You are on page 1of 7

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.2016.2533565, IEEE
Transactions on Power Delivery

Protection Coordination Index Enhancement


Considering Multiple DG Locations Using FCL
Ł. Huchel, H. H. Zeineldin, Senior Member, IEEE and Ehab F. El-Saadany, Senior Member, IEEE

Abstract—Increasing the penetration level of distributed the time dial and pickup current setting, are decision variables
generation (DG) may lead to improper protection coordination. of the problem. Furthermore, in [8], a mixed integer non-
Previous work introduced a Protection Coordination Index (PCI) linear programming (MINLP) formulation has been proposed
that can quantify the impact of installing a DG at individual where the pickup current setting is considered as a discrete
locations. This paper extends on previous work and proposes a variable. Based on the formulation, several optimization tech-
generalized approach for calculating PCI considering multiple
DG locations simultaneously. Furthermore, the paper proposes
niques (both deterministic and heuristic methods) are utilized
a solution for enhancing the PCI using fault current limiters to solve the protection coordination problem. Such techniques
(FCL). The problem is formulated as a mixed integer non-linear include, two-phase simplex [9], sequential quadratic problem
programming (MINLP) problem to determine the optimal FCL [7], genetic algorithm [10], particle swarm optimization [11]
size and location as well as enhanced PCI. and evolutionary algorithm [12].
Optimal location of DG units and their respective impact
Keywords—Protection Coordination, Distributed Generation, on the protection scheme is presented in [13]. In [2], a
Genetic Algorithm, Penetration Level. protection coordination index (PCI) is proposed to quantify
the impact of installing a DG at a specific location on the
protection coordination. The approach, presented in [2], only
I. I NTRODUCTION considers single DG installations, whereas planning studies
AXIMUM penetration level of distributed generation need to address scenarios involving the installation of DG
M (DG) in a system is limited by several factors such
as system ratings, power quality standards, and the designed
units at multiple locations simultaneously. Furthermore, the
analysis presented in [2] reveals that the maximum DG
protection schemes. Typically, protection schemes are de- penetration at a specific location is affected by the capacity
signed based on the conventional radial system structure of DG installed at other locations. Accordingly, the PCI
where the power flows from the main substation to the loads index should be calculated considering a more generalized
in one direction [1]. Therefore, protection schemes based scenario with multiple DG installations. Application of fault
on overcurrent relays (OCRs) are utilized. One of the main current limiters (FCL) may be proposed to limit the fault
features that should be maintained by any successful protec- current contribution of DG. With fixed settings of protection
tion scheme is its selectivity. This is commonly known as devices, the installation of FCL limits the magnitude of the
protection coordination, which states that all the primary and fault current and can allow for further DG installation. In
backup protective devices are coordinated within a specific the literature, the influence of FCL on the performance of
time interval [2]. However, the introduction of DG units to protection devices has been discussed thoroughly. In [14] FCL
radial systems jeopardizes the performance of the OCRs. is utilized to minimize the short circuit current, consequently,
reducing the influence on protection coordination in a system.
The integration of DG units in radial systems changes In [15], the reactance value is calculated, to minimize the total
the magnitude and direction of the line fault currents. More- operating time of protection relays with respect to protection
over, due to the increased popularity of DG integration to coordination constraints. Similarly, in [16], the authors pro-
distribution systems and the recent drive towards smart grids, pose the application of FCL to a distribution system equipped
distribution systems are becoming more of a meshed structure with DOCRs to accommodate increasing DG penetration. The
[3]. Unlike radial systems, meshed systems cannot be pro- approach presented in [16] optimizes the size of FCL to
tected solely by OCRs due to the bi-directional flow of fault minimize the contribution of DG to the fault current. Fault
currents. Commonly, directional overcurrent relays (DOCR) current analysis for various types of DG is performed in
are utilized for protecting meshed distribution systems [4]. [17]-[19]. Synchronous generators have been shown to have
Several protection coordination problem formulations significant influence on fault currents and consequently on
have been proposed in the literature. In [5] and [6], the the performance of protection devices [17], [18] and [19].
problem is formulated as a linear programming (LP) problem Consequently, all of the DG units applied in the presented
by pre-setting the pickup current settings. Moreover, in [7], work are synchronous based.
the protection coordination problem is formulated as a non- This paper expands the proposed algorithm presented
linear programming (NLP) problem where the relay settings, in [2], by proposing a modified problem formulation for
determining the PCI considering the installation of DG units
This work was supported and funded by the Masdar Institute of Science at multiple locations simultaneously. New constraints are
and Technology.
Ł. Huchel and H. H. Zeineldin are with The iEnergy Center and the proposed that take into account various combinations of DG
Electrical Engineering and Computer Science Department, Masdar Institute capacities while determining the PCI. In addition, an optimal
of Science and Technology, Abu Dhabi, UAE. H. H. Zeineldin is currently on solution that relies on the installation of magnetic FCL is pre-
leave from the faculty of Engineering, Cairo University, Giza 12316, Egypt. sented to enhance the PCI levels at candidate locations. The
Ehab F. El-Saadany is with the ECE Department, University of Waterloo
and currently on leave with the PI, UAE (E-mails: huchellukasz@gmail.com,
proposed algorithm guarantees proper protection coordination
hzainaldin@masdar.ac.ae and ehab@waterloo.ca). for different system scenarios where the DG capacities can

