Electric Power Systems Research 186 (2020) 106417

Contents lists available at ScienceDirect

Electric Power Systems Research

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

Mitigation of DGs Impact on Variable-Topology Meshed Network Protection T

System by Optimal Fault Current Limiters Considering Overcurrent Relay
Coordination
⁎
A. Elmitwally, M.S. Kandil, E. Gouda, A. Amer
Elec. Eng. Dept., Mansoura University, Mansoura 35516, Egypt

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

Keywords: - In this paper, the optimal deployment of fault current limiters (FCLs) in a variable-topology distribution net-
Multiple topologies work with distributed generators (DGs) is analyzed. FCLs are used to mitigate the negative impacts of DGs on
Fault current limiter, Multi-objective overcurrent protection system of the distribution network including escalated electromagnetic stresses, costly
optimization replacement of some protective devices, and protection miscoordination. A proposed multi-objective optimi-
Relay re-setting, DGs
zation scheme is applied to maximize the mitigation effect of FCLs and minimize their total cost. The scheme
determines the optimal locations of the FCLs and the associated optimal impedance values. Meanwhile, a par-
