This article has been accepted for inclusion in a future issue of this journal.

Content is final as presented, with the exception of pagination.

IEEE TRANSACTIONS ON SMART GRID 1

A Communication-Assisted Overcurrent Protection

Scheme for Radial Distribution Systems With
Distributed Generation
Vassilis C. Nikolaidis, Member, IEEE, Evangelos Papanikolaou, and Anastasia S. Safigianni, Senior Member, IEEE

AbstractConventional overcurrent protection schemes for elements have been proposed in [3], but this method can-
radial distribution systems usually attempt to coordinate not check the coordination between protective relays after
a recloser at the beginning of the feeder with the fuses on the connecting each DG in the network. Coordination between
laterals. The integration of distributed generation in distribu-
tion systems leads to problems related to protection coordination reclosers and fuses has been obtained in [4] with the use of
that are difficult to be solved by applying conventional protection synchronized measurements and off-line design calculations.
techniques. This paper proposes an efficient communication- Funmilayo and Butler-Purry [5] proposed to replace the fuses
based protection scheme that implements common directional on the laterals where DG is connected with numerical reclosers
overcurrent relays instead of reclosers at the line, assisted and relays in order to maintain coordination between the over-
by intertripping and blocking transfer functions. The pro-
posed protection strategy guarantees selectivity regardless of current protection devices. This methodology has been tested
whether the generating units are connected to the network in two template distribution systems and it seems to work ade-
or not, and can be designed retaining either the fuse-blowing quately, despite the fact that some fuse fatigue issues could not
or fuse-saving philosophy. Meaningful conclusions are derived be avoided.
from the application of the scheme on a test distribution Directional overcurrent relays are mainly used in ring-type
system.
distribution systems. However, directional overcurrent protec-
Index TermsDirectional elements, distributed genera- tion was introduced to radial distribution systems [6], [7] due
tion (DG), distribution systems, overcurrent protection, protec- to the massive DG integration experienced the last years.
tion coordination, signal transfer
Directional overcurrent relays with communication capabilities
are used in a radial network in [8] to minimize the number of
disconnected DG units in case of faults. Reclosing operations
I. I NTRODUCTION
are also performed to restore the system.
CONVENTIONAL overcurrent protection scheme
A designed for radial distribution lines is usually based
on the use of a recloser, at the beginning of the feeder,
Inherent fault direction discrimination and fault location
is obtained if distance relays are applied in distribution
networks [9], [10]. Dedicated fault location algorithms based
which is coordinated with the downstream protection means on neural networks [11], [12] have been recently proposed
(overcurrent relays and/or sectionalizers) on the main line for the design of protection schemes in distribution systems
and with the fuses on the laterals [1]. It is also common to with DG. However, the speed and the complexity of those
apply a reclosing circuit breaker (CB) at the beginning of the methods make them unfavorable in terms of protection needs.
feeder with one or two reclosers at the midpoint and the end There are also current-based algorithms proposed for the
of the feeder. design of protection schemes in radial distribution systems
It is a challenging task to coordinate the recloser(s) in with DG. The calculation of the steady state fault current
a radial distribution line with fuses on the laterals when magnitudes from steady state network equivalent reduction is
distributed generation (DG) units are present at the line. proposed in [13], while the current phase angle comparison is
Microprocessor-based reclosers with adaptive capabilities have used in [14] as a criterion for a pilot relaying scheme.
been proposed in [2] for achieving protection coordination in The differential protection principle has been successfully
such a distribution network. This scheme applies effectively, tested as a unit protection scheme in microgrids [15][17].
but it suggests that the DG unit will be disconnected before In [18], the same principle has been tested in modern urban
the first reclosing operation. On the other hand, nonadaptive distribution systems taking advantage from the IEC-61850
microprocessor-based reclosers in conjunction with directional Standard, while Karady et al. [19] deployed a pilot-based
differential protection scheme on the IEEE 34-bus test dis-
Manuscript received July 31, 2014; revised November 13, 2014 and tribution system.
January 17, 2015; accepted March 2, 2015. Paper no. TSG-00773-2014.
The authors are with the Department of Electrical and Computer This paper proposes an efficient communication-based over-
Engineering, Democritus University of Thrace, Xanthi 67100, Greece (e-mail: current protection scheme for distribution lines with DG.
[email protected]). Section II presents a detailed description of this scheme.
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. A comparison with different protection schemes is included
Digital Object Identifier 10.1109/TSG.2015.2411216 in the same section, while the limitations of the method and
1949-3053 c 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE TRANSACTIONS ON SMART GRID

Fig. 1. Primary protection zones.

