This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
1

An Algorithm to Mitigate the Infeed Effect on

Overreaching Distance Zones Settings
Kleber M. Silva, Senior Member, IEEE, Jéssica S. G. Pena, Helon D. M. Braz, Member, IEEE,
and Mladen Kezunovic, Life Fellow, IEEE

Abstract—This paper revisits the analysis of the well-known on the network parameters. The method described in [6] is
infeed effect on transmission lines distance protection. A new based on fault studies, and takes into account single-level
adaptive algorithm to overcome this drawback is proposed, which contingencies and infeed currents. The new setting strategy
requires solely impedance data and the operational status of
power apparatus to perform overreaching distance zones settings. presented in [7] uses input signals from local relays along
A comparison of the proposed strategy and the traditional with command data from the distance relay at the remote
one is performed. The obtained results demonstrate that the end transmitted using teleprotection system. In [8], a novel
proposed settings ensure the correct operation of overreaching multi-agent system based methodology for power system
zones. Thereby, the backup coverage is improved significantly, protection coordination is described. In [9], an adjustment
besides avoiding distance zones overlapping between adjacent
lines. It reveals the proposed strategy usefulness and value from factor is suggested and the probability of a fault being detected
the practical point of view, since it can be implemented with correctly when the relay operates is determined. In [10], a
technology readily available on the market. synchrophasor-based algorithm, which uses the active power
Index Terms—Distance protection, infeed current effect, over- measured on buses to eliminate the effects of infeed currents
reaching distance zones. is reported. In [11], an adaptive setting of the 3rd distance
zone based upon synchronized measurements is outlined.
Adaptive strategies for distance protection to update the
I. I NTRODUCTION operating characteristic according to changes in the system
ARGE interconnected grids require bulk power transmis- topology are proposed in [12]–[19]. In [12], an expert system
L sion over very long overhead lines. To avoid widespread
blackouts, faults must be cleared quick and selectively, requir-
method to determine the relay settings during power system
topology changes is elaborated. In [13], a probabilistic model
ing various levels of protection system redundancy [1]. Among to adjust all distance zones is developed. In [14], operating
different redundancy strategies, the use of overreaching dis- characteristics are determined using artificial neural networks.
tance zones to provide remote backup protection is widespread In [15], a new methodology based on the analysis of events and
[2]. Since distance protection suffers from inaccuracies caused their consequences is discussed. A method based on network
by infeed conditions, overreaching zones may underreach and mapping obtained using information from the actual system
provide smaller backup coverage than expected [3]. This effect topology is proposed in [16]. In [17], the adequacy of the
can be even worse for transmission lines operating in parts distance relay settings by comparing the apparent impedance
of the grid with large concentration of infeed sources. To of the actual topology and a pre-defined threshold is evalu-
overcome this drawback, the infeed effect could be taken ated. A wide-area backup protection algorithm that gathers
into account when settings are determined, but it traditionally distance relays status is presented in [18]. In [19], a wide-
requires extensively analysis of system topologies and opera- area protection algorithm, defined by an objective function
tional conditions to compute the different infeed currents levels based on the performance of various distance relays zones
[4]. That is the why infeed effect is usually disregarded during and telecommunication information is mentioned.
overreaching distance zones setting calculation procedures. Our paper revisits the infeed effect on transmission lines
There are some improved algorithms reported on the lit- distance protection and suggests an algorithm to overcome
erature to overcome the infeed effect on distance protection this drawback, by taking into account changes in the system
performance [5]–[11]. In [5], the impedance seen by the relay topology. It requires solely impedance data and operational
is calculated using a weighted equation whose terms depend status of power apparatus to determine the overreaching zones
settings. It is formulated here for the 2nd zone only, but
This work was supported in part by the Federal District Research Foun- it can be extended to the 3rd zone straightforwardly. A
dation (FAP-DF), in part by Brazilian National Council for Scientific and
Technological Development (CNPq) and in part by the Coordination for the simplified optimization process is also proposed to improve
Improvement of Higher Education Personnel (CAPES). the backup coverage on adjacent lines. A comparison of
Kleber M. Silva and Jéssica S. G. Pena are with the Department of Electrical the proposed settings and the traditional ones is performed
Engineering at University of Brasilia (UnB), 70910-900 Brasilia, Brazil. (e-
mail: [email protected], [email protected]). with and without contingencies in a real power system by
Helon David M. Braz. is with the Department of Electrical Engineering means of EMTP simulations and experimental tests using a
at Federal University of Paraiba (UFPB), 58051-970 João Pessoa, Brazil. (e- commercial available relay. The obtained results demonstrate
mail: [email protected]).
Mladen Kezunovic is with Department of Electrical & Computer Engineer- that the proposed settings improve the backup protection and
ing, Texas A&M University, USA. (e-mail: [email protected]) also avoids distance zones overlapping in adjacent lines.

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
2

II. P ROBLEM S TATEMENT

Consider the general power system topology shown in Fig.
1. The transmission lines starting in buses from 3 to N
terminate at the bus 2 . For a distance relay located at at bus
r , a solid fault is placed at the fictitious bus 1 , which is
inserted in the line between buses m and 2 . In the example
shown in Fig. 1, bus r is taken as bus 3 , and bus m
corresponds to bus 4 . Thereby, the voltage at the relay point
can be computed as [4]:


V R = Z32 I 32 + hZ42 I 32 + I 52 + · · · + I N 2 , (1)
where h is the percentage of the line 4 − 2 correspondent
to the line segment 1 − 2 (see Fig. 1); Z32 and Z42 are the Fig. 1. General power system topology.
series impedance of the lines 3 − 2 and 4 − 2 , respectively;
and I ij is the infeed current coming from the bus i towards
bus j . where the superscript (s) = (0), (1) or (2) stands for the zero,
Since the current at the relay point is I 32 , the apparent positive and negative sequence components, respectively; h is
impedance seen by the relay is computed as: the percentage of the line m − 2 from bus 1 to bus 2 (see
(s) (s)
Fig. 1); Zr2 and Zm2 are (s)-sequence series impedance of
 