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

vary, at the various locations, up to the maximum possible where i and j represent the fault locations and the relay iden-
capacity. The problem is formulated as a MINLP and is tifiers, respectively. Ipj denotes the pick-up current setting
solved using genetic algorithm. The approach is tested on of relay j, while ISCij is the short circuit current passing
the distribution part of the IEEE 14-bus system. through relay j for fault location i and is a function of DG
capacity (SDG ) and reactance of the limiter (XF CL ). A and
II. M ODIFIED F ORMULATION FOR C ALCULATING THE B are constants of particular type of time curve, T DSj is
M AXIMUM DG P ENETRATION C ONSIDERING M ULTIPLE a relay j time dial setting. Constraints on the relay settings,
DG L OCATIONS relay operating time and protection coordination constraints
The PCI is defined as the rate of change of DG penetration are included and can be found in [2].
levels with respect to the rate of change of CTI. The PCI index
at each bus may be expressed as below: A. Constraints to Account for Multiple DG Locations
ΔP One major challenge, while formulating the optimization
P CI = − (1) model for maximizing the DG penetration at the predefined
ΔCT I
where ΔP represents the DG penetration level that can be locations, is the inclusion of constraints that account for mul-
achieved at a particular location and ΔCT I is the corre- tiple DG simultaneously. The reason is that while maximizing
sponding decrease in coordination time interval (CTI). By the DG at the various locations, the solution must guarantee
referring to (1), PCI indicates the decrease in the coordination that, for any possible DG capacity combinations (below the
time interval as a result of the interconnection of a DG at a maximum optimal DG capacity), the protection system is
specific bus. The amount of CTI decrease will depend on the coordinated. Thus, the protection coordination problem has to
DG capacity to be installed. DGs installed at buses with high be optimized for all possible sets of DG capacities between
PCI values will have less impact on protection coordination 0 and SDGi . In order to do so, a new variable v is defined
compared to DGs installed at busses with lower PCI values. which corresponds to the number of protection coordination
The negative sign in (1) compensates the negative sign in violations attained for each particular combination set of DG
ΔCT I which decreases as capacity increases. capacities and should be constrained as follows:
As mentioned earlier, the formulation provided in [2], N umberof violations = v = 0 (4)
for calculating the maximum DG penetration, focuses on
individual locations within a system. Thus, the maximum
DG penetration is determined at each bus without taking V=0
into account the effect of allocating DG at other locations. for a=0:step
. size:SDG1
As shown in [2], interconnecting a DG at a specific lo- .
cation can impact the amount of DG capacity that can be .
installed at another candidate location considering protection for a=0:step size:SDGk
coordination. For example, if the maximum penetration level
that will affect the CTI by 0.01 seconds at two individual Create a Ybus(SDGk,XFCL);
buses is determined to be 5 and 10 MVA, installing both Perform Fault Analysis;
simultaneously will result in a higher decrease in the CTI.
This will be highlighted in Section IV (refer to Table I). Due if t pijab -t bijab ≤ CTI
to the non-linearity of the problem, it is worthy to note that V=V+1;
maximizing the DG penetration at multiple candidate loca-
end
tions simultaneously does not guarantee optimality for other
combinations of DG sizes. The aforementioned problems will end
be illustrated more through various case studies presented
end
in Section IV. Thus, new constraints need to be added to
account for the simultaneous influence of all candidate DG Fig. 1: Pseudo code representation for violation constraint
locations. Furthermore, additional constraints are added to
take into account optimal sizing and allocation of FCL for The pseudo code developed in Fig. 1 determines the num-
maximizing the DG penetration and consequently PCI. In ber of possible violations. The step size represents the relative
order to calculate the PCI, the maximum DG penetration change in DG capacity. For each DG set, the impedance
is maximized at all candidate locations simultaneously as matrix is calculated and fault analysis is performed. Finally,
follows: the relay operating times are determined and the protection
K
coordination constraint is checked for violations and v is
M aximize P = SDGl (2)
incremented with every violation detected. The optimization
l=1
algorithm will optimally determine SDGi such that constraint
where l represents the DG candidate location, K is the set (4) is satisfied.
of all locations and SDGi is the DG i capacity. The tripping
time of DOCRs are usually governed by an inverse function
of the fault current. In this study, the inverse time-current B. Constraints to Account for FCL Sizing and Allocation
relay characteristic used by all the DOCRs follows the IEC The algorithm determines the optimal size and location
255-3 standard as presented in [2]. of FCL to maximize the DG penetration and consequently
A the PCI. The FCL will have an impact on the impedance
tij (SDG , XF CL ) = T DSj  B (3) matrix, the fault current level as given in (3) and the number
ISCij (SDG ,XF CL )
Ipj − 1 of violations as given in Fig. 1. In this study, the FCLs
0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