the economic feasibility of the required investment are also the polarizing voltage [20] for all the possible faults magni-
discussed. Section III presents the results derived from the tudes and locations along the feeder. The proposed protection
application of the protection scheme on a test distribution scheme indicates that every relay Rjk should be set to clear
system, while Section IV summarizes the derived conclusion. faults occurring in the forward direction within their primary
protection zone. In other words, every overcurrent directional
II. P ROPOSED OVERCURRENT P ROTECTION S CHEME relay is set to react only for faults flowing into their primary
protected zone.
This paper proposes a simple overcurrent protection scheme
Overlapping is applied when defining the primary protection
that is based on the use of numerical directional overcurrent
zones of the relays. In particular, the primary protection zone
relays for the protection of a radial distribution line with DG
of the relay Rjk is formed in such a way that it covers the
units instead of relying primarily to a recloser at the begin-
line segment Sjk plus an additional length of the nearby line
ning of the line. Any modern intelligent electronic device
section Sj+1,k+1 in the relays forward direction. For example,
having common communication capabilities can be used for
the relay R23 is set to cover the whole line segment S23 plus
the implementation of this scheme. The proposed protection
a% of the length of the subsequent section S34 . On the other
scheme can be designed retaining either the fuse-blowing (FB)
hand, the opposite relay R32 is set to cover the whole line
or fuse-saving (FS) philosophy as if a common reclosing
segment S23 plus a% of the length of the subsequent section
scheme was in operation.
S12 in its forward direction. Of course, the primary protection
The following assumptions have been made in the analysis.
zone of the relay overlaps that of the fuse in the adjacent
1) Only synchronous DG units have been considered.
lateral. The overlapping protection zones in both directions
2) The DG units are connected to the main trunk.
are shown with dashed lines in Fig. 1.
3) The DG units operate in a constant power factor mode.
Note that at the marginal positions (e.g., at buses B1 and B6)
the overlapping principle cannot be applied as straightforward
A. Design Philosophy as it is applied for the line segments. Regarding the relay R21
The design logic of the protection scheme proposes direc- the maximum overreaching percentage b% is determined from
tional overcurrent relays to be applied at both ends of every the need to form a protection zone that overlaps the primary
line section, as shown in Fig. 1. Every relay in Fig. 1 is repre- differential protection zone of the transformer and that of the
sented by the symbol R having subscripts which indicate the bus B1. Similarly, the maximum overreaching percentage e%
forward fault direction for the particular relay. In other words, of the relay R56 is determined from the need to form a pro-
for the directional overcurrent relay Rjk a forward fault corre- tection zone that simply overlaps the primary protection zone
sponds to a short-circuit current that flows from buses j to k. of the fuse F6.
It is evident that current and voltage measurements are needed Each relay has to be set with two definite-time (DT) stages
from any line segment terminals as inputs to the directional for its phase and two DT stages for its earth overcurrent ele-
element of the relays. Therefore, individual current and voltage ment. The first-stage DT setting is used for sensing forward
transformers are needed to be installed at those places. phase/earth faults occurring everywhere within the primary
In order to set the directional element accurately, a design protection zone of the relay. A time-delayed, second-stage DT
study should be performed to define the expected range of setting is applied as a backup function in case the primary
angles formed between the measured short-circuit currents and protection function of the downstream relay fails to operate.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NIKOLAIDIS et al.: COMMUNICATION-ASSISTED OVERCURRENT PROTECTION SCHEME FOR RADIAL DISTRIBUTION SYSTEMS 3

Fig. 2. Two-step zone overlapping scheme and communication path.

Thus, if a relay fails to operate for a phase/earth fault within Such a communication path, shown with dashed and dotted
its primary zone, the upstream relay looking into the same lines in the one-line diagram in Fig. 2, could be an Ethernet
direction is expected to operate after a sufficient coordination network that interconnects the relays through single-mode
time interval (CTI). fiber optic cables and switches [21].
This two-step zone overlapping scheme is shown in Fig. 2. Suppose now that a fault occurs within the primary protec-
Every differently colored line in the diagrams indicates the tion zone of the relay R34 . Without any DG present in the line,
two DT stages of the phase/earth overcurrent element of the this fault will simultaneously be sensed from all the upstream
relays. The fastest DT setting corresponds to a tripping time relays looking to the forward direction; that are the relays
equal to tRjk,1 or tRjk,2 , expected for any fault occurring within R12 , R23 , and R34 . Selectivity requires that only the shortest
the primary protection zone of the relay Rjk (obviously tRjk,1 section of the power system should be disconnected due to
and tRjk,2 refer to the first-stage tripping time of relays looking a fault. Thus, only the relay R34 must trip its CB leaving the
to opposite directions and in general they could be set equal). upstream distribution system in operation.
The slower DT setting corresponds to a constant tripping time The backup protection zone, meaning the second stage of
(t3 for forward and t4 for reverse relays, respectively), that is the relay R12 , marginally reaches the faulted section S34 ,
coordinated with an adequate CTIRjk in relevance to the first- whereas R23 second stage reach does clearly cover the whole
stage time delay, expected substantially for any fault occurring section S34 . Under specific circumstances, those relays could
within the primary zone of the neighboring relay. trip both, in order to clear the assumed fault. Hence, it is a mat-
The relay R65 is set with a second, time-delayed DT element ter to prevent the tripping of the relays R12 and R23 in order
(shown with dashed line in Fig. 2) for its own primary pro- to preserve selectivity. If the fault is located in a distance that
tection zone because none neighboring relay can backup R65 is larger than the a% length of the line section S34 , as mea-
for faults in its primary zone. The same is assumed for the sured from bus B3, the relay R12 cannot trip. As for the relay
relay R12 , although the overcurrent element of the transformer R23 , the CTI between the relays R34 first-stage time-delay and
or bus differential relay could be used as a backup protection. the relays R23 second-stage time-delay, provides the relay R34
The overlapping logic described earlier is necessary in order enough time to trip first. There is only a selectivity issue if the
to ensure that any relay Rjk will operate for a fault occur- fault occurs within the overlapping part of the primary zones
ring everywhere inside the line segment Sjk , even at its limits. of the relays R23 and R34 . In that case both relays would trip
Since fault currents have similar magnitudes when appearing and selectivity would be lost. The backup relay R12 would not
in the proximity of a protection zone limit, an underreach- trip since the relays R23 and R34 would clear the fault.
ing scheme would negatively affect the dependability of the To prevent this problem, the next rule is applied to all the
proposed method for those faults. On the other hand, the relays looking to the same direction.
security of the proposed protection scheme is guaranteed 1) Whenever the first-stage phase/earth overcurrent element
by configuring a communication path between the relays. of a relay picks up, a blocking signal is sent to the
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE TRANSACTIONS ON SMART GRID

