Electrical Power and Energy Systems 105 (2019) 454–469

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

A pilot-based distance protection scheme for meshed distribution systems T

with distributed generation
⁎
Aristotelis M. Tsimtsiosa, George N. Korresb, Vassilis C. Nikolaidisa,
a
Department of Electrical and Computer Engineering, Democritus University of Thrace, Xanthi 67100, Greece
b
School of Electrical and Computer Engineering, National Technical University of Athens, Iroon Polytechniou 9, 15780 Athens, Greece

A R T I C LE I N FO A B S T R A C T

Keywords: This paper proposes a pilot-based distance protection scheme for meshed distribution systems with a high pe-
Distance protection netration of distributed generation. The proposed scheme assumes distance relays installed at the opponent ends
Distributed generation of each main line segment. A forward distance element is enabled in each relay to protect the main line segment,
Meshed distribution system while, a reverse distance element is enabled in each relay for protecting adjacent buses and laterals. The relays
Pilot protection
protecting a main line segment or a bus/lateral communicate in a permissive logic, ensuring protection sensi-
tivity and security at the same time. The scheme provides efficient primary and backup protection against faults
of any type, even with a considerable fault resistance, and under weak-infeed conditions. It is also appropriate
for both the grid-connected and the islanded mode of system operation. Coordination of the distance protection
scheme with the lateral protection means is also addressed. Existing numerical distance relay technology is
considered, to enhance the applicability of the proposed scheme. Offline setting of the distance relays is required,
which has to be performed once. Only one setting group is ultimately extracted for each relay, suitable for
different fault/system conditions, avoiding the need for adaptive protection techniques. The proposed scheme is
applied to a test meshed distribution system and conclusions are drawn. Comparison with directional over-
current and differential protection shows the advantages of the designed distance-based protection scheme.

1. Introduction operators (DSOs) and DG producers [3]. In fact, such multi-relay

The integration of distributed generation (DG) into distribution
systems imposes the reconsideration of conventional overcurrent pro-
tection techniques, due to several issues encountered under the pre-
sence of DG [1]. This is further necessitated by the prospect of oper-
ating distribution networks in a looped/meshed configuration, which
aims to enhance their reliability, but also to improve their voltage
profile, to reduce power losses, and to increase their DG hosting ca-
pacity [2].
In order to maximize the reliability of distribution networks, which
is the main purpose of looped/meshed operation, this concept requires
the installation of circuit breakers, operated by relays, at both ends of
each line segment, and, usually, the use of communication means to
ensure protection efficiency under various conditions concerning the
mode of system operation (i.e. grid-connected vs. islanded), the DG
contribution and the faults occurring (e.g. fault type/position, fault
resistance etc.). Although this prospect requires a considerable invest-
ment, the latter can be considered justified, taking into account the
operational/maintenance benefits gained by the distribution system
communication-assisted protection schemes have been considered for
real-world applications, even for looped distribution systems without
DG [4,5], to enhance their reliability.
So far, multi-relay schemes using directional overcurrent relays
have been mainly considered in the literature as a solution for looped/
meshed distribution systems. Within this context, optimization algo-
rithms are a promising tool for setting directional overcurrent relays
protecting such networks [6–9]. Communication-based adaptive di-
rectional overcurrent protection is applied to deal either with DG-
connection and system-configuration changes [10,11], or with the
transition from the grid-connected to the islanded mode of system op-
eration [11]. Pilot-based directional overcurrent protection schemes
could also be applicable to actual looped distribution systems without
DG [4,5], or to looped microgrids [12].
Alternatively, advanced differential protection schemes have been
considered for protecting each line segment in looped [13] or meshed
[14,15] distribution systems, while, communication-assisted multi-
relay schemes, applying alternative protection principles, are also
proposed for such applications [16–18].

⁎
Corresponding author.
E-mail address: vnikolai@ee.duth.gr (V.C. Nikolaidis).