9
introduced to the system are limited by an upper bound set
R2 R5
to 1.5 as shown below:
10 R1 R3 R6 8

0 ≤ XFk,k
CL ≤ 1.5Ω ∀k, k
 F1 F3
(5) 11 R7
R16 F2 R12
 12
XFk,k
R11
where CLis the reactance value of the limiter installed F8 R13
F6
between buses k and k  next to bus k.
F7 F4
R10
R4 F5
Moreover, the utilization of FCL increases the impedances R15
R14
G
of the systems lines. This may influence the steady state 2
R9 R8
6
voltage level and cause violation of the grid codes limits 7
[20]. Thus, power flow analysis is performed to be able to 5
introduce another set of constraints that limit the voltage at 4
each bus within the acceptable limits. This set of constraints
is represented as follows according to the EN 50160 standard:
1

0.9 ≤ Vl ≤ 1.1 ∀l (6) G

where Vl is the steady state voltage level in bus l. 3


Fig. 2: Single line diagram of 14-bus system for protection coordination
III. SYSTEM AND SIMULATION SETUP studies.

This section describes the test system under study and the
developed algorithm. The first subsection presents in detail the
distribution part of the modified IEEE 14-bus system as well One major challenge in the proposed formulation is the depen-
as the various case studies. The latter subsection gives insight dence of the Ybus on variables (discrete and continuous) such
into the algorithm structure and the utilized solvers. as the DG capacity, FCL size and location. The Zbus (inverse
of the Ybus ) needs to be calculated to perform the fault
analysis. It is worthy to note that developing mathematical
A. Description of the Test System and Case Studies expressions for the Zbus is complex, similarly, due to its
All simulations are performed on the distribution part of dependence on the decision variables. Thus, genetic algorithm
the modified IEEE 14-bus system. The single line diagram (GA) is utilized for solving the proposed optimization model
is given in Fig. 2 . System is supplied by the generators due to its capability of solving such complex and non-linear
connected to the buses 1 and 2. The distribution part of the problems. The main steps included in the algorithm are given
system is fed through two transformers connected at bus 6 as follows:
and 7. Detailed data of the system with the connected loads 1. The relay settings (TDS and Ip), line data (given in [2] and
are given in [21]. In this study, all of the considered DG units [21]) as well as step size and CTI decrease are inputted as
are synchronous based with 0.0967p.u sub-transient reactance a parameters.
proportional to the installed capacity. Moreover, the installed 2. Initialize SDGi , XF CL and FCL location.
DG units are connected to the grid through step-up transform- 3. Perform steps defined in the pseudo code (Fig. 1).
ers 0.48kV/33kV with 0.05p.u sub-transient reactance. This 4. Perform the load flow analysis and determine the steady
paper focuses on synchronous-based DG units only since the state voltage level.
contribution to short circuit current of inverted-based DG is
not significant (due to the existing control schemes which 5. Calculate the value of the objective function and fitness
limit the amount of fault current). Furthermore, the system function.
is equipped with 16 DOCRs denoted by R1-R16. Bolted 6. If any of the stopping criteria is satisfied calculate the PCI
symmetrical fault analysis are conducted at the midpoint of values. If not, perform selection, mutation and reproduc-
each line in the distribution system. Fault locations are marked tion of previous population in order to obtain new one and
in Fig. 2 as F1-F8. The base power is equal to 100MVA while go back to the point 3.
the base voltage is 33kV. Fig. 3 presents the flowchart of the proposed algorithm.
In this paper, three case studies have been conducted to In this paper, MATLAB GA function is utilized. Typically,
validate the proposed algorithm. Basically, case 1 is based on in the GA, the decision variables of the problem are defined
the approach, proposed in [2], where the PCI is determined as chromosomes in which each consists of an encoded binary
for an individual DG candidate location. Moreover, case 2 string. An initial population is generated consisting of a
determines the PCI value for multiple candidate DG locations group of randomly initialized chromosomes. In this paper,
simultaneously. Finally, case 3 solves the PCI optimization the population size for both optimization models are assumed
problem while taking into account the installation of FCLs in to be 6500 individuals. The fitness of each chromosome in
the system. the population is evaluated based on the objective function
(maximizing the DG capacity at pre-specified locations). After
assessing the chromosome in the current population, the
B. Description of the Proposed Algorithm
next generation is produced by applying the mutation and
Conventionally, in protection coordination studies, the crossover on some of the chromosomes with the best fitness
admittance matrix (Ybus ) is a parameter that is used for deter- value. This process is repeated until any of the pre-defined
mining the fault currents. The problem presented in Section II stopping criteria are reached. In this paper, the algorithm is
is a mixed integer non-linear programming (MINLP) problem terminated when either a relative change in the fitness value
in which the pseudo code presented in Fig. 1 is incorporated. is lower than the pre-specified tolerance, or the number of
0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