first-stage phase/earth element of all the upstream relays. for islanding conditions), irrelative if DG is present or not on
The second-stage overcurrent function of the upstream the line. The latter will be further addressed in the following
relay stays unaffected from this blocking signal. sections.
2) At the time the relay trip its CB, an intertripping signal
to the CB at the opposite side of the same line section
will be sent without any intentional delay to completely B. Setting the Relays
disconnect the faulted line section. A detailed fault analysis is needed in order to set appropri-
The intertripping of the opposite CBs is decided in order ately the phase and earth overcurrent elements of the relays
to speed up the fault clearing. Otherwise, the tripping time at every line section. This procedure includes the calculation
determined from the DT characteristic of the opposite relays of the minimum and maximum short-circuit currents expected
should pass for the line section to be totally disconnected. In to be seen from every relay of interest for all possible phase
the grid-connected operation mode, the opposite relays could and earth fault conditions. To be more specific, the simulation
actually be neglected because of the intertripping, but they are of all common short-circuit types (three-phase, double-phase,
set as a local backup function in case the intertripping fails. double-phase-ground, and single-phase), occurring within the
However, the opposite (reverse) relays are absolutely necessary primary and secondary protection zone of every relay (that
if the distribution line operates in the island-mode. Assume, for is in their forward direction), has to be performed considering
example, that the interconnection of the line with the transmis- minimum and maximum short-circuit conditions of the sources
sion system is lost and that a DG unit is operating, connected (equivalent source and connected DG units), any possible
at bus B6. In that case, only the reverse relays can clear faults combination of DG connection locations and fault resistance.
occurring on the line. To explain this procedure a little more, let us consider the
The time needed for the blocking signal to transfer between relay R34 . Initially, short-circuit currents have to be calculated
the relays before any of them trips is of great importance for and stored for faults occurring within its primary protection
setting the relays appropriately. In fact, the total time delay for zone, thus meaning within a distance range from just close-in
a signal to be transmitted over a 10 km long, 100 Mb/s fiber to the relay R34 up to a% of the next line segment S45 . All
cable is of the order of 1540 ms [22], including the wire line possible fault and network conditions described earlier should
latency, the switches latency and the processing time of sub- be taken into account in this analysis.
cycle relays. Note that this delay refers to the worst network The maximum (Ipf,R34 max ) and minimum (I min ) of all
pf,R34
topology in terms of communications delays which is the cas- the phase fault currents calculated following the described
caded or ring type one, where the signals should transfer over methodology define the range of short-circuit current mag-
the 10 km line section length in a cascaded manner. For any nitudes that can be sensed from the phase element of the
other Ethernet network topology, the transfer delay is signifi- relay R34 within its primary zone, under all possible net-
cantly lower. Thus, selectivity is not affected since the relays max and I min
work and fault conditions. Ief,R34 ef,R34 are, respectively,
are set with a time delay that is larger than the transfer delay. the maximum and minimum short-circuit currents for earth
To sum up, if the Ethernet network is specified adequately in faults occurring within the primary protection zone of the relay
terms of architecture, redundancy, availability, performance, R34 . As an example, the largest phase fault current within the
and cost [21], reliable operation is expected meaning that all primary zone of the relay R34 has been calculated for a three-
packets will be transferred between the relays (through the phase fault close-in to the relay in case that two DG units
switches) without any problem. Therefore, false or undesired are connected on bus B3, while the smallest one for a single-
blocking/tripping signals are not expected, except for the case phase fault occurring at the a% distance of the line segment
where a physical cut occurs. S45 when two DG units are connected on the buses B1 and B4.
Assume again that a fault occurs within the primary pro- Obviously, if one knows the minimum short-circuit magni-
tection zone of the relay R34 . Immediately after the relay R34 tudes expected to be seen from any relay within its primary
senses the fault it will send a blocking signal to the relays zone, then he can directly determine the first-stage (Ipu1 ) and
R23 and R12 first-stage element. At the same time, it will send second-stage (Ipu2 ) overcurrent setting (in primary A) of the
a tripping signal to the opposite CB. Hence, the faulted line phase/earth overcurrent element, respectively, of the relay Rjk .
section S34 will be disconnected, while the fastest element of This is graphically shown in Fig. 3, where the two DT curves
all the upstream relays is blocked. are illustrated without discriminating between phase and earth
To summarize, if a fault occurs everywhere within a pri- fault elements of the relay Rjk . Fig. 4 depicts the overall
mary protection zone, the assigned relay will pick-up and at protection and communication logic of the relays.
the same time it will sent a blocking signal to the first-stage It is evident that the described fault analysis is an off-line
overcurrent element of all the upstream relays looking to the procedure, since one should beforehand know the rated capac-
same direction. When the time delay of the assigned relay ity of the DG units and their location, the generator step-up
expires, it will trip its CB and simultaneously a second trip- transformer characteristics, and the minimum and maximum
ping signal will be transferred to the CB at the opposite side penetration level. However, once these data are known there
of the line section. If the assigned relay fails to assert a trip- is no need to collect online information about the status of the
ping order, the backup relay operates after a sufficient CTI DG units (connected/disconnected, connection points, etc.).
to clear the fault. Note that this protection scheme operates If there appears a large difference in the short-circuit cur-
effectively for phase and earth faults in both directions (even rents between the grid-connected and the islanded network
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NIKOLAIDIS et al.: COMMUNICATION-ASSISTED OVERCURRENT PROTECTION SCHEME FOR RADIAL DISTRIBUTION SYSTEMS 5

