 , (3)
step, which consists in performing the digital acquisition where SLP defines the inclination of the differential char-
of the CT currents using anti-aliasing filters and analog- acteristic curve; k0 corresponds to a preset threshold
to-digital converters, as described in Tavares and Silva (pickup current). In this application, k0 is computed tak-
(2014); the current phasor estimation, used to extract ing into account the capacitive current, that arises as a
the module and angle of the fundamental component of spurious differential current.
local and remote currents; and the differential analysis,
that implements logics of the percentage- and alpha-plane The differential characteristic curve is obtained from the
differential protections. solution of (3), with the definitions of the operating and
restriction zones (Fig. 3).

Figure 1. The differential relay block diagram.

2.1 Percentage Differential Protection Algorithm Figure 3. Characteristic curve of the percentage differential
protection.Medeiros et al. (2016)
Fig. 2 depicts the traditional configuration of a percentage 2.2 Alpha-Plane Differential Protection Algorithm
differential protection scheme, composed by two CTs, two
restriction coils (RC) and one operating coil (OC). Besides The alpha-plane differential protection is based on the
the traditional operating coil, the restriction coil has been comparison of the ratio between IˆL and IˆR . Traditionally,
incorporated into the relay, which has the main function the algorithm is defined by a quantity , which is expressed
of reinforcing the actuation torque of the relay when an by the ratio between the currents of the local and remote
internal fault occurs and weakens it for external faults. terminals, as follows:
(Molas et al., 2012)
According to Fig. 2, CTs have inverted polarities with each K̂ = IˆR /IˆL . (4)
other, providing the following analysis based on Kirchhoff
current law in the equivalent circuit, in nominal loading Fig. 4 depicts the alpha-plane differential characteristic,
or external fault situations, the current phasors (IˆL ) and which consists in a circular region centered in the origin
of the complex plane with its respective operation and After the analog filtering, the signals were subsampled
restriction zones. According to Fig. 4, when the value K̂ to a sampling rate of 960 Hz, value commonly used in
exceeds the limits of the restriction zone and achieves the commercial relays which corresponds to 16 samples per
operation zone, a internal fault is detected. cycle of 60 Hz, and then the phasors were estimated by
means of the one-cycle Fourier algorithm.
Since in external faults or in nominal loading conditions IˆL
is approximately equal to IˆR , but in the opposite direction, The effect of the communication channel was considered
taking into account a delay in the samples sent from each
as depicted in Fig. 2 (a), the value K̂ computed through
terminal in accordance with the communication protocol
(4) will be close to the point (-1, 0) in the complex plane,
described in Hou and Dolezilek (2010). The times consid-
i.e, at the restriction zone (Fig. 4). By the other hand,
ered to represent the message processing in the communi-
during an internal fault condition, the real part of K̂ cation device and the transmission of the signal through
becomes positive for internal faults, and negative within the optical fiber in a length of 200 km (length of the line)
circumference R2 = 1/R1 for outfeed faults (Fig. 4). The were 3 ms and 2 ms, respectively, in accordance with Costa
region called “rainbow” was choosed according to Roberts et al. (2017). Therefore, the total communication delay was
et al. (2001) and the operation and restriction zones were 5 ms.
defined according to Benmouyal and Lee (2004), taking
into account the effects of the communication delays, Tables 1 and 2 describes the databases with the fault
capacitive currents and CT saturation. parameters that were varied to performance assessment
of the differential protections, considering the presence
and absence of CT saturation. All simulated cases were
obtained by changing only one parameter at time, while
the others were maintained in their default values, namely:
fault inception angle of θf = 0◦ , fault resistance of rf = 0
Ω, and fault location at df = 100 km.
Therefore, the total number of internal fault cases is
1008, considering the combination of 4 types of fault, 101
fault resistance values, 25 loading angles, and 2 satura-
tion conditions. The total cases of external faults is 176,
Figure 4. Characteristic curve of the alpha-plane differen- considering 4 types of faults, 11 fault resistance values, 2
tial protection. application sites and 2 saturation conditions.
3. METHODOLOGY Table 1. Configuration for internal faults
Simulation variables Values
The modeling of the 230 kV power system shown in Fig. 5, Fault type AG, AB, ABG, ABC.
was performed using the Matlab/Simulink software. The Fault resistance rf (Ω) 0 < rf < 1000 (∆rf = 10Ω)
protected transmission line (T12), was represented by dis- System loading δl (◦ ) -120, -110o , ..., 110o , 120o .
tributed parameter model of Bergeron. Coupling capacitor Total 1008 scenarios
voltage transformers (CCVTs) and current transformers
Table 2. Configuration for external faults
(CTs) were positioned in both line terminals and their
Simulation variables Values
models can be seen in IEEE-PSRC (2004). A typical 60
Fault type AG, AB, ABG, ABC.
db SNR (Signal-to-Noise-Ratio) was considered at the
Fault resistance rf (Ω) 0 < rf < 500 (∆rf = 50Ω)
measured signals to make them more realistic. The system-
Fault location df External local and remote terminal
to-line impedance ratio (SIR) was choosed to be equal to Total 176 scenarios
0.1 at local and remote terminals, and the power system
loading angle was equal to -15, which corresponds to a
moderate power. 4. RESULTS