Input relay settings, step the amount of reduction in the PCI value of bus 8 is 100%
size and CTI
with respect to case 1. Thus, it is clear that the amount of
reduction in the DG penetration level is dependent on the DG
Initialize S DG, FCL locations considered.
size and location

TABLE I: OPTIMAL PCI VALUES FOR SCENARIOS WITH TWO PRE-


DEFINED DG LOCATIONS
Follow steps in Fig. 1.
Case study PCI
Bus 6/12 Bus 10/8 Bus 11/9
case 1 64.38/17.75 17.5/13.25 30.25/7.75
Selection, mutation and reproduction
Perform load flow analysis case 2 64/1.5 17.5/0 27.73/5.75
of previous population

Solve the problem using GA In order to enhance the optimal PCI values obtained
from the proposed method (case 2), the problem is solved
considering FCL (case 3). In this study, one FCL is introduced
to the system. The location and size of this FCL are among
No
Any of the stopping the decision variables of this problem. Table II presents
criteria satisfied? the optimal PCI values that are obtained from the proposed
Yes
method without FCL (case 2) and the proposed method with
FCL (case 3) for three pre-defined DG locations. In addition,
Calculate PCI values the optimal size and location of the FCL obtained in case
Fig. 3: Flowchart of the proposed algorithm 3 are provided in Table II. The first scenario presented in
Table II considers buses 7, 8, and 10 as the candidate DG
locations. As can be seen, the PCI value at bus 7 increases
generations is exceeded. by 140% with respect to case 2. Moreover, it is clear that
in case 2, no DG units can be installed on buses 8 and 10
(PCI value equal to zero). Yet, by introducing a 0.097 p.u
IV. OPTIMAL PROTECTION COORDINATION FCL (case 3), it is possible to install DG units at buses 8 and
INDEX FOR MULTIPLE DG LOCATIONS 10 with capacities corresponding to the PCI values of 3.75
CONSIDERING FCL and 20.62, respectively. Hence, from the results, it is clear
This section presents the obtained results and underlines that by introducing FCLs, the PCI values at each location are
the improvements achieved by applying the proposed algo- increased significantly compared to case 2. Similar observa-
rithm. The optimal PCI values obtained with and without tions are obtained for the scenarios considering DG locations
FCL are presented. The PCI is defined as the rate of change at busses 6, 9, 11 and 6, 8, 12, respectively (Table II).
of DG penetration with respect to decrease in CTI [2]. The
latter subsection, illustrates the relationship between the DG TABLE II: OPTIMAL PCI VALUES, FCL SIZE AND LOCATION FOR
penetration level at each location and the number of installed SCENARIOS WITH THREE PRE-DEFINED DG LOCATIONS- CASE 2
AND CASE 3
FCLs.
Case 2 Case 3
Scenario FCL’s
PCI PCI
A. Maximizing the DG Penetration Level without FCL React., [p.u] Location
Bus:7/8/10 47.37/0/0 0.097 6-11 113.75/3.75/20.62
Table I presents the optimal PCI values that are obtained Bus:6/9/11 61.49/0/3.62 0.139 7-8 95.12/4.5/27.37
for case 1 and case 2 (generalized approach). By maximizing Bus:6/8/12 64/0/1.52 0.137 7-12 104.87/22/3.87
DG capacity at individual locations (case 1), the PCI values
at bus 6 and 12 are 64.38 and 17.75. In terms of DG capacity,
these values correspond to 0.6438 MW and 0.1775 MW,
B. Influence of Different Generation Conditions on Protection
respectively, for a 0.01 decrease in CTI. Thus, up to 0.6438
Coordination
MW can be installed at bus 6 assuming that there is no DG
to be installed at bus 12. Similarly, up to 0.1775 MW can Although, the introduction of FCL to the system has
be installed at bus 12 assuming that there is no DG to be improved significantly the PCI value of each pre-defined
installed at bus 6. For case 2, the PCI for bus 6 and 12 are 64 DG location when considered simultaneously, the problem
and 1.5, respectively, and similarly for a 0.01 decrease in CTI formulation does not take into account the different generation
this corresponds to 0.64 MW and 0.015 MW, respectively. In condition of each DG unit. For example, with a 0.097 p.u FCL
this case, up to 0.64 MW and 0.015 MW can be connected at located in line 6-11 and with DG capacity of 9.1MW, 0.3MW
busses 6 and 12 simultaneously. It can be seen that the optimal and 1.65MW at buses 7, 8 and 10 respectively. The protection
PCI values obtained with the proposed method (case 2) are coordination will not hold for any DG combinations that
lower than in case 1. This means that the DG penetration are below 9.1MW, 0.3MW and 1.65MW. There will exist
level decreases when both locations are considered at once. protection coordination violations. To further generalize the
Moreover, as can be seen, from Table I, for buses 6 and 12, solution, the various DG combinations should be taken into
the amount of reduction in PCI in bus 6 is 0.59% with respect account and protection coordination should be satisfied for all
to case 1 while for bus 12 it is 91.5%. On the other hand, combinations. In practice, a DG unit can be either connected
it can be observed that for buses 8 and 10, the PCI value of or disconnected due to transient changes such as, fault isola-
bus 10 is the same for both cases (case 1 and case 2) while tion actions or maintenance activities. Furthermore, depending
0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