D. Economic Evaluation
The protection equipment requires a considerable invest-
ment which depends on the desired reliability level of the
protection system. Even if the simplest level of reliability
is assumed (i.e., without any redundancy in the protec-
tion devices), installing relays and CTs/VTs on every line
section definitely increases the implementation cost of the
protection system. If this cost is treated as a network rein-
forcement expense charged only to the distribution system
Fig. 3. Overcurrent phase/earth elements setting.
operator (DSO), then this investment will certainly prove
economically infeasible.
However, the required investment should be evaluated by
also taking into account the operational and maintenance bene-
fits expected to be gained from the DSO and the DG producers
due to the protection scheme operation [26]. For example, the
cost of electricity supply interruptions, the cost of increased
network losses, the outage cost of DG units, as well as power
quality or contract/regulatory penalties are critical measures
for the DSO and the DG producers and should be included
in the economical evaluation of the protection scheme. Note
that all those costs are closely related with the operation of
the protection system [27], [28].
On the other hand, policy makers continue to motivate
Fig. 4. Internal relay protection and communication logic.
the stakeholders for DG investments in distribution systems.
operation, two different setting groups should be available Nowadays, the common practice for many DSOs is to demand
to the relays. The relay will switch to the appropriate set- the immediate disconnection of the DG units in case of a fault.
ting group based on a signal received from an islanding As long as the number of the producers continues to increase,
detection relay or on a substation breaker status signal [23]. this will not be acceptable so easily, because the producers will
Note, that in this case, two different time-overcurrent settings not agree to invest if they are forced to lose their production
should be defined for each phase/earth element: 1) for the due to any single fault on the line.
grid-connected; and 2) for the islanded network operation. The communications infrastructure requires a relatively high
investment as well, which is determined from the desired level
of speed and redundancy. Note, however, that a communication
C. Coordination Issues
network is crucial in distribution networks not only for pro-
The time settings of the phase and earth overcurrent ele- tection reasons but also for implementing various automation
ments must be selected in such a way that coordination operations. Distribution automation may significantly reduce
between the relays and between the relays and the fuses on the the operating costs in a distribution system [29].
laterals is ensured. For example, in a FB scheme, a sufficient It is evident that, for all the above mentioned reasons, a cost-
first-stage phase/earth DT delay tRjk,1 and tRjk,2 is needed in benefit study [30] must be conducted prior to the final decision
order to provide the fuses enough time to blow first. Thus, concerning the investment. A compromise between the desired
if any fault occurs on a lateral, the fuse will blow prior to cost and reliability of the overall protection scheme is neces-
any relay operation on the main trunk. On the other hand, in sary in that case. Within this context, we conceptualize that
an FS scheme, the relays are supposed to operate faster that the required investment could be shared to both beneficial par-
the fuses. In this case, a faster DT delay or an instantaneous ties (DSO and DG producers) in order to compensate for the
setting must be used for the first-stage phase/earth overcurrent obtained benefits: 1) the increased reliability of the distribu-
element. Accordingly, the time delay of the relays in the high- tion system and 2) the increased energy production for the
voltage to medium-voltage substation should be modified for producers by minimizing the possibility of unwanted tripping
coordination, if needed. of the generating units. This could be legislated as an ancillary
Usually, every DG unit connected to a distribution network service, imposed proportionally to both parties.
is protected by a great variety of protective relays [24]. Among
them, the voltage/frequency, the overcurrent, and the intercon-
nection relay seem to play the most important role regarding E. Limitations of the MethodAdvantages
coordination issues with the line protection. Based on inter- and Disadvantages
national standards [25] and the present practice, these devices In this paper, we considered the radial line configuration
are intended to immediately disconnect the DG units when because this is the most commonly encountered distribution
a fault appears. In the island-mode, the network is not oper- network configuration and because it is simple enough for
able without DG units, while in the grid-mode the scenarios illustrating the proposed method. In a ring-type network, as
with different penetration levels have been investigated. that shown in Fig. 5, the same protection scheme can be
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE TRANSACTIONS ON SMART GRID

TABLE I
R ECLOSER T IME -OVERCURRENT S ETTINGS

Fig. 5. Meshed distribution network configuration.

