See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320366572

Setting Zero-Sequence Compensation Factor in Distance Relays Protecting

Distribution Systems

Article  in  IEEE Transactions on Power Delivery · June 2018

DOI: 10.1109/TPWRD.2017.2762465

CITATIONS READS

9 2,006

2 authors:

Aristotelis M. Tsimtsios Vassilis C. Nikolaidis

Democritus University of Thrace Democritus University of Thrace
20 PUBLICATIONS   44 CITATIONS    40 PUBLICATIONS   290 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Special Issue of IJEPES - Harnessing Smart Grid Technologies for Combating Modern Power Systems Stability Issues through Feedback Control View project

Fault location in distribution systems View project

All content following this page was uploaded by Aristotelis M. Tsimtsios on 08 November 2018.

The user has requested enhancement of the downloaded file.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

Setting Zero-Sequence Compensation Factor in

Distance Relays Protecting Distribution Systems
Aristotelis M. Tsimtsios, Student Member, IEEE, and Vassilis C. Nikolaidis, Member, IEEE

 the performance of the distance relay is tested by using real-

Abstract--Reliable protection of modern distribution systems is world event report data. In [6], a multi-zone distance protec-
challenging due to various technical difficulties, mainly related to tion scheme for radial distribution systems with DG is pre-
the need for more efficient use of existing networks (e.g. shorter sented, taking into account the infeed effect as well as the co-
interruption durations and fewer affected customers, integration
ordination of the distance relay with other protection means in
of distributed generation, meshed network operation etc.). Such
challenges require, among others, the reconsideration of conven- the system. The application of distance protection to a distri-
tional protection techniques. A potential solution under examina- bution system with DG is examined in [7] but since, in this
tion concerns the adoption of distance protection for distribution work, DG is always connected to the substation bus, coordina-
feeders, due to its advantages compared to overcurrent protec- tion issues are not actually dealt with. In [8], the authors inves-
tion. However, as distance relays are mainly designed for trans- tigate the replacement of overcurrent relays with distance re-
mission networks, there are several issues to deal with in distri-
lays in an actual distribution network with DG and study the
bution applications, such as the proper setting of the zero-
sequence compensation factor (K0). The latter issue is critical in infeed/outfeed effect as well as the load variation effect on the
order to guarantee correct relay operation during single-line- performance of the distance relays. Coordination between the
ground faults. This paper examines the effect of K0 on the opera- line distance relays and distance relays used at the DG units
tion accuracy of distance relays protecting inhomogeneous distri- (instead of common undervoltage elements) to enhance fault-
bution feeders. Theoretical analysis, as well as investigation of ride-through capability, is further checked. Distance protection
various influencing factors, result in the determination of a pat-
is also considered in microgrid applications [9], [10].
tern, which relates relay accuracy to K0. Based on this analysis, a
simple methodology for setting K0 properly is proposed, which is A significant issue that requires thorough investigation is
implementable with commercially available relays. The method- the proper setting of zero-sequence compensation factor (K0)
ology is applied on a test distribution feeder and meaningful con- so that a distance relay installed in a distribution feeder can
clusions are derived. protect against single-line-ground (SLG) faults efficiently.
Although this issue is important for transmission [11] and sub-
Index Terms--Distance protection, distribution systems, relay transmission [12] systems as well, it is less challenging there,
accuracy, zero-sequence compensation factor.
due to the line homogeneity at high voltage levels, despite the
fact that mutual coupling could make it quite demanding [13].
I. INTRODUCTION
On the contrary, in distribution systems, proper K0 setting

M ODERN distribution systems have to operate reliably

into a challenging technical, economic, and regulatory
framework. Among other technical issues that need to be re-
could be much more complicated due to the inhomogeneity of
distribution lines as for their impedance data. In [14], an opti-
mized value of K0 is extracted for distance relays protecting
solved in this context, protection philosophy should also be distribution systems, through an optimization algorithm,
reconsidered in a way to obtain greater controllability and en- achieving discrimination between different distance relays.
hance efficiency, reliability, and quality of service [1] in dis- However, discrimination between the zones of each individual
tribution systems. Under such a demanding protection envi- distance relay is not always achieved.
ronment, the application of non-conventional, highly-reliable As a further step by the authors in the examination of ap-
protection concepts in distribution systems is investigated. plying distance protection to distribution systems, this paper
A solution being considered concerns the application of analyses the effect of K0 on the operation of distance relays
distance relays to distribution systems. Inherent directionality protecting inhomogeneous radial feeders without DG and pro-
and independency from external system conditions [2] make poses a methodology for setting K0 properly in order to
distance protection an attractive option for distribution net- achieve efficient protection against SLG faults. The proposed
works [3], even as a solution for the coordination of protection methodology can also be used for setting K0 properly when
means in multi-recloser schemes [4]. DG is connected to the feeder, as part of a preliminary attempt
Major issue for the proper application of distance protec- to deal with the infeed effect. This methodology is applicable
tion in distribution systems with intermediate sources (DG) is using commercially available distance relays, without any
the infeed effect [2]. The latter is explicitly analysed in [5] and need for intervention in the relay logic.
The organization of this paper is as follows. Section II in-
A. M. Tsimtsios and V. C. Nikolaidis are with the Department of Electri- cludes a brief theoretical description of the basic K0 factor
cal and Computer Engineering, Democritus University of Thrace, Xanthi
67100, Greece (e-mails: [email protected]; [email protected]). aspects. In Section III, theoretical analysis of the effect that K0