on the utility’s preference, different DG capacities at each imization problem is solved using the proposed formulation.
pre-defined location can be installed. Thus, it is necessary to In order to investigate the influence of adding multiple FCLs
check the protection coordination of all the DOCRs pairs, for on the PCI value at each pre-defined DG locations, three
all possible combinations of DG capacities installed at each scenarios are considered. Tables V, VI, and VII present the
pre-specified location between zero and the corresponding three scenarios in which each consider three pre-defined DG
optimal capacity (determined in case 3). In this study, the locations; namely buses 6, 8 and 12, 6, 9 and 12, and 6, 10
step size is set to 20% of the respective maximum optimal and 11, respectively. Each table provides the optimal FCLs
DG capacity. Table III provides the number of protection sizes and locations, for each additional FCL, corresponding to
coordination violations that occur due to the consideration of each scenario, respectively. Moreover, Fig. 4, 5 and 6 illustrate
different generation conditions. The violations shown in this the relationship between the optimal PCI value, for each pre-
table correspond to the three scenarios previously presented defined DG location, and the number of installed FCLs in
in Table II. It is clear that the number of violations depends each scenario, respectively.
on the examined DG locations and their respective optimal
capacities. For three DG locations and 22 back-up schemes TABLE V: OPTIMAL FCLS SIZES AND LOCATIONS FOR EACH NUM-
the maximum number of violations is equal to 4752. BER OF FCLS IN A SCENARIO WITH BUSES: 6, 8 AND 12

Buses: 6-8-12
TABLE III: PROTECTION COORDINATION VIOLATIONS DUE TO No. of limiters
FCL 1 2 3 4
DIFFERENT GENERATION CONDITIONS
FCL on line: 7-12 - - -
1 Reactance,(p.u) 0.100 - - -
Scenario: No. of violations
FCL on line: 7-12 7-8 - -
Bus:7/8/10 7 2 Reactance,(p.u) 0.099 0.109 - -
Bus:6/9/11 6 FCL on line: 7-12 7-8 9-6 -
Bus:6/8/12 108 3 Reactance,(p.u) 0.076 0.135 0.021 -
FCL on line: 7-12 7-8 9-6 11-6
4 Reactance,(p.u) 0.063 0.139 0.0191 0.023
In order to eliminate these violations, equation 4 and
pseudo code is introduced into the proposed optimization
formulation (refer to subsection II.A). The same scenarios
that are presented in Table II are examined again using the
proposed approach. Table IV presents the optimal PCI values Bus 6 Bus 8 Bus 12
that are obtained from the proposed method considering 150
different generation conditions for the three DG locations
100
PCI

