IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO.

3, SEPTEMBER 2013 1523

A Protection Coordination Index for Evaluating

Distributed Generation Impacts on Protection for
Meshed Distribution Systems
H. H. Zeineldin, Senior Member, IEEE, Yasser Abdel-Rady I. Mohamed, Senior Member, IEEE,
Vinod Khadkikar, Member, IEEE, and V. Ravikumar Pandi, Member, IEEE

AbstractDepending on the capacity, type and location, dis- multi-objective index which takes into account real and reac-
tributed generation (DG) can have an impact on protection tive power losses, voltage, conductor capacity, and short circuit
coordination of directional over-current relays for looped distri- levels. Such indices can be used in identifying where the DG
bution systems. In this paper, a new index is proposed, protection
coordination index (PCI), which can serve as an effective measure has least impact on the distribution network.
when planning the protection of meshed distribution systems with New techniques are needed and essential to determine the
DG. A two-phase non-linear programming (NLP) optimization maximum amount of DG that could be installed without re-
problem is proposed to determine the PCI by optimally calculating quiring major changes to the distribution network [6]. In [7],
variations in the maximum DG penetration level with changes in the maximum allowable DG penetration level considering the
the protection coordination time interval. Furthermore, the influ-
ence of connecting a DG at a certain location on the system PCIs IEEE 519 standard harmonic limits was studied. Closed form
is examined. The presented analysis is tested on the distribution equations were derived for radial distribution systems with uni-
section of the IEEE 14-bus and IEEE 30-bus systems. The PCI form, linearly increasing or linearly decreasing load patterns. In
can serve as an efficient index for distribution system planners: [8], the maximum amount of active power that can be supplied
(i) to determine the best DG candidate locations for utility owned by the DG at each bus of a radial distribution system, taking
DG and (ii) to evaluate the impact of a customer owned DG,
considering distribution system protection. into account voltage violations, was determined using repeti-
tive power flow studies. In addition, an index is proposed to
Index TermsDistributed generation, faults, optimization, pro- help utility managers in identifying which DGs are responsible
tection coordination.
for the voltage violation. Analytical expressions for simple ra-
dial distribution system were developed, in [6], to determine the
allowable DG penetration considering conductor ampacity and
I. INTRODUCTION
voltage rise.
Protection device coordination for distribution systems can
be affected, as well, by the integration of DG [1]. Synchronous
C URRENT Distribution systems can accommodate lim-
ited number of distributed generation (DG) due to voltage
profile and short circuit variations which consequently affect
based DG has a much more profound effect on protection co-
ordination than inverter based DG [9]. In [2], the impacts of
inverter based DG on re-closer/fuse coordination were inves-
the power quality and protective relaying [1]. The situation be-
tigated with inverter based DG and a DG interface control was
comes more severe with the increase in the DG penetration level
proposed to mitigate such impact. In [10], [11], the maximum al-
[2]. In [3], it is shown that the dynamic behavior and transient
lowable DG penetration level for radial distribution systems was
stability of a power system can become a concern with increased
calculated considering protection coordination. Due to the radial
DG penetration. The amount of DG penetration can be limited
nature of the system, a simple approach relying on the protec-
by the conductor ampacity, voltage regulation and short cir-
tive device characteristic can be implemented to determine the
cuit currents [4]. In [5], DG impacts were evaluated using a
allowable penetration level at each bus. In [12], a method that
relies on optimally locating fault current limiters to minimize
Manuscript received October 01, 2012; revised February 07, 2013 and April
the impacts of DG on protection coordination in radial distribu-
14, 2013; accepted May 12, 2013. Date of publication June 10, 2013; date of
current version August 21, 2013. This work was supported by the Masdar Insti- tion systems was proposed and solved using particle swarm op-
tute of Science and Technology , Abu Dhabi, UAE. Paper no. TSG-00661-2012. timization. Due to the integration of DG and the recent drive to-
H. H. Zeineldin is with Masdar Institute of Science and Technology,
wards smart grids, distribution systems are expected to be more
Abu Dhabi, UAE, and also with Cairo University, Giza, Egypt (e-mail:
hzainaldin@masdar.ac.ae). of the meshed structure [13][15]. In [16], a reliability assess-
V. Khadkikar and V. R. Pandi are with the Masdar Institute of Science ment algorithm is developed considering meshed distribution
and Technology, Abu Dhabi, UAE (e-mail: vkhadkikar@masdar.ac.ae;
systems. In [17], a new hybrid structure allowing both radial
vpandi@masdar.ac.ae).
Y. A.-R. I. Mohamed is with the Department of Electrical and Computer and meshed operation of the distribution system is proposed to
Engineering, University of Alberta, Edmonton, AB, Canada T6G 2V4 (e-mail: increase DG penetration. Identifying the impact of DG pene-
yasser_rady@ieee.org).
tration for looped distribution systems on protection coordina-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. tion problem with directional over-current relays is more com-
Digital Object Identifier 10.1109/TSG.2013.2263745 plicated and has not been addressed in previous literature.