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

has on the distance relay operation is performed, whereas the TABLE I

K0 VALUES CORRESPONDING TO SEVERAL ACSR NOMINAL CROSS-SECTIONS
findings are verified through simulations in a test distribution
Nominal cross- Z1 Z0
feeder. In Section IV, a specific pattern, describing the effect section (mm2) (Ω/km) (Ω/km)
K0
of K0 on the relay operation, is determined. Based on this, a 95 0.215 + 0.334j 0.363 + 1.556j 1.030  26o
proper K0 setting methodology is presented and applied to a 50 0.404 + 0.386j 0.552 + 1.584j 0.720  39o
test feeder without DG units. An extension of this methodolo- 35 0.576 + 0.397j 0.724 + 1.595j 0.580  48o
gy to deal with the infeed effect if DG units are connected to 16 1.268 + 0.422j 1.416 + 1.620j 0.300  65o
the feeder is also presented in this section. The conclusions of
this work are drawn in Section V. III. EFFECT OF K0 ON THE RELAY ACCURACY
A. Theoretical Analysis
II. ZERO-SEQUENCE COMPENSATION FACTOR (K0)
In case of a distance relay protecting a line or a line seg-
Distance relays are designed to determine the positive- se- ment, which is homogenous in terms of conductor type and
quence impedance up to the fault position, by measuring the cross-section, setting K0 is simple, as only the Z1 and Z0 values
voltages and currents at the relay location [2]. Therefore, the of the line to protect are needed. It is worth mentioning that, in
reach setting of each phase and ground zone of a distance re- this case, the proper K0 factor is independent of the line length,
lay is set based solely on the positive-sequence impedance as according to the definitions of the previous section, only the
between the relay and the endpoint of the line segment to pro- p.u. values of Z1 and Z0 are needed for its calculation. If, how-
tect. In case of phase faults, the aforementioned setting is ade- ever, the protected line consists of segments with different Z1
quate for the relay to locate the fault accurately, as the phase and Z0 values, the determination of K0 factor for efficient pro-
distance element calculates the actual positive-sequence im- tection against SLG faults is rather complicated. In other
pedance up to the fault position. However, in case of SLG words, the impedance calculated by a ground distance element
faults, this reach setting cannot guarantee correct operation of in case of a SLG fault is inaccurate if K0 is incompatible with
the distance relay, as the impedance calculated by the ground the protected inhomogeneous line.
distance element is not equal to the positive-sequence imped-
The latter can be shown with the help of Fig. 1, which de-
ance up to the fault position due to the zero-sequence current
picts a radial overhead line with laterals as well as its repre-
effect. In order for the ground element to operate correctly, the
sentation on the complex impedance plane. The main trunk
zero-sequence current is compensated by the zero-sequence
consists of segments (LA1, LA2,….,LAn) of the same length,
compensation factor (K0), which stands for a complex value
conductor type and nominal cross-section, which are repre-
included in the relay settings.
sented on the R-X plane by the positive-sequence impedance
K0 factor is mainly expressed as:
vectors ZA1, ZA2….ZAn respectively. For all the line laterals,
K 0  ( Z 0  Z 1) / ( KZ 1) (1) the same conductor is considered which is of the same length
but smaller nominal cross-section compared to the conductor
where Z1 and Z0 stand for the actual positive- and zero- of the main trunk segments. Laterals LB1, LB2,….,LBn are rep-
sequence line impedance from the relay location up to the resented on the R-X plane by the positive-sequence impedance
fault position respectively, whereas K can be equal to 1 or 3, vectors ZB1, ZB2….ZBn respectively. Because of the smaller
depending on the relay design [15]. cross-sections, ZB vectors have a greater magnitude and small-
Alternatively, for several relays, K0 is expressed as the er angle compared to ZA vectors. The whole line is protected
complex ratio of Z0 to Z1 [16]. K0 can also be expressed by by a distance relay (namely DR) installed at the head of seg-
being split into a real and an imaginary part giving a reactive ment LA1, i.e. at the departure of the feeder. Only one ground
and a resistive zero-sequence compensation factor [16]. distance zone is shown in Fig. 1, which is set to reach up to
It is clear that each K0 value refers to a specific line type, the endpoint of lateral LB1, covering also a part of lateral LB2.
having specific characteristics, as each line type corresponds It is assumed that the time delay of this zone is set, so that the
to unique values for Z1 and Z0. Table I presents all the ACSR distance relay coordinates properly with the protection means
conductor nominal cross-sections used in the Greek distribu- (e.g. main fuses fB1 and fB2) of laterals LB1 and LB2.
tion system, the corresponding per unit (p.u.) Z1 and Z0 values, Let us focus on the two segments LA1 and LB1. If K0 is set
as well as the corresponding K0 factor, calculated using (1) for according to the positive- and zero-sequence impedance of
K = 3. It is apparent that, as the conductor nominal cross- segment LA1, then, for a SLG fault anywhere in this segment,
section decreases, K0 decreases in magnitude and increases in the ground distance element will calculate the positive-
angle. This behaviour of K0, shown above for ACSR conduc- sequence impedance up to the fault position accurately. Never-
tors, is generic and holds for all common conductor types used theless, in case of a SLG fault occurring anywhere in segment
in overhead distribution systems. LB1 (e.g. at point SC1’ in Fig. 1), the aforementioned K0 setting
will result in an inaccurate ground distance calculation and
consequently, in an inaccurate fault location. It should be not-
ed that this problem cannot be solved if K0 is set based on the
impedance data of segment LB1, as segment LA1 interposes
between the relay location and segment LB1. The latter K0 set-