TABLE IV: OPTIMAL PCI VALUES, FCL SIZE AND LOCATION FOR
SCENARIOS WITH THREE PRE-DEFINED DG LOCATIONS- MODI-
FIED CASE 3 50

Case 3
Scenario FCL’s 0
React., [p.u] Location
PCI 0 1 2 3 4
Number of Limiters
Bus:7/8/10 0.131 9-8 81.5/0.75/0
Bus:6/9/11 0.094 7-12 99.62/0.25/3.72 Fig. 4: Optimal PCI value corresponding to the number of installed FCLs
Bus:6/8/12 0.1 7-12 106.87/0.06/1.73 for buses 6, 8 and 12

In comparison with Table II the sum of optimal PCI


values in each scenario attained in Table IV is lower than TABLE VI: OPTIMAL FCLS SIZES AND LOCATIONS FOR EACH
its corresponding PCI values in Table II. Yet, the individual NUMBER OF FCLS IN A SCENARIO WITH BUSES: 6, 9 AND 12
PCI value at each DG location can either decrease (refer to Buses: 6-9-12
Table II-case 3 and Table IV, Scenario 6, 9 and 11, bus 9) or No. of limiters
FCL 1 2 3 4
increase (Scenario 6, 8 and 12, bus 6). The amount of change FCL on line: 7-12 - - -
1 Reactance,(p.u) 0.091 - - -
in each PCI depends on the pre-specified DG location as well
FCL on line: 7-12 7-8 - -
as mutual influence of other DG locations. Moreover, different 2 Reactance,(p.u) 0.097 0.101 - -
optimal FCL sizes are obtained (Table IV) in each scenario FCL on line: 7-12 7-8 9-6 -
3 Reactance,(p.u) 0.07 0.138 0.0187 -
with respect to their corresponding values in Table II-case 3.
FCL on line: 7-12 7-8 9-6 11-12
In addition, the optimal FCL location can either remain the 4 Reactance,(p.u) 0.061 0.139 0.0179 0.019
same (refer to scenario bus 6/8/12 in Table II-case 3 and Table
IV) or change (refer to the other two scenarios) depending
From the presented optimal FCLs sizes and locations,
on the investigated pre-defined DG locations. Although, the
PCI values at each DG location is reduced by introducing the two main observations can be deduced. The first observation
additional set of constraints to the proposed approach, proper is that the optimal location of each installed FCL does not
protection coordination is guaranteed for different generation change by the addition of another FCL to the system. For
conditions. example, in Table V (Scenario with buses 6,8 and 12), the
optimal location for installing the first FCL is always on the
line 7-12, irrespective to the addition of other FCLs. Similar
C. Optimal Protection Coordination Index Considering Mul-
observations are inferred from other scenarios presented in
tiple FCLs
Tables VI and Table VII. The second observation is that
In this subsection the influence of multiple FCLs on the from Tables VI and Table VII, it can be seen that there are
PCI value is investigated. In a similar manner, the PCI max- common optimal FCL locations for all examined scenarios.
0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

case. The PCI values tend to decrease at a specific location


Bus 6 Bus 9 Bus 12 when other DG locations are simultaneously considered. The
150 proposed algorithm determined the optimal size and location
of FCL that can enhance the PCI values. The results show
100 that the proposed algorithm is capable of eliminating any
violations in protection coordination while achieving higher
PCI