applied without any modification, except that the computa- exchange signals, which is a requirement similar to that of
tional effort for the off-line studies increases. However, in the protection scheme proposed in this paper. Thus, the cost
a meshed network (shown in Fig. 5 if the dashed lines are of a differential protection scheme is comparable to that of
connected), the coordination of the relays becomes a very com- the directional overcurrent scheme since the additional cost of
plicated task since a large combination of off-line studies is installing the VTs should be compared with the cost of adding
needed. In some cases it may be hard to find unique relay set- one more differential relay in each section.
tings, appropriate for all possible network topologies formed
by the different circuit loops and DG connection points. For III. C ASE S TUDY
those cases, it may be considered impractical to apply the pro- A. Test System Description
posed protection scheme. It is evident that, if permitted from
The proposed overcurrent protection method has been tested
the operational planning point of view, a radial network recon-
on the distribution system of Fig. 1, which consists of one
figuration would certainly simplify the operation practices and
radial 20 kV, 50 Hz, and 50 km long overhead line, equally
would improve the performance of the protection scheme.
divided in five segments; a 95 mm2 aluminium conductor steel
Obviously, the size and complexity of the distribution net-
reinforced (ACSR) conductor is used on segments S12 , S23 ,
work affects the computational effort required for the off-line
and S34 , while a 50 and 35 mm2 ACSR conductor is used
computations. However, this is a task that must be performed
on segments S45 and S56 respectively. The total line load
only once, that is after a DG unit is going to be connected to
is 7.5 MW and 1.48 MVAR. The load L1, added to the
or disconnected from the network, which does not happen very
bus B1, represents the total power consumed on an identical
often. On the other hand, planning studies restrict the DG pen-
feeder supplied from the same substation. The transmission
etration to a certain level and the same is true for the number
grid is represented by an equivalent source, having a maxi-
of the possible connection points. Thus, even in the most com-
mum and minimum short-circuit power of 435 and 250 MVA,
plex distribution networks, the number of DG units that can be
respectively. A conventional round-rotor synchronous machine
connected to the line is actually limited and not all buses can
with the nominal data 4.08 MW, 10.5 kV, and 50 Hz,
potentially facilitate a DG unit. Other restraints imposed due
cos n = 0.85 has been assumed as a DG unit. Note, how-
to financial and environmental reasons (including the energy
ever, that the DG units operate with a unity power factor in
resources location if RES are considered instead of conven-
the grid-connected network operation, while in the islanded
tional DG generators) minimize further the possible locations
network one of the DG units controls the voltage. A 25 MVA,
for DG connection.
20 kV (YN)/10.5 kV (D), and 50 Hz step-up transformer has
One could think to apply differential relays instead of over-
been used in all the DG units. Standard models available in
current ones to implement the protection scheme. Although
Digsilent Power Factory 15 have been used for representing
a differential protection scheme is possible and reliable in gen-
the described system.
eral, it does not simplify the implementation and maintenance
of the protection system in distribution systems because the
differential function must be properly set in too many line B. RecloserFuse Coordination
sections. On the other hand, overcurrent protection is always The main feeder is originally protected by a reclosing relay
required as a backup scheme. Hence, the overcurrent ele- SEL-351R, located at the beginning of the line. This recloser
ments of the differential relays should additionally be set for performs one fast and two delayed operations with a reclos-
the distribution system to be protected effectively. Moreover, ing cycle (open intervals) of 0.5510 s. The settings of the
some well-known problems related with the differential protec- phase and earth overcurrent elements of the recloser are shown
tion principle (CTs saturation, line charging currents, unequal in Table I. There is no need to supervise the fault direc-
burden, inrush currents, harmonic content, etc.) are not so tion because originally no DG unit is connected to the line.
pronounced with overcurrent relays. Hence, only the fault contribution from the external grid is
Regarding the implementation costs, if long lines are con- expected. The line laterals are protected by expulsion fuses
sidered, applying differential protection to every line section having a nominal rating of 25 A.
means that two differential relays are needed, one in every Fig. 6 shows the coordination between the recloser
line section terminal. Then, a dedicated and fast communi- SEL-351R and one of that fuses with the help of a
cation link is required for allowing the differential relays to timeovercurrent plot. The green and blue curves correspond
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NIKOLAIDIS et al.: COMMUNICATION-ASSISTED OVERCURRENT PROTECTION SCHEME FOR RADIAL DISTRIBUTION SYSTEMS 7

Fig. 6. Coordination graphs for the recloserfuse scheme (without DG).

Fig. 8. Nuisance feeder tripping when DG is present.

Table II summarizes the minimum calculated phase (I )