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

ting would be proper only if the relay was installed at the be-
ginning of segment LB1 (i.e. protecting only this segment). In
order to achieve accurate impedance calculation for a SLG
fault at point SC1’, K0 should be set according to the imped-
ance data of the fictitious segment LC1, whose positive-
sequence impedance is represented by the vector ZC1 on the R-
X plane of Fig. 1. However, in this case, the relay will locate
the fault accurately only if the fault occurs exactly at point
SC1’, whereas the problem will remain unsolved for the rest of
the line.
It can be derived that, when a distance relay protects a line
consisting of segments with different impedance data, it is not
possible to determine a single value for K0 which results in
accurate fault distance calculation for all the possible SLG
fault situations in the protected line. This phenomenon can be
analysed based on the general expression of the impedance
calculated by the relay in case of a phase-a SLG fault [2]:
Fig. 1. Distance relay protecting a line consisting of different segments.
Za  Va / [ Ia (1  K 0 )] (2)
where Va and Ia are the phase-a voltage and current measured The above conclusions can be further clarified with a more
by the relay respectively. detailed analysis using Fig. 1. We examine now three fault
We could express (2) as: cases, in which SLG faults occur in the main trunk and in the
first lateral LB1, at the same distance from the relay location.
Za  Znc / (1  K 0 ) (3) In particular, the following fault cases are examined:
where Znc stands for the non-compensated impedance, namely  Faults SC1 and SC1’, at distance d1 from the relay location
the calculated impedance up to the fault position without the  Faults SC2 and SC2’, at distance d2 from the relay location
consideration of K0, which would lead the relay to an inaccu-  Faults SC3 and SC3’, at distance d3 from the relay location
rate fault distance calculation. where d1 < d2 < d3. Note that all SLG faults (six in total) shown
Considering again the example of Fig. 1 and based on (3), in Fig. 1 with their actual fault position (symbol SC) are inde-
the impedance calculated by the relay for a phase-a SLG fault pendent of each other and do not occur at the same time.
at point SC1’ of segment LB1 will be: Due to the differences in the conductor cross-sections, the
Za  [ ZA1nc / (1  K 0 )]  [ ZB1 ' nc / (1  K 0 )] (4) impedance vector (dashed lines) up to the fault position for a
fault in the lateral is of greater magnitude and smaller angle
where ZA1nc is the non-compensated impedance from the relay
compared to the corresponding vector for a fault in the main
location to the endpoint of LA1 and ZB1’nc is the non-
trunk, at the same distance from the relay location. Further-
compensated impedance from the endpoint of LA1 to the fault
more, it is apparent that as the distance from the relay location
position SC1’ (i.e. the non-compensated calculated impedance
increases, the divergence of the impedance vector for a lateral
for sub-segment LB1’). We assume that, in (4), K0 has been set
fault, from the corresponding vector for a fault in the main
according to the impedance data of segment LA1, i.e. the relay
trunk, increases. Assuming that K0 is set according to the im-
can operate accurately only for SLG faults occurring in this
pedance data of the main trunk, then, as the lateral conductor
segment (and the rest of the main trunk).
length up to the fault position increases, the compatibility of
The right part of (4) is the sum of two terms. The first term
K0 decreases, so does the relay accuracy. This can be related to
is ZA1nc/(1+K0), where the calculated impedance for segment
the previous interpretation of (4). In general, one could say
LA1 is “corrected” by being multiplied by 1/(1+K0), with K0
that the inaccuracy rate of a ground distance element is pro-
being compatible with the impedance data of this segment.
portional to the percentage of the total conductor length up to
The second term is ZB1’nc/(1+K0), where the calculated imped-
the fault position that corresponds to different impedance data
ance for sub-segment LB1’ is intended to be “corrected” based
compared to the impedance data based on which K0 is set.
on a K0 value, which is incompatible with the impedance data
of LB1’. Hence, it would be reasonable to say that ZB1’nc/(1+K0) B. Simulation Results
is the term that affects the impedance calculation accuracy of The previous analysis is performed for the radial 20 kV, 50
the ground distance element and, subsequently, the fault dis- Hz, overhead distribution line, shown in Fig. 2, which is based
tance calculation accuracy. That means that the inaccuracy on typical data of the Greek distribution system, as received
rate of the ground distance element is proportional to the mag- by the Hellenic Electricity Distribution Network Operator S.A.
nitude of ZB1’nc. Since the impedance magnitude of a line is (HEDNO S.A.). All segments of the main trunk consist of
proportional to its length, we could say that the inaccuracy rate 5-km long, 95-mm2 ACSR conductors, whereas the laterals
of a ground distance element increases as the length of the consist of 3-km long, 16-mm2 ACSR conductors. The total
protected segment, whose impedance data are incompatible to line load is 3.32 MVA (2.86 MW). The transmission grid is
the K0 setting, increases. represented by an equivalent source with a maximum short-

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

circuit power of 2085.1 MVA at 150 kV. The feeder is sup-

plied from the external transmission grid through a 150/20 kV
bulk distribution transformer. The distribution system is mod-
elled using DIgSILENT PowerFactory 2016.
A particular distance relay model was chosen from the
software database to replicate the relay that is considered be-
ing installed at the head of the line. The relay model used pro- Fig. 2. Test distribution feeder.
vides four independent quadrilateral phase elements and four
independent quadrilateral ground elements. The quadrilateral
characteristic is selected due to its advantages compared to
others (e.g. mho), especially regarding fault resistance cover-
age. In addition, the relay model used provides four independ-
ent polarizing elements, which allow the user to set a different
K0 value for each ground element, whereas it applies expres-
sion (1) with K = 3 for the calculation of K0.
At first, the variation of the relay accuracy (calculated fault
position) with regard to the actual fault position is examined.
For this purpose, four SLG faults are simulated at points P1,
P2, P3 and P4 across lateral L1 (shown in Fig. 3), corresponding
to a distance equal to 0%, 20%, 50%, and 100% from the be-
ginning of L1 respectively. F1, F2, F3 and F4 in Fig. 3 stand for
the positions, at which the relay locates the faults occurring at
P1, P2, P3 and P4 respectively, if K0 is set according to the im- Fig. 3. Variation of the relay accuracy with regard to the actual fault position.
pedance data of the main trunk. As can be seen, in the first
case, point F1 coincides with the actual fault position P1, be-
cause P1 belongs to the main trunk. However, as the actual
fault position moves away from P1 and consequently, from the
main trunk, the divergence of the calculated fault position
from the actual one increases, i.e. the accuracy of the relay
decreases. The above observations agree with the findings of
the previous theoretical analysis.
At this point, it is interesting to examine how the relay ac-
curacy varies with the change of K0, for a SLG fault at a given
position. Fig. 4 shows the different positions where the relay
locates a SLG fault occurring at the midpoint of segment L12,
depending on the K0 setting. In particular, five cases are illus-
trated in Fig. 4, depending on how K0 is set:
1) K0 is set according to the impedance data of the main
trunk, being equal to 1.03  26o. The relay locates the
fault accurately, at the midpoint of segment L12 (point C1).
Fig. 4. Variation of the relay accuracy with regard to the K0 value.
2) The magnitude of the accurate K0 is increased by 0.7, i.e.
the K0 setting is equal to 1.73  26o. The relay locates the
Table II shows the positive-sequence impedance values,
fault inaccurately, at point C2.
corresponding to points C1-C5 of Fig. 4, as calculated by the
3) The magnitude of the accurate K0 is decreased by 0.7, i.e.
ground distance element of the relay. As can be derived by
the K0 setting is equal to 0.33  26o. The relay locates the
Fig. 4 and Table II, by increasing the magnitude of K0, the
fault inaccurately, at point C3.
relay calculates an impedance of smaller magnitude and
4) The angle of the accurate K0 is increased by 20o, i.e. the
smaller angle compared to the impedance of the actual fault
K0 setting is equal to 1.03  46o. The relay locates the
position (C1), whereas, by decreasing the magnitude of K0,
fault inaccurately, at point C4.
exactly the opposite is valid. In addition, by increasing the
5) The angle of the accurate K0 is decreased by 20o, i.e. K0
angle of K0, the relay calculates an impedance of smaller angle
setting is equal to 1.03  6o. The relay locates the fault in-
and greater magnitude compared to the impedance of the actu-
accurately, at point C5.
al fault position, whereas, by decreasing the angle of K0, the
opposite is valid. The above results show that the variation of
the fault location point with regard to the K0 value follows a
specific pattern. The generalization of this pattern assists in
coming up with a solution for setting K0 factor properly.