The following results show the behavior of the phase and

sequence units for single-phase faults, emphasizing that
their behavior can be extended to the analysis of other
types of faults.

4.1 External faults

Fig. 6 depicts the trajectory of the operating points

of phase and sequence units, in a single line-to-ground
external fault in the presence and in the absence of CT
saturation.

Figure 5. Power system model. With respect to the percentage differential protection, the
units presented higher sensitivity during CT saturation.
The currents were sampled with a sampling rate of 15.36 According to Figs. 6(a), (c) and (e), the operation points
kHz, and filtered through a second-order Butterworth anti- related to the external fault without CT saturation re-
aliasing low-pass filter, with cutoff frequency of 960 Hz. mained at the restriction zone during all the event, whereas
 CT altered the trajectory in order to increase the restric-
tion current, due to the restriction coils, however there was
no reduction in the operating current, therefore the entire
trajectory remained in the restriction zone. With respect
to the alpha-plane method and unlike the percentage dif-
ferential method, CT saturation caused a change in the
magnitudes of the currents such that the operation points
moved to the restriction region [Figs. 7(b), (d), and (f)]
in all units. At the end of the CT saturation period, the
points returned to the operation region. Therefore, relay
tripping was delayed due to the CT saturation.
15 6
4
10 2 zoom

Im(K)
^
0

Iop
5 -2
Saturation -4 Saturation
initiation initiation
0 -6
0 1 2 3 -8 -6 -4 -2 0 2
^
Re(K)
Ires

5 6
4
4
2 zoom

Im(K)
3

^
0
Iop

2 -2
Saturation
1 Saturation -4 initiation
initiation
0 -6
0 0.5 1 -6 -4 -2 0 2
^
Re(K)
Ires

5 6
Figure 6. External fauts: (a) Percentage 87LA; (b) Alpha- 4
plane 87LA; (c) Percentage 87LG; (d) Alpha-plane 4
2 zoom
Im(K)
87LG; (e) Percentage 87LQ; (f) Alpha-plane 87LQ. 3
^

0
Iop

2 -2
the most of the operation points moved to the operation Saturation
zone during external fault with CT saturation, which could 1 Saturation -4 initiation
initiation
cause relay misoperation. This fact is expected since the 0 -6
CT secondary signals are distorted during CT saturation, 0 0.5 1 -6 -4 -2 0 2
^
Re(K)
causing the measurements at the line terminals very differ- Ires
ent from each other and allowing the differential protection Without saturation
to act improperly. With saturation

Regarding the alpha-plane method, the 87LG and 87LQ

units did not show significant variation during CT sat- Figure 7. Internal single-phase-to-ground fault: (a) Per-
uration, and the 87LA was more affected. However, all centage - 87LA unit; (b) Alpha-plane - 87LA unit; (c)
operating points of the 87LA unit remained in the restric- Percentage - 87LG unit; (d) Alpha-plane - 87LG unit;
tion zone and no trip was provided. During external faults, (e) Percentage - 87LQ unit; (f) Alpha-plane - 87LQ
even under saturation condition, the ratio K moves in the unit.
negative direction within the restriction zone. 4.3 System load analysis