https://doi.org/10.1016/j.ijepes.2018.08.022
Received 5 April 2018; Received in revised form 19 July 2018; Accepted 18 August 2018
0142-0615/ © 2018 Elsevier Ltd. All rights reserved.
A.M. Tsimtsios et al.
The advantages of distance relays (e.g. inherent directionality and
non-direct dependency on short-circuit currents [19]) render them a
promising option for modern distribution systems, although their ap-
plication is still challenging [20]. So far, distance protection schemes
proposed for distribution systems mainly regard radial networks.
Distance protection is applied to grid-connected radial distribution
systems without DG in [21,22]. The reach of the distance zones is
empirically set in these applications, which, in general, might prove
unreliable if DG and various fault/system conditions are also con-
sidered.
The presence of DG renders the application of distance protection
more challenging. The authors of [23] consider a distance relay for
protecting a grid-connected radial feeder with DG, which coordinates
with other protection means. Moreover, the underreach effect of DG
infeed is dealt with through an overreaching distance zone. However,
the resistive reach of the distance zones is set based on a rule of thumb,
so, the actual effect of fault resistance is not taken into account. Dis-
tance-relaying applications for grid-connected radial feeders with DG
are also examined by the authors as part of [24,25]. Multiple distance
zones are applied, which are suitably extended to address the DG infeed
effect, while, coordination with lateral fuses is addressed. Nevertheless,
similarly to the abovementioned paper, the effect of fault resistance is
still empirically considered. In [26], distance relays are intended to
replace the undervoltage relays of DG units and some of the feeder
overcurrent relays, in a grid-connected radial feeder. The issue of dis-
tance relay coordination with lateral protection is also mentioned.
However, the performance of the distance relays is adversely affected
by infeed/outfeed currents and fault resistance. Distance protection is
applied to radial feeders with DG, as part of [27,28]. Mho-based dis-
tance elements are used, which are empirically set, taking into account
the infeed constants. Fault resistance effect is not actually dealt with,
while, coordination with lateral protection is also not considered. In
[29], an adaptive distance protection scheme is applied to a grid-con-
nected radial feeder with DG. Only solid phase faults are considered,
while, coordination with lateral protection is not addressed in this work
either.
Alternative techniques have also been proposed, intending to deal
with the effect of DG infeed and/or fault resistance. In [30], a distance
protection scheme, applied to a grid-connected/islanded radial feeder,
is proposed, where the DG infeed effect is compensated by calculating
the infeed error term through current measurements. However, this
compensation would not be directly applicable under the effect of fault
resistance, which is not taken into account in this work. In addition,
coordination of the distance relays with lateral protection means is not
addressed. A measurement-based infeed-current compensation metho-
dology is also used in [31]; however, fault resistance is not dealt with in
this case as well. Coordination of backup main line distance protection
with primary main line overcurrent protection is also addressed, al-
though the latter does not hold for coordination with lateral protection.
In [32], the distance relays protecting a grid-connected radial feeder
against single-line-ground faults are set through an optimization algo-
rithm, considering a specific range of fault resistance values. Never-
theless, the efficiency of this algorithm can be limited due to system
changes and/or under the presence of DG, while, coordination with
common lateral protection means is not taken into account. A pro-
mising distance-based approach for a grid-connected radial distribution
network with DG is presented in [33]. This approach is fault-location-
oriented, aiming to compensate for impedance calculation errors due to
DG infeed and fault resistance. Communication means and multi-point
measurements are required for this purpose. However, the accuracy of
this methodology can be limited when load and fault resistance are
high, while, the methodology is not applicable against single-line-
ground faults (i.e. the most frequent faults).
Despite the fact that pilot-based distance protection has been tra-
ditionally applied to transmission systems [34], its proper application/
adaptation to modern distribution systems could be considered for
455
Electrical Power and Energy Systems 105 (2019) 454–469
addressing several of the aforementioned issues. A pilot-based distance
protection scheme for a grid-connected/islanded microgrid is proposed
in [35,36], which is, however, applicable only to a radial configuration.
Moreover, specific guidelines for setting the impedance characteristics
used against various fault/system conditions are not given. Faults with
a considerable resistance are not simulated, thus, the effect of fault
resistance is not actually dealt with. Coordination of the distance relays
with other protection means encountered in distribution systems (e.g.
fuses) is also not addressed. A permissive underreaching transfer trip
(PUTT) logic is applied as part of [37], but only for the sake of accel-
erating the operation of one of the two distance relays protecting a line
segment in a radial microgrid, on condition that this relay has already
detected the fault inside its second distance zone. Multiple setting
groups are stored for protection against various system conditions
(adaptive logic), while, Mho-based distance zones are used, whose
settings are extracted based on a rule of thumb. The latter might prove
unreliable for various resistive faults. Moreover, the scheme is tested
only for low-resistance three-phase faults, while, coordination with
lateral protection means is not addressed.
Another concept considering a distance relay as a protection ele-
ment in a grid-connected, radial distribution network has been recently
proposed in [38]. However, in this work, the distance relay is only
examined as the interconnection relay of a wind farm, while, the au-
thors investigate the effect of wind power intermittency on the relay’s
performance, solely in terms of polarization using the pre-fault voltage.
Other impedance-based protection schemes for radial distribution fee-
ders with DG have also been proposed, as part of a differential [39,40]
and an inverse-time-impedance [39] protection principle though;
hence, actual/common distance protection is not examined in these
papers.
The advantages of distance protection could become more promi-
nent in looped/meshed distribution networks with DG, although their
consideration for actual applications might still prove problematic [41].
Despite the significant issues to be addressed in this direction, relevant
research efforts are limited. In [35], a central distance-based protection
scheme is considered for an islanded, ring-type microgrid; however, the
issues mentioned previously for the radial application of the same work
hold in this case as well. Distance protection is considered for a grid-
connected ring-type network in [42], although intermediate DG infeed
is not taken into account. Coordination with lateral protection means is
also not addressed. In [43] and [44], distance protection is, respec-
tively, applied to a meshed and a looped distribution system with DG,
using multiple setting groups (adaptive logic) against different system
conditions. Fault resistance effect and coordination with lateral pro-
tection means are not addressed in these papers. A preliminary attempt
by the authors to apply distance protection to a looped distribution
network with DG is also included in [25].
As part of [45,46], the authors also analyze the considerable effect
of zero-sequence compensation, DG infeed and fault resistance speci-
fically on distance relays protecting distribution systems. These factors
might lead to severe impedance calculation errors, compromising pro-
tection dependability and security if their concurrent effect is not
properly considered during relay setting. However, the latter has not
been sufficiently addressed in the literature so far. Note that, although
[45,46] refer to radial feeders, the aforementioned influencing factors
must be considered for setting distance relays in looped/meshed dis-
tribution networks as well.
In this work, a distance-based protection scheme for meshed dis-
tribution systems with DG is developed, intending to address all the
aforementioned issues encountered in the literature, concerning the
application of distance protection to distribution systems. For the sake
of enhancing the applicability of the proposed solution, existing dis-
tance relay technology (i.e. tested by vendors in real-world applica-
tions) is exploited, which is, however, properly adapted to the needs of
the present application. The designed distance scheme is suitably set to
provide efficient protection against solid and non-solid (with a
A.M. Tsimtsios et al.
considerable fault resistance) faults of any type, under high DG pene-
tration, while, weak-infeed conditions are also dealt with. Permissive
logic is applied to ensure protection speed, sensitivity and security at
the same time. Coordination with lateral protection means is also ad-
dressed. Moreover, the proposed scheme is efficient in both the grid-
connected and the islanded mode of system operation, without using
any complex adaptive logic.
The organization of this paper is as follows. Section 2 includes a
brief theoretical analysis of the basic influencing factors affecting dis-
tance relay operation. In Section 3, the generic meshed distribution
system configuration adopted in this paper is briefly described, while, a
thorough description of the proposed protection philosophy is also
given. The detailed setting procedure of the distance relays is addressed
in Section 4. Section 5 presents the simulation results obtained from the
application of the proposed scheme to a test system. In Section 6, the
advantages of the proposed scheme compared to other distance-based
schemes are described, while, a comparison with a directional-over-
current-based and a differential-based protection scheme is also in-
cluded. Conclusions are drawn in Section 7.
2. Factors affecting operation of distance relays
Distance relays calculate the positive-sequence impedance of the
protected circuit up to the fault position, based on local voltage and
current measurements. In fact, a distance relay calculates six impedance
values, three for phase fault loops (a-b, b-c, c-a) and three for ground
fault loops (a-g, b-g, c-g). The calculated impedance for a phase (e.g. a-
b) and a ground (e.g. a-g) fault loop is, respectively, equal to [19]:
Za − b = (Va−Vb) (Ia−Ib )
Za − g = Va [Ia + K 0·(3·I0)]
(1)
where Va, Vb, Ia and Ib is the phase-a voltage, phase-b voltage, phase-a
current and phase-b current, respectively, measured by the relay. K0 is
the zero-sequence compensation factor, required to compensate for
zero-sequence current (I0) during ground faults, and in that case is
calculated as [19]:
K 0 = (Z0−Z1) (3·Z1) (2)
where Z0 and Z1 is the positive- and zero-sequence impedance, re-
spectively, up to the fault position.
Major influencing factors for distance relays are the intermediate
infeed effect, which appears when power sources are connected be-
tween the relay and the fault position, and the fault resistance effect.
Both may result in inaccurate apparent impedance calculation by the
distance relay [34].
The latter is briefly explained with the help of Fig. 1a, which depicts
a faulty part of a distribution feeder, protected by a distance relay (DR).
In order to analyze the infeed effect solely, a DG unit is assumed con-
nected between the DR and the fault position F (Case 1 in Fig. 1a). In
Fig. 1. (a) Infeed/fault resistance effect, (b) Impedance seen by the distance relay.
456
Electrical Power and Energy Systems 105 (2019) 454–469
case of a solid short-circuit, the positive-sequence impedance up to
point F, seen by DR, will be:
Z = +Z
ZA B + 
(IB
/ )·Z
IA B
ZF Zadd1 (3)
where ZA is the positive-sequence impedance from the relay location up
to the DG connection point, ZB is the positive-sequence impedance from
the DG connection point up to the fault position F, IA is the short-circuit
current flowing from the external grid and IB is the infeed current
flowing from the DG unit.
To analyze the fault resistance effect solely, a fault with resistance
RF is assumed at point F (Case 2 in Fig. 1a). For the sake of generality, it
is assumed that the fault is also fed by a reverse current (IR ), which
could flow from DG units downstream to point F and/or from the grid
in case of a looped/meshed network topology. Note that the infeed
effect is not examined now, since, in that case, DG units would be as-
sumed connected between the relay and point F. The apparent positive-
sequence impedance up to point F will be:
Z = ZF + 
(IR
/
IF )·
RF
Zadd2 (4)
where ZF is the positive-sequence impedance from the relay location up
to the fault position F, and IF is the short-circuit current flowing from
the grid.
It is apparent that, in both cases (Case 1 and Case 2), the relay sees
an additional impedance (Zadd1 and Zadd2 , respectively), apart from the
actual impedance ZF up to the fault position. Depending on the relation
between the phase angles of IB , IA for Case 1 and IR , IF for Case 2, the
above phenomena can lead to underreach or overreach of the distance
relay [46]. Fig. 1b shows the impedance seen by a distance relay on the
complex impedance (R-X) plane for a fault downstream or upstream to
its location, in case of underreach (ZC1 and ZC1′, respectively) or over-
reach (ZC2 and ZC2′, respectively). The underreach/overreach effect
results from the additional calculated impedance Zadd due the separate
or even the combined effect of the above-described factors. It appears
that, for downstream faults (forward direction), the relay can see the
fault in the first or even in the fourth quadrant, in case of a very intense
overreach (dashed arrow) [46]. Respectively, for upstream faults (re-