1949-3053 2013 IEEE

1524 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 3, SEPTEMBER 2013

In this paper, a protection coordination index is defined and

proposed for use in determining the impact of integrating DG
(synchronous based type) on the protection of meshed distribu-
tion system. A two-phase nonlinear programming optimization
problem is proposed for calculating the PCI and is applied to
the power distribution network of the IEEE 14-bus and IEEE
30-bus system. The PCI is calculated by optimally determining
the rate of change of the DG penetration level with respect to
the change in coordination time interval. The usefulness of the
proposed PCI for distribution system planners in evaluating the
impacts of both utility owned and customer owned DG is dis-
cussed and analyzed. Fig. 1. Proposed two-phase problem formulation.

II. PROPOSED APPROACH TO DETERMINE THE PROTECTION

COORDINATION INDEX protection coordination study, the proposed approach involves
an additional algorithm (Phase II) to calculate the PCI.
Due to the bi-directional fault current flow, directional over-
current relays (OCR) are used for protecting looped distribu- A. Phase I: Conventional Protection Coordination
tion systems. The operation time of a directional OCR is an Formulation
inverse function of the short-circuit current flowing through it.
This function is defined by two parameters, namely the time-dial The settings of the relay are commonly calculated by formu-
settings of the relay and the pickup current , which lating an optimization model where the main objective is to min-
is the minimum value of current above which the relay will imize the sum of relay operating time subject to protection
start to operate. The relay time-current characteristic can be ex- coordination, relay setting and relay operating time constraints.
pressed as follows: The main optimization variables are the relay settings which in-
clude the and as well as the total relay operating time.
(1) This model will be denoted as the CPC formulation and can be
expressed as follows:

where is the relay identifier and is the fault location identifier. (2)
and are constants that vary with the type of OCR which are
set to 0.14 and 0.02, respectively. The term represents the
relay short circuit current and represents the relay pickup where is the total number of relays and is the total number
current. of fault locations investigated. Variables and represent the
The protection coordination index, PCI, is defined as the rate primary relay operating time and the backup relay operating
of change of the maximum DG penetration level with respect time for a fault at location , respectively. The CPC formulation
to the rate of change of coordination time interval CTI. To de- includes protection coordination constraints such that in case a
termine the PCI, a two-phase protection coordination optimiza- primary relay fails to isolate the fault in its zone, a backup relay
tion model is proposed for looped distribution systems with will operate. To assure proper coordination, a minimum gap in
directional over-current relays. In Phase I, the settings of the time between the operation of primary and backup relays
relays are determined using the conventional protection coor- , known as the CTI, needs to be maintained. The protection
dination (CPC) optimization model and with no DG present coordination constraint can be expressed as follows:
in the system [18]. Various methods have been proposed to
optimally solve the CPC model which includes the DICOPT (3)
solver in GAMS [19], genetic algorithms [20], evolutionary al-
gorithm [21], MATLAB optimization toolbox [22] and particle In addition to the above, there are upper and lower bound con-
swarm optimization [23]. The outputs of phase I are the optimal straints on the relay settings and relay operating time which can
relay settings which include the time-dial and the pickup current be expressed as follows:
settings. These settings are inputted to Phase II as parameters
(4)
where the main objective is to determine the maximum allow-
able DG penetration level considering protection coordination (5)
and relay operating time constraints. The DG penetration level (6)
is optimally calculated considering various CTI values to deter-
mine the PCI at each DG candidate location. Fig. 1 presents a where and are the lower and upper limits on
block diagram of the proposed two-phase formulation. As can the relay pickup current setting and and
be seen, the output of Phase I, which includes the optimal relay are the lower and upper limits on the relay TDS setting which are
settings, is an input to Phase II where the main objective is to set to 0.1 and 11, respectively. Parameter represents the
maximize the DG penetration. In comparison to a conventional maximum relay operating time which will depend on the utility
ZEINELDIN et al.: A PROTECTION COORDINATION INDEX FOR EVALUATING DG IMPACTS 1525

operator design criterion. For numerical relays, can be set as

a continuous variable.