4.2 Internal single-phase-to-ground fault with CT saturation This section assess the impact of the system loading vari-
ation on the performance of the two protection methods.
For this analyze, it was assumed the remote and local bars
Fig. 7 depicts the trajectory of the operating points of had voltages equal to 16 40o p.u. and local bar with 16 δ o
L
phase and sequence units, in a single line-to-ground in- p.u. The phase of the voltage of the local bar was varied
ternal fault in the presence and in the absence of CT between −120o and 120o in order to verify the effects of
saturation. The fault was simulated with df = 100 km, the load system variation. The power flow moves from bars
rf = 0 Ω, and θ = 0o . After CT saturation initiation, the with larger angles to bars with smaller angles and it is
trajectory performed by the operating points in the oper- proportional to the angular opening between them.
ation and restriction zones of both percentage differential
and alpha-plane protection methods is affected. Regarding An idea about the performances of the percentage- and
the percentage differential method, the saturation of the alpha-plane differential protection methods regarding the
system load variation is depicted in Fig. 8. For sake of
illustration, the most representative point of operation in
the post-fault regime was represented for each analyzed
case.

Figure 9. Percentage differential - Parametric trajectory

for single-phase faults: (a) Phase units; (b) Zero
sequence unit; (c) Negative sequence unit.
Figure 8. System loading - Internal AG fault: (a) Per-
centage - Phase units; (b) Alpha-plane - Phase units;
(c) Percentage - Sequence units; (d) Alpha-plane -
Sequence units.

According to Figs. 8(a) and (b), for δL > 40, the power
flow direction is from local terminal to remote terminal and
the operating current increases. Otherwise, if δL < 40, the
power flow is in the opposite direction, causing a reduction
of the operating current and in the ratio K̂, according to
(4). The success rates for the phase units of the percentage-
and alpha plane methods were 100% and 75%, respectively.
Regarding the sequence units, both methods were not
affected by the system loading variation, ensuring success
rate of 100% [Figs. 8(c) and (d)].

Figure 10. Fault detection time for fault impedance varia-

4.4 Detection time analysis tion: (a) 87LA unit; (b) 87LG and 87LQ units.

The fast action of protection is desirable in an electrical

power system as it prevents and reduces damage to equip- Fig. 11 shows the parametric analysis of the same cases
ment and ensures the stability of the system. considering the alpha-plane differential protection scheme.
Fig. 9 depicts the most representative post-fault points for The 87LA unit has a lower sensitivity when compared to
single-phase-to-ground faults simulated with variation of the percent differential protection phase unit, since the
the fault resistance parameter, for the phase and sequence fault detection limit is 550 Ω. However, the sequence units
units of the percentage differential protection method. In do not lose sensitivity, ensuring 100% of the detection
a general way, the sensibility of the units decreased with cases simulated. Fig. 12 depicts the relationship between
the increase of the fault resistance. However, the sequence the fault impedance and the detection time. The detection
units were less affected than the phase unit. The effect of time increases with the increase of the fault impedance for
delay in relay operation for these faults is depicted in Fig. all units.
10. According to Fig. 10, the greater the fault resistance,
the longer the detection time and the delay in operation.
It is noteworthy that the percentage differential protection
has not been able to identify faults with resistances greater
than 885 Ω.
 safety and reliability of a differential protection system for
transmission lines, where the percentage differential pro-
tection presented limitations mainly in the identification
of high impedance faults, whereas the alpha-plane method
was the most robust and presented the fastest response.
ACKNOWLEDGMENT