verse direction) the relay can see the fault in the third or the second
quadrant. Such intense overreach effects may be encountered under
specific fault conditions (concerning fault type and fault position) if
fault resistance is considerable. The concurrent effect of K 0 during
ground faults plays a crucial role in the appearance of these overreach
phenomena, in distribution system applications of distance relaying
[46].
It is obvious that if the above-described factors are not properly
taken into account when designing a distance protection scheme
(especially for distribution applications), the reliability of the latter will
be doubtful. These factors are addressed by the proposed protection
A.M. Tsimtsios et al.
Fig. 2. Examined system and proposed protection scheme.
scheme, described in the following.
3. Proposed distance protection scheme
3.1. Generic meshed distribution system
The proposed protection scheme is demonstrated with the help of
the generic meshed overhead distribution system of Fig. 2. Two pairs of
crossed interconnection line segments are assumed to increase the
network’s complexity. This system can operate in both grid-connected
and islanded mode. Inverter-interfaced generating units (photovoltaic
plants PV1 and PV2 ), as well as conventional synchronous generators
(SG1 and SG2 ), are assumed connected to the positions shown in Fig. 2.
Fuses are assumed protecting laterals Li (i = 1, …, 10) . Fuses are con-
sidered instead of overcurrent relays, as they are the most encountered
(low-cost) protection means for distribution system laterals, while, they
are more difficult to coordinate with main feeder relay(s), due to their
non-settable characteristic.
3.2. Description of the proposed protection scheme
3.2.1. Dealing with faults occurring in the main line segments
Each main line segment Li, j (i, j = 0, …, 10, i ≠ j ) connecting buses
Bi and Bj is protected by a numerical distance relay DRi, j installed at Bi
and a numerical distance relay DRj, i installed at Bj (see Fig. 2).
457
Electrical Power and Energy Systems 105 (2019) 454–469
Through the above implementation, distance relays are never af-
fected by intermediate infeed during forward faults, since no DG units
are connected between buses Bi and Bj . Note that, forward faults are
recognized by the relays as fault currents flowing into the corre-
sponding main line segment. Even if a DG unit was tapped at an in-
termediate point of a main line segment, the latter would be divided at
the point of DG connection and distance relays would again be installed
at the endpoints of the new segments formed, without any DG con-
nected between them. This practice is due to avoid unnecessary dis-
connection of DG units during main line faults, which is critical to
motivate DG investments in distribution systems. However, inter-
mediate DG infeed effect is encountered when designing bus and lateral
protection, and will be dealt with as part of the corresponding proposed
protection scheme, described later.
Two forward distance elements (FDEs) are enabled in each distance
relay, i.e. a phase forward distance element (PFDE) and a ground for-
ward distance element (GFDE). PFDEi, j and PFDEj, i of distance relays
DRi, j and DRj, i , respectively, are responsible for protecting line segment
Li, j against forward line-line-line (LLL) and double-line (LL) faults,
whereas, GFDEi, j and GFDEj, i of distance relays DRi, j and DRj, i respec-
tively, are assigned for forward single-line-ground (SLG) and double-
line-ground (LLG) faults. It has to be mentioned that in several com-
mercial distance relay models, LLG faults can be detected and cleared
by both phase and ground distance elements. However, from the
viewpoint of the methodology proposed here, this is not problematic,
A.M. Tsimtsios et al.
Fig. 3. Overall protection and communication logic of the proposed scheme.
since the concurrent trip of phase and ground distance elements does
not affect the scheme’s efficiency. This will become clearer later in this
subsection, after the proposed protection philosophy is further de-
scribed. Unless necessary, we will not specifically refer to phase or
ground distance elements in the rest of this paper, as the proposed
methodology is equally applied to both.
The reach of FDEi, j (resp. FDEj, i ) is initially set to cover beyond Bj
(resp. Bi ), meaning that both FDEs overreach the total line length of the
protected segment (e.g. by 20% of this length), to compensate for any
relay calculation or measuring errors during faults in segment Li, j . Note
that this reach extension is performed only for the latter reason and not
for detecting faults in the adjacent line segment (i.e. for backup pro-
tection). The tripping direction of each FDE and its assigned main line
part is shown with solid arrows in Fig. 2. A quadrilateral characteristic
is applied to all distance elements.
The previous paragraph (and Fig. 2) outlines just the basic setting
philosophy of the proposed scheme. In case of solid faults occurring in
segment Li, j , both FDEi, j and FDEj, i will calculate the impedance up to
the fault position accurately, as distance elements are not affected by
any external changes (e.g. transmission grid changes, transition from
grid-connected to islanded mode etc.), while, infeed/fault resistance
effect is not present. Therefore, if we considered only solid faults, the
above-described basic setting philosophy would be adequate, i.e. the
reach of FDEi, j /FDEj, i would simply be set according to the positive-
sequence impedance of Li, j (also considering 20% overreach for calcu-
lation/measuring errors). Since, however, the proposed scheme deals
with more realistic fault conditions, i.e. fault resistance is considered as
well, the additive term Zadd2 of (4) is taken into account for setting the
distance elements properly. In particular, when needed, the reach of
FDEi, j /FDEj, i is further extended to compensate for the effect of fault
resistance, so that every possible resistive fault in Li, j is located inside
the corresponding distance zones. The amount of the required extension
is determined through proper simulation study, as described later. Al-
though reach extension (by 20% or more) can result in FDEi, j /FDEj, i
sensing faults occurring outside Li, j , permissive logic will prevent them
from tripping undesirably, ensuring protection security.
The aforementioned permissive logic is designed as follows. Once
FDEi, j of a relay DRi, j confirms that a fault is detected inside its forward
distance zone (FDZi, j ), it sends a permission signal to FDEj, i of DRj, i and
vice versa. The signal is transferred through a communication path (e.g.
a fiber optic cable) linking opponent FDEs. Note that PFDEi, j (resp.
GFDEi, j ) exchanges permission signals only with PFDEj, i , (resp. GFDEj, i ).
458
Electrical Power and Energy Systems 105 (2019) 454–469
We assume that fault-confirmation and permission signals remain ac-
tive for a proper time duration, to ensure correct operation of the
protection scheme.
Ultimately, FDEi, j trips circuit breaker CBi, j at DRi, j location if the
following conditions hold simultaneously:
• FDE i, j confirms that a fault lies inside FDZi, j .
• FDE i, j receives permission from FDEj, i .
Hence, for faults inside line segment Li, j , FDEi, j and FDEj, i trip CBi, j
and CBj, i , respectively, isolating only this segment. For faults outside
Li, j , at least one of the above FDEs does not sense the fault, thus un-
desired trips are avoided.
From the viewpoint of tripping time, FDEi, j and FDEj, i are set with
an instantaneous time delay (tFi, j and tFj, i , respectively), ensuring rapid
fault clearance, as only this unintentional instantaneous time delay (i.e.
the response time of the relays), which is less than 50 ms for common
relays, should elapse for a trip order to be issued. Note that commu-
nication latency is negligible (4.9 μs/km for fiber optic cable), thus it
does not practically affect the tripping time of the relays. Of course, the
overall fault clearing time (tC) is the sum of the tripping (response) time
of the relays, plus the interrupting time of the CBs (tCB).
It is worth mentioning that, after a distance relay DRi, j calculates the
impedance of each fault loop during a fault (see Section 2), it has to
determine the type of the fault as well as the faulted phase. This is to
take into account the correct impedance value for checking if the latter
is detected inside a distance zone or not, as well as to decide between
single- or three-pole trip. This can be easily addressed, since most
common commercial numerical distance relays include advanced phase
selection functions. The latter are based on several criteria, such as the
impedance magnitude of each fault loop, the relation between the
phase angles of the current/voltage sequence components etc.
[34,47,48].
The protection and communication logic of a FDEi, j (described
previously) is included in Fig. 3. To sum up, after the relay calculates
the impedance of the fault loop indicated by the phase selection func-
tion, it checks if it lies inside FDZi,j. In that case (and after the time
required for the relay to respond expires), a permission signal is sent to
FDEj,i. If, at the same time, a permission signal is received from FDEj,i,
CBi,j is tripped and the breaker-failure function (described later) is in-
itiated. Obviously, FDEj, i operates similarly.
A.M. Tsimtsios et al.
Fig. 4. Pilot-based protection scheme for faults in the first line segment, under
weak-infeed conditions.
3.2.2. Special protection treatment of segments L0,1 and L0,10
A potential issue during the grid-connected mode of system opera-
tion, concerns faults occurring close to the main substation, i.e. in
segments L0,1 and L0,10 . In the general case where fault resistance is
expected, and concerning only FDE1,0 (resp. FDE10,0 ), the term IR/IF of
(4) may become considerably large, due to the weak current IF flowing
from bus B1 (resp. B10 ). This would subsequently result in a large re-
quired extension of FDZ1,0 (resp. FDZ10,0 ) when setting the relay, which
may not be applicable with commercial relays, due to their limited
reach setting range. Although this issue could be addressed by the relay
vendors, a dedicated solution is proposed below, to keep up with
commercially-available relay specifications.
As the operation of FDE1,0 and FDE10,0 is not reliable for faults close
to the main substation, a combination of the directional comparison
blocking (DCB) principle with the permissive logic is proposed for
protecting segments L0,1 and L0,10 . This functional combination is illu-
strated in Fig. 4, focusing on L0,1, while it is equally applied for pro-
tecting L0,10 . In brief, FDE0,1 is set as all FDEs, with the exception of not
requiring permission to trip from the opponent distance element FDE1,0 ,
which is kept disabled in DR1,0 . Instead, a reverse blocking distance
element (RDEb1,0 ) is enabled in DR1,0 . This element supervises the
feeder downstream to B1 (in the reverse direction of DR1,0 , as shown
with the solid arrow in Fig. 4) and blocks FDE0,1 whenever a fault oc-
curs inside this area. In this way, undesired trips of FDE0,1 due to the
extension of its reach beyond B1 (as explained before), are avoided.
The main particularity of the pilot-based protection solution pro-
posed here concerns the fact that, since FDE1,0 is disabled, RDEb1,0 is
also responsible for tripping CB1,0 in case of a fault inside L0,1 . For this
reason, RDEb1,0 receives a permission signal from FDE0,1, once the latter
confirms that a fault lies inside FDZ0,1 (i.e. after tF0,1), while it trips CB1,0
only if:
• ItIt has received permission from FDE . 0,1
• has not detected a fault inside its reverse blocking zone RDZb 1,0 .
RDEb1,0 is set with instantaneous time delay (tR1,0 ), as FDE0,1. Hence,
the overall fault clearing time (tC) in this case will be equal to the re-
sponse time of the relay plus the interrupting time of the CB (tCB).
The trip logic of FDE0,1 and RDEb1,0 is also shown in Fig. 3. To sum
up, the respective relays initially calculate the impedance of the fault
loop indicated by the phase selection function. If the impedance cal-
culated by DR1,0 lies inside RDZb1,0, RDEb1,0 sends a blocking signal to
FDE0,1. If the opposite happens, and RDEb1,0 receives a permission
signal from FDE0,1, then (after the unintentional time delay tR1,0 ex-
pires), CB1,0 is tripped and the breaker-failure function (described later)
is initiated. Note that, in order for the aforementioned permission signal
to be sent, FDE0,1 has in the meantime confirmed that the impedance
calculated by DR0,1 lies inside FDZ0,1 (after the expiration of the relay’s
response time). In the latter case, FDE0,1 also trips CB0,1 and initiates
the breaker-failure function (described later).
3.2.3. Dealing with faults occurring in buses or laterals
For the primary protection of each bus Bi and the backup protection
of the corresponding lateral Li (primarily protected by fuse Fi ), the,
already installed, distance relays adjacent to bus Bi are exploited, to
459
Electrical Power and Energy Systems 105 (2019) 454–469
avoid increasing the scheme’s cost by installing dedicated relay(s) for
bus/lateral protection. Specifically, in each relay DRi, j adjacent to Bi/Li,
apart from the FDEi, j activated for main line protection, a reverse dis-
tance element RDEi, j is activated for bus and lateral protection. RDEi, j is
responsible for protecting from bus Bi up to the endpoint of Li . For
instance, RDE1,0 , RDE1,2 and RDE1,9 of distance relays DR1,0 , DR1,2 , and
DR1,9 , respectively, are used to protect bus B1 and lateral L1, up to its
endpoint. In fact, the reach of RDEi, j is initially set to cover beyond Li
endpoint (e.g. increased by 20%, as happens for FDEi,j) to compensate
for possible relay calculation or measuring errors, during faults in Li .
The trip direction of each RDEi, j and its assigned line part is shown in
Fig. 2 with dashed arrows. Note that, concerning DR1,0 and DR10,0 , the
reverse blocking elements RDEb1,0 and RDEb10,0 , described in the pre-
vious paragraph, are independent of RDE1,0 and RDE10,0 , protecting
B1/L1 and B10 /L10 , respectively.
All the RDEs protecting Bi /Li form a group of adjacent RDEs, re-
ferred as GARDEi . Permissive trip logic is applied, i.e. RDEi, j trips CBi, j