B. Phase II: DG Maximization Formulation

In Phase II, the relay settings ( and ) are treated as
fixed parameters with values calculated from Phase I. The pro-
posed approach calculates the PCI by optimally determining the
change in the maximum achievable penetration level with CTI.
In Phase II, the main objective is to maximize the DG penetra-
tion as follows:

(7)

where is the DG MVA rating to be installed at bus and

denotes the number of DG. Similarly, constraints (3) & (6) are
included in the problem formulation. One important difference
between Phase II and Phase I is the short circuit calculation pro-
cedure. In Phase I, since there are no DG units in the distribution
system, the short circuit currents are calculated and are inputted
as a parameter in (1). On the contrary, for phase II, the short cir-
cuit currents vary depending on the DG location and capacity.
Inserting DG at certain locations affect the bus admittance ma- Fig. 2. Power distribution system of the IEEE 30-bus system under study.
trix which in turn changes the bus impedance matrix resulting in
variations in short circuit current values. Thus, the short circuit
current in (1) is a function of the DG capacity as follows: studies analyzed, for which the PCI is calculated, is presented.
Furthermore, a flowchart detailing the various components of
(8) the proposed algorithm is presented and explained.
In addition, an upper limit is set on the DG capacity that can be
installed, at each DG bus, as follows,
A. Test System Under Study
(9)
The single-line diagram of the power distribution system of
the IEEE 30-bus test system, where power is fed from more than
where represents the upper limit on the DG capacity
one point, is depicted in Fig. 2. The power distribution system of
to be installed at each location which is set to 10 MVA. The
the IEEE 14- bus system is given in Appendix I. Detailed system
value of will depend on the utility planner design
parameters can be found in [24]. The distribution system is fed
preference. The DG penetration level in (7) is calculated for
through three 50 MVA 132 kV/33 kV transformers connected
different values of CTI and the rate of change of P with
at buses 2, 8 and 12. Ten buses have been selected as candidate
respect to the rate of change of CTI is calculated to de-
locations for DG installation and these buses include 2 to 10 and
termine the PCI. Thus, the PCI index can be defined as follows:
bus 12. All DG units are connected to the system through a 480
V/33 kV step-up transformer with 0.05 p.u transient reactance.
(10)
The system is equipped with 28 directional over-current relays.
Similarly, the short circuit currents are calculated by conducting
The units of PCI are in MVA/s. The PCI is defined such that
a bolted three-phase midway fault on each line. The fault loca-
a positive value indicates that a reduction in CTI will result in
tions are denoted as F15 to F30 as shown in Fig. 2. For both
an increase in the DG penetration. Higher values for PCI at cer-
systems, the chosen DG technology is a synchronous type with
tain locations indicate that higher DG penetration levels can be
0.0967 p.u transient reactance based on its capacity. It is as-
achieved with less impact on protection coordination (or in other
sumed that all relays are identical and have the standard inverse
words with small changes in CTI).
relay curves with the following parameters 0.14 and 0.02 for
and , respectively.
III. SYSTEM AND SIMULATION SETUP As stated in [25], DG can be either owned by the utility or by
The proposed two-phase problem formulation is applied a customer and in both cases there could be an impact on pro-
to the distribution network of the IEEE 14-bus and the IEEE tection coordination. For utility owned DG, the PCI can be used
30-bus power systems. For brevity, the paper will focus on ana- in identifying the best location on the distribution system which
lyzing in details the results of the IEEE 30-bus system whereas would result in minimum impact on protection coordination. For
some of the analysis and simulation results for the IEEE 14-bus customer owned DG, there is no control over the location of the
system will be provided. In this section, a description of the case DG as it depends mainly on the customer location. In such case,
1526 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 3, SEPTEMBER 2013

the PCI can be used as a measure for utility operators to deter-

mine if changes and modifications are needed for the protection
system as a result of a customer owned DG interconnection. In
this paper, the applicability of the PCI in planning the distri-
bution system protection considering both utility and customer
owned DG is examined. In addition, the sensitivity of the PCI
value at a specific location as a result of changes in location and
size of DG at other locations is investigated.