This work was supported by the National Council for

Scientific and Technological (CNPq) and by the Coordina-
tion for the Improvement of Higher Education Personnel
(CAPES), Brazil.
Figure 11. Alpha-plane - Parametric trajectory for single-
phase faults. (a) Phase units; (b) Zero and negative REFERENCES
sequence units.
Benmouyal, G. and Lee, T. (2004). Sequence-current
differential elements. SEL Journal of Reliable Power.
Costa, F.B., Monti, A., Lopes, F.V., Silva, K.M.,
Jamborsalamati, P., and Sadu, A. (2017). Two-
terminal traveling-wave-based transmission-line protec-
tion. IEEE Transactions on Power Delivery, 32(3),
1382–1393. doi:10.1109/TPWRD.2016.2574900.
Hall, I., Beaumont, P.G., Baber, G.P., Shuto, I., Saga,
M., Okuno, K., and Uo, H. (2003). New line current
differential relay using gps synchronization. In 2003
IEEE Bologna Power Tech Conference Proceedings,,
volume 3, 8pp. doi:10.1109/PTC.2003.1304464.
Hou, D. and Dolezilek, D. (2010). Iec 61850 – what it can
and cannot offer to traditional protection schemes. SEL
Journal of Reliable Power, 1(2), 1–11.
IEEE-PSRC (2004). Emtp reference models for transmis-
sion line relay testing.
Kasztenny, B., Benmouyal, G., Altuve, H., and Fischer,
Figure 12. Fault detection time for fault impedance varia- N. (2014). Tutorial on operating characteristics of
tion: (a) 87LA unit; (b) 87LG and 87LQ units. microprocessor-based multiterminal line current differ-
ential relays. Line Current Differential Protection: A
5. CONCLUSION Collection of Technical Papers Representing Modern So-
lutions.
This paper presented a comparison study between the per- Medeiros, R.P., Costa, F.B., , and Almeida, M.A.D.
formances of two transmission line differential protection (2016). Assessment of the main phasor-based power
algorithms: the percentage and alpha-plane methods. Sev- transformer differential protection elements. SBSE - VI
eral simulated internal and external faults, with variations Simpósio Brasileiro de Sistemas Elétricos.
of the fault type, fault resistance, and system loading, in Mills, D.L. (1991). Internet time synchronization: the
the presence and in the absence of CT saturation, was network time protocol. IEEE Transactions on Commu-
considered. In a general way, both strategies presented a nications, 39, 1482–1493.
fast fault detection. However, the fault resistance variation Molas, E.C., Silva, K.M., and Kusel, B.F. (2012). Teoria
greatly affects the permanent regime point of the phase de comparadores aplicada na avaliação da atuação da
units of both methods and of the sequence units of the proteção diferencial de linhas de transmissões no plano
percentage method, a fact not observed in the alpha plane alfa. XIX CBA Congresso Brasileiro de Automática.
sequence units, therefore according to this feature, the Paithankar, Y.G. and Bhide, S. (2011). Fundamentals of
alpha-plane method reached a 100% fault detection rate, power system protection. PHI Learning Pvt. Ltd.
while the percentage method reached 95.3%. It is notewor- Roberts, J., Tziouvaras, D., Benmouyal, G., and Altuve,
thy that the greater the fault resistance, the greater the H. (2001). The effect of multiprinciple line protection on
time spent for its identification. dependability and security. In proceedings of the 55th
The studies of saturation and loading of the system point Annual Georgia Tech Protective Relaying Conference,
out the need for extra routines that could assist in increas- Atlanta, GA.
ing safety in order to identify the saturation of CTs during SEL, S.E.L. (2011). Sel-411l: Advanced line differen-
external faults mainly in the percentage method most tial protection, automation, and control system. URL
affected by this aspect. For the case of the variation of the https://selinc.com/products/411L/.
loading of the system, it is possible to rotate the restriction Tavares, K.A. and Silva, K.M. (2014). Evaluation of
and operation zones for the alpha plane method, that was power transformer differential protection using the atp
affected by this aspect. software. IEEE Latin America Trans., 12(2), 161–168.
doi:10.1109/TLA.2014.6749533.
The study performed in this work is important for the Ziegler, G. (2012). Numerical Differential Protection:
understanding of methodologies and aspects that affect the Principles and Applications. John Wiley & Sons.