only if:
• It sees a fault between (and including) B and L endpoint.
i i
• It receives permission from the rest RDEs of GARDE . i
As mentioned earlier, the reach of each RDEi, j is initially set to cover
at least 120% of Li length; however, there might be cases where this
reach is further extended during the setting procedure. Now, such cases
may arise not only due to the fault resistance effect (as happens when
setting FDEs for the main line protection) but also due to intermediate
infeed currents (see Section 2) flowing from DG units and/or other
feeders connected to Bi . Similarly to FDEs, the required additional reach
extension of RDEs, to compensate for the infeed/fault resistance effect,
is determined through proper simulation study, as described later.
RDEs should also coordinate with the lateral fuses. Therefore, the
maximum total clearing time (tTC,max ) of the lateral fuses is critical for
setting RDEs properly. This time depends on the fuse’s type and rated
current and the fault/system conditions. As tTC,max of a fuse Fi can reach
high values, it is up to the protection engineer to decide whether tRi of
the RDEs constituting GARDEi will be set greater than tTC,max for the
needs of relay-fuse coordination (as indicated later in Subsection 4.2),
or a lower tRi setting will be considered, to limit through-fault damage.
Note that in the latter case, relay-fuse coordination for a number of
fault cases will be “sacrificed” for the sake of protection speed. Ob-
viously, if fuse Fi fails to clear a fault in Li, the overall fault clearing time
(tC) will be equal to tRi plus the interrupting time of the CBs (tCB). It is
worth mentioning that the replacement of lateral fuses with relays
would make the protection design study much simpler. However, it is
here considered preferable to maintain fuses for lateral protection in
order not to increase the scheme’s cost and in order to consider the most
challenging scenario for setting the relays.
The trip logic of a RDEi, j is similar to that of a FDEi, j and is also
shown in Fig. 3. To sum up, after the relay calculates the impedance of
the fault loop indicated by the phase selection function, it checks if it
lies inside RDZi,j. In that case (and after the time delay set for co-
ordination with the corresponding lateral fuse expires), a permission
signal is sent to the rest relays of GARDEi. If a permission signal is also
received from the relays of GARDEi, CBi,j is tripped and the breaker-
failure function (described later) is initiated.
From what is described in this subsection, concerning both FDEs and
RDEs, it is derived that the proposed scheme ensures protection sensi-
tivity and security at the same time. The latter is further clarified with
the help of Fig. 5, which illustrates a main line segment Li,j (re-
presenting any main line segment of a meshed distribution system),
along with its adjacent buses, main line segments and laterals. For
simplicity, in this example, we focus solely on the protection of segment
Li,j, bus Bi and lateral Li. Fig. 5 also illustrates the communication paths
for permission signal exchange between the FDEs enabled in DRi,j, DRj,i
to protect Li,j, as well as the respective paths of the RDEs constituting
A.M. Tsimtsios et al.
Fig. 5. Explanation of the protection and communication philosophy of the proposed scheme.
GARDEi, enabled in DRi,j, DRi,h, DRi,l to protect Bi/Li. Let us assume fault
SC1 occurring in Li,j. In that case, both FDEi,j and FDEj,i will see the fault,
thus they will trip the CBs at their location, after they exchange per-
mission signals. However, as explained before, the FDEs might need to
extend further than their opponent bus (by even more than the initial
overreach of 20%), for sensitivity purposes. For instance, FDEi,j might
erroneously see fault SC1 as if it occurred in Lj,k instead of Li,j (under-
reach effect), due to the effect of fault resistance (see Section 2). By
properly extending the reach of FDEi,j, the latter will be able to protect
against such fault cases in Li,j. Even if this extension can also lead to
FDEi,j sensing a fault actually occurring further than Bj, e.g. fault SC2,
the permissive logic described previously will guarantee protection
security in that case. In other words, although FDEi,j will send permis-
sion to FDEj,i, the opposite will not happen, as FDEj,i will not see the
fault. Hence, none of these distance elements will ultimately trip in-
correctly. Remember that both the opponent FDEs must receive per-
mission signals from each other, in order for their CBs to be tripped.
RDEs operate similarly, and, in fact, are also properly extended to
deal with underreach phenomena, due to the effect of fault resistance
and/or the effect of intermediate DG connected to bus Bi (see Section
2). In case of fault SC3 occurring in lateral Li, all the RDEs of GARDEi
will see the fault and trip the CBs at their location, after they exchange
permission signals. Note that protection security is ensured in this case
as well. For instance, fault SC1 can be sensed by RDEi,h and RDEi,l, as the
latter are set to protect towards Bi/Li (i.e. in their reverse direction).
However, RDEi,j will not see this fault (since it has occurred in its op-
posite direction), thus it will not send permission signals to RDEi,h and
RDEi,l. Remember that all the RDEs of a GARDE must exchange per-
mission signals in order for their CBs to be tripped.
3.3. Backup protection
For the case where a CB fails to open, although a trip order has been
issued by the respective relay, a breaker-failure (BF) protection scheme
is designed. The latter ensures that if a CB fails to open after receiving a
trip signal by its corresponding relay, its adjacent CBs will be tripped by
the same relay. In general, BF protection functions are included in
several distance relay models [34]; however, the design of this logic is
here suitably adapted to the needs of the proposed protection scheme.
Specifically, as part of this work, BF is designed as follows: Once
FDEi, j or RDEi, j of DRi, j issues a trip order to CBi, j , a breaker-failure
initiation (BFI) signal is simultaneously asserted (see Fig. 3). We assume
that the BFI signal remains active for a proper time duration. After-
wards, the state of CBi, j is checked for a time period equal to tBF . Ty-
pically, tBF lasts for 7–15 cycles [49], to compensate for the CBi, j in-
terrupting time plus any other delays associated with the relay. If CBi, j
460
Electrical Power and Energy Systems 105 (2019) 454–469
remains closed after tBF expires, DRi, j trips the adjacent CBs. In the
meantime, a re-trip command is issued to CBi, j , as a second attempt to
trip this CB, in order to avoid opening the adjacent CBs. The re-trip time
delay (tRT ) is set to 1–2 cycles, just to avoid misoperation due to nui-
sance input (e.g. noise) [49]. Fig. 6 shows the designed BF trip logic of a
relay DRi, j . Note that the proposed BF function discriminates between
the BFI signal coming from FDEi, j (BFIFi, j ) or RDEi, j (BFIRi, j ), so as to trip
the proper adjacent CBs (CBFi, j and CBRi, j , respectively).
At this point, let us provide an example of the BF function. Once
FDE1,2 issues a trip command to CB1,2 and a BFI signal is asserted, a trip
command is re-issued to CB1,2 after tRT . If CB1,2 remains closed after
tBF > tRT , then DR1,2 trips CB1,0 and CB1,9 . In the special case where
FDE0,1 (resp. FDE0,10 ) fails to trip CB0,1 (resp. CB0,10 ), CB0,10 (resp. CB0,1)
and the CB at the substation transformer secondary trip instead after tBF
elapses. Since, tripping the CB at the substation transformer secondary
would lead to islanded operation, breaker redundancy could also be
considered as an option just for CB0,1 (resp. CB0,10).
In order to preserve their functionality, modern numerical relays
can continuously perform self-diagnosis in the form of watchdog cir-
cuitry and check for internal failures [19]. Hence, it is anytime known
whether each relay is able to provide primary/backup protection, or a
reparation action is required. Moreover, the operability of commu-
nication channels can be checked through regular signal exchange be-
tween the corresponding relays (e.g. through GOOSE messages).
In any case, the overcurrent element of the substation transformer’s
differential relay can be set as an additional backup protection (besides
BF) for the whole distribution system. The latter has to be set with a
time delay greater than the greatest time delay of the FDEs and RDEs
used, plus the tBF set in the distance relays. If the faulted network is
isolated from the external grid and, for any reason, both primary and BF
Fig. 6. Breaker-failure trip logic of a distance relay DRi,j.
A.M. Tsimtsios et al.
protection fail to clear the fault, the network will be de-energized after
the protection means of the DG units (which are out of the scope of this
work) disconnect the latter.
4. Detailed setting procedure of distance relays
For setting distance elements as part of the proposed methodology,
a proper fault-simulation study is conducted, taking into account all the
possible combinations of expected marginal fault/system conditions. In
this way, the relay setting procedure considers any infeed/fault re-
sistance effect expected during the grid-connected/islanded mode of
system operation, as well as any other factor that could affect the relays’
efficiency. A fixed setting group for each relay is finally determined,
protecting the system efficiently under any expected fault/system
condition, while, adaptive or other complex setting strategies are
avoided.
4.1. Setting FDEs for main line protection
To set FDEi,j, faults of all common types (i.e. LLL, LL, LLG, SLG) are
simulated in front of DRi, j and at the opponent endpoint of the protected
main line segment, i.e. at the marginal points of the line part assigned to
FDEi,j. Assuming the grid-connected mode of system operation, all the
simulations are performed for both zero and maximum DG production,
as well as for zero and maximum fault resistance. Assuming the is-
landed mode of system operation, all the simulations are performed for
zero and maximum fault resistance, but only for a fixed DG penetration
level permitting the islanded operation. This procedure is further clar-
ified with the help of Fig. 7a, which illustrates a main line segment Li,j
and its adjacent lateral Li. For simplicity, only the protection means
installed in Li,j and Li are shown. For setting FDEi,j, a number of fault-
simulation scenarios SFT,PL,RF,OM is considered for each position SCFDEi,j
(i.e. in front of DRi,j) and SCFDEj,i (i.e. in front of DRj,i). FT stands for the
fault type (LLL, LL, LLG, or SLG), PL stands for the DG penetration level
(zero or maximum), RF stands for fault resistance (zero or maximum
expected) and OM stands for the operation mode of the examined
system (grid-connected or islanded). Remember that zero PL is not
taken into account when OM is islanded. From the faults simulated at
each position, e.g. at position SCFDEi,j, a number of values (combina-
tions) results for the positive reactance +XFDEi,j,FT,PL,RF,OM, the negative
reactance −XFDEi,j,FT,PL,RF,OM, and the positive resistance
+RFDEi,j,FT,PL,RF,OM, calculated by the relay.
Fig. 7b depicts a generic quadrilateral FDEi, j protecting line segment
Li, j . Similar characteristic/settings also hold for RDEs/RDEbs, which are
set later. Setting each FDEi, j (and RDEi, j as explained later) is based on
the abovementioned simulation results and the determination of the
following settings:
1. Positive reactance reach ( + XFDEi, j )
2. Negative reactance reach (−XFDEi, j )
3. Positive resistive reach ( + RFDEi, j )
Fig. 7. (a) Fault-simulation strategy for setting FDEi,j/RDEi,j, (b) Typical quadrilateral FDEi,j settings.
461
Electrical Power and Energy Systems 105 (2019) 454–469
4. Angle of the down zone blinder (θd, FDEi, j )
5. Angle of the right zone blinder (θr , FDEi, j )
6. Angle of the left zone blinder (θl, FDEi, j )
Specifically, for determining settings 1–3 from the previous list, the
following maximum magnitudes, seen by each phase (resp. ground)
distance element during phase (resp. ground) faults, are extracted from
the simulations:
a. Maximum positive reactance + Xp,max (resp. + Xg,max ), being equal
to:
+ Xp,max (resp. + Xg,max )
= max{ + XFDEi, j, FT , PL, RF , OM , + XFDEj, i, FT , PL, RF , OM } (5)
b. Maximum negative reactance |−X |p,max (resp. |−X |g,max ), being equal
to:
|−X |p,max (resp. |−X |g,max ) = max{|−XFDEi, j, FT , PL, RF , OM |,
|−XFDEj, i, FT , PL, RF , OM |} (6)
c. Maximum positive resistance + Rp,max (resp. + Rg,max ), being equal
to:
+ Rp,max (resp. + Rg,max )
= max{ + RFDEi, j, FT , PL, RF , OM , + RFDEj, i, FT , PL, RF , OM } (7)
In (5)–(7), FT stands for a LLL or LL (resp. LLG or SLG) fault.
The ultimate settings +XPFDEi,j/+XGFDEi,j, |−XPFDEi,j|/|−XGFDEi,j|,
and + RPFDEi,j/+RGFDEi,j of phase/ground FDEi,j will be, respectively,
equal to:
+ XPFDEi, j = ( + Xp,max ) + ε , + XGFDEi, j = ( + Xg,max ) + ε
|− XPFDEi, j | = (|−X |p,max ) + ε , |−XGFDEi, j | = (|−X |g,max ) + ε
+ RPFDEi, j = ( + Rp,max ) + ε , + RGFDEi, j = ( + Rg,max ) + ε (8)
where ε (ε > 0) is an optional safety margin, to further ensure fault
detection.
Note that, concerning GFDEs, K 0 is set based on the positive- and
zero-sequence impedance of the protected main line segment. Also note
that, although only forward faults are of interest, −XFDEi, j and θd, FDEi, j
should be set properly, as it is possible for a forward fault to be detected
by FDEi, j in the fourth quadrant of the R-X plane (see Section 2). Angle
θd, FDEi, j (setting 4) could a priori be set equal to 90° (i.e. on the −X
axis), to ensure fault detection. Since the + RFDEi, j setting alone pro-
vides maximum fault resistance coverage, as it is based on the max-
imum resistance seen by the relay during the simulations, angle θr , FDEi, j
of the right blinder (setting 5) is set to 90°. The latter setting is here
disengaged from the traditional distance relay setting philosophy,
where the angle of the right zone blinder is set equal to the impedance
angle of the protected conductor. In addition, since forward faults are
not expected to lie in the second quadrant of the R-X plane, angle θl, FDEi, j
A.M. Tsimtsios et al. Electrical Power and Energy Systems 105 (2019) 454–469