 N  the relay line r − 2 and the faulted line m − 2 , respectively;
VR  X I n2  (s)
ZR = = Z32 + hZ42  1 + , (2) and I n2 is the (s)-sequence infeed current coming from bus
IR 
 I 
n=5 32  n towards bus 2 , which can be computed as:
| {z }
M (s) (s)
" #
(s) (s)
where M is named as infeed coefficient [4]. Usually, it is a (s) Vn −V2 zn1 − z21 (s) (s) (s)
I n2 = (s)
=− (s)
I f = Γn2 I f , (4)
near real positive number, that causes the distance relay to Zn2 Zn2
underreach. Traditionally, protection engineers consider some
with
critical operational conditions to include the infeed effect on (s) (s)
(s) zn1 − z21
settings. Nevertheless, since it is challenging to known infeed Γn2 = , (5)
(s)
currents a priory for different operational conditions (i.e., with Zn2
and without contingencies), there is no guarantee that distance (s) (s)
zone overlapping in adjacent lines is completely overcome, where zn1 and z21 stands for the elements (n,1) and (2,1)
(s)
mainly for deeply connected power grids. Therefore, for the of the (s)-sequence bus impedance matrix, respectively; Zn2
sake of simplicity, protection engineers may disregard the is the (s)-sequence series impedance of the line n − 2 ; and
(s)
infeed effect on settings to avoid relay overreaching. As a I f is the (s)-sequence fault current at the fault point.
consequence, the backup coverage on adjacent lines may Substituting (5) in (3) yields:
reduce tremendously depending on infeed current levels [4].  
To provide remote backup protection, large 3rd distance (s)
N (s)
Γn2  (s) (s)
 (s) (s)
X
zone can be used to cover the adjacent lines entirely [20]. V R,m = Zr2 + hZm2 Γ I .
(s)  r2 f
(6)
This approach can restrict the amount of load a circuit can n=3 Γr2
n̸=m
carry under emergency conditions due to relays tripping,
which leads to an unintentionally operation during widespread Aiming to understand how phase-to-phase (PP) and phase-
disturbances causing cascading relay tripping, as happened in to-ground (PG) distance units perform during faults, consider
the North American power grid blackout in 2003 [21]. As a the three-phase and single-phase-to-ground fault analysis dis-
result, several countries, among them Brazil, have standardized cussed next.
the use of local backup protection principle for the entire
transmission system such that just the 1st and 2nd distance
A. Phase-to-Phase (PP) Distance Units
zones have been used.
In order to analyze the PP distance units response, consider
III. P ROPOSED A LGORITHM a three-phase fault taking place at the faulted line m - 2 . Ac-
(a) (1)
From (2), and based on the general power system topology cording to the fault analysis fundamentals [22], V R,m = V R,m
shown in Fig. 1, one can write the sequence voltages at the (b) (1)
and V R,m = a2 V R,m , such that:
relay point for a fault taking place at bus 1 in between the  
line m - 2 , as follows: N (1)
(a) (b)  (1) (1)
X Γn2 
(1) (1)
N V R,m −V R,m = Zr2 + hZm2 (1) 
1 − a2 Γr2 I f ,
(s) (s) (s) (s)
n=3 Γr2
(s)
X
V R,m = Zr2 I r2 + hZm2 I n2 , (3) n̸=m
n=3 (7)
n̸=m

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
3

where the superscript (a) and (b) stand for phases quantities, From the distance protection basics, the apparent impedance
(a)
and the complex constant a = 1∠120◦ . Likewise, I R,m = seen by the phase-to-ground AG unit is computed as [4]:
(1) (b) (1) (a)
I R,m and I R,m = a2 I R,m . Thus, from (4) one can obtain: (ag) V R,m
ZR,m = (a) (0)
. (15)
(a) (b)
(1) (1)
I R,m − I R,m = 1 − a2 Γr2 I f . (8) I R,m + Kr2 I R,m
From the distance protection basics, the apparent impedance Therefore, by substituting (12) and (14) in (15) yields:
seen by the phase-to-phase AB unit is computed as [4]: N (1) (0)
2Γn2 + (1 + Km2 ) Γn2
(ag) (1) (1)
X
(a) (b) ZR,m = Zr2 + hZm2 (1) (0)
, (16)
(ab) V R,m − V R,m n=3 2Γr2 + (1 + Kr2 ) Γr2
ZR,m = (a) (b)
. (9) n̸=m
I R,m − I R,m
where it can be seen that (16) depends on the line impedance
Therefore, by substituting (7) and (8) in (9) yields: and the (s)-sequence bus impedance matrix only.
N (1) Likewise, the response of the other phase-to-ground distance
(ab) (1) (1)
X Γn2
ZR,m = Zr2 + hZm2 (1)
, (10) units for the different single-phase-to-ground faults can be
n=3 Γr2 obtained straightforwardly in the same.
n̸=m