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

TABLE II Let us refer to the example of Fig. 3 again. ZP2, ZP3 and ZP4
CALCULATED IMPEDANCE FOR EACH FAULT LOCATION POINT
stand for the impedance vectors up to the fault positions P2, P3
Fault location Impedance Z Reactance X Resistance R
point (Ω primary) (Ω primary) (Ω primary)
and P4 respectively. ZP2’, ZP3’ and ZP4’ represent the imped-
C1 2.979  57.12o 2.502 1.617 ance vectors up to the fault position for SLG faults at the same
C2 2.220  54.00o 1.796 1.305 distance from the relay location compared to the faults at P2,
C3 4.487  63.45o 4.014 2.006 P3 and P4 respectively, but in a fictitious 16-mm2 conductor
C4 3.145  47.11o 2.304 2.141 starting from the relay location. It can be seen that as the fault
C5 2.915  67.13o 2.686 1.133 position moves away from the main trunk across the lateral L1,
the fault impedance vector differs considerably from that of a
IV. EXAMINATION OF THE K0 SETTING PROBLEM SLG fault at the same distance in the main trunk (increases in
magnitude and decreases in angle). Actually, it tends to be-
As mentioned in the previous section, in case of a distance
come similar to the fault impedance vector corresponding to a
relay protecting an inhomogeneous feeder, it is not possible to
SLG fault at the same distance in the fictitious 16-mm2 con-
achieve absolute fault distance calculation accuracy for every
ductor. In fact, the (theoretical) ideal K0 value for accurate
SLG fault by using a single K0 setting, at least with the current
fault location approaches the K0 value of a 16-mm2 conductor,
relay technologies. However, that might not be necessary. Ac-
i.e. decreases in magnitude and increases in angle. Since K0 is
tually, what should be preserved is the relay never to locate a
set based on the impedance data of the main trunk (having a
fault outside the desirable distance zone due to under-reach
95-mm2 cross-section conductor), it is of greater magnitude
phenomena. That means that relay inaccuracy could be al-
and smaller angle (see Table I) compared to the ideal K0 value
lowed, as long as the fault is located inside the desirable dis-
for a fault anywhere in lateral L1.
tance zone. Nevertheless, in order to achieve this goal, a spe-
In general, situations such as the above constitute a combi-
cific pattern for the K0 effect should be determined first. Oth-
nation of Cases 1 and 4 described previously. Based on that,
erwise, this procedure would be much more complicated and
the ground distance element will calculate an impedance of
considerable computational effort would be needed.
smaller magnitude compared to the actual impedance up to the
A. Determination of the K0 effect pattern fault position. However, Cases 1 and 4 are contradictive re-
The terms Znc and (1+K0), used in (3), can be expressed in garding the angle change of K0, and by extension, the change
their polar form as: of angle c. Case 1 (resp. Case 4) indicates a decrease (resp.
Znc  Aa increase) in angle c of the calculated impedance compared to
(5) the actual impedance. Nevertheless, no matter whether the
(1  K 0)  Bb
variation of angle c is positive or negative, it always results in
Therefore, (3) can be alternatively expressed as: a lower calculated reactance than that of the actual fault posi-
Za  C c (6) tion. Bearing in mind that in order for this to be valid, the
where C = A/B and c = a – b. combination of Cases 1 and 4 should result in decrease of
Fig. 5a shows the vector (1+K0), as well as its components magnitude C (increase of magnitude B) and/or decrease of
10 and K0. As can be derived from this vector diagram: angle c (increase of angle b), it is explained as follows:
1) If only the magnitude |K0| of K0 increases, then magnitude Table III shows the percentage variation of magnitude B
B and angle b increase both. This results in a decrease in and angle b of term (1+K0), when moving from a lower to a
magnitude C and angle c (as A and a remain constant). greater ACSR nominal cross-section (i.e. from CSi to CSi+1, i =
2) If only the magnitude |K0| of K0 decreases, then magnitude 1,…,4), which gives a combination of Cases 1 and 4 as for the
B and angle b decrease both. This results in an increase in change of K0. As can be seen, when the variation of angle b is
magnitude C and angle c. negative (which tends to increase angle c and thus the calcu-
3) If only the angle g of K0 increases, then angle b increases, lated reactance) the variation of magnitude B (always positive)
whereas magnitude B decreases. This results in a decrease is greater than that of angle b. Therefore, the tendency towards
in angle c and an increase in magnitude C. the reduction of the calculated reactance is always dominant.
4) If only the angle g of K0 decreases, then angle b decreas- Note that in the case of moving from CS1 to CS2, the percent-
es, whereas magnitude B increases. This results in an in- age variation of angle b is greater than that of magnitude B;
crease in angle c and a decrease in magnitude C. however, as this variation is positive, it further strengthens the
It can be seen that, in general, the K0 setting affects the re- tendency towards the reduction of the calculated reactance.
lay accuracy following a specific pattern.
TABLE III
PERCENTAGE (%) VARIATION OF MAGNITUDE AND ANGLE OF TERM (1+K0)
Nominal Moving from CSi to CSi+1
1+K0
CSi cross-section % variation % variation
(mm2) (B  b)
of B of b
CS1 16 1.16  13.57o +25.00 +27.12
CS2 35 1.45  17.25o +11.72 -6.09
Fig. 5. (a) Vector diagram of (1+K0) and its components, (b) Variation of the CS3 50 1.62  16.20o +22.22 -18.58
calculated reactance with regard to the change of vector C  c. CS4 95 1.98  13.19o - -