and residual earth (3Io ) currents sensed from the relays for
faults at the remotest end of their primary protection zone
in the grid-connected and in the islanded network. All pos-
sible DG connection points have been considered. Due to
space limitations the results for the relays looking to the
opposite direction (from the line to the external grid) are not
included here.
The minimum of all the remotest phase (respectively, resid-
ual earth) fault currents seen by the relay Rjk determines the
overcurrent setting Ipu1 of its first-stage (fastest) DT element,
while the minimum of all the remotest phase (respectively,
Fig. 7. Overcurrent protection blinding when DG is present. residual earth) fault currents seen by the downstream relay
Rj+1,k+1 determines the pickup current setting Ipu2 of the
second-stage (slowest) DT element of the relay Rjk . Those
to the minimum melting (MM) and total clearing (TC) char- values have been highlighted in Table II for every relay of
acteristics of the fuse, respectively. An FS logic is followed. interest.
Without DG on the line, the coordination between the Table III concentrates the two-step overcurrent settings
recloser and the fuses is guaranteed for any fault occurring determined for every relay following the above mentioned
everywhere on the line. However, coordination is lost when procedure. Due to inaccuracies expected from the current
DG is present. For example, suppose that one DG unit is con- transformers, these settings have been derived by reducing the
nected to the bus B4, as shown in Fig. 2. Fig. 7 depicts that highlighted values of Table II by 10%. Depending on the relay
under minimum short-circuit conditions in the transmission specifications, the closest available settings to that of Table II
grid, the recloser (both its phase and earth overcurrent element) should actually be selected.
cannot even sense a single-phase-ground fault occurring on bus Note that, since intertripping is applied between the over-
B6. Indeed, on a time-overcurrent plot there is no intersec- current relays installed at the opposite ends of the same line
tion of the vertical fault current lines with any of the recloser section, the relays looking to the grid (Rjk , j > k) actually have
curves. On the other hand, the recloser trips undesirably for not to be set in the grid-connected mode. Those relays will be
a three-phase fault at the beginning of the neighboring feeder intertripped from the forward relays (Rjk , k > j). Hence, the
(Fig. 8) if maximum fault conditions exist in the transmission settings shown with parentheses in Table III can be considered
system. as redundant in case the intertripping fails.
Next, the appropriate time delays tRjk1 , tRjk,2 , t2 , and t4
should be determined to ensure coordination between the
C. Application of the Proposed Protection Scheme relays and the fuses. For a FB scheme, the maximum TC
As described in the previous sections, the proposed protec- time tTCmax of every fuse F (i = 2, . . . , 6) must be calculated
Fi i
tion scheme suggests applying directional overcurrent relays at initially. Then, in every related relay-fuse pair, the first-stage
both ends of every line section instead of using the recloser at tripping time of the relay must be set higher than the sum of
the beginning of the line. The total active load is 7.5 MW, so the maximum TC time of that fuse and the minimum required
a maximum of two DG units can simultaneously be connected CTI (CTImin ). The latter is taken 0.3 s. The constant DT delays
to the line if a penetration level close to 100% is desired. t2 and t4 are determined by adding the CTImin to the maximum
Taking this constraint into account, the short-circuit analysis max and tmax , respectively).
of all the first-stage time delays (tRjk,1 Rjk,2
described in Section II-B has been conducted. In other words, (1) must be satisfied. Of course, coordination
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE TRANSACTIONS ON SMART GRID

TABLE II
R EMOTEST P HASE (I ) AND R ESIDUAL E ARTH (3Io ) S HORT-C IRCUIT C URRENTS S ENSED F ROM THE R ELAYS IN T HEIR P RIMARY Z ONE

TABLE III
T IME -OVERCURRENT S ETTINGS

between the relays on the main line must also be ensured derived from the simulations that, in the grid-connected net-
work, the maximum blowing time of the fuses is experienced
tRjk,1 = tTC
max
Fi
+ CTImin when minimum short-circuit conditions hold on the transmis-
tRjk,2 = tTC
max
Fi
+ CTImin sion system and none DG unit is in operation. In the islanded
network, the maximum TC time is obtained for one DG unit
t2 = t max
+ CTImin
Rjk,1 connected to the line.
t4 = tmax + CTImin . (1) Table III includes the time delay settings obtained from (1).
Rjk,2
The timeovercurrent plots for the grid-connected and the
The maximum TC time of a fuse is determined by simulat- islanded mode are shown in Figs. 9 and 10, respectively.
ing a number of possible short-circuits at the remotest end of These figures refer to the forward relays, while only the
the fuse laterals under different network conditions. It has been TC characteristic of the T25 A fuse is shown. Coordination
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

NIKOLAIDIS et al.: COMMUNICATION-ASSISTED OVERCURRENT PROTECTION SCHEME FOR RADIAL DISTRIBUTION SYSTEMS 9

Fig. 9. Relays-fuse coordination for grid-mode operation (FB scheme). Fig. 11. Relays-fuse coordination for grid-mode operation (FS scheme).

Fig. 12. Relaysrelays coordination for grid-connected operation.

Fig. 10. Relays-fuse coordination for islanded operation (FB scheme).