B. Algorithm Implementation
The two-phase protection coordination optimization
problem, presented in Section II, is implemented in MATLAB
and solved using the fmincon function which uses the reduced
gradient approach (first order optimality) for solving con-
strained non-linear optimization problems. In Phase I, the relay
settings are optimally determined for the system considering
no interconnected DG. The main algorithm steps for Phase I
are as follows.
1) Construct the system impedance matrix with no
DG.
2) Calculate the midway three phase short circuit currents
passing through all relays in both primary and backup
modes.
3) Perform Phase I of the optimization model and obtain op-
timal and values.
The output of Phase I ( and ) is inputted to Phase II
as parameters. For Phase I, the system impedance matrix, as
well as the short circuit currents, is independent of the opti-
mization variables ( and ). For Phase II, since the ob-
jective is to calculate the PCI by determining the rate of change
of DG capacity at selected candidate locations, both the system
impedance matrix and short circuit currents will vary with vari-
ations in DG capacity and location. The main steps for Phase II
are as follows.
1) Set initial values for CTI and select the DG location.
2) Construct the system impedance matrix which will be a
function of .Calculate the three phase short circuit
currents passing through all relays in both primary and
backup modes.
3) Perform Phase II of the optimization model and obtain
at the selected bus location. Fig. 3. Proposed two-phase problem formulation.
4) Change the CTI values and repeat steps 2 to 5.
5) Calculate the PCI at the selected bus location .
Fig. 3 presents a flowchart of the proposed two-phase opti- and a maximum relay operating time of 2.5 s. The setting,
mization problem. for the majority of relays, hits the lower bound which is 0.1 s. In
addition, it can also be seen from Table II that the coordination
IV. PROTECTION COORDINATION INDEX ANALYSIS constraint presented in (3) is binding for the majority of fault lo-
The conventional protection coordination (Phase I of the pro- cations. For example, for a fault at F19, the difference between
posed algorithm) is applied to the power distribution system of the backup relay (R9) and primary relay (R7) is 0.3 seconds
the IEEE 30-bus system to minimize the total operating time which is equal to the CTI limit. The results presented in Table I
as in (2) by choosing the optimal relay settings and operating will be inputted to Phase II of the optimization problem to de-
time. Table I presents the optimal relay settings for the IEEE termine the proposed PCI.
30-bus case considering no DG interconnection. The relay op- Similarly, the conventional protection coordination algo-
erating times (primary and backup) are given in Table II. Table II rithm is applied on power distribution system of the IEEE the
also presents the backup and primary relay pairs for faults F15 14-bus system to minimize the total operating time as in (2)
to F30. by choosing optimal relay settings. The optimal relay settings
Depending on the fault location, there can exist more than one for the IEEE 14-bus case considering no DG interconnection is
backup for a relay. The results were obtained for s given in Appendix II.
ZEINELDIN et al.: A PROTECTION COORDINATION INDEX FOR EVALUATING DG IMPACTS 1527

TABLE I TABLE II
OPTIMAL RELAY TDS AND IP SETTINGS FOR IEEE 30-BUS CASE OPTIMAL PRIMARY AND BACKUP RELAY OPERATING TIMES

A. Application of PCI to Meshed Distribution Networks

The DG maximization problem, presented in Section II-B,
is solved using the optimal relay settings presented in Table I.
The CTI is varied and for each value of CTI, the maximum DG
penetration level at each possible candidate DG location is op-
timally determined. It is worthy to note that the choice of CTI
depends on the protection system designer and can range from
0.2 to 0.3 s. Fig. 4 presents the maximum percentage DG pen-
etration level obtained by the reduced gradient approach at se-
lected buses versus the CTI values. The percentage DG penetra-
tion was calculated by determining the ratio of DG penetration
in MVA with respect to the total system MVA (150 MVA for the
IEEE 30 bus system and 120 MVA for the IEEE 14 bus system).
As defined earlier, the PCI is the rate of change of the maximum
DG penetration level with respect to CTI. The average slope of
the curve presented in Fig. 4 is calculated and Table III presents
the PCI values obtained. Locations with higher PCI result in
lower DG impact on protection coordination.
The DG penetration versus CTI curves, presented in Fig. 4,
can be used in indentifying the best location to install a utility
owned DG. For example, if the utility wants to install a 5 MVA
DG and assuming that the possible candidate locations are
busses 2, 3, 6, 9, 10, and 12, then the best location to install the
DG would be bus 12. As seen from the curves, installing a 5
MVA DG at location 12 will result in the least change in CTI
( 0.275 s). For such case, the utility might not need to modify
the relay settings since the impact is minimal. On the contrary,
if the possible candidate DG locations are 5 and 6 only, bus
5 would be the best option since it has a higher PCI. But, by
referring to Fig. 4, a 5 MVA DG at bus 5 would result in a
significant impact on protection coordination
and in such case, the relay settings must be modified to accom-
modate the DG. by 0.5/89.69 which will result in a CTI of 0.294 seconds. In-
For customer owned DG, utility operators have no control stalling the same DG capacity at bus 6 will change the CTI by
over the location and size of the DG to be installed. It is pri- 0.5/12.3175 resulting in an overall CTI of 0.259. The PCI can
marily dictated by the customer preference and location. The thus help utility planners and designers to assess impact of DG
DG penetration versus CTI curves as well as the PCI can pro- integration as well as identify the best candidate locations for
vide utility operators with a fast and accurate assessment of the DG that would have least impact on protection coordination.
impact of a customer owned DG. For example, for a customer The same analysis is applied to the distribution system of the
requesting to install 0.5 MVA DG at bus 9, the CTI will change IEEE 14-bus system. The DG penetration versus CTI curves are
1528 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 3, SEPTEMBER 2013

Fig. 4. Variation of DG penetration with CTI for the IEEE 30 bus system. Fig. 6. Variation of DG penetration over CTI with preinstalled DG for the IEEE
30 bus system.