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

Based on the above explanation, the decrease of magnitude

C is always more intense than the possible increase of angle c.
In fact, when the combination of Cases 1 and 4 results in an
increase of angle c, this increase is not enough to compensate
for the decrease of magnitude C and lead to a greater calculat-
ed reactance compared to that of the actual fault position. This
can be further clarified with the example of Fig. 5b, which Fig. 6. Ground distance zones for the examined distribution feeder.
illustrates the impedance vectors Cact  cact and Ca  ca corre-
sponding to the actual and the calculated fault position respec- Moreover, the fuse-blowing principle is adopted in [6],
tively, on the R-X plane. In this example, it is assumed that Ca meaning that faults occurring in a lateral are let to be cleared
and ca both differ by 50% (decrease and increase respectively) by the lateral fuse before the distance relay trips. This way, the
compared to Cact and cact respectively. As analysed before, the relay serves as primary protection for the main line and as
variation in angle c is always less intense. However, even in backup protection for the line laterals, which are primarily
this case, where the vector magnitude and angle are equally protected by the fuses. In order for this to be ensured, the time
varied, the calculated reactance (+Xa) is considerably lower delay of each phase/ground distance zone is set greater (in-
than the reactance of the actual fault position (+Xact). creased by the proper Coordination Time Interval - CTI) than
To sum up, it is generally fair to say that when a SLG fault the maximum possible total clearing time of any main fuse(s)
occurs at a point that corresponds to an impedance vector of included in this zone. In addition, coordination between two
greater magnitude and smaller angle compared to the imped- successive zones is achieved, meaning that the time delay tk of
ance vector based on which K0 is set, the relay calculates a zone k is always lower than the time delay tk+1 of zone k+1.
lower reactance value compared to the reactance of the actual Assuming now that the reach and time delay settings of the
fault position. Similar analysis showed that exactly the oppo- ground elements of a distance relay, protecting a distribution
site is also valid. Just to mention, in the example of Fig. 3, for feeder, have already been set based on [6], K0 for each ground
the SLG faults at P2, P3 and P4, the ground distance element element is set following the next steps:
calculates, in each case, an impedance of smaller magnitude 1) The areas Ak-1,k, formed in the complex impedance plane
and greater angle (resulting to a lower reactance), compared to between +Xk-1 and +Xk, where +Xk is the forward reac-
the actual impedance up to the fault position. tance reach of ground distance zone k, are determined.
B. Proposed K0 setting methodology The first area A01 is the area between the R-X origin (relay
location) and the forward reactance reach of the first
Major problem for setting a ground distance element pro-
ground distance zone.
tecting an inhomogeneous line part is that a single K0 cannot
2) The positive-sequence impedance (Z1) and the zero-
be compatible with all the possible fault cases. This can result
sequence impedance (Z0) of the main trunk, from the relay
in undetected ground faults, if the fault is located outside the
location up to the beginning of the lateral inside each area
desirable zone, or in undesired zone trips due to under-/over-
Ak-1,k which is closest to the relay, are calculated. These
reach. This work proposes a simple methodοlogy to deal with
calculations give a pair of Z1,k and Z0,k values defining a
this problem, based on the previously derived pattern for the
reference point for each ground distance element k.
effect of K0 on the relay accuracy. The proposed methodology
3) For each ground distance element k, K0 is calculated using
mainly considers the calculated reactance variation and is im-
the Z1,k and Z0,k values calculated for this element. If the
plementable using the current relay technologies and logics;
first ground distance zone is set to cover a part of the
however, it requires that the relay allows the determination of
main line between the relay location and the first lateral,
a different K0 setting for each ground distance element, a fea-
K0 for this zone element is calculated based on the p.u. Z1
ture provided by several well-established relay manufacturers.
and Z0 values of the protected segment.
In principle, the proposed methodology is based on the set-
The rationale behind the proposed methodology is ex-
ting rules originally described in [6]. To briefly illustrate these
plained through Fig. 7, which shows a part of a distribution
setting rules, let us consider again the radial overhead line
line with four laterals on the R-X plane. A distance relay at the
depicted in Fig. 2. Fuses f1 - f4 protect the main laterals,
head of the feeder protects the whole line. Consider the area
whereas fuses f1.r - f4.r protect the remotest tapped distribution
Ak-1,k between the +Xk-1 and +Xk settings of ground distance
transformer connected to each lateral. Coordination between
elements k-1 and k respectively, shown with solid horizontal
the main fuse and the fuse at the primary of the tapped trans-
lines in Fig. 7. The +Xk setting of zone k is defined as the up-
formers is always assumed. According to the setting philoso-
per boundary (UB) of this zone. Lateral Lk-1 is assigned to
phy of [6], the zone 1 phase and ground distance elements of
zone k-1, laterals Lk1 and Lk2 are assigned to zone k and lateral
the relay are set to cover 85% of the main line between the
Lk+1 is assigned to zone k+1 [6].
relay location and the first lateral. The higher-step distance
The first condition that should be fulfilled by ground dis-
zones are set to protect up to the endpoint of one or more lat-
tance element k is this element never to locate a SLG fault
erals, i.e. up to the remotest distribution transformer(s) fed by
occurring in the segments D-U1, D-U2 or D-U3 outside of zone
the lateral(s), without overreaching it (them). Fig. 6 shows the
k, i.e. above the UB (+Xk) of that zone. Let us consider the
feeder with ground distance zones graphically illustrated on it.
reference point S, which corresponds to the beginning of the

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

lateral inside the area Ak-1,k which is closest to the relay. Line e
separates the feeder into the parts upstream and downstream to
S. If K0 factor for element k is calculated based on the Z1 and
Z0 values from the relay location up to S, it is guaranteed that
any SLG fault inside area Ak-1,k will be located inside this area
(thus inside zone k). This is based on the derived pattern for
the effect of K0 on the calculated reactance, in combination
with the fact that, in principle, the nominal cross-section of the
conductor that constitutes the segments of a distribution feeder
is descending as we move towards the end of the feeder.
Specifically, if a SLG fault occurs anywhere downstream to
S, the Z1 vector up to the fault position will always be of
greater magnitude and smaller angle compared to the corre-
Fig. 7. Explanation of the proposed methodology.
sponding vector up to S. Since K0 has been set based on the
latter vector, the calculated reactance for a fault anywhere in Fig. 8 depicts the zone characteristics of the ground dis-
the segments S-U1, S-U2 or S-U3, will be lower than the reac- tance elements of the relay. The first zone is set to protect the
tance of the actual fault position. Therefore, the fault point 85% of L01 instantaneously. The second and the third zone are
calculated by element k will always lie between line e and the set to protect up to the endpoint of laterals L1 and L2 respec-
UB of zone k. Similarly, if a SLG fault occurs anywhere up- tively, whereas their time delays are set to be higher than the
stream to S, the Z1 vector up to the fault position will always highest ground fault clearing time of the corresponding main
be of smaller magnitude and greater angle compared to the lateral fuse. The fourth zone is set to protect up to the endpoint
corresponding vector up to S. Thus, the fault point calculated of lateral L4 (covering also lateral L3), with a time delay that is
by element k will always lie under line e. higher than the highest ground fault clearing time of the main
The second condition that should be fulfilled is ground dis- fuses at laterals L3 and L4. The positive resistive reach (+Rk) of
tance element k never to locate a SLG fault occurring in a lat- each zone k has been set equal to four times the positive reac-
eral Lk+j,,m (j, m ≥ 1), that is assigned to a higher-step element tance reach (+Xk). This is a good assumption for considering
k+j, inside zone k. If this happens, selectivity problems be- fault resistance in a medium voltage distribution network [3].
tween the relay and the lateral’s fuse may occur (due to the Based on the method-implementation steps described in the
lower time delay of element k compared to that of element previous subsection, the Z1 and Z0 values of the main line up
k+j). The possibility of such undesired trips is lower, as closer to buses B1, B2 and B3 (reference points), are calculated. Then,
the reference point S is placed to lateral Lk+1,m. This is be- by using (1), the K0 setting values for each ground distance
cause, then, the K0 setting value of element k will be as close element are calculated. The results of the above calculations
as possible to the ideal K0 value for a SLG fault in lateral are given in Table IV. Note that the K0 setting for the first el-
Lk+1,m. However, at the same time, the reference point S should ement is equal to the K0 value corresponding to the 95-mm2
be upstream to all the laterals assigned to element k, so that ACSR conductor (Κ0 = 1.03  26o).
every SLG fault in the segments S-U1, S-U2 or S-U3 is locat- The above K0 settings are applied to each ground distance
ed under the UB of zone k, according to the previous explana- element and SLG faults are simulated at each point where the
tion. Therefore, the consideration of the reference point S be- main trunk or the laterals cross the +Xk characteristic of each
ing the intersection point of the main line and the lateral inside ground zone k (i.e. the UB of each zone). The exact fault posi-
area Ak-1,k, which is the closest to the relay, gives an optimized tions (points E) are given in Fig. 8. The reactance, calculated
solution for setting K0 for ground distance element k. It has to by each ground distance element for those faults, is given in
be noted that such undesired trips of a lower-step zone due to Table V. From all the calculated reactances, those correspond-
a SLG fault occurring in a lateral assigned to a higher-step ing to faults that are assigned to the corresponding ground
zone have never been observed whenever the proposed meth- distance element of the first column are highlighted with bold
odology has been applied, even after extensive simulations in Table V. The non-bolded results refer to reactance values
performed on different distribution line configurations. that are calculated by ground distance elements, which are not
C. Application of the proposed methodology assigned to trip for the corresponding faults.
The distribution feeder shown in Fig. 2 is considered again. Table V gives also the reactance values of the UB (+Xk)
For the sake of generality, the main trunk is not assumed ho- that should not be violated by zone k element in each case. As
mogenous in this example, but with different ACSR conductor can be seen, in all cases, the bolded calculated reactances are
cross-sections along the line; 95-mm2 for L01, 50-mm2 for L12, below the corresponding UB value. Since this is valid for
35-mm2 for L23 and L34. The line is protected by the previous- faults occurring at the zones’ UB and based on the previous
ly described distance relay, which is installed at its departure analyses, any fault in the examined feeder will be located in-
and is set according to [6]. side the desirable zone. Note that in all fault cases, the fault is
located inside the assigned zone, as well as inside all the high-
er-step zones, but never inside a lower-step zone. Therefore,
the desirable ground distance element always trips first.