positions SCRDEi,j and SC′RDEi,j of Fig. 7a (marginal points of the line

part protected by RDEi,j). The reach settings of RDEi,j are based on
the maximum calculated values resulting from these simulations, as
described for setting FDEi,j.
• The time delay tRi of RDEi, j is set to coordinate with fuse Fi of Li (see
Subsection 3.2). For this purpose, tTC,max for every fuse Fi (resulting
from the lowest short-circuit current flowing through Fi) is de-
termined by simulating faults of all types at the remotest point of Li ,
considering all the DG-contribution and system-operation scenarios.
Then, tRi for all the RDEs of GARDEi is set as:

tRi ⩾ tTC,max + CTI (10)

where CTI is a proper Coordination Time Interval.

Note that DRi, j locates reverse faults in the second or the third
quadrant of the R-X plane, as they occur upstream to its location (see
Fig. 1b). Therefore, a fault is determined as reverse, if the following
holds for the angle θz of the calculated impedance:

90o < θZ < 270o (11)

Fig. 8. Generic quadrilateral phase/ground FDEi,j, after performing the pro-
posed setting procedure. Nevertheless, the impedance values opposite to the calculated ones
are determined to extract RDEi, j settings, since, in all common com-
mercial distance relays, reverse elements are set like being forward.
of the left blinder (setting 6) is set to zero. Based on the above angle- Compensation factor K 0 of GRDEi, j is set based on the impedance of
setting philosophy, a fault is determined as forward, if the following the conductor where DRi, j is installed. The impedance calculation error
holds for the angle θz of the calculated impedance: due to the effect of K 0 , accompanied with the effect of infeed currents
−90o < θZ < 90o (9) and fault resistance, is anyway compensated [45,46] by the distance
zone reach settings extracted from the above-described simulation
Finally, as mentioned in Section 3, no intentional time delay is set
study, since K0 is taken into account during the simulations, along with
for FDEs.
all the other influencing factors.
Fig. 8 illustrates the generic form of a quadrilateral phase/ground
Finally, undesired trips of phase RDEs under heavy normal load
FDEi,j, after being set according to the above-described philosophy. The
conditions must be avoided (FDEs are never affected as at least one of
dotted arrows stand for the impedance calculation errors during the
the opponent FDEs in a main line segment does not see the load current
specific faults, occurring at the marginal points of Li,j (SCFDEi,j and
in its trip direction). To deal with this issue, a load encroachment (LE)
SCFDEj,i), which give the maximum calculated positive reactance/re-
element is enabled for each phase RDE. This element, included in sev-
sistance and negative reactance. The final FDEi,j reach settings can re-
eral commercial distance relays [23], blocks trip command if the cur-
main equal to the respective maximum calculated values, or can be
rent flowing over the segment of interest is within its normal load va-
further extended by ε (dashed lines in Fig. 8), to enhance dependability.
lues. The latter are determined through pre-fault load-flow simulations,
Note that, in the example of Fig. 8, the choice of which fault gives each
considering all the possible DG-connection cases and line segment/
maximum value is indicative. Moreover, a single fault could determine
lateral configurations in the grid-connected/islanded mode. Note that
more than one reach settings of FDEi,j.
cases where LE elements prevent phase RDEs from clearing LLL faults
It is worth mentioning that the value of the calculated reach settings
are expected to be very rare.
depend on various factors concerning the specific system examined,
such as the length/cross-section of the protected conductors, the short-
circuit contribution of the utility (during the grid-connected mode) and
4.3. Setting distance relays in segments L0,1 and L0,10
the DG units, the system configuration etc. Moreover, the effect of
several influencing factors (e.g. fault resistance) is unpredictable.
As explained in Subsection 3.2, the reach settings of RDEb1,0 have to
Hence, a simulation study (as that proposed in this work) is a quite
be greater than the (over)reach settings of FDE0,1 (see Fig. 4). There-
reliable tool to take into account all the particularities of the specific
fore, the following must hold:
network under consideration. Note that simulation studies considering
marginal fault/system conditions constitute the common methodology + XRDEb1,0 > [(+ XFDE 0,1)−( + XL0,1)]
used by protection engineers for setting protective relays in real-world
+ RRDEb1,0 > [(+ RFDE 0,1)−( + RL0,1)]
applications. In addition, as mentioned before, by considering ε, it is
|− XRDEb1,0 | > |(−XFDE 0,1 )−( + XL0,1)| (12)
further ensured that all the expected faults will be detected by the ex-
amined distance element, compensating for possible calculation/mea-
suring errors. Since, at the same time, protection security is guaranteed where + XRDEb1,0 , + RRDEb1,0 and −XRDEb1,0 is the positive reactance
by the permissive logic described in Section 3, ε can be set as great as reach, positive resistive reach and negative reactance reach, respec-
desired, to ensure dependability. tively, of RDEb1,0 , + XFDE0,1, + RFDE0,1 and −XFDE0,1 is the positive re-
actance reach, positive resistive reach and negative reactance reach,
4.2. Setting RDEs for bus and lateral protection respectively, of FDE0,1 and + XL0,1, + RL0,1 is the positive-sequence
reactance and resistance, respectively, of L0,1 conductor. Similarly to
The same procedure as that described in Subsection 4.1 is followed FDEs/RDEs, the angle of RDEb1,0 left blinder is set equal to zero, the
for setting RDEi, j , meaning to determine settings 1–6 for this element. angle of RDEb1,0 right blinder is set equal to 90° and the angle of
The only differences here are that: RDEb1,0 down blinder is also set equal to 90°, to ensure fault detection.
RDEb10,0 is set similarly. No intentional time delay is set for RDEbs.