TABLE III
PCI FOR THE IEEE 30 BUS CASE

Fig. 7. Variation of DG penetration over CTI with preinstalled DG for the IEEE
14 bus system.

tion, the effect that an installed DG at a specific location might

have on the PCI at other locations is examined. In order to ana-
lyze the interaction between DG units installed in the system and
its effect on PCI, pre-installed DG of fixed rating is considered
to be connected already in the system. In the IEEE 30-bus test
Fig. 5. Variation of DG penetration with CTI for the IEEE 14 bus system. system, a pre-installed DG of 5 MVA rating is connected at bus
10. The DG Penetration versus CTI curves at two selected buses
TABLE IV 2 and 5 are shown in Fig. 6. It can be seen that the impact on the
PCI FOR THE IEEE 14 BUS WITH 7 DG CANDIDATE LOCATIONS protection coordination could be positive or negative. The PCI
at bus 2 would decrease when a customer, at bus 10, installs a 5
MVA DG. The PCI at bus 2 has decreased from 29.3738 (with
no DG at bus 10) to 26.8263 (with a 5 MVA DG). On the con-
trary, for bus 5, the PCI increased from 17.4536 to 18.823 when
a 5 MVA DG is installed at bus 10.
Similarly, in the case of IEEE 14-bus system, a preinstalled
DG of 2.5 MVA rating is considered at bus 4. The DG penetra-
given in Fig. 5. Table IV presents the PCI at the various buses on tion versus CTI curves at buses 3 and 5 are shown in Fig. 7. The
the system. From the results, it can be observed that bus 1 has the protection coordination index at bus 3 increased from 17.8174
maximum PCI and thus a DG installed at that location will have to 20.4275 and also increased at bus 5, from 17.5670 to 18.4565.
least impact on protection coordination. On the contrary, bus 6 Furthermore, variations in PCI magnitude, at selected loca-
has the least PCI value which indicates that adding a DG at such tions, as a result of the variation in the capacity of pre-installed
location will have greater impact on protection coordination. DG at bus 10 are shown in Fig. 8. It can be seen that the PCI
level is decreasing at bus 2 with the increase in capacity of
B. Impact on PCI due to Installed DG pre-installed DG connected at bus 10. On the contrary, the PCI
Since the fault current magnitudes will vary with the integra- level at bus 5 is increasing with the increase in the capacity of
tion of DG, it is expected that the PCI will also vary. In this sec- pre-installed DG connected at bus 10. The same analysis has
ZEINELDIN et al.: A PROTECTION COORDINATION INDEX FOR EVALUATING DG IMPACTS 1529

Fig. 8. Variation in PCI values over pre-installed DG capacity for the IEEE 30 Fig. 10. Impact of pre-installed DG location and capacity on the PCI of bus 3
bus system. for the IEEE 14 bus system.

Fig. 11. Impact of pre-installed DG location and capacity on the PCI of bus 5
Fig. 9. Variation in PCI over pre-installed DG capacity for the IEEE 14 bus
for the IEEE 14 bus system.
system.

has increased and thus reducing the possible allowable DG pen-

been applied to the power distribution system of the IEEE 14 etration to avoid further degradation of the protection coordina-
bus system. It has been noticed that for the case under study, tion.
for all buses, there is an increase in the PCI as the DG capacity For this case study, there are two main factors that can im-
increases. Fig. 9 presents the variation of the PCI with DG ca- pact the PCI value which include the pre-installed DG location
pacity at busses 3 and 5. It can be seen that the PCI level is and capacity. Figs. 10 and 11 highlight the impact of both the
increasing at bus 3 as well as bus 5 with the increase in the ca- pre-installed DG location and size on the PCI of bus 3 and 5,
pacity of pre-installed DG connected at bus 4. By referring to respectively. For brevity, the results for the IEEE 14 bus system
Fig. 9 and using (10), with no DG installed at bus 4, a 1 MVA are presented but similar results can be obtained for the IEEE 30
DG installed at bus 3 would result in a drop of 0.056 s in CTI bus system. Each bus PCI varies differently depending on the
and thus the protection system would be coordinated with a CTI location and size of the pre-installed DG. In general, the PCI at
equal to 0.2438 s. With a 2.5 MVA DG installed at bus 4, a 1 bus 3 increases in the presence of installed DG at other loca-
MVA DG at bus 3 would drop the CTI to 0.2512 s. This indi- tions. Bus 3 is most affected by the presence of a DG at bus 4.
cates that installing a DG at bus 4 might further allow more DG For bus 5, a pre-installed DG at bus 6 or 7 will have significant
to be installed at bus 3 with lower impact on the protection co- impact on the PCI of bus 5. On the contrary, the presence of DG
ordination. at busses 2, 3, 4 and 1 have minimal effect on the PCI of bus
The results shown in Fig. 8 and 9 indicate that installing a DG 5. The results also show that the PCI can increase and decrease
at a specific location can either have positive or negative impact with the change in pre-installed DG capacity (refer to busses 5
on the PCI at other locations on the system. In other words, for and 6 in Fig. 11).
cases where the PCI increases (for example, as in the case of
Bus 5 in Fig. 8), this indicates that more DG penetration can be V. DISCUSSION
achieved with lower impact on protection coordination. On the In the literature, protection coordination of directional over-
other hand, a decrease in the PCI at a candidate bus indicates current relays is commonly analyzed and performed considering
that the impact of installing a DG, on protection coordination, bolted faults [18][23]. Thus, the impact of fault impedance
1530 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 3, SEPTEMBER 2013