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

to the K0 setting of zone 1 element (i.e. that corresponding to

the main line segment where the relay is installed, according
to the traditional K0 setting philosophy).
The simulation results are given in Table VI. As can be
seen, for the conventional (uniform) K0 setting, zone 3 trips
undesirably in case of a fault at B3. Referring to Fig. 8, the
fault at B3 (inside zone 4) is located by zone 3 element inside
zone 3. However, this is avoided if the proposed K0 setting is
Fig. 8. Ground zone characteristics of the distance relay. used. Note that due to the coincidence of the uniform K0 set-
ting with the proposed K0 setting for zone 2, the corresponding
TABLE IV
CALCULATED K0 VALUES FOR EACH GROUND DISTANCE ELEMENT element calculates the same reactance values in each case,
ACCORDING TO THE PROPOSED METHODOLOGY whereas it never trips undesirably for a SLG fault at B2 or B3.
Ground
Z1 up to the Z0 up to the It results that the traditional K0 setting philosophy is not
reference point reference point K0 suitable for distribution system applications, as it can lead to
element (k)
(Ω primary) (Ω primary)
undesired zone trips, affecting selectivity between the distance
2 1.075 + 1.670j 1.815 + 7.780j 1.030  26o
3 3.095 + 3.600j 4.575 + 15.700j 0.860  34o
relay and the protection means of the line laterals. This is be-
4 5.975 + 5.585j 8.195 + 23.675j 0.740  40o
cause K0 is uniform for the whole line and not suitably special-
ized for each individual ground zone, so as to avoid undesired
TABLE V trips (see Subsection IV.B). On the contrary, the proposed K0
CALCULATED REACTANCE FOR SLG FAULTS AT BOUNDARY POSITIONS can ensure the detection of a fault inside the desirable zone
Calculated reactance and, at the same time, eliminate the danger of undesired trips.
(Ω primary) UB (+Xk)
Ground (Ω primary) TABLE VI
E1 E2 E3 E4 E5 CHECK FOR UNDESIRED TRIPS DUE TO SLG FAULTS IN LATERALS
element (k)
1 2.78 2.41 4.34 4.05 7.03 1.42 Calculated Calculated
reactance with reactance with the
2 2.78 2.41 4.34 4.05 7.03 2.97 conventional K0 proposed K0 UB (+Xk)
3 2.98 2.51 4.62 4.25 7.35 4.90 (Ω primary) (Ω primary) (Ω primary)
4 3.18 2.64 4.89 4.48 7.70 8.87 Ground
B2 B3 B2 B3
element (k)
D. Comparison with the traditional K0 setting philosophy 2 3.33 4.88 3.33 4.88 2.97
To date, significant experience concerning K0 setting is re- 3 3.33 4.88 3.56 5.18 4.90
trieved from transmission line applications. However, as
E. Advantages of the proposed distance scheme compared to
transmission lines are in their vast majority homogeneous as to overcurrent protection
their conductor type and cross-section, the traditional K0 set-
ting philosophy indicates that K0 is set simply based on the Distance protection provides inherent directionality and in-
dependency from external system conditions compared to
p.u. positive- and zero-sequence impedance of the conductor
overcurrent protection [2]. Apart from those advantages, the
where the distance relay is installed. That means that a uni-
ability of distance relays to coordinate efficiently with other
form K0 setting is determined for the whole line. Nevertheless, protection means in the system (e.g. fuses), while, at the same
if this setting is applied to an inhomogeneous distribution line, time, achieving the minimum possible fault clearing time, ren-
protected by a multi-zone distance scheme as that of Fig. 8, ders the application of distance protection to distribution sys-
there could be undesired zone trips due to the loss of discrimi- tems a promising solution.
nation between the ground distance zones. Let us compare the distance-based protection scheme
In the following, the possibility of an undesired zone trip shown in Fig. 8 with a typical overcurrent one, where a very
due to a SLG fault in a lateral assigned to a higher-step zone is common recloser is installed at the head of the line. Since a
checked, by applying the traditional and the proposed K0 set- fuse-blowing philosophy is adopted in the distance-based pro-
ting methodology. For this purpose, SLG faults are simulated tection scheme, the same philosophy is adopted in the reclos-
at the beginning of every lateral that follows the UB of a er-fuses coordination. Remember that according to [6], the
ground distance zone. In the examined distribution system, time delay tk of each ground distance zone k is set equal or
SLG faults are applied at buses B2 and B3, from which laterals greater to the largest possible total clearing time (tTCmax) of the
L2 and L3 depart, so as to check whether those faults are locat- main fuse at the lateral(s) protected by this zone, plus the re-
ed outside zone 2 and zone 3 respectively (as desired), by zone quired CTI, which is here considered equal to 0.3 s. 30T fuse
2 and zone 3 elements respectively. This check is not per- links are assumed at all the line laterals.
formed for lateral L1, since zone 1 and zone 2 elements have Table VII shows the largest total clearing times tTCmax of the
the same K0 setting, meaning that every fault in lateral L1 will main fuses, resulting from ground fault simulations at the end-
point of each lateral. It also shows the time delay setting tk of
be located by zone 1 between bus B1 and the UB of zone 2.
each ground distance element k. Zone 1 is set with an almost
Moreover, simulations performed with the proposed K0 set-
instantaneous time delay. Zones 2 and 3 are coordinated with
ting, are compared to those conducted with a uniform K0 set-
the largest total clearing time of the main fuse f1 and f2 respec-
ting for all ground distance elements. The latter is taken equal tively. Zone 4 is coordinated with the largest total clearing