PCI values. Furthermore, it is noticed that among the inves-


50
tigated scenarios, there are some specific locations where the
installation of FCLs will improve the DG penetration level
0 irrespective of the examined DG locations.
0 1 2 3 4
Number of Limiters
Fig. 5: Optimal PCI value corresponding to the number of installed FCLs
for buses 6, 9 and 12 R EFERENCES
[1] H.-C. Jo, S.-K. Joo, and K. Lee, “Optimal placement of superconduct-
TABLE VII: OPTIMAL FCLS SIZES AND LOCATIONS FOR EACH ing fault current limiters (sfcls) for protection of an electric power
NUMBER OF FCLS IN A SCENARIO WITH BUSES: 6, 10 AND 11 system with distributed generations (dgs),” Applied Superconductivity,
IEEE Transactions on, vol. 23, no. 3, pp. 5 600 304–5 600 304, June
Buses: 6-10-11 2013.
No. of limiters
FCL 1 2 3 4 [2] H. Zeineldin, Y.-R. Mohamed, V. Khadkikar, and V. Pandi, “A protec-
FCL on line: 7-12 - - - tion coordination index for evaluating distributed generation impacts
1 Reactance,(p.u) 0.086 - - - on protection for meshed distribution systems,” Smart Grid, IEEE
FCL on line: 7-12 7-8 - -
2 Transactions on, vol. 4, no. 3, pp. 1523–1532, Sept 2013.
Reactance,(p.u) 0.07 0.041 - -
FCL on line: 7-12 7-8 11-12 - [3] C. Gandioli, M. Alvarez-Herault, P. Tixador, N. Hadjsaid, and D. Med-
3 Reactance,(p.u) 0.048 0.107 0.053 - ina, “Innovative distribution networks planning integrating super-
FCL on line: 7-12 7-8 11-12 9-8 conducting fault current limiters,” Applied Superconductivity, IEEE
4 Reactance,(p.u) 0.054 0.091 0.078 0.072
Transactions on, vol. 23, no. 3, pp. 5 603 904–5 603 904, June 2013.
[4] W. El-Khattam and T. Sidhu, “Resolving the impact of distributed
renewable generation on directional overcurrent relay coordination: a
Bus 6 Bus 10 Bus 11 case study,” Renewable Power Generation, IET, vol. 3, no. 4, pp. 415–
425, December 2009.
150
[5] H. Askarian, Abyaneh, M. Al-Dabbagh, H. Kazemi, Karegar,
S. Hesameddin, Hossein Sadeghi, and R. Abul Jabbar Khan, “A new
100 optimal approach for coordination of overcurrent relays in intercon-
PCI

nected power systems,” Power Engineering Review, IEEE, vol. 22,


no. 6, pp. 60–60, June 2002.
50
[6] A. Noghabi, H. Mashhadi, and J. Sadeh, “Optimal coordination of
directional overcurrent relays considering different network topologies
0 using interval linear programming,” Power Delivery, IEEE Transac-
0 1 2 3 4 tions on, vol. 25, no. 3, pp. 1348–1354, July 2010.
Number of Limiters
Fig. 6: Optimal PCI value corresponding to the number of installed FCLs [7] D. Birla, R. Maheshwari, and H. Gupta, “A new nonlinear directional
for buses 6, 10 and 11 overcurrent relay coordination technique, and banes and boons of near-
end faults based approach,” Power Delivery, IEEE Transactions on,
vol. 21, no. 3, pp. 1176–1182, July 2006.
[8] A. Urdaneta, R. Nadira, and L. Perez Jimenez, “Optimal coordination
These optimal FCL locations are lines 7-8, and 7-12. This of directional overcurrent relays in interconnected power systems,”
means that for this test system there are optimal FCL locations Power Delivery, IEEE Transactions on, vol. 3, no. 3, pp. 903–911, Jul
in which the PCI values are improved regardless of the pre- 1988.
defined DG locations. From Fig. 4 (Scenario with buses 6, 8 [9] B. Chattopadhyay, M. Sachdev, and T. Sidhu, “An on-line relay coor-
and 12), the PCI value for bus 6 increased by 66% when one dination algorithm for adaptive protection using linear programming
FCL is introduced on the line 7-12, while the installation of technique,” Power Delivery, IEEE Transactions on, vol. 11, no. 1, pp.
165–173, Jan 1996.
the second FCL on the line 7-8, caused an additional increased
in the PCI value by only 8%. By introducing the third and [10] R. Chabanloo, H. Abyaneh, S. Kamangar, and F. Razavi, “A new
fourth FCL, the relative improvement in PCI value at bus 6 genetic algorithm method for optimal coordination of overcurrent
and distance relays considering various characteristics for overcurrent
reduces to 1.3% and 1.07%, respectively. However, the PCI relays,” in Power and Energy Conference, 2008. PECon 2008. IEEE
value at buses 8 and 12 is improved significantly when two 2nd International, Dec 2008, pp. 569–573.
FCLs are installed. Thus, there is an optimal number of FCL [11] M. M. Mansour, S. Mekhamer, and N.-S. El-Kharbawe, “A modified
beyond which minimal improvement in PCI can be achieved. particle swarm optimizer for the coordination of directional overcurrent
Similar observations are deduced for the other scenarios from relays,” Power Delivery, IEEE Transactions on, vol. 22, no. 3, pp.
Fig. 5 and Fig. 6. 1400–1410, July 2007.
[12] C. W. So and K. K. Li, “Time coordination method for power system
protection by evolutionary algorithm,” Industry Applications, IEEE
V. CONCLUSIONS Transactions on, vol. 36, no. 5, pp. 1235–1240, Sep 2000.