Fig. 12. Impact of fault resistance on PCI value for the IEEE 30 Bus system.

is not taken into consideration while determining the optimal

relay settings, specifically for directional over-current relays.
The fault impedance can range from low values (in the range of
milliohms [2], [26]) up to higher values in the range of ohms [2].
In the previous sections, the PCI was determined considering
bolted three phase faults. In this section, a preliminary study that
discusses the possible impact of faults, with low impedance, on
the PCI is presented. Fig. 12 shows the value of PCI at each
bus for a three phase fault considering various fault resistance
values for the IEEE 30 bus system. First, it can be seen that as the
fault resistance increases, the PCI value increases. A higher PCI
would indicate that more DG penetration is possible and thus,
for the system under study, considering bolted faults would pro-
vide a conservative measure for calculating the PCI. Secondly, Fig. 13. Power distribution system of the IEEE 14-bus system under study.
the fault resistance does not have an impact on the rank of each
bus in terms of PCI. In other words, bus 12 remains the optimal of dynamic loads can impact the DG penetration level as well
location for installing a DG while bus 6 remains the least at- as PCI.
tractive for installing a DG. For the system under study, higher
values of fault resistance will result in protection coordination VI. CONCLUSIONS
failure which is expected since the fault current magnitudes will This paper proposes a new protection coordination index
be significantly affected. In [2], for radial systems, it has been (PCI) for quantifying impacts of interconnecting DG on
shown that the fault impedance has an impact on re-closer/fuse protection coordination. The PCI is calculated by optimally
coordination. The impact of higher fault resistance on protection determining the rate of change of DG penetration level with
coordination of meshed distribution system and its mitigation is respect to the coordination time interval. A two-phase opti-
out of the scope of this paper and will be investigated in details mization problem is proposed, implemented and tested on the
in future work. power distribution side of the IEEE 14-bus and IEEE 30-bus
The impact of a specific DG technology on the PCI will de- systems. The results show that the proposed PCI can serve as
pend primarily on the DG interface. The fault current magni- an effective measure for utility operators to determine the best
tudes from DG with a synchronous machine interface are much location for allocating a utility owned DG with minimal impact
higher than an inverter-based DB unit. Commonly, the interface on the protection coordination. Furthermore, the PCI can also
control of inverter based DG are equipped with current limiters help in determining the extent to which the system protection
which can limit the fault current during fault conditions [9]. coordination is affected with the installation of customer owned
Thus, it is expected that the PCI for inverter based DG will be DG and thus identifying any necessities for modifying the relay
even higher than the values calculated for synchronous-based settings as a result of the DG integration. Lastly, the results
DG. In addition, in protection coordination studies [18][23], showed that a DG installed at a certain location can either
all loads are assumed to be static loads. Static loads do not have increase or decrease the PCI value at other locations.
fault current contribution and for this reason the effect of load
is commonly neglected in protection coordination studies. On APPENDIX I
the contrary, dynamic loads (such as motors) can contribute to The single-line diagram of the IEEE 14-bus looped distribu-
fault current. The effect of dynamic loads has not been consid- tion system is shown in Fig. 13. The detailed data for the IEEE
ered in this paper but it is worthy to mention that the presence 14-bus test system can be obtained from [24]. The distribution
ZEINELDIN et al.: A PROTECTION COORDINATION INDEX FOR EVALUATING DG IMPACTS 1531