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

time out of those calculated for fuses f3 and f4 at laterals L3 1) A 1.0 MVA, 0.4 kV, 50 Hz round-rotor synchronous gen-
and L4 respectively. Finally, Table VII includes the tripping erator (SG), operating with a unity power factor.
time tREC of the recloser, for the same simulated ground faults. 2) A set of four 250 KVA, 0.4 kV, 50 Hz induction genera-
It is shown that, for the same SLG faults, the recloser set with tors (IGs).
inverse time-overcurrent phase and ground characteristics 3) A set of two 500 KVA, 0.4 kV, 50 Hz inverter-interfaced
(Fig. 9) acts with an increased time delay compared to that of photovoltaic (PV) systems operating with a unity power
the distance relay. The difference in the tripping time between factor. The maximum steady state short-circuit contribu-
the recloser and the distance relay increases as the fault dis- tion of each PV system is limited to its nominal current.
tance from the relay location increases. Bear also in mind that SLG faults are simulated at the endpoints of laterals L1, L2
the tripping time of the recloser varies with respect to the fault
and L4, which correspond to the UB of ground distance zones
location due to the inverse-time characteristic, whereas the
2, 3, and 4 respectively, in order to calculate the increased +Xk
tripping time of each ground distance element remains always
setting for those zones, which addresses the infeed effect.
constant, in the lowest possible value. Moreover, the tripping
time of the recloser varies with respect to the short-circuit ca- Zone 1 remains unaffected. Each fault is simulated twice; at
pacity of the external grid, unlike the distance relay. first for the (theoretical) ideal K0 corresponding exactly to the
fault position, in order to extract the pure effect of the inter-
mediate infeed, and secondly for the K0 setting of Table IV.
The results of this analysis are presented in Table VIII.
Each value in the fourth column of Table VIII corresponds to
the reactance calculated by each ground distance element for
the ideal and the proposed K0 setting, when DG production is
present in the line. These values can be compared to those of
the third column, corresponding to zero DG production, in
order to calculate the required increase in the +Xk setting and
achieve discrimination between the ground distance zones.
It is shown that when SG or IG type DG units are connect-
ed to the line, the +Xk setting of all zones, except the first one,
Fig. 9. Conventional recloser-fuses scheme. has to be increased either if the ideal or the proposed K0 is
assumed. However, when the proposed K0 setting is applied,
TABLE VII the required increase is reduced. Concerning the connection of
TRIPPING TIME OF GROUND DISTANCE ELEMENTS COMPARED TO THAT OF A
TYPICAL RECLOSER
PV type DG units, the infeed effect is negligible due to their
Ground Lateral(s) where
low short-circuit contribution. Therefore, when the ideal K0 is
Examined tTCmax tk tREC considered, the required increase of the +Xk setting for zones
element ground faults are
fuse (fk) (s) (s) (s)
(k) simulated 2, 3 and 4, is very low in this case. On the contrary, when the
1 - - 0.055 proposed K0 is considered, not only there is no need to in-
2 L1 f1 0.265 0.565 0.567
crease the +Xk setting of the ground distance zones, but also
3 L2 f2 0.378 0.678 0.729
4 L3, L4 f3, f4 0.752 1.052 1.196 the calculated reactance becomes lower than the initial +Xk
setting (i.e. the reactance corresponding to the actual fault po-
F. Setting the distance relay considering DG sition). In the latter case, the proposed methodology seems to
It is a fact that (2)-(6) are not valid if intermediate sources be fully efficient in terms of addressing the infeed effect.
are present in a distribution line. Therefore, the proposed K0 DG intermittence is not a problem in the above-mentioned
setting alone cannot guarantee discrimination between the approach, since when DG production is zero (e.g. during night
ground distance zones due to the infeed effect [2], which usu- for PV systems), the reactance seen by a ground element for a
ally results in an increased reactance calculated by the DR SLG fault will be lower than the actual one, thus it will defi-
[16]. However, the infeed effect can be compensated by in- nitely lie inside the assigned zone.
creasing the forward reactance reach setting +Xk of each Based on the results of the above analysis, the proposed K0
ground distance zone suitably, even if the proposed K0 is taken setting facilitates the optimized adjustment of the relay set-
into account, which is calculated without considering DG. In tings to the infeed effect conditions. A more detailed study of
this way, in case of DG being present in the line, the imped- setting the ground distance elements rigorously, when DG is
ance that a ground distance element will calculate, using the present, is planned for future work.
proposed K0, for a SLG fault occurring anywhere in its as-
signed line part, will lie inside the respective (desirable) zone
area.
The latter conclusion will be verified by considering the
distribution feeder of Fig. 2, having now two generating plants
connected through step-up transformers to buses B1 and B3
respectively (Fig. 10). Three cases are considered as for the
type of the DG units composing the plants:
Fig. 10. Connection of DG plants to the examined distribution line.