between the relays and between the relays and the fuses is DG present in the distribution line this can be consid-
guaranteed for any fault. ered to some extent inevitable. There is one more option
A similar procedure can be followed if an FS scheme is available to improve further the performance of the pro-
desired, but now the minimum tMM min of all the MM times
Fi
tection scheme; replacing the fuses on the laterals with
must be calculated for every fuse Fi . Then, in every related overcurrent relays. Such a revamping adds an extra cost
relayfuse pair, the tripping time of the relay must be set lower to the DSO, but the overall improvement can compensate
min = 0.3 s.
min to that of fuse by at least the CTI
than the tMM Fi
this cost.
We choose a first-stage DT delay for the relays equal to Based on the above, we decided to replace every fuse on
0.1 s. With this choice a very fast communication means is the laterals with an overcurrent relay having: 1) an extremely
needed so as the 15 ms transfer time between the relays is inverse phase element which picks-up for currents larger than
guaranteed. It has been observed that for various lateral faults, 125% of the lateral line load; 2) a DT phase element which
coordination between the relays and the 25 A fuses cannot be trips for currents larger than 180 A that is 20% less than the
obtained due to the low MM times. Hence, in order to apply minimum calculated fault current at the remotest lateral end;
the FS design, the replacement of the existing fuses with ones and 3) a DT earth element which picks-up for currents larger
having a larger rating has been examined. This introduced one than 30% of the lateral line load.
additional constraint: the new fuse should melt for the global A fast response of 0.15 s is decided for the phase and
minimum fault current Imin at the laterals. earth DT elements of the lateral relays, which subsequently
The investigation showed that coordination is achieved leads to a first-stage DT setting of the relays in the main line
when a 125 A rated fuse is selected (Fig. 11), except for the equal to 0.5 s if assuming a CTImin of at least 0.3 s. The
single-phase fault at BL6 which under minimum short-circuit second-stage is set with a time delay equal to 0.8 s. Note
conditions produces a current flow equal to 230 A. The global that the overcurrent settings of the main line relays remain
minimum of all other lateral fault currents is Imin = 340 A. unchanged.
The time-overcurrent plot for the forward looking relays in
D. Protection Scheme Improvement the grid-connected network is shown in Fig. 12. The close-in
Adding relays and CBs to every line section definitely three-phase fault at fuse F2 is depicted in the same figure, as
increases the cost of the protection scheme. However, with a simple example of protection coordination.
 This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE TRANSACTIONS ON SMART GRID