This paper proposes an algorithm for enhancing the PCI [13] J. Wong, P. Baroutis, R. Chadha, R. Iravani, M. Graovac, and X. Wang,
“A methodology for evaluation of permissible depth of penetration
values considering multiple DG locations. In addition to the of distributed generation in urban distribution systems,” in Power
conventional protection coordination constraints, an additional and Energy Society General Meeting - Conversion and Delivery of
algorithm is incorporated to account for the multiple DG Electrical Energy in the 21st Century, 2008 IEEE, July 2008, pp. 1–8.

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for 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.2016.2533565, IEEE
Transactions on Power Delivery

[14] R. Chabanloo, H. Abyaneh, A. Agheli, and H. Rastegar, “Overcurrent Ehab F. El-Saadany (SM05) was born in Cairo,
relays coordination considering transient behaviour of fault current Egypt, in 1964. He received the B.Sc. and M.Sc.
limiter and distributed generation in distribution power network,” degrees in electrical engineering from Ain Shams
Generation, Transmission Distribution, IET, vol. 5, no. 9, pp. 903– University, Cairo, Egypt, in 1986 and 1990, respec-
911, September 2011. tively, and the Ph.D. degree in electrical engineering
from the University of Waterloo, Waterloo, ON,
[15] W. Najy, H. Zeineldin, and W. Woon, “Optimal protection coordination
Canada, in 1998. Currently, he is a Professor with
for microgrids with grid-connected and islanded capability,” Industrial
the Department of Electrical and Computer Engi-
Electronics, IEEE Transactions on, vol. 60, no. 4, pp. 1668–1677,
neering, University ofWaterloo. His current research
April 2013.
interests include distribution system operation and
[16] W. El-Khattam and T. Sidhu, “Resolving the impact of distributed control, distributed generation, smart-grid applica-
renewable generation on directional overcurrent relay coordination: a tions, self-healing mechanisms, power quality, and microelectromechanical-
case study,” Renewable Power Generation, IET, vol. 3, no. 4, pp. 415– systems micropower generators. Dr. El-Saadany is the holder of the Canada
425, December 2009. Research Chair in Smart Distribution Systems in recognition of his contri-
butions to the areas of distributed generation and microgrids.
[17] J. Keller and B. Kroposki, “Understanding fault characteristics of
inverter-based distributed energy sources,” National Renewable Energy
Laboratory,Golden, CO, 2010.
[18] N. Nimpitiwan, G. Heydt, R. Ayyanar, and S. Suryanarayanan, “Fault
current contribution from synchronous machine and inverter based
distributed generators,” Power Delivery, IEEE Transactions on, vol. 22,
no. 1, pp. 634–641, Jan 2007.
[19] H. Yazdanpanahi, W. Xu, and Y. W. Li, “A novel fault current control
scheme to reduce synchronous dg’s impact on protection coordination,”
Power Delivery, IEEE Transactions on, vol. 29, no. 2, pp. 542–551,
April 2014.
[20] “Ieee guide for identifying and improving voltage quality in power
systems,” IEEE Std 1250-2011 (Revision of IEEE Std 1250-1995), pp.
1–70, March 2011.
[21] Univ. washington. seattle, wa, mar. 2006. [Online]. Available:
http://www.ee.washington.edu/research/pstca/

Ł. Huchel received the B.Sc. degree in Electrical


Power Engineering from the Silesian University of
Technology in Gliwice, Poland, in 2013. In 2015
he graduated from the Electrical Engineering and
Computer Science department in Masdar Institute
of Science and Technology. He is currently working
on power system protection with penetration of
distributed generation. His current interests include
power system protection and stability, application
of optimization techniques to the problems related
to power system operation and control.

H. H. Zeineldin (M06SM13) received the B.Sc.


and M.Sc. degrees in electrical engineering from
Cairo University, Cairo, Egypt, in 1999 and 2002,
respectively, and the Ph.D. degree in electrical and
computer engineering from the University of Water-
loo, Waterloo, ON, Canada, in 2006. He was with
Smith and Andersen Electrical Engineering, Inc.,
where he was involved with projects involving dis-
tribution system design, protection, and distributed
generation. He was then a Visiting Professor at the
Massachusetts Institute of Technology, Cambridge,
MA, USA. He is with the Faculty of Engineering, Cairo University, and
currently he is a Professor with the Masdar Institute of Science and Tech-
nology,Masdar City, UAE. He is currently an Editor of the IEEE TRANS-
ACTIONS ON ENERGY CONVERSION and IEEE TRANSACTIONS ON
SMART GRIDS. His current interests include power system protection,
distributed generation, and deregulation.

0885-8977 (c) 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

You might also like