0885-8977 (c) 2017 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 publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPWRD.2017.2762465, IEEE
Transactions on Power Delivery

10

TABLE VIII [3] J. Chang, L. Gara, P. Fong and Y. Kyosev, “Application of a multifunc-
+Xk SETTING DETERMINED IN EACH CASE TO ADDRESS THE INFEED EFFECT tional distance protective IED in a 15KV distribution network,” in Proc.
No DG 2 DG plants (at B1 and B3) 66th Annu. Conf. Protective Relay Engineers (CPRE 2013), College Sta-
tion, TX, 2013, pp. 150-171.
Ground Initial +Xk +Xk for +Xk for the [4] A. Enayati and T. H. Ortmeyer, “A novel approach to provide relay
DG
element setting ideal K0 proposed K0 coordination in distribution power systems with multiple reclosers,” in
type
(k) (Ω primary) (Ω primary) (Ω primary) Proc. 2015 North Amer. Power Symp. (NAPS), Charlotte, NC, 2015.
2 2.97 3.67 3.05 [5] A. Sinclair, D. Finney, D. Martin, and P. Sharma, “Distance protection
SG 3 4.90 6.73 6.09 in distribution systems: How it assists with integrating distributed re-
4 8.87 14.30 13.57 sources,” IEEE Trans. Ind. Appl., vol. 50, no. 3, pp. 2186–2196,
2 2.97 3.72 3.08 May/Jun. 2014.
IG 3 4.90 6.84 6.18 [6] V. C. Nikolaidis, C. Arsenopoulos, A. S. Safigianni and C. D. Vournas,
4 8.87 14.63 13.88 “A distance based protection scheme for distribution systems with dis-
2 2.97 2.99 2.45 tributed generators,” in Proc. 2016 Power Syst. Computation Conf.
PV 3 4.90 4.96 4.38 (PSCC), Genoa, Italy, 2016.
4 8.87 9.02 8.25 [7] I. Chilvers, N. Jenkins and P. Crossley, “Distance relaying of 11kV
circuits to increase the installed capacity of distributed generation,” IEE
Proc.-Generation, Transmission and Distribution, vol. 152, no. 1, pp.
V. CONCLUSION 40-46, Jan. 2005.
[8] K. Pandakov, H. K. Høidalen and J. I. Marvik, “Implementation of dis-
This paper examines the effect of K0 on the operation accu- tance relaying in distribution network with distributed generation,” in
racy of distance relays protecting distribution feeders. Accord- Proc. 13th Int. Conf. Develop. Power Syst. Protection 2016 (DPSP), Ed-
ing to the theoretical analysis performed and based on the re- inburgh, UK, 2016.
[9] H. Lin, C. Liu, J. M. Guerrero and J. C. Vásquez, “Distance protection
sults extracted from the examined test cases, the K0 effect fol- for microgrids in distribution system,” in Proc. 41st Annu. Conf. IEEE
lows a specific pattern, considering conductors commonly Ind. Electron. Soc. (IECON 2015), Yokohama, Japan, 2015, pp.731-736.
used in distribution networks. In particular, if K0 has been set [10] L. Jin, M. Jiang and G. Yang, “Fault analysis of microgrid and adaptive
distance protection based on complex wavelet transform,” in Proc. 2014
based on a specific positive- and zero-sequence impedance Int. Power Electron. Appl. Conf. Expo. (PEAC), Shanghai, China, 2014.
(reference impedance corresponding to a reference position) [11] V. Sellwood, U. Klapper, M. Kruger and S. Kaiser, “A new technique
and a fault occurs in a position (downstream to the reference for setting distance protection and fault location by measurement of
transmission line system impedance characteristics,” in Proc. 8th IEE
position), which corresponds to a positive-sequence imped- Int. Conf. AC and DC Power Transmission, London, UK, 2006.
ance of greater magnitude and smaller angle compared to the [12] M. El-Hadidy, H. Farouk and B. Soudy, “Zero sequence compensation
reference impedance, the ground distance element locates the factor effect on distance protection ground reach (practical study),” in
Proc. CIGRE 2012 Paris Session, Paris, France, 2012.
fault in a position of lower reactance compared to that of the [13] Y. Hu, D. Novosel, M. M. Saha and V. Leitloff, “An adaptive scheme
actual fault position, i.e. over-reach appears. On the contrary, for parallel-line distance protection,” IEEE Trans. Power Del., vol.
if a fault occurs upstream to the reference position, under- 17, no. 1, pp. 105-110, Jan. 2002.
[14] C. Jecu, B. Raison, R. Caire, P. Alibert, P. Deschamps, O. Chilard and S.
reach may appear. If not considered properly, those phenome- Grenard, “Protection scheme based on non communicating relays de-
na may lead to incorrect relay operation. ployed on MV distribution grid,” in Proc. 2013 PowerTech, Grenoble,
By exploiting this pattern, the authors develop a methodol- France, 2013.
ogy for determining a single K0 factor for each ground distance [15] J. G. Andrichak and G. E. Alexander, “Distance relays fundamentals,”
General Electric Co, Malvern, PA, Rep. GER-3966.
element, which leads the element to locate any SLG fault oc- [16] G. Ziegler, Numerical Distance Protection: Principles and Applications.
curring in its assigned line part, inside the corresponding (de- Erlangen, Germany: Publicis Publishing, 2011.
sirable) distance zone. The proposed methodology can be ap-
plied offline, does not require considerable computational ef- VIII. BIOGRAPHIES
fort, and is efficient for distribution systems without DG, in Aristotelis M. Tsimtsios (S’17) received the Diploma of Electrical and Com-
contrast to the traditional K0 setting philosophy. If intermedi- puter Engineering and the M.Sc. degree in Energy Systems and Renewable
Energy Sources from the Department of Electrical and Computer Engineering,
ate DG sources are considered, it is observed that the proposed Democritus University of Thrace, Xanthi, Greece, in 2013 and 2015 respec-
K0 reduces the infeed effect, making its handling easier. tively. He is currently pursuing the Ph.D. degree at the same Department. His
research interests include power system protection, distributed generation and
power system reliability. Mr. Tsimtsios is a member of the Technical Cham-
VI. FUTURE WORK ber of Greece.
Using the findings of this work, the authors intend to pro-
Vassilis C. Nikolaidis (M’2011) received the five-year Diploma of Electrical
ceed with a deeper examination of additional parameters, such and Computer Engineering from the Department of Electrical and Computer
as the effect of fault resistance and DG. In addition, several Engineering, Democritus University of Thrace, Xanthi, Greece, in 2001, the
other setting criteria considering the coverage and time delay M.Sc. degree in Energy Engineering and Management from National Tech-
nical University of Athens (NTUA), Athens, Greece, in 2002, and the Doctor
of distance relay zones will be examined. The ultimate goal is of Engineering from NTUA, in 2007. Since 2008 he has been working as a
to propose complete guidelines for the optimized setting of power systems consulting engineer. Currently he is an Assistant Professor
distance relays protecting distribution feeders with DG. with the Department of Electrical and Computer Engineering, Democritus
University of Thrace, Greece. His research interests mainly deal with power
system protection, control, stability, and transients.
VII. REFERENCES
[1] J. A. Momoh, Electric Power Distribution, Automation, Protection, and
Control. Boca Raton, FL: CRC Press, 2007.
[2] J. M. Gers and E. J. Holmes, Protection of Electricity Distribution Net-
works. London, UK: IET, 2004.

0885-8977 (c) 2017 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.
View publication stats