where, one can see that (10) depends on the line impedance C. 2nd Zone Settings for PP and PG Distance Units
and the (s)-sequence bus impedance matrix only.
Considering faults in the different lines connected to the
Similarly, it is possible to determine that for the other phase-
bus 2 , and varying the percentage h of the faulted line
to-phase units the response will be the same. PP
(see Fig. 1), one can calculate the 2nd zone setting Z2nd of
PP distance units using (10). Thereby, the effective backup
B. Phase-to-Ground (PG) Distance Units PP
protection coverage CR,m for faults in the different lines can
Aiming to analyze the PG distance units response, consider be obtained taking into account the infeed effect by solving:
a single-phase-to-ground fault in phase A, taking place at the (1) (1)
PP PP
faulted line m - 2 . The phase voltage at the relay point can Z2nd = Zr2 + CR,m Zm2 GP P
R,m , (17)
| {z }
be written as: PP
(a) (0) (1) (2)
ZR,m
V R,m = V R,m + V R,m + V R,m PP
  where ZR,m is the apparent impedance seen by the PP distance
N (0) unit of the relay for a fault taking place at the percentage
 (0) (0)
X Γn2  (0) (0) PP
= Zr2 + hZm2 Γ I +
(0)  r2 f
CR,m of the line m - 2 ; m ̸= r and the coefficient GP P
R,m is
n=3 Γr2 computed as:
n̸=m N (1)
 
PP
X Γn2
N (1)
X Γ  (1) (1) (11) GR,m = (1)
. (18)
n=3 Γr2
 (1) (1) n2
Zr2 + hZm2 Γ I +
(1)  r2 f n̸=m
n=3 Γr2

n̸=m
 Thereby, the total effective backup protection coverage for
N (2)
all the adjacent lines connected to bus 2 can be obtained as:
 (2) (2)
X Γn2  (2) (2) N
Zr2 + hZm2 Γ I .
(2)  r2 f X
PP PP
n=3 Γr2 T CR,m = CR,m . (19)
n̸=m
m=3
m̸=r
Considering negative sequence impedances are equal to the
positive ones, and the relationship of sequence currents at the According to (16), one can calculate the 2nd zone setting
(0) (1) (2) PG
fault point I f = I f = I f , one can rewrite (11) as: Z2nd for PG distance units taking into account the infeed
h i effect, considering phase-to-ground faults in the different lines
(a) (1) (1) (1) (0)
V R,m = Zr2 I f 2Γr2 + (1 + Kr2 ) Γr2 connected to the bus 2 , and changing the percentage h where
PG
N h the fault may occur. Thus, the effective coverage CR,m for
(1) (1)
X (1) (0)
i (12) faults in different lines can be determined as:
+hZm2 I f 2Γn2 + (1 + Km2 ) Γn2 ,
PG PG (1) (1)
n=3
n̸=m
Z2nd = Zr2 + CR,m Zm2 GP G
R,m , (20)
| {z }
(0) (1) (0) (1) PG
Zr2 − Zr2 Zm2 − Zm2 ZR,m
where Kr2 = (1)
and Km2 = (1)
.
Zr2 Zm2 where ZR,mPG
is the apparent impedance seen by the PG
From (4) one can obtain:
distance unit of the relay for a phase-to-ground fault taking
(a) (0) (1) (2) (0) (0) (1) (1) PP
I R,m = I R,m + I R,m + I R,m = Γr2 I f + 2Γr2 I f . (13) place at the percentage CR,m ̸ r and
of the line m - 2 ; m =
PG
(0)
the coefficient GR,m is computed as:
Adding Kr2 I R,m on both sides of (13), it is possible to
N (1) (0)
rewrite (13) as: X 2Γn2 + (1 + Km2 ) Γn2
h i GP G
R,m = (1) (0)
. (21)
(a) (0) (1) (1) (0) 2Γr2 + (1 + Kr2 ) Γr2
I R,m + Kr2 I R,m = I f 2Γr2 + (1 + Kr2 ) Γr2 . (14) n=3
n̸=m