• Faults are simulated at bus B and at the endpoint of lateral L , i.e. at

i i

462
 A.M. Tsimtsios et al.

Table 1
Maximum calculated reactance/resistance (in Ω pri) by FDE/RDE of each relay during phase-/ground-fault simulations.
Mode Distance Calculated value DR0,1, DR1,0, DR1,2, DR2,1, DR2,3, DR9,8 DR3,2, DR3,4, DR4,3, DR4,5, DR7,6 DR5,4, DR5,6, DR1,9, DR2,10, DR3,7, DR4,8,
element DR0,10 DR10,0 DR10,9 DR9,10 DR8,9 DR8,7 DR7,8 DR6,7 DR6,5 DR10,2 DR9,1 DR8,4 DR7,3

GC FDEi,j +Xp,max/+Xg,max 2.28/ 1.67/ 2.14/ 1.67/ 1.94/14.96 1.97/ 1.67/ 2.21/ 1.67/15.48 1.67/ 1.67/ 2.14/ 1.67/ 1.67/ 2.21/
13.58 66.73 20.00 47.54 63.26 20.79 44.85 67.33 24.61 20.00 47.54 20.79 44.85
|−X|p,max/|−X|g,max 0/11.40 46.16/ 0/13.48 0.32/ 0.01/7.86 1.15/ 0.05/8.77 0/22.54 0/5.64 0.02/ 0/11.10 0/13.48 0.32/ 0.05/8.77 0/22.54
4024 25.78 35.83 59.46 25.78
+Rp,max/+Rg,max 18.77/ 3001/ 27.49/ 59.67/ 21.78/ 79.45/ 29.43/ 53.35/ 17.00/ 82.59/ 27.66/ 27.49/ 59.67/ 29.43/ 53.35/
59.52 5717 74.71 141.66 78.02 189.06 123.03 126.36 39.24 286.35 64.52 74.71 141.66 123.03 126.36
RDEi,j +Xp,max/+Xg,max – 5.24/ 12.66/ 6.97/ 16.88/ 5.42/ 11.26/ 6.43/ 16.92/ 3.39/ 5.63/ 12.66/ 6.97/ 11.26/ 6.43/
17.03 66.74 26.19 88.98 18.76 61.85 27.02 93.55 19.30 32.11 66.74 26.19 61.85 27.02
|−X|p,max/|−X|g,max – 0/13.14 2.05/ 0.08/ 2.88/38.51 0.33/7.64 0.49/ 0.55/ 1.02/63.52 0/7.33 0.04/ 2.05/ 0.08/ 0.49/ 0.55/
28.23 15.32 24.78 11.83 12.85 28.23 15.32 24.78 11.83

463
+Rp,max/+Rg,max – 28.50/ 98.04/ 43.00/ 130.72/ 33.40/ 87.15/ 46.34/ 136.23/ 26.14/ 43.77/ 98.04/ 43.00/ 87.15/ 46.34/
69.38 165.34 88.00 220.46 89.70 147.00 144.93 353.35 44.12 73.54 165.34 88.00 147.00 144.93

ISL FDEi,j +Xp,max/+Xg,max 1.67/ 1.67/ 1.67/ 1.67/ 4.28/36.24 1.67/ 1.97/9.20 1.67/ 1.67/10.47 1.67/ 1.67/ 1.67/ 1.67/ 1.97/9.20 1.67/
36.29 13.92 31.36 15.64 11.09 13.88 14.06 15.56 31.36 15.64 13.88
|−X|p,max/|−X|g,max 0/25.36 0/7.46 0/22.41 0/9.26 0/28.59 0.11/5.97 0/35.07 0.01/9.00 0/5.46 0.02/ 0/11.10 0/22.41 0/9.26 0/35.07 0.01/9.00
59.46
+Rp,max/+Rg,max 59.89/ 20.93/ 53.51/ 24.84/ 56.74/ 18.79/ 49.03/ 25.13/ 16.96/ 82.59/ 27.70/ 53.51/ 24.84/ 49.03/ 25.13/
117.76 40.39 105.06 48.22 146.80 33.72 184.35 47.11 31.80 286.35 56.73 105.06 48.22 184.35 47.11
RDEi,j +Xp,max/+Xg,max – 12.66/ 5.07/ 11.24/ 3.50/15.40 16.30/ 4.97/ 10.84/ 16.92/ 3.39/ 5.63/ 5.07/ 11.24/ 4.97/ 10.84/
56.53 22.62 49.24 58.73 20.65 13.40 20.11 14.48 23.54 22.62 49.24 20.65 13.40
|−X|p,max/|−X|g,max – 0/27.56 0/11.01 0.11/ 0.34/7.65 0/30.76 0.25/ 0.15/ 1.02/63.52 0/7.14 0.04/ 0/11.01 0.11/ 0.25/ 0.15/
24.57 10.77 38.07 12.85 24.57 10.77 38.07
+Rp,max/+Rg,max – 98.04/ 39.22/ 87.54/ 28.83/ 91.71/ 39.74/ 79.77/ 136.23/ 26.07/ 43.77/ 39.22/ 87.54/ 39.74/ 79.77/
140.23 56.07 124.73 38.51 176.12 54.51 227.02 353.35 36.25 66.01 56.07 124.73 54.51 227.02
Electrical Power and Energy Systems 105 (2019) 454–469
A.M. Tsimtsios et al. Electrical Power and Energy Systems 105 (2019) 454–469

5. Application to a test distribution system
5.1. Test system description
The proposed protection scheme is applied on a 20 kV, 50 Hz, me-
shed overhead distribution system, having exactly the same config-
uration as that of Fig. 2. The system is modelled with DIgSILENT
PowerFactory, based on realistic data provided by the Hellenic Elec-
tricity Distribution Network Operator S.A (HEDNO S.A.). 5-km long 95-
mm2 and 5-km long 16-mm2 ACSR conductor is considered for all the
main line segments Li, j and all the line laterals Li , respectively. Each
lateral is protected by a 30 T-fuse Fi .
The external transmission grid is represented by an equivalent
source with a maximum short-circuit power of 2085.1 MVA at 150 kV.
The system is normally supplied from the transmission grid through a
150/20 kV bulk distribution transformer but is also able to operate as
an island. Two 2-MW-rated PV plants, PV1 and PV2 , operating with unity
power factor, are connected through a step-up transformer to buses B2
and B9 , respectively. The maximum steady-state short-circuit con-
tribution of each PV plant is limited to its nominal load current. Two
synchronous generators, SG1 and SG2 , are also considered connected
through a step-up transformer to buses B4 and B7 , respectively. SGs have
a maximum production of 2 MW (with unity power factor) in the grid-
connected mode, whereas, their active and reactive output is suitably
adjusted to support the islanded operation. The total system load is 9.08
MVA (8.17 MW). A maximum DG penetration level (PL) of 98% is
achieved in the grid-connected mode if all DG units are simultaneously
in operation, whereas, a PL of 103% is required for the islanded op-
eration.
5.2. Study procedure and assumptions
The fault-simulation procedure described in Section 4 is performed
for the test system. That means that phase/ground faults of different
types are simulated at critical positions for setting FDE/RDE/RDEb of
each relay, taking into account fault resistance, DG infeed, and the
mode of system operation, i.e. grid-connected vs. islanded mode. The
following assumptions additionally hold in the present application:
• As for the simulation of phase (LLL and LL) faults, the maximum
fault resistance considered during the simulations is indicatively
taken equal to five times the initial (basic) positive reactance reach
of a distance element, based on the relevant assumption of [23].
Based on Subsection 3.2, the minimum required positive reactance
reach in primary ohms (Ω pri) of all FDEs and RDEs in the test
system (which might be further extended depending on the simu-
lation results) is 2 Ω and 2.53 Ω, respectively (i.e. 120% of the
protected conductor’s reactance). This results in a maximum fault
resistance of 10 Ω and 12.65 Ω, respectively. However, for
simplicity, a uniform maximum fault resistance of 10 Ω is con-
sidered for simulating phase faults in the present application.
• As for the simulation of ground (SLG and LLG) faults, we consider a
maximum fault resistance of 40 Ω, so as to test the applicability of
the proposed protection scheme for greater fault resistance values
(as expected during ground faults). The latter fault resistance value
is considered as a standard value by several utilities in distribution
system protection design studies [50]. Note that the fault resistance
of 40 Ω is indicative, and is here considered during the simulations
as a representative fault resistance example, for the sake of de-
monstrating the setting procedure of the proposed scheme. The
proposed methodology can be similarly applied for higher fault re-
sistance values, without any modification or additional challenge,
on condition that the relay model used is not restrictive concerning
its reach setting range. As mentioned earlier, the latter is up to the
relay designer.
• The two feeders, connecting B0 to B5 and B0 to B6 are identical,
while, the same holds for the interconnection line segments (e.g.
L1,9 , L2,10 etc.). This results in identical settings between symmetrical
distance relays (e.g. DR 0,1 and DR 0,10 , DR1,0 and DR10,0 etc.).
Although this assumption is made to give a clearer illustration of the
results, the proposed methodology can be similarly applied to non-
identical feeders. In any case, the required fault-simulations have
been performed in the whole network, to ensure the validity of the
relay settings.
• The relay model assumed in this study (chosen from the database of
the software used) provides independent phase and ground quad-
rilateral elements, settable as forward, reverse or non-directional.
5.3. Simulation results and resulting relay settings
Table 1 summarizes the maximum values of list a-c in Subsection
4.1, in both the grid-connected (GC) and the islanded (ISL) mode of
system operation, as calculated by the assigned FDEs/RDEs of each pair
of symmetrical relays, after simulating all the phase-/ground-fault cases
indicated in Section 4. As described in Section 4, these cases refer to
different fault types, fault resistance, DG infeed, and mode of system
operation (i.e. grid-connected/islanded). Note that, during the simula-
tions, it was also confirmed that the faults are detected by the relays in
the correct direction. The bolded values in Table 1 correspond to the
maximum values calculated by FDEs/RDEs in each column, i.e. the
maximum values of list a-c calculated by FDEi, j /RDEi, j of each relay
DRi, j , considering all the simulated fault cases.
Based on the latter values, the final relay settings of Table 2 are
extracted. Specifically, for each distance relay, settings 1–3 from the list
of Subsection 4.1 are given for the corresponding phase/ground FDE
(PFDE/GFDE) and phase/ground RDE (PRDE/GRDE). Table 2 does not
include settings for FDE1,0 and FDE10,0 , as they will be individually
determined later. To consider a safety margin ε for certain fault