system is fed through two 60 MVA 132 kV/33 kV transformers [10] S. Chaitusaney and A. Yokoyama, Prevention of reliability degrada-
connected at buses 1 and 2. All seven buses have been selected tion from recloser-fuse miscoordination due to distributed generation,
IEEE Trans. Power Del., vol. 23, no. 4, pp. 25452554, Oct. 2008.
as candidate locations for DG installation. All DG units are con- [11] T. K. Abdel-Galil, A. E. B. Abu-Elanien, E. F. El-Saadany, A. Girgis,
nected to the system through a 480 V/33 kV step-up transformer Y. A.-R. I. Mohamed, M. M. A. Salama, and H. H. M. Zeineldin,
with 0.05 p.u transient reactance. The system is equipped with Protection Coordination Planning With Distributed Generation CETC
Number 2007-149/2007-09-14, Sep. 2007.
16 directional over-current relays, two on each side of a line. [12] H. H. Zeineldin, E. F. El-Saadany, M. M. Salama, A. H. K. Alaboudy,
The short circuit currents are calculated by conducting a bolted and W. L. Woon, Optimal sizing of thyristor-controlled impedance
three-phase midway fault on each line. The fault locations are for smart grids with multiple configurations, IEEE Trans. Smart Grid,
vol. 2, no. 3, pp. 528537, Sep. 2011.
denoted as F8 to F15 as shown in Fig. 13. [13] J.-H. Teng, Unsymmetrical short-circuit fault analysis for weakly
meshed distribution systems, IEEE Trans. Power Syst., vol. 25, no.
1, pp. 96105, Feb. 2010.
APPENDIX II [14] Y. Qiang, J. A. Barria, and T. C. Green, Communication infrastruc-
tures for distributed control of power distribution networks, IEEE
Trans. Ind. Inform., vol. 7, no. 2, pp. 316327, May 2011.
[15] Y. Li, D. Czarkowski, and F. de Leon, Optimal distributed voltage
regulation for secondary networks with DGs, IEEE Trans. Smart Grid,
TABLE V vol. 3, no. 2, pp. 959967, Jun. 2012.
OPTIMAL RELAY TDS AND IP SETTINGS FOR IEEE 14-BUS CASE [16] W. Zhuding, F. Shokooh, and Q. Jun, An efficient algorithm for
assessing reliability indexes of general distribution systems, IEEE
Trans. Power Syst., vol. 17, no. 3, pp. 608614, Aug. 2002.
[17] M.-C. Alvarez-Herault, D. Picault, R. Caire, B. Raison, N. HadjSaid,
and W. Bienia, A novel hybrid network architecture to increase DG
insertion in electrical distribution systems, IEEE Trans. Power Syst.,
vol. 26, no. 2, pp. 905914, May 2011.
[18] W. El-Khattam and T. Sidhu, Restoration of directional overcurrent
relay coordination in distributed generation systems utilizing fault cur-
rent limiter, IEEE Trans. Power Del., vol. 23, no. 2, pp. 576585,
2008.
[19] H. H. Zeineldin, Optimal coordination of microprocessor based
directional overcurrent relays, in Proc. Canadian Conf. Electrical
and Computer Engineering (CCECE 2008), May 47, 2008, pp.
000289000294.
REFERENCES [20] C. W. So, K. K. Li, K. T. Lai, and K. Y. Fung, Application of genetic
algorithm for overcurrent relay coordination, in Proc. IEE Develop-
[1] M. Biserica, Y. Besanger, R. Caire, O. Chilard, and P. Deschamps, ments in Power System Protection Conf., 1997, pp. 6669.
Neural networks to improve distribution state estimation-volt var [21] C. W. So and K. K. Li, Time coordination method for power system
control performances, IEEE Trans. Smart Grid, vol. 3, no. 3, pp. protection by evolutionary algorithm, IEEE Trans. Ind. Appl., vol. 36,
11371144, Sep. 2012. no. 5, pp. 12351240, Sep./Oct. 2000.
[2] H. Yazdanpanahi, W. L. Yun, and W. Xu, A new control strategy to [22] P. P. Bedekar and S. R. Bhide, Optimum coordination of directional
mitigate the impact of inverter-based DGs on protection system, IEEE overcurrent relays using the hybrid GA-NLP approach, IEEE Trans.
Trans. Smart Grid, vol. 3, no. 3, pp. 14271436, Sep. 2012. Power Del., vol. 26, no. 1, pp. 109119, Jan. 2011.
[3] J. G. Slootweg and W. L. Kling, Impacts of distributed generation [23] M. M. Mansour, S. F. Mekhamer, and N. E.-S. El-Kharbawe, A mod-
on power system transient stability, in Proc. 2002 IEEE Power Engi- ified particle swarm optimizer for the coordination of directional over-
neering Society Summer Meeting, Jul. 2525, 2002, pp. 862867. current relays, IEEE Trans. Power Del., vol. 22, no. 3, pp. 14001410,
[4] J. Wong, P. Baroutis, R. Chadha, R. Iravani, M. Graovac, and X. Wang, Jul. 2007.
A methodology for evaluation of permissible depth of penetration of [24] Univ. Washington. Seattle, WA, Mar. 2006 [Online]. Available:
distributed generation in urban distribution systems, in Proc. 2008 http://www.ee.washington.edu/research/pstca/
IEEE Power and Energy Society General MeetingConversion and [25] R. Dugan, T. McDermott, and G. Ball, Planning for distributed gener-
Delivery of Electrical Energy in the 21st Century, Jul. 2024, 2008, ation, IEEE Ind. Appl. Mag., vol. 7, no. 2, pp. 8088, Mar./Apr. 2001.
pp. 18. [26] P. K. Nayak, A. K. Pradhan, and P. Bajpa, A fault detection technique
[5] L. F. Ochoa, A. Padilha-Feltrin, and G. P. Harrison, Evaluating dis- for the series-compensated line during power swing, IEEE Trans.
tributed generation impacts with a multiobjective index, IEEE Trans. Power Del., vol. 28, no. 2, pp. 714722, Apr. 2013.
Power Del., vol. 21, no. 3, pp. 14521458, Jul. 2006.
[6] R. A. Shayani and M. A. G. de Oliveira, Photovoltaic generation pen- H. H. Zeineldin (M06SM13) received the B.Sc.
etration limits in radial distribution systems, IEEE Trans. Power Syst., and M.Sc. degrees in electrical engineering from
vol. 26, no. 3, pp. 16251631, Aug. 2011. Cairo University, Cairo, Egypt, in 1999 and 2002,
[7] A. Bhowmik, A. Maitra, S. M. Halpin, and J. E. Schatz, Determina- respectively, and the Ph.D. degree in electrical
tion of allowable penetration levels of distributed generation resources and computer engineering from the University of
based on harmonic limit considerations, IEEE Trans. Power Del., vol. Waterloo, ON, Canada, in 2006.
18, no. 2, pp. 619624, Apr. 2003. He was with Smith and Andersen Electrical En-
[8] H. M. Ayres, W. Freitas, M. C. De Almeida, and L. C. P. Da Silva, gineering Inc., where he was involved with projects
Method for determining the maximum allowable penetration level involving distribution system design, protection, and
of distributed generation without steady-state voltage violations, IET distributed generation. He was then a Visiting Pro-
Gener., Transmiss. & Distrib., vol. 4, no. 4, pp. 495508, Apr. 2010. fessor at the Massachusetts Institute of Technology
[9] N. Nimpitiwan, G. T. Heydt, R. Ayyanar, and S. Suryanarayanan, (MIT), Cambridge. He is currently an Associate Professor with the Masdar In-
Fault current contribution from synchronous machine and inverter stitute of Science and Technology, Abu Dhabi, UAE, and a faculty member in
based distributed generators, IEEE Trans. Power Del., vol. 22, no. 1, the Electrical Power Department, Cairo University. His research interests in-
pp. 634641, Jan. 2007. clude power system protection, distributed generation, and micro-grids.
1532 IEEE TRANSACTIONS ON SMART GRID, VOL. 4, NO. 3, SEPTEMBER 2013