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
4

The premise taken is that, for the k-th iteration, hm,k has a
polynomial relation with the maximum backup coverage Cm,k
among the adjacent lines. Therefore, a second order Lagrange
interpolation polynomial has been applied as follows [23]:
(1 − Cm,k−2 ) (1 − Cm,k−1 )
hm,k = hm,k−3
(Cm,k−3 − Cm,k−2 ) (Cm,k−3 − Cm,k−1 )
(1 − Cm,k−3 ) (1 − Cm,k−1 )
+ hm,k−2
(Cm,k−2 − Cm,k−3 ) (Cm,k−2 − Cm,k−1 )
(1 − Cm,k−3 ) (1 − Cm,k−2 )
Fig. 2. Proposed settings versus boundary buses impedances variations. + hm,k−1 .
(Cm,k−1 − Cm,k−3 ) (Cm,k−1 − Cm,k−2 )
(23)
Thus, the total effective percentage coverage for all the The 2nd distance zone setting is computed for hm,k [using
adjacent lines connected to bus 2 can be determined as: (10) or (16)], and its effective backup coverage for each
adjacent line is calculated [using (17) or (20)]. Thereby, Cm,k
N
PG
X
PG is taken as the largest backup coverage among adjacent lines,
T CR,m = CR,m . (22)
creating an ordered pair (hm,k , Cm,k ). Then (23) is applied
m=3
m̸=r to k + 1, and so on. The optimization process continues until
It is noteworthy that, for both PP and PG distance units, the optimal hm,k is obtained subject to the constraint Cmax .
GR,m depends on CR,m and this relation might be quite Special countermeasures must be taken for k ≤ 2: for k = 0,
nonlinear, specially for deeply connected power grids. It is (hm,0 , Cm,0 ) = (0 , 0) and for k = 1, hm,1 is assumed equal
observed in Fig. 2, that shows some apparent impedance to a initialization value hini arbitrarily. Then, the 2nd distance
curves seen by the relay for different adjacent lines. Therefore, zone setting is computed for hini , and the maximum backup
different strategies could be used to solve (17) and (20), coverage Cm,1 is taken as the largest coverage among adjacent
obtaining the effective backup coverage CR,m related to Z2nd . lines, creating the ordered pair (hm,1 , Cm,1 ). For k = 2, in
Traditionally, when infeed effect is taken into account during turn, the percentage hm,2 is calculated using the first order
settings calculation, protection engineers use short-circuit pro- Lagrange polynomial as [23]:
grams to compare |ZR,m | and |Z2nd | graphically for some 1
hm,2 = hm,1 . (24)
critical operational conditions known beforehand. The load Cm,1
current are usually disregarded. Thereby, Z2nd is obtained to
Then, the 2nd zone setting is computed, and the maximum
mitigate distance zones overlapping [4]. Nevertheless, even
backup coverage Cm,2 is taken as the largest effective coverage
though this method is widespread, its use in an automatic
in adjacent lines, creating the ordered pair (hm,2 , Cm,2 ).
procedure to compute the effective coverage considering all the
In order to avoid convergence problems, hm,k for each
system topologies may lead to large errors, because for meshed
iteration must be confined to a limit defined as:
grids the fault current direction may change throughout the
system depending on the fault location. As a result, |ZR,m | hm,k−1 − ∆hmax < hm,k < hm,k−1 + ∆hmax , (25)
may be smaller than |Z2nd |, but ZR,m is outside of the
distance operation characteristic. To overcome this drawback, where ∆hmax is the maximum increment considered.
alternatively, in our paper we evaluate whether ZR,m lies At the end of the whole optimization process, the optimal
inside the operation characteristic using cosine-type phase hm and its correspondent 2nd zone setting for each adjacent
comparators [22]. Thereby a more reliable estimation of CR,m line are obtained, as well as their effective backup coverage
can be obtained, because the comparator output is more similar in each circuit.
PP PG
to that of commercial available relays. The proposed settings Z2nd and Z2nd are those that lead
PP
To define the best relay settings Z2nd PG
and Z2nd , a simplified to the largest total effective percentage coverage, which are
optimization method is applied: the sum of backup coverage in computed using (19) for PP units and using (22) for PG units.
the all adjacent lines is maximized and restricted to a specified
maximum coverage of an adjacent line in particular, to avoid E. Proposed Algorithm Application
distance zones overlapping.
As one can see, the proposed algorithm requires only
impedance data and the operational status of the power ap-
D. Simplified Optimization Process paratus. Then, the simplified optimization process described
For the sake of simplicity, superscripts PP and PG are in the Sec. III-D can be carried out, resulting in the optimal
omitted, since the same optimization process is applied for 2nd zone settings that lead to the maximum backup protection
both units. The goal is to calculate the percentage hm in the coverage for adjacent lines. It can be used for both offline and
adjacent line m - 2 used to compute the 2nd zone settings online applications straightforwardly.
that maximize the effective backup coverage for all adjacent For offline applications, the proposed algorithm gives sup-
lines, subject to the given constraint Cmax , which stands for port to protection engineers, providing optimal settings cal-
the maximum coverage considered for backup protection. culation. Contingency scenarios of nearby equipment can be

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
5

AGL RIB RC1 BGI CGD

AGL_RIB_L1 RIB_RC1_L1 RC1_BGI_L1
NTD
NTD_CGD_L1
RC1_BGI_L2
1x100MVA 3x100MVA
2x100MVA
NTD_CGD_L2
RC2_BGI_L1
AGL_RC1_L1
2x100MVA 1x100MVA
NTD_CGD_L3

2x100MVA PRD NTD_CGD_L4

RC1_PRD_L1 PFE
2x100MVA
PFE_CGD_L1
CGD38 STD38
MRR
MRR_PFE_L1 PFE_CGD_L2 CGD_STD_L1
1x100MVA 2x55MVA
GNN
CGD_STD_L2
MRR_RC1_L1 RC1_GNN_L1
Relay Point MRD
3x100MVA 2x100MVA
MRR_GNN_L1 MRD_GNN_L1
Transmission lines
Evaluated in RC2
Case 2 MRR_RC2_L1 MRD_GNN_L2 2x100MVA Transmission lines
Evaluated in
Case 1
MRR_RC2_L2 RC2_PRD_L1 MRD_GNN_L3 2x100MVA

AGL_RC2_L1 RC2_GNN_L1 CGD_GNN_L1

GNN_CGD_L1 Legend
Transmission Line
Relay Point Two winding
AGL_CGD_L1 230 kV
Power Transformer
138 kV
TAC Three winding 69 kV
AGL_TAC_L1 TAC_CGD_L1 Power Transformer 13,8 kV

AGL_TAC_L2 TAC_CGD_L2 Thévenin

Equivalent

AGL_TAC_L3 Load

2x100MVA

Fig. 3. Evaluated Power System.

considered during the optimization process to prevent 2nd zone mutual impedances. Then, power equipment impedance data
overlapping of adjacent lines. Moreover, permanent topology were used to calculate the 2nd distance zone setting by means
changes may occur over time, and new apparatus may be of the proposed algorithm. Finally, faults were simulated
connected to the grid. The proposed algorithm can also be using ATP/ATPDraw to assess the actual backup coverage
used to monitor the effectiveness of distance relays settings performance of the obtained settings.
continuously. Settings obtained by the proposed algorithm consider the
Following the concept of wide-area adaptive protection maximum constraint of 60% (i.e., Cmax = 0.6) for the effec-
applications, the proposed algorithm can also be integrated tive backup coverage. Their results were compared to those
into a feedback loop toward relays to change their settings calculated using the traditional procedure, without taking into
online dynamically. Aiming to do so, the operational status account infeed effect and considering 30% of the impedance
of the power apparatus provided by SCADA can be used to of the smallest adjacent line [4]. Two different cases were
decide whether relay settings must be updated. Moreover, it is evaluated. In each one, the effective coverage of both PP
noteworthy that it must be done only for those contingencies and PG units were computed for each infeed circuit, with
that lead to large overreach and may cause 2nd zone overlap- and without contingencies in the system. Then, faults were
ping in adjacent lines, as discussed next. simulated in ATP/ATPDraw using time step of 1 µs, exactly
on the maximum backup coverage zone obtained using the
IV. P ROPOSED A LGORITHM E VALUATION proposed algorithm for each infeed circuit. A third order low-
Aiming to evaluate the proposed algorithm, the power pass anti-aliasing Butterworth filter with cutoff frequency at
system depicted in Fig. 3 was modeled in software 180 Hz was applied to voltages and currents at the relay point
ATP/ATPDraw. It corresponds to part of the power system (see Fig. 3). The filtered signals were resampled at 16 samples
of a Brazilian utility. Power transformers were represented per cycle of 60 Hz, and then phasors were estimated [25]. The
using the saturable transformer model, and transmission lines apparent impedance seen by the relay for each distance unit
were modeled as fully transposed using the Bergeron model is shown in the RX diagram along with Mho characteristics
[24]. Load and shunt reactors were modelled as lumped com- of 2nd distance zones adjusted using proposed settings and
ponents. Thévenin’s equivalents were included in boundary the traditional ones. As another test, ATP/ATPDraw simulated
buses (i.e., AGL, RC1 and RC2), considering both self and signals were converted into COMTRADE files, thereby exper-

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
6

(a) (b)
Fig. 4. Case 1 − Effective coverage obtained by the proposed and traditional settings considering the system without contingency: (a) PP units; (b) PG units.

(a) (b) (c) (d)

Fig. 5. Case 1 − Performance comparison of 2nd distance zone settings computed using the proposed algorithm and the traditional procedure, considering
different types of faults: (a) AG, (b) BC, (c) BCG and (d) ABC.

imental evaluation were carried out using a playback test with was adjusted using the traditional procedure. In Fig. 5(a),
a commercial available relay. the apparent impedance computed by the ZAG distance unit
for an AG solid fault is shown. On the other hand, Figs.
5(b) and 5(c) depict the apparent impedance seen by the
A. Case 1
ZBC distance unit for BC and BCG solid faults, respectively.
In Case 1, lines from buses AGL, GNN, NTD, PFE, and Fig 5(d), in turn, shows the apparent impedance seen by the
TAC to the bus CGD were considered (see transmission ZAB distance unit for a three-phase solid fault. As it can be
lines highlighted in blue color on Fig. 3). In this case, the observed, the apparent impedance settled down on the light
relay is located at the terminal AGL of the transmission line gray Mho characteristic boundary in all cases, revealing the
AGL_CGD_L1. proposed settings improves distance protection performance.
The proposed algorithm was applied to the system topology Conversely, if traditional settings procedure is used, the relay
shown in Fig. 3, without contingencies. It converged in 65 underreaches and does not provide effective backup protection.
iterations for PP units and in 61 iterations for PG units. The
effective coverage obtained using the proposed and traditional Aiming to further evaluate the effectiveness of the proposed
settings are illustrated in Fig. 4(a) and 4(b) for PP and PG and traditional settings, the contingency of each infeed circuit
units, respectively. As one can see, the proposed settings lead was assessed, and then the backup protection coverage in
to much larger backup coverage of adjacent lines. Indeed, as the remain circuits were computed. The obtained results are
discussed in Sec. II, since the traditional procedure evaluated shown in Figs. 6(a) and 6(b) for PP units, respectively, whereas
here disregard the infeed effect, and bus CGD concentrates a the results for PG units are shown in Figs. 7(a) and 7(b),
large number of circuits, the backup coverage provided by the respectively. These figures depict the percentage coverage
traditional settings is quite reduced. matrices for contingency analysis. Their rows represent the
The performance of the distance relay for faults simulated effective coverage of the relay settings for infeed circuits,
using ATP/ATPDraw are shown in Figs. 5. Each color used regarding one of them is under contingency. For example,
to plot the apparent impedance represents results for a fault the first row of matrices shown in Figs. 6 and 7 represents
simulated in a particular adjacent line, exactly on the maxi- the effective coverage of the proposed and traditional settings
mum backup coverage computed using the proposed algorithm for each infeed circuit, considering the line TAC_CGD_L1 is
[see Figs. 4(a) and 4(b)]. For example, the blue color is under contingency. Whenever the effective coverage is smaller
used to plot the apparent impedance seen by the relay for than the maximum constraint of 60%, the correspondent cell
faults in the line TAC_CGD_L1, whereas the green color is is filled in green color. Otherwise, cells are filled in red color,
taken for faults in the line GNN_CGD_L1, and so on. The indicating the maximum constraint was violated, except the
light gray Mho characteristic represents the 2nd distance zone main diagonal cells, in which a blue color and the letter X
obtained using the proposed settings, whereas the dark one were used to represent these elements must be disregarded.

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
7

(a) (b)
Fig. 6. Case 1 − Percentage coverage matrix for PP distance units in the case of a single contingency on infeed circuits: (a) proposed settings computed
without considering contingencies; (b) traditional settings.

(a) (b)
Fig. 7. Case 1 − Percentage coverage matrix for PG distance units in the case of a single contingency on infeed circuits: (a) proposed settings computed
without considering contingencies; (b) traditional settings.

(a) (b)
Fig. 8. Case 1 − Results of the new proposed setting for PP units considering contingencies during the optimization process: (a) effective coverage for the
system without contingency (b) percentage coverage matrix in the case of contingencies on infeed circuits.

By the analysis of Fig. 6(a), one can see that the proposed optimization process, resulting new settings that avoid 2nd
settings lead to quite larger effective coverage for PP distance zones overlapping. The effective coverage obtained using this
units in comparison with the traditional ones. However, they new setting for PP units are illustrated in Fig. 8(a), and the
overreach remote bus of adjacent lines during contingencies correspondent percentage coverage matrix is shown in Fig.
in lines TAC_CGD_L1, TAC_CGD_L2 and GNN_CGD_L1, 8(b). As observed, the overreach was completely overcome,
revealing this setting must be updated. In offline applica- and the effective coverage of the proposed setting is still quite
tions, contingencies could be taken into account during the larger than the one obtained using the traditional setting. Alter-

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
8

(a) (b)
Fig. 9. Case 2 − Effective coverage obtained by the proposed and traditional settings considering the system without contingency: (a) PP units; (b) PG units.

(a) (b) (c) (d)

Fig. 10. Case 2 − Performance comparison of 2nd distance zone settings computed using the proposed algorithm and the traditional procedure, considering
different types of faults: (a) AG, (b) BC, (c) BCG and (d) ABC.

natively, an adaptive protection scheme could be implemented to plot the apparent impedance represents results for a fault
for online applications. Therefore, whenever contingencies in simulated at the maximum backup coverage computed using
lines TAC_CGD_L1, TAC_CGD_L2 and GNN_CGD_L1 are the proposed algorithm for each adjacent line. For example,
detected by the analysis of SCADA data, new settings for the the blue color is used to plot the apparent impedance seen
system operating with each one of them out of service could by the relay for faults in the line RC2_GNN_L1, whereas the
be send to the relay. Also, since these scenarios are known green color is taken for faults in the line CGD_GNN_L1, and
beforehand, settings could be computed offline. Thereby, one so on. The light gray Mho characteristic represents the 2nd
can understand that the proposed settings could always provide distance zone obtained using the proposed settings, whereas
the optimal backup protection coverage. the dark one was adjusted using the traditional settings. In
Conversely, besides the proposed setting for PG units lead to Fig. 10(a), the apparent impedance computed by the ZAG
larger backup protection coverage than the traditional setting, distance unit for an AG solid fault is shown. On the other
there is no overreach larger than 60%, even when contin- hand, Figs. 10(b) and 10(c) depict the apparent impedance
gencies in infeed circuits happens. Therefore, in the case of seen by the ZBC distance unit for BC and BCG solid faults,
PG units, the proposed setting computed without taking into respectively. Fig 10(d), in turn, shows the apparent impedance
consideration contingencies does not need to be updated. seen by the ZAB distance unit for a three-phase solid fault.
As it can be observed, the apparent impedance settled down
on the light gray Mho characteristic boundary in all cases,
B. Case 2
revealing the proposed settings improves distance protection
In Case 2, lines from buses RC1, RC2, CGD, MRR and performance. Conversely, if the traditional settings procedure
MRD to the bus GNN were considered (see transmission lines is used, the relay underreaches and does not provide effective
highlighted in pink color on Fig. 3). In this case, the relay is backup protection.
located at the terminal RC1 of the line RC1_GNN_L1.
The proposed algorithm was applied to the system topol- The percentage coverage matrices for contingency analysis
ogy shown in Fig. 3, without considering contingencies. It of the proposed and traditional settings are shown in Figs.
converged in 61 iterations for PP units and in 54 iterations for 11(a) and 11(b) for PP distance units, respectively, whereas
PG units. The effective coverage obtained using the proposed the ones for PG units are shown in Figs. 12(a) and 12(b),
settings and the traditional ones are illustrated in Fig. 9(a) and respectively. It can be seen that the proposed settings lead
9(b) for PP and PG units, respectively. As one can see, the to larger effective coverage for both PP and PG units in
proposed settings lead to larger backup coverage of adjacent comparison with the traditional settings. However, both units
lines (at least the double in all cases). overreach remote bus of adjacent lines during contingencies
The performance of the distance relay for faults simulated in lines RC2_GNN_L1, CGD_GNN_L1 and MRR_GNN_L1,
using ATP/ATPDraw are shown in Figs. 10. Each color used revealing these settings must be updated. In offline applica-

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
9

(a) (b)
Fig. 11. Case 2 − Percentage coverage matrix for PP distance units in the case of a single contingency on infeed circuits: (a) proposed settings computed
without considering contingencies; (b) traditional settings.

(a) (b)
Fig. 12. Case 2 − Percentage coverage matrix for PG distance units in the case of a single contingency on infeed circuits: (a) proposed settings computed
without considering contingencies; (b) traditional settings.

tions, contingencies could be taken into account during the the boundary buses AGL, RC1 and RC2 (see Fig. 3) were
optimization process, resulting in new settings that avoid 2nd varied from −50% to +50%, and the magnitude variation
zones overlapping. The backup coverage obtained using this on settings obtained using the proposed algorithm for both
new setting for PP and PG units are illustrated in Figs. 13(a) PP and PG units was observed, as shown in Fig. 15. As
and 13(b), respectively. The correspondent percentage cover- demonstrated, even with a quite large variation of impedances
age matrices are shown in Figs. 14(a) and 14(b), respectively. in the boundary buses, the variation on settings ranges between
As observed, the overreach was overcome for both PP and ±1.35%. Indeed, as a rule of thumb, it is recommended that
PG units and the effective coverage of the proposed settings monitored system boundaries are defined two buses away
is still larger than the one obtained using the traditional from the relay bus whenever it is possible. Thereby, these
settings. Alternatively, an adaptive protection scheme could equivalents have no significant influence on the proposed
be implemented for online applications. Therefore, whenever settings so their impact can be disregarded and their values
contingencies in lines RC2_GNN_L1, CGD_GNN_L1 and for just one specific operational condition are required (e.g.,
MRR_GNN_L1 are detected by the analysis of SCADA data, for heavy load condition). Admittedly, the impedance data and
new settings previously calculated offline for the system with operational status of power equipment in the monitored system
each one of them out of service could be send to the relay. are more important to the proposed algorithm.
Thereby, the proposed settings could always provide the op-
timal backup protection coverage, besides avoiding 2nd zone
D. Experimental Evaluation
overlapping.
The performance of the proposed settings was also evaluated
in a commercially available relay, considering case 1 and 2
C. Boundary Buses Impedances Influence described in Sec. IV-A and IV-B. To prove the usefulness of
Aiming to evaluated the impact of impedances of boundary the proposed method, quadriletaral characteristic was used for
buses on the proposed settings, a sensitivity analysis was both PP and PG 2nd distance units instead of mho characteris-
carried out. In each case, Thévenin equivalent impedances in tic. To do so, only reactances of the proposed settings and the

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
10

(a) (b)
Fig. 13. Case 2 − Effective coverage of the proposed settings, considering contingencies on infeed circuits, and of the traditional settings: (a) PP distance
units; (b) PG distance units.

(a) (b)
Fig. 14. Case 2 − Percentage coverage matrix for the proposed settings computed considering contingencies on infeed circuits: (a) PP distance unit; (b) PG
distance unit.

For offline applications, the proposed algorithm gives sup-

port to protection engineers, providing optimal settings calcu-
lation. Outage scenarios in nearby equipment can be taken into
account during the optimization process. On the other hand, it
can also be used for online applications in the context of wide-
area adaptive protection schemes. Aiming to do so, the relay
can change its settings depending on whether contingencies
are detected by SCADA data analysis. It is noteworthy that
the settings only need to be changed if contingency takes
Fig. 15. Proposed settings versus boundary buses impedances variations. place in a few infeed circuits. Since these scenarios are known
beforehand, settings could be computed offline. Moreover,
there is no need to make changes for contingencies in faraway
traditional ones were taken into account. The ATP/ATPDraw lines. Thereby, one can understand that the proposed settings
simulated signals were converted in COMTRADE files and a could always provide the maximum backup coverage for each
playback test was carried out using a conventional relay test adjacent line.
set. The obtained results reaffirmed the computer simulations,
such that all faults were detected only when the proposed set- It is noteworthy that the proposed method disregards the
tings were used, guaranteeing the maximum backup coverage load current effect, likewise the settings procedure traditionally
for each adjacent lines. used by protection engineers. As a result, admittedly, the
distance protection function may underreach or overeach,
depending on the the loading condition. To overcome this
V. C ONCLUSIONS
drawback, the proposed algorithm could be improved by
In this paper, the well-known infeed effect on distance taking into account the loading current. Furthermore, the
protection was revisited and an adaptive algorithm to compute proposed method proposes a correction on the 2nd zone set-
2nd distance zone settings was proposed. It requires only tings, rather than distance polarizing quantities (e.g, zero and
impedance data and operational status of power apparatus, negative sequence quantities), what could improve even more
such that voltages and currents are not required. It can be used the distance function performance. Indeed, such improvements
for both offline and online applications straightforwardly. will be presented in future researches.

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2021.3128131, IEEE
Transactions on Power Delivery
11

The effectiveness of the proposed algorithm was assessed [19] S. A. E. Mousavi, R. M. Chabanloo, M. Farrokhifar, and D. Pozo,
by means of faults simulated in an actual power system of a “Wide area backup protection scheme for distance relays considering
the uncertainty of network protection,” Electric Power Systems Research,
Brazilian utility using ATP/ATPDraw software. The apparent vol. 189, p. 106651, 2020.
impedance seen by the relay algorithm for each case was [20] NERC Report - System Protection and Control Task Force, “Rationale
plotted in the RX diagram along with Mho characteristics for the use of local and remote (zone 3) protective relaying backup
systems,” Princeton, NJ, EUA, Tech. Rep., 2005.
of 2nd distance zones adjusted using the proposed settings [21] NERC Steering Group Report, “Technical analysis of the august 14,
and traditional settings. It was observed that the apparent 2003, blackout: What happened, why, and what did we learn?” Prince-
impedance settled down on the Mho characteristic boundary ton, NJ, EUA, Tech. Rep., 2004.
[22] P. M. Anderson, Power System Protection. Piscataway, New Jersey,
adjusted using proposed settings in all cases, revealing the EUA: John Wiley & Sons Inc., 1999.
proposed algorithm improves tremendously the backup protec- [23] W. H. Press, S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery,
tion coverage by taking the infeed effect into account, while Numerical Recipes: The Art of Scientific Computing. Cambridge, UK:
Cambridge University Press, 2007.
avoiding distance zones overlapping between adjacent lines. [24] ATP - Alternative Transient Program, Leuven EMTP Center, Herverlee,
The same results were observed in the experimental evaluation Belgium, 1987.
using a commercially available relay. This highlights the [25] K. M. Silva and F. A. O. Nascimento, “Modified dft-based phasor
estimation algorithms for numerical relaying applications,” IEEE Trans.
proposed algorithm usefulness and value from the practical on Power Delivery, vol. 33, no. 3, pp. 1165–1173, Jun. 2018.
point of view, since it can be implemented with the digital
technology readily available on the market.
Kleber M. Silva (M’05, SM’20) received the B.Sc.,
M.Sc., and Ph.D. degrees in electrical engineering
from the University of Campina Grande, Brazil,
R EFERENCES in 2004, 2005, and 2009, respectively. Since 2009,
[1] IEEE PSRC Working Group I19, “Redundancy considerations for pro- he has been a professor with the University of
tective relaying systems,” USA, Tech. Rep., 2019. Brasilia, Brazil, and the Head of the Power System
Protection Group. From 2019 to 2020, he was a
[2] IEEE PSRC Working Group D4, “Application of overreaching distance
visiting professor at the Texas A&M University,
relays,” USA, Tech. Rep., 2009.
College Station, TX, US. His research interests fo-
[3] W. A. Elmore, Protective Relaying Theory And Applications. New
cus on power system protection, fault location, and
York, US: CRC Press, 2003.
electromagnetic transients. He is an Editor for the
[4] G. Ziegler, Numerical Distance Protection: Principles and Applications,
IEEE Transactions on Power Delivery and Member of SC B5–Protection and
2nd ed. Berlin, Germany: Siemens, 2006.
Automation Committee of Cigre Brazil.
[5] Y. Xia, K. Li, and A. David, “Adaptive relay setting for stand-alone
digital distance protection,” IEEE Transactions on Power Delivery,
vol. 9, no. 1, pp. 480–491, 1994. Jéssica S. G. Pena received the B.Sc. degree in
[6] T. Sidhu, D. Baltazar, R. Palomino, and M. Sachdev, “A new approach energy engineering and the M.Sc. degree in electrical
for calculating zone-2 setting of distance relays and its use in an adaptive engineering from the University of Brasilia (UnB),
protection system,” IEEE Transactions on Power Delivery, vol. 19, no. 1, Brasilia, Brazil, in 2013 and 2016, respectively.
pp. 70–77, 2004. She is currently working toward the Ph.D. degree
[7] M. Gilany, A. M. Al-Kandari, and J. Madouh, “A new strategy for in electrical engineering at the UnB. Her research
determining fault zones in distance relays,” IEEE Transactions on Power interests include power system protection, energy
Delivery, vol. 23, no. 4, pp. 1857–1863, 2008. quality and electrical generation.
[8] M. A. Haj-ahmed and M. S. Illindala, “Intelligent coordinated adaptive
distance relaying,” Electric Power Systems Research, vol. 110, pp. 163–
171, 2014.
[9] J. Ma, C. Liu, and J. S. Thorp, “A wide-area backup protection algorithm
based on distance protection fitting factor,” IEEE Transactions on Power Helon D. M. Braz (M’ 17) received the B.Sc. degree
Delivery, vol. 31, no. 5, pp. 2196–2205, 2016. in electrical engineering from the Federal Univer-
[10] N. A. Al-Emadi, A. Ghorbani, and H. Mehrjerdi, “Synchrophasor-based sity of Paraíba (UFPB), Campina Grande, Brazil,
backup distance protection of multi-terminal transmission lines,” IET in 2001. He later received his M.Sc. and Ph.D.
Gen., Trans. & Distrib., vol. 10, no. 13, pp. 3304–3313, 2016. degrees from the Federal University of Campina
[11] P. Regulski, W. Rebizant, M. Kereit, and S. Schneider, “Adaptive reach Grande (UFCG), Campina Grande, Brazil, in 2003
of the 3rd zone of a distance relay with synchronized measurements,” and 2010, respectively. He worked at Eletrobras
IEEE Transactions on Power Delivery, vol. 36, no. 1, pp. 135–144, 2021. Chesf on power systems planning studies for seven
[12] J. Wang and J. Trecat, “Rsvies-a relay setting value identification expert years and currently, he is a Professor at UFPB.
system,” Elec. Pow. Sys. Research, vol. 37, no. 2, pp. 153–158, 1996. His research activities are focused on optimization
[13] J. Pinto De Sa, J. Afonso, and R. Rodrigues, “A probabilistic approach methods applied in power systems.
to setting distance relays in transmission networks,” IEEE Transactions
on Power Delivery, vol. 12, no. 2, pp. 681–686, 1997.
[14] K. Li, L. Lai, and A. David, “Stand alone intelligent digital distance Mladen Kezunovic (S’77-M’80–SM’85–F’99-
relay,” IEEE Trans. Power Systems, vol. 15, no. 1, pp. 137–142, 2000. LF’17) has been with Texas A&M University,
[15] K. El-Arroudi, G. Joos, D. McGillis, and R. Brearley, “Comprehensive College Station, TX, USA for 35 years, where he
transmission distance protection settings using an intelligent-based anal- holds titles of Regents Professor, Eugene E. Webb
ysis of events and consequences,” IEEE Transactions on Power Delivery, endowed Professor, and Site Director of “Power
vol. 20, no. 3, pp. 1817–1824, 2005. Engineering Research Center” consortium. He is
[16] P. Kundu and A. K. Pradhan, “Enhanced protection security using the also the Principal of XpertPower Associates™, a
system integrity protection scheme (sips),” IEEE Transactions on Power consulting firm specializing in power systems data
Delivery, vol. 31, no. 1, pp. 228–235, 2016. analytics for the last 30 years. His expertise is
[17] M. Tasdighi and M. Kezunovic, “Automated review of distance relay set- in protective relaying, computational intelligence
tings adequacy after the network topology changes,” IEEE Transactions for automated power system disturbance analysis,
on Power Delivery, vol. 31, no. 4, pp. 1873–1881, 2016. predictive data analytics for outage management, and smart grids. Dr.
[18] M. Chen, H. Wang, S. Shen, and B. He, “Research on a distance relay- Kezunovic is an IEEE Life Fellow, and a CIGRE Fellow, Honorary and
based wide-area backup protection algorithm for transmission lines,” Distinguished Member. He is a Registered Professional Engineer in Texas.
IEEE Transactions on Power Delivery, vol. 32, no. 1, pp. 97–105, 2017.

0885-8977 (c) 2021 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on December 03,2021 at 07:18:57 UTC from IEEE Xplore. Restrictions apply.