Table 2
Reactance/resistance reach settings (in Ω pri) and time data of the phase/ground FDE/RDE of each relay.
Distance Settings/Time data DR0,1, DR1,0, DR1,2, DR2,1, DR2,3, DR3,2, DR3,4, DR4,3, DR4,5, DR5,4, DR5,6, DR1,9, DR2,10, DR3,7, DR4,8,
element DR0,10 DR10,0 DR10,9 DR9,10 DR9,8 DR8,9 DR8,7 DR7,8 DR7,6 DR6,7 DR6,5 DR10,2 DR9,1 DR8,4 DR7,3

FDEi,j +XPFDEi,j/+XGFDEi,j 3/37 – 3/32 2/48 5/37 2/64 2/21 3/45 2/16 2/68 2/25 3/32 2/48 2/21 3/45
|−XPFDEi,j|/|−XGFDEi,j| 0/26 – 0/23 1/26 1/29 2/36 1/36 1/23 0/6 1/60 0/12 0/23 1/26 1/36 1/23
+RPFDEi,j/+RGFDEi,j 60/ – 54/ 60/ 57/ 80/ 50/ 54/ 17/40 83/ 28/65 54/ 60/142 50/ 54/
118 106 142 147 190 185 127 287 106 185 127
tFi,j (s) 0.04 – 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04
tC = tFi,j + tCB (s) 0.1 – 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1

RDEi,j +XPRDEi,j/+XGRDEi,j – 13/57 13/67 12/50 17/89 17/59 12/62 11/28 17/94 4/20 6/33 13/67 12/50 12/62 11/28
|−XPRDEi,j|/|−XGRDEi,j| – 0/28 3/29 1/25 3/39 1/31 1/25 1/39 2/64 0/8 1/13 3/29 1/25 1/25 1/39
+RPRDEi,j/+RGRDEi,j – 99/ 99/ 88/ 131/ 92/ 88/ 80/ 137/ 27/45 44/74 99/ 88/125 88/ 80/
141 166 125 221 177 147 228 354 166 147 228
tRi (s) – 0.58 0.58 0.60 0.60 0.63 0.63 0.65 0.65 0.69 0.69 0.58 0.60 0.63 0.65
tC = tRi + tCB (s) – 0.64 0.64 0.66 0.66 0.69 0.69 0.71 0.71 0.75 0.75 0.64 0.66 0.69 0.71

464
A.M. Tsimtsios et al.
Fig. 9. (a) Reactance calculated by DR0,1, (b) Resistance calculated by DR0,1.
detection (see Section 4), the settings of Table 2 result from the cor-
responding bolded values of Table 1, rounded up to the next integer.
Note that a relatively low ε is considered here for illustration purposes,
i.e. in order to keep discernible the correspondence of the bolded values
of Table 1, to the respective settings of Table 2. Of course, a much
greater ε could be normally considered.
Furthermore, the offline fault-simulations that gave the ultimate
relay settings of Table 2 were repeated in the time-domain, in order to
further check the validity of the relay settings. It was found that the
magnitudes of the reactance/resistance values calculated by the relays
(after their response time) were always below the corresponding reach
settings of Table 2, so the relays performed well. Two representative
examples of time-domain relay calculations are given in Fig. 9a and
Fig. 9b, which illustrate the reactance (X) and resistance (R), respec-
tively (in secondary Ω), calculated by DR0,1 during a 40-Ω SLG fault in
front of DR1,0, in the islanded mode of system operation. The fault is
assumed occurring at t = 0, while, a 2-cycle time-window is considered,
equal to the response time of the relay used. This specific fault de-
termined the −XGFDE0,1 and the +RGFDE0,1 setting of GFDE0,1 (shown
with dashed line in Fig. 9a and Fig. 9b respectively). It appears that the
magnitudes of the reactance and resistance calculated by the relay are
slightly (due to the low ε considered previously) lower than the cor-
responding relay settings. Hence, the relay proves efficient against these
marginal fault cases. Just to mention, as part of this study, the relay
calculations were based on a sampling frequency of 1000 Hz, con-
sidering 20 samples/cycle, and a full-cycle Discrete Fourier Transform
(DFT) filter. The current and voltage transformation ratio is 60 and 200,
respectively.
Table 2 also includes the tRi settings of all RDEs, as resulted ac-
cording to Subsection 4.2 and considering a CTI of 0.3 s. An example of
relay-fuse coordination is illustrated in Fig. 10a. Specifically, Fig. 10a
shows the minimum-melting and total-clearing curve of fuse F1 (30T-
fuse), as well as its tTC,max (0.275 s), which results from the less severe
fault simulated at the endpoint of lateral L1 (leading to a short-circuit
current IL1,min of 623 A flowing through F1). In this particular case, this
was a SLG fault at the endpoint of L1, in the grid-connected mode of
system operation, with PL = 0. By setting the time delay tR1 of GARDE1
relays according to (10), i.e. equal to 0.58 s, we ensure that, in case of a
fault in L1, these relays will let F1 clear the fault first.
Moreover, Table 2 gives the tripping (response) time tFi,j (uninten-
tional time delay) of each FDEi,j, as well as the overall fault clearing
time (tC) corresponding to each FDE and RDE (see Subsection 3.2 for tC
explanation), during any fault in its assigned line part. Note that CBs
with an interrupting time (tCB) of 60 ms (3 cycles) are considered,
while, as mentioned previously, the response time of the particular
relay model chosen for the simulation study is 40 ms (2 cycles). It has to
465
Electrical Power and Energy Systems 105 (2019) 454–469
be mentioned that the response time of a relay varies from model to
model and it could be even lower. The time delay tBF of the BF function
is set to 15 cycles in all the relays, to ensure correct BF operation (see
Subsection 3.3).
Concerning settings 4–6 from the list of Subsection 4.1, they are
equally adopted for all FDEs/RDEs, according to the angle-setting phi-
losophy described in Section 4:
• The angle of the right/down zone blinder (θ /θ ) is set equal to 90°.
r d
• The angle of the left zone blinder (θ ) is set equal to zero.
l
Focusing on FDE1,0 and FDE10,0 , a high negative reactance and po-
sitive resistive reach setting is required, as realized from the italicized
values in Table 1. This is particularly due to the weak current flowing
through DR1,0/DR10,0 during faults in L0,1/L0,10, in the grid-connected
mode, with PL = 0. Such great reach values are not settable, based on
the reach setting range of common commercial distance relays. Al-
though, as mentioned before, this could be addressed by the relay de-
signers, in order to take into account the most restrictive case, FDE1,0
and FDE10,0 are replaced by RDEb1,0 and RDEb10,0 , respectively (see
Subsection 3.2), set according to (12). The rest values of Table 1 are
settable in several commercial distance relays.
Table 3 includes the positive reactance reach ( + XRDEb1,0 ), negative
reactance reach magnitude, (|−X |RDEb1,0 ) and positive resistive reach
( + RRDEb1,0 ) of phase and ground RDEb1,0 (PRDEb1,0 and GRDEb1,0 , re-
spectively), set according to (12). Note that +XL0,1 and +RL0,1 of (12)
are equal to 1.67 Ω and 1.075 Ω, respectively. These settings are equally
applied to phase/ground RDEb10,0 , which is symmetrical to RDEb1,0 .
The rest RDEbs’ settings are determined according to Subsection 4.3 and
are listed below:
• The angle of the right/down zone blinder (θ /θ ) is set equal to 90°.
r d
• The angle of the left zone blinder (θ ) is set equal to zero.
l
No intentional time delay is set for RDEb, i.e. it trips after 2 cycles
(response time of the relay).
Finally, Table 4 shows the maximum load current (IL,max ) flowing on
each line segment Li, j , determined through a proper pre-fault load-flow
simulation study (see Subsection 4.2). Specifically, for determining
IL,max values, load flow simulations were conducted in the grid-con-
nected/islanded mode of system operation, for a variety of cases. These
cases resulted from consecutively disconnecting main line segments or
laterals in the examined network. In addition, in each one of the re-
sulting cases during the grid-connected mode of the system, sub-cases
were formed by considering all the possible connections of the DG
units.
A.M. Tsimtsios et al. Electrical Power and Energy Systems 105 (2019) 454–469

Fig. 10. (a) Coordination between fuse F1 and the relays of GARDE1, (b) Phase and ground zone characteristics of DR1,2.

Table 3 regarding several influencing factors and proper relay setting are noted.
Settings of phase/ground reverse blocking elements. This work proposes a distance-based protection scheme for meshed
Distance elements +XRDEb1,0 (Ω pri) |−X|RDEb1,0 (Ω pri) +RRDEb1,0 (Ω pri)
distribution systems, intending to address all the protection issues that
could be encountered in practice.
PRDEb1,0/GRDEb1,0 2/36 2/28 59/117

The impedance setting ZLE of the LE element, enabled for the phase
RDE of both relays installed at the opponent ends of each line segment
Li,j, is also included in Table 4. This setting is calculated based on the
corresponding IL,max, increased by a safety margin of 25%. Note that,
since the same current magnitude IL,max flows in symmetrical line seg-
ments, the corresponding relays have the same ZLE setting. Remember
that FDEs do not require a LE element, since, at least one of the oppo-
nent FDEs in a main line segment does not see the load current in its trip
direction. Also note that in the cases where the resulting ZLE is greater
than the positive resistive reach of the corresponding phase RDE, a LE
element does not have to be ultimately enabled for this RDE.
The representative case of DR1,2 is chosen in Fig. 10b, to illustrate
the resulting phase (solid lines)/ground (dashed lines) zone character-
istic of FDE1,2 and RDE1,2 , and the corresponding LE element.
6. Advantages of the proposed protection scheme and comparison
with other protection principles
6.1. Advantages of the proposed protection scheme compared to other
distance-based schemes
This work intends to address all the issues encountered so far in the
literature, concerning the application of distance protection to dis-
tribution systems with DG (see Section 1). In the following, the main
contributions of this paper, compared to relevant research efforts (as
those cited in Section 1), are summarized.
6.1.1. An efficient distance protection scheme for meshed distribution
systems is proposed
So far, the vast majority of research efforts, dealing with the topic of
distance protection for distribution systems, regard radial network
configurations. There are limited papers applying distance protection to
looped/meshed networks, but, even in these cases, significant issues
6.1.2. All the factors affecting distance relay operation are addressed
To ensure that the applied distance protection scheme is reliable
under realistic system conditions, the concurrent effect of all the factors
influencing distance relay operation in distribution systems (e.g. DG
infeed, fault resistance and K0), must be properly considered [46] (also
see Section 2). So far, a complete solution in this direction has not been
proposed as part of distance protection applications for distribution
systems. Several relevant research efforts intend to deal with the effect
of DG infeed, but they do not address (at least properly) the effect of
fault resistance. The latter is critical, as the nature of its effect is un-
predictable and can lead to intense underreach or overreach phe-
nomena. Especially in case of ground faults, due to the concurrent effect
of K0, the overreach effect of fault resistance can be very intense, so, it
requires special attention [46]. On the other hand, a few proposed
approaches which intend to deal with the fault resistance effect (non-
empirically), either display limited efficiency under the presence of DG
and various system changes [32], or they are not applicable against SLG
faults [33], which, besides they are the most frequent, they are the main
cause of intense overreach phenomena, as mentioned previously.
The proposed methodology takes into account the concurrent
(overall) effect of the aforementioned influencing factors, by con-
ducting a proper simulation study, considering various combinations of
DG-production, fault-type, fault-resistance and system-operation sce-
narios at the same time. Remember that protection sensitivity against
all these factors is achieved by suitably setting (extending) distance
zones, so that the most severe impedance calculation errors, observed
for each relay, are covered. At the same time, protection security is
ensured through the permissive logic applied. The above philosophy
constitutes the main difference of the proposed scheme compared to
other distance protection schemes which intend to subtract impedance
calculation errors from the calculated impedance [30,33]. Specifically,
although it seems a promising solution, such a compensation of im-
pedance calculation errors is difficult to be applied against all fault
types and/or fault conditions. On the other hand, the proposed ap-
proach “allows” the appearance of these errors, and deals with them

Table 4
Maximum load in each segment and load encroachment settings.
Line segment (Li,j) L0,1, L0,10 L1,2, L10,9 L2,3, L9,8 L3,4, L8,7 L4,5, L10,9 L5,6 L1,9, L2,10 L3,7, L4,8

IL,max (A) 279 126 167 70 55 28 126 70

ZLE (Ω pri) 33 73 55 132 168 330 73 132

466
A.M. Tsimtsios et al.
Table 5
Number of examined cases where DOCRs failed to trip.
DOCRi,j DOCR1,0, DOCR10,0 DOCR1,2, DOCR10,9 DOCR2,1, DOCR9,10
Ip,ph (A) 349 158 158
Ip,g (A) 84 38 38
TNOC 8 4 6 12
through a proper relay setting and communication logic.
6.1.3. A fixed setting group is determined for each relay
So far, several distance-relaying applications for distribution sys-
tems, which consider both the grid-connected and the islanded mode of
system operation [27,30,31,35,36,43], have not extensively studied the
suitability of distance relay settings under a variety of system and fault
conditions (e.g. different fault types, different fault resistance values
etc.). Furthermore, in some distance-based applications, multiple set-
ting groups are applied against different system conditions (adaptive
logic) [29,37,43,44].
As part of the proposed methodology, a single setting group is de-
termined for each relay, suitable for all the expected system conditions.
Hence, an adaptive logic is avoided, maintaining protection simplicity.
6.1.4. A solution against weak-infeed conditions is proposed
Research efforts dealing with the application of distance protection
to distribution systems have not provided a detailed solution against the
problem of insufficient distance zone reach sensitivity due to weak-in-
feed conditions so far. This problem is described in Subsection 3.2.2. A
problem of the same nature is noted for a pilot-based directional
overcurrent protection scheme in [12], even for a solid fault. However,
the solution proposed in this case leads to the disconnection of the
whole distribution system.
In this work, the aforementioned issue is dealt with, while, at the
same time, selectivity is maintained; in other words, the fault is cleared
by disconnecting only the faulted line segment (see Subsection 3.2.2).
6.1.5. Coordination with lateral protection is ensured, also providing
backup protection for the laterals and maintaining maximum protection
speed for the main line
It is a fact that many studies applying distance protection to dis-
tribution systems do not consider the coordination of distance relays
with other protection means encountered in distribution systems (as
overcurrent relays or fuses). Nevertheless, actual distribution networks
include laterals protected by separate means (usually fuses). Normally,
main line protection should coordinate with lateral protection and
serve as backup protection for the laterals. Furthermore, in other ap-
plications where coordination of distance zones with lateral protection
is addressed (e.g. [24–26]), the distance zones’ time delay has to be
properly adjusted for this purpose, also affecting protection speed
against main line faults.
As part of the proposed scheme, separate distance elements are
enabled in each relay for the primary protection of the main line (FDE)
and the backup protection of its adjacent lateral (RDE). In this way,
FDEs can be set to operate instantaneously, while, RDEs can be set with
a proper time delay to coordinate with lateral fuses, without affecting
protection speed in the main line.
6.1.6. A solution applicable with existing distance relay technology is
proposed
It is a fact that most distance protection concepts addressed in the
literature apply common distance relay technology. However, as de-
scribed, significant issues are not dealt with. Some studies intend to
include alternative functions [30,31,33] or characteristics [32] in dis-
tance relays, to improve their performance; however, significant issues
are noted in those cases as well.
Electrical Power and Energy Systems 105 (2019) 454–469
DOCR2,3, DOCR9,8 DOCR3,2, DOCR8,9 DOCR4,5, DOCR7,6 DOCR1,9, DOCR10,2 DOCR2,10, DOCR9,1
209 209 69 158 158
50 50 17 38 38
10 2 4 6
The methodology presented in this work uses existing distance relay
technology, so as to enhance the applicability of the proposed scheme.
At the same time, all the relevant issues concerning the application of
distance protection to distribution systems are addressed. Note that the
specifications (e.g. reach setting limitations) of commercially-available
distance relays are also taken into account in this work.
6.2. Comparison with a directional overcurrent protection scheme
A critical advantage of distance relays is that distance zones can be
suitably extended to protect against faults with a considerable fault
resistance, whereas, this is not always possible with directional over-
current relays, due to sensitivity limitations. Assume, for instance, that
each DRi, j in Fig. 2 is replaced by a directional overcurrent relay
DOCRi, j . More specifically, assume that each (forward or reverse) phase
distance element is replaced by a (forward or reverse, respectively)
definite-time phase overcurrent element, which picks-up for currents
larger than 125% of the corresponding line segment’s maximum load
(see Table 4). In addition, assume that each (forward or reverse) ground
distance element is replaced by a (forward or reverse, respectively)
definite-time ground overcurrent element, which picks-up for residual
currents larger than 30% of the corresponding line segment’s maximum
load. After simulating exactly the same faults as those simulated for
setting the distance relays (see Section 4), it was observed that some
DOCRs failed to trip, unlike the corresponding distance relays which
performed well. Table 5 shows the pairs of symmetrical DOCRs which
failed to trip, the total number of examined cases (TNOC) where this
happened for each DOCR pair, and the pick-up current of the phase and
ground element of each DOCR (Ip, ph and Ip, g , respectively). It has to be
noted that, in several cases of Table 5, DOCRs failed to trip even for
solid faults.
6.3. Comparison with a differential protection scheme
It is a fact that differential protection could also be considered as a
reliable solution for meshed distribution systems, providing sensitivity
and security. However, the main drawback of differential schemes is
their inability to inherently provide backup protection [51]. For in-
stance, let us assume that the distance-based scheme protecting each
main line segment in Fig. 2 is replaced by a typical differential scheme.
Although the differential scheme protecting a segment Li,j is expected to
perform well against a variety of faults in Li,j, it cannot simultaneously
serve as primary protection for its adjacent buses Bi, Bj and backup
protection for its adjacent laterals Li, Lj. That means that separate
protection schemes (differential or other) would be required for bus/
lateral protection, which is not so practical. Instead, the proposed dis-
tance-based scheme, protecting each segment Li,j (using FDEs), exploits
the multiple distance elements provided by a common commercial
distance relay, settable in the forward or reverse direction, to protect
adjacent buses and laterals as well (using RDEs), without installing a
separate relay.
Also note that, every protection scheme (even a differential one)
considered for protecting laterals, should coordinate with the corre-
sponding lateral fuse, thus it should be set with a proper time delay.
Within this context, a differential scheme would not differ from the
proposed distance-based scheme in terms of protection speed. In other
words, both these schemes would trip instantaneously in case of a main
467
A.M. Tsimtsios et al.
Table 6
Maximum fault resistance (in Ω) sensed by the distance and the differential scheme during SLG faults at the midpoint of main line segments.
Mode Applied scheme L0,1, L0,10 L1,2, L10,9 L2,3, L9,8
GC Distance 242 400 274
Differential 217 216 216
ISL Distance 290 336 217
Differential 207 207 208
line fault, and with the same time delay in case of a lateral fault.
Furthermore, since communication latency is negligible, it does not
affect protection speed either in the distance-based or in the differential
scheme.
Despite the aforementioned drawback of the differential scheme
concerning bus/lateral protection, it would be interesting to compare
the performance of the proposed distance scheme to that of a differ-
ential one, against faults in main line segments. It is true that the dif-
ferential scheme is expected to perform well against all the main line
faults simulated previously for setting the distance-based scheme, as it
can display adequate sensitivity. Hence, a substantial comparison of
these two schemes could be performed for (ground) faults with high
resistance. For this comparison, actual relay sensitivity limits have been
considered for both the distance and the differential element, based on
the specifications of a typical commercial multifunctional relay of a
well-established manufacturer. Specifically, the ground distance ele-
ment applied has a maximum settable reactive and resistive reach of
500 secondary Ω, and a maximum measured phase and sequence cur-
rent sensitivity of 5%In, where In is the corresponding current trans-
former’s (CT) nominal current. The differential element applied has a
minimum differential pick-up current of 20%In. Note that a current
transformation ratio of 60 is considered for all the distance/differential
relays and a voltage transformation ratio of 200 is considered for all the
distance relays.
Table 6 shows the maximum fault resistance accompanying a SLG
fault at the midpoint of each main line segment of the system of Fig. 2,
which can be sensed by the distance and the differential scheme. Faults
were simulated in both the grid-connected (GC) and the islanded (ISL)
mode of system operation, considering DG. Actually, Table 6 includes
the maximum fault resistance sensed in each case, for which the cor-
responding sensitivity limits of the previous paragraph are not violated.
It is apparent that the distance-based scheme can protect against faults
with higher resistance in the vast majority (90%) of the simulated fault
cases. Note, however, that this percentage might vary depending on the
particular relay model installed (i.e. the sensitivity limits of the relay)
and the fault/system conditions.
7. Conclusion
This paper proposes a distance protection scheme for meshed dis-
tribution systems with DG, aiming to disconnect only the faulted line
part, avoiding unnecessary disconnection of consumers and/or DG
units. Separate distance elements are enabled in each relay for main
line protection and bus/lateral protection, ensuring maximum protec-
tion speed for main line faults and relay-fuse coordination for lateral
faults. Permissive pilot protection allows the extension of the distance
zones, in order to achieve efficient protection under the effect of DG
and fault resistance, while, undesired trips due to overreach are
avoided; hence, the proposed protection scheme guarantees protection
sensitivity and security at the same time. An efficient solution against
weak-infeed conditions, which normally affect relay sensitivity, is also
proposed. The offline relay setting procedure is conducted only once,
and the resulting settings are suitable for both the grid-connected and
the islanded mode of system operation, as well as for all the expected
fault/system conditions, avoiding the use of adaptive logic. Since the
proposed scheme is based on existing distance relay technology and
468
Electrical Power and Energy Systems 105 (2019) 454–469
L3,4, L8,7 L4,5, L7,6 L5,6 L1,9 L2,10 L3,7 L4,8
306 186 452 400 400 306 306
216 216 215 216 216 216 216
292 181 437 336 336 292 292
209 208 208 207 207 209 209
takes into account specifications of commercially-available distance
relays, its applicability is enhanced. The advantages of the designed
distance-based protection scheme, compared to a directional-over-
current-based and a differential-based protection scheme, as well as to
other proposed distance protection schemes, are also addressed.
To sum up, the main contributions of the proposed distance-based
protection scheme are that (i) it constitutes an efficient distance pro-
tection solution for meshed distribution systems, (ii) it deals with the
concurrent effect of all the factors affecting distance relay operation,
(iii) it applies a fixed setting group to each relay, suitable for different
fault/system conditions, (iv) it deals with weak-infeed conditions, (v) it
coordinates with lateral protection means, maintaining maximum
protection speed for the main line, and (vi) it is based on existing dis-
tance relay technology and realistic relay specifications.
Acknowledgements
The authors would like to thank HEDNO S.A. (Xanthi area) and
especially the Director V. Sarris for providing them with all the ne-
cessary data to model the test distribution system as part of this work.
This work was supported by the General Secretariat for Research
and Technology (GSRT)/Hellenic Foundation for Research and
Innovation (HFRI).
References
[1] Bollen MH, Hassan F. Integration of distributed generation in the power system.
New York: Wiley-IEEE Press; 2011.
[2] Celli G, Pilo F, Pisano G, Allegranza V, Cicoria R, Iaria A. Meshed vs. radial MV
distribution network in presence of large amount of DG. In: IEEE PES PSCE, New
York; 2004. p. 709–14.
[3] Gellings C. Estimating the costs and benefits of the smart grid. A preliminary esti-
mate of the investment requirements and the resultant benefits of a fully functioning
smart grid. Electric Power Research Institute, Rep. 1022519, Palo Alto, CA; 2011.
[4] Fairman JR, Zimmerman K, Gregory JW, Niemira JK. International drive distribu-
tion automation and protection. Pullman, WA: SEL Inc.
[5] Lauria S, Codino A, Calone R. Protection system studies for ENEL Distribuzione's MV
loop lines. In: PowerTech, Eindhoven, Netherlands; 2015. p. 1–6.
[6] Alam MN, Das B, Pant V. An interior point method based protection coordination
scheme for directional overcurrent relays in meshed networks. Int J Electric Power
Energy Syst 2016;81:153–64.
[7] Yazdaninejadi A, Nazarpour D, Golshannavaz S. Dual-setting directional over-cur-
rent relays: an optimal coordination in multiple source meshed distribution net-
works. Int J Electric Power Energy Syst 2017;86:163–76.
[8] Saleh KA, Zeineldin HH, Al-Hinai A, El-Saadany EF. Optimal coordination of di-
rectional overcurrent relays using a new time-current-voltage characteristic. IEEE
Trans Power Del 2015;30(2):537–44.
[9] Ojaghi M, Mohammadi V. Use of clustering to reduce the number of different setting
groups for adaptive coordination of overcurrent relays. IEEE Trans Power Del
2018;33(3):1204–12.
[10] Papaspiliotopoulos VA, Korres GN, Kleftakis VA, Hatziargyriou ND. Hardware-in-
the-loop design and optimal setting of adaptive protection schemes for distribution
systems with distributed generation. IEEE Trans Power Del 2017;32(1):393–400.
[11] Liu Z, Su C, Høidalen HK, Chen Z. A multiagent system-based protection and control
scheme for distribution system with distributed-generation integration. IEEE Trans
Power Del 2017;32(1):536–45.
[12] Che L, Khodayar ME, Shahidehpour M. Adaptive protection system for microgrids:
protection practices of a functional microgrid system. IEEE Electrific Mag
2014;2(1):66–80.
[13] Liu X, Shahidehpour M, Li Z, Liu X, Cao Y, Tian W. Protection scheme for loop-
based microgrids. IEEE Trans Smart Grid 2017;8(3):1340–9.
[14] Aghdam TS, Karegar HK, Zeineldin HH. Variable tripping time differential protec-
tion for microgrids considering DG stability. IEEE Trans Smart Grid [in press].
[15] Kar S, Samantaray SR, Zadeh MD. Data-mining model based intelligent differential
microgrid protection scheme. IEEE Syst J 2017;11(2):1161–9.
A.M. Tsimtsios et al.
[16] Li X, Dyśko A, Burt GM. Traveling wave-based protection scheme for inverter-
dominated microgrid using mathematical morphology. IEEE Trans Smart Grid
2014;5(5):2211–8.
[17] Bukhari SBA, Zaman MSU, Haider R, Oh YS, Kim CH. A protection scheme for
microgrid with multiple distributed generations using superimposed reactive en-
ergy. Int J Electric Power Energy Syst 2017;92:156–66.
[18] Zhang Z, Xu B, Crossley P, Li L. Positive-sequence-fault-component-based blocking
pilot protection for closed-loop distribution network with underground cable. Int J
Electric Power Energy Syst 2018;94:57–66.
[19] Gers JM, Holmes EJ. Protection of electricity distribution networks. London, UK:
IET; 2004.
[20] Hooshyar A, Iravani R. Microgrid protection. Proc IEEE 2017;105(7):1332–53.
[21] Chang J, Gara L, Fong P, Kyosev Y. Application of a multifunctional distance pro-
tective IED in a 15KV distribution network. In: 66th Annu Conf Protective Relay
Engineers, College Station, TX; 2013. p. 150–71.
[22] Enayati A, Ortmeyer TH. A novel approach to provide relay coordination in dis-
tribution power systems with multiple reclosers. In: North Amer Power Symp,
Charlotte, NC; 2015. p. 1–6.
[23] Sinclair A, Finney D, Martin D, Sharma P. Distance protection in distribution sys-
tems: how it assists with integrating distributed resources. IEEE Trans Ind Appl
2014;50(3):2186–96.
[24] Nikolaidis VC, Arsenopoulos C, Safigianni AS, Vournas CD. A distance based pro-
tection scheme for distribution systems with distributed generators. In: PSCC,
Genoa, Italy; 2016. p. 1–7.
[25] Tsimtsios AM, Nikolaidis VC. Application of distance protection in mixed overhead-
underground distribution feeders with distributed generation. In: DPSP 2018,
Belfast, UK. p. 1–6.
[26] Pandakov K, Høidalen HK, Marvik JI. Implementation of distance relaying in dis-
tribution network with distributed generation. In: DPSP 2016, Edinburgh, UK.
p. 1–7.
[27] Lin H, Liu C, Guerrero JM, Vásquez JC. Distance protection for microgrids in dis-
tribution system. In: 41st Annu Conf IEEE Ind Electron Soc (IECON 2015),
Yokohama, Japan; 2015. p. 731–6.
[28] Uthitsunthorn D, Kulworawanichpong T. Distance protection of a renewable energy
plant in electric power distribution systems. In: POWERCON, Hangzhou, China;
2010. p. 1–6.
[29] Ma J, Li J, Wang Z. An adaptive distance protection scheme for distribution system
with distributed generation. In: 5th Int Conf Critical Infrastructure, Beijing, China;
2010. p. 1–4.
[30] Biswas S, Centeno V. A communication based infeed correction method for distance
protection in distribution systems. In: North Amer Power Symp, Morgantown, WV;
2017. p. 1–5.
[31] Jin L, Jiang M, Yang G. Fault analysis of microgrid and adaptive distance protection
based on complex wavelet transform. In: PEAC, Shanghai, China; 2014. p. 360–4.
[32] Jecu C, Raison B, Caire R, Alibert P, Deschamps P, Chilard O, Grenard S. Protection
scheme based on non communicating relays deployed on MV distribution grid. In:
PowerTech, Grenoble, France; 2013. p. 1–6.
[33] Pandakov K, Høidalen HK. Distance protection with fault impedance compensation
for distribution network with DG. In: IEEE PES ISGT-Europe, Torino, Italy; 2017.
469
Electrical Power and Energy Systems 105 (2019) 454–469
p. 1–6.
[34] Ziegler G. Numerical distance protection: principles and applications. Erlangen,
Germany: Publicis Publishing; 2011.
[35] Elkhatib M, Ellis A, Biswal M, Brahma S, Ranade S. Protection of renewable-
dominated microgrids: challenges and potential solutions. Sandia National
Laboratories, Rep. SAND2016-11210, Albuquerque, NM; 2016.
[36] Elkhatib M, Ellis A. Communication-assisted impedance-based microgrid protection
scheme. In: IEEE PES General Meeting, Chicago, IL; 2017. p. 1–5.
[37] Lin H, Guerrero JM, Vásquez JC, Liu C. Adaptive distance protection for microgrids.
In: 41st Annu Conf IEEE Ind Electron Soc (IECON 2015), Yokohama, Japan; 2015.
pp. 725–30.
[38] El-Arroudi K, Joós G. Performance of interconnection protection based on distance
relaying for wind power distributed generation. IEEE Trans Power Del
2018;33(2):620–9.
[39] Huang W, Nengling T, Zheng X, Fan C, Yang X, Kirby BJ. An impedance protection
scheme for feeders of active distribution networks. IEEE Trans Power Del
2014;29(4):1591–602.
[40] Xu M, Zou G, Xu C, Sun W, Mu S. Positive sequence differential impedance pro-
tection for distribution network with IBDGs. In: POWERCON, Wollongong,
Australia; 2016. p. 1–5.
[41] Biller M, Jaeger J, Mladenovic I, Schacherer C, Wolter D, Stoetzel M. Protection
systems in distribution grids with variable short-circuit conditions. In: DPSP 2016,
Edinburgh, UK. p. 1–5.
[42] Chilvers I, Jenkins N, Crossley P. Distance relaying of 11kV circuits to increase the
installed capacity of distributed generation. IEE Proc – Gen Transm Distrib
2005;152(1):40–6.
[43] El Naily N, Saad SM, Elsayed RE, Aomura SA, Mohamed FA. Planning & application
of distance relays coordination for IEC microgrid considering intermediate in-feed
factor. In: 9th Int Renewable Energy Congr, Hammamet, Tunisia; 2018. p. 1–6.
[44] Liu Z, Høidalen HK, Saha MM. An intelligent coordinated protection and control
strategy for distribution network with wind generation integration. CSEE J Power
Energy Syst 2016;2(4):23–30.
[45] Tsimtsios AM, Nikolaidis VC. Setting zero-sequence compensation factor in distance
relays protecting distribution systems. IEEE Trans Power Del 2018;33(3):1236–46.
[46] Nikolaidis VC, Tsimtsios AM, Safigianni AS. Investigating particularities of infeed
and fault resistance effect on distance relays protecting radial distribution feeders
with DG. IEEE Access 2018;6:11301–12.
[47] Kasztenny B, Cambell B, Mazereeuw J. Phase selection for single-pole
tripping–Weak infeed conditions and cross-country faults. In: 27th Annu Western
Protective Relay Conf, Spokane, WA; 2000. p. 1–19.
[48] Costello D, Zimmerman K. Determining the faulted phase. In: 63rd Annu Conf
Protective Relay Engineers, College Station, TX; 2010. p. 1–20.
[49] Xue Y, Thakhar M, Theron JC, Erwin DP. Review of the breaker failure protection
practices in utilities. In: 65th Annu Conf Protective Relay Engineers, College
Station, TX; 2012. p. 260–8.
[50] Grigsby LL. Electric power generation, transmission and distribution. Boca Raton,
FL: CRC Press; 2012.
[51] Ziegler G. Numerical differential protection: principles and applications. Erlangen,
Germany: Publicis Publishing; 2012.