Yasser Abdel-Rady I. Mohamed (M06SM11) been an Assistant Professor at Masdar Institute of Science and Technology, Abu
was born in Cairo, Egypt, on November 25, 1977. Dhabi, UAE. From April 2010 to December 2010, he was a visiting faculty at
He received the B.Sc. (Hons.) and M.Sc. degrees in Massachusetts Institute of Technology, Cambridge, MA. His research interests
electrical engineering from Ain Shams University, include applications of power electronics in distribution systems and renewable
Cairo, in 2000 and 2004, respectively, and the Ph.D. energy resources, grid interconnection issues, power quality enhancement, ac-
degree in electrical engineering from the University tive power filters, and electric vehicles.
of Waterloo, Waterloo, ON, Canada, in 2008.
He is currently an Associate Professor with the
Department of Electrical and Computer Engineering,
University of Alberta, AB, Canada. His research V. Ravikumar Pandi (M10) received the B.E.,
interests include dynamics and controls of power degree from Madurai Kamaraj University, Tamil
converters; microgrids; distributed and renewable generation; modeling, Nadu, India, in 2003, the M. Tech., degree from An-
analysis and control of smart grids; electric machines and motor drives. namalai University, Tamil Nadu, India, in 2005, and
Dr. Mohamed is an Associate Editor of the IEEE TRANSACTIONS ON the Ph.D. degree from Indian Institute of Technology
INDUSTRIAL ELECTRONICS and a Guest Editor of the IEEE TRANSACTIONS Delhi, New Delhi, India, in 2010.
ON INDUSTRIAL ELECTRONICS (Special Section on Distributed Generation Currently he is a Postdoctoral Researcher with the
and Micro-grids). He is a registered Professional Engineer in the province of Masdar Institute of Science and Technology, Abu
Alberta. His biography is listed in Marques Whos Who in the World. Dhabi, UAE. His research interest includes power
system optimization, power quality, protection,
transmission pricing, artificial intelligence, and
evolutionary algorithms.
Vinod Khadkikar (S06M09) received the B.E.
degree from the Government College of Engineering,
Dr. Babasaheb Ambedkar Marathwada University,
Aurangabad, India, in 2000, the M. Tech. degree
from the Indian Institute of Technology (IITD), New
Delhi, India, in 2002, and the Ph.D. degree from the
cole de Technologie Suprieure (E.T.S.), Montral,
QC, Canada, in 2008, all in electrical engineering.
From December 2008 to March 2010, he was a
Postdoctoral Fellow at the University of Western On-
tario, London, ON, Canada. Since April 2010, he has