IV. C ONCLUSION [18] O. Rintamaki and J. Ylinen, Communicating line differential protection
for urban distribution networks, in Proc. 2008 China Int. Conf. Elect.
This paper proposes a communication-based protection Distrib. (CICED), Guangzhou, China, pp. 15.
scheme for radial distribution lines with DG. The scheme [19] G. G. Karady, A. Rogers, and V. Iyengar, Feasibility of fast pilot pro-
implements common directional overcurrent relays, assisted tection for multi-load distribution systems, in Proc. 2013 IEEE Power
Energy Soc. Gen. Meeting Conf., Vancouver, BC, Canada, pp. 15.
by intertripping and blocking transfer functions, and guaran- [20] K. Zimmerman and D. Costello, Fundamentals and improvements for
tees selectivity between the relays in the main trunk and the directional relays, in Proc. 63rd Annu. Conf. Prot. Relay Eng., College
fuses at the laterals under all possible operating conditions. If Station, TX, USA, 2010, pp. 112.
[21] IEEE Power System Relaying Committee, Relaying
relays are installed at the laterals instead of fuses, protection Communications Subcommittee. (Jan. 14, 2013). Communications
reliability can further increase. Technology for Protection Systems. [Online]. Available:
http://www.pes-psrc.org/Reports/Apublications_ new_format.htm
[22] Network Protection and Automation Guide, AREVA, Paris, France,
2005, p. 500.
R EFERENCES [23] S. P. Chowdhury, S. Chowdhury, and P. A. Crossley, Islanding protec-
[1] P. M. Anderson, Power System Protection. Piscataway, NJ, USA: IEEE tion of active distribution networks with renewable distributed genera-
Press, 1998, p. 1330. tors: A comprehensive survey, Elect. Power Syst. Res., vol. 79, no. 6,
[2] S. M. Brahma and A. A. Girgis, Microprocessor-based reclosing to pp. 984992, Jun. 2009.
coordinate fuse and recloser in a system with high penetration of dis- [24] D. Hornak and N. H. Chau, Distributed Generation Interconnections:
tributed generation, in Proc. 2002 IEEE Power Energy Soc. Winter Protection, Monitoring and Control Opportunities. Highland, IL, USA:
Meeting Conf., vol. 1. New York, NY, USA, pp. 453458. Basler Elect, 2002.
[3] A. Zamani, T. Sidhu, and A. Yazdani, A strategy for protection coor- [25] IEEE Standard for Interconnecting Distributed Generation: How Could
dination in radial distribution networks with distributed generators, in It Help My Facility? IEEE Standard 1547-2003, Nov. 2003.
Proc. 2010 IEEE Power Energy Soc. Gen. Meeting Conf., Minneapolis, [26] C. Gellings, Estimating the costs and benefits of the smart grid.
MN, USA, pp. 18. A preliminary estimate of the investment requirements and the resultant
[4] S. M. Brahma and A. A. Girgis, Development of adaptive protection benefits of a fully functioning smart grid, Elect. Power Res. Inst.,
scheme for distribution systems with high penetration of distribu- Palo Alto, CA, USA, Tech. Rep. 1022519, Mar. 2011. [Online].
tion generation, IEEE Trans. Power Del., vol. 19, no. 1, pp. 5663, Available: http://www.epri.com/abstracts/Pages/ProductAbstract.aspx?
Jan. 2004. ProductId=000000000001022519
[5] H. B. Funmilayo and K. L. Butler-Purry, An approach to mitigate the [27] C. A. P. Meneses and J. R. S. Mantovani, Improving the grid operation
impact of distributed generation on the overcurrent protection scheme and reliability cost of distribution systems with dispersed generation,
for radial feeders, in Proc. 2009 IEEE/PES Power Syst. Conf. Expo., IEEE Trans. Power Syst., vol. 28, no. 3, pp. 24852496, Aug. 2013.
Seattle, WA, USA, pp. 111. [28] M. E. Samper and A. Vargas, Investment decisions in distribution
networks under uncertainty with distributed generationPart I: Model
[6] I. Xyngi and M. Popov, An intelligent algorithm for the protection
formulation, IEEE Trans. Power Syst., vol. 28, no. 3, pp. 23312340,
of smart power systems, IEEE Trans. Smart Grid, vol. 4, no. 3,
Aug. 2013.
pp. 15411548, Sep. 2013.
[29] M. Lehtonen and S. Kupari, A method for cost benefit analysis of
[7] P. Mahat, Z. Chen, B. Bak-Jensen, and C. L. Bak, A simple adap-
distribution automation, in Proc. 1995 Int. Conf. Energy Manage. Power
tive overcurrent protection of distribution systems with distributed
Del., vol. 1. Singapore, pp. 4954.
generation, IEEE Trans. Smart Grid, vol. 2, no. 3, pp. 428437,
[30] V. Giordano, I. Onyeji, G. Fulli, M. S. Jimenez, and C. Filliou,
Sep. 2011.
Guidelines for conducting a cost-benefit analysis of smart
[8] M. Dewadasa, A. Ghosh, and G. Ledwich, Protection of distributed grid projects, European Commission, Joint Res. Centre, Inst.
generation connected networks with coordination of overcurrent relays, Energy Transport, Brussels, Belgium, Tech. Rep. EUR 25246 EN,
in Proc. 37th Annu. Conf. IEEE Ind. Elect. Soc. (IECON), Melbourne, 2012. [Online]. Available: http://ec.europa.eu/energy/en/content/
VIC, Australia, 2011, pp. 924929. guidelines-conducting-cost-benefit-analysis-smart-grid-projects
[9] A. Sinclair, D. Finney, D. Martin, and P. Sharma, Distance protec-
tion in distribution systems: How it assists with integrating distributed
resources, IEEE Trans. Ind. Appl., vol. 50, no. 3, pp. 21862196,
May/Jun. 2014. Vassilis C. Nikolaidis (M11) received the Diploma degree in electrical and
[10] C. Jecu et al., Protection scheme based on non communicating relays computer engineering from the Democritus University of Thrace, Xanthi,
deployed on MV distribution grid, in Proc. 2013 PowerTech, Grenoble, Greece, in 2001, and the M.Sc. degree in energy engineering and management
France, pp. 16. and the Ph.D. degree in engineering from the National Technical University
[11] S. Javadian, M. R. Haghifam, and N. Rezaei, A fault location and of Athens, Athens, Greece, in 2002 and 2007, respectively.
protection scheme for distribution systems in presence of dg using MLP From 2008 to 2012, he was a Power Systems Consulting Engineer
neural networks, in Proc. 2009 IEEE Power Energy Soc. Gen. Meeting with PROTASIS SA, Athens, Greece. He is currently a Lecturer with the
Conf., Calgary, AB, Canada, pp. 18. Department of Electrical and Computer Engineering, Democritus University
[12] H. Zayandehroodi, A. Mohamed, H. Shareef, and M. Farhoodnea, of Thrace. His current research interests include power system protection,
A novel neural network and backtracking based protection coordination control, stability, and transients.
scheme for distribution system with distributed generation, Int. J. Elect.
Power Energy Syst., vol. 43, no. 1, pp. 868879, 2012.
[13] J. Ma, X. Wang, Y. Zhang, Q. Yang, and A. G. Phadke, A novel adaptive Evangelos Papanikolaou received the D.Eng. degree in electrical and com-
current protection scheme for distribution systems with distributed gen- puter engineering from the Democritus University of Thrace, Xanthi, Greece,
eration, Int. J. Elect. Power Energy Syst., vol. 43, no. 1, pp. 14601466, in 2013.
2012. His current research interests include power system protection and energy
[14] N. El Halabi, M. Garca-Gracia, J. Borroy, and J. L. Villa, Current phase systems.
comparison pilot scheme for distributed generation networks protection,
Appl. Energy, vol. 88, no. 12, pp. 45634569, 2011.
[15] E. Sortomme, S. S. Venkata, and J. Mitra, Microgrid protection using Anastasia S. Safigianni (M97SM11) received the D.Eng. and Ph.D.
communication-assisted digital relays, IEEE Trans. Power Del., vol. 25, degrees from the Democritus University of Thrace, Xanthi, Greece, in 1981
no. 4, pp. 27892796, Oct. 2010. and 1988, respectively, both in electrical engineering.
[16] M. Dewadasa, A. Ghosh, and G. Ledwich, Protection of microgrids She is currently an Associate Professor with the Department of Electrical
using differential relays, in Proc. 21st Australas. Univ. Power Eng. and Computer Engineering, Democritus University of Thrace. Her current
Conf. (AUPEC), Brisbane, QLD, Australia, 2011, pp. 16. research interests include power systems and electrical installations, short-
[17] E. Sortomme, J. Ren, and S. S. Venkata, A differential zone protection circuit losses and forces in metal enclosed arrangements, power systems
scheme for microgrids, in Proc. 2013 IEEE Power Energy Soc. Gen. planning and optimization, lighting systems, influence of extra low frequencies
Meeting Conf., Vancouver, BC, Canada, pp. 15. electric and magnetic fields on human beings, and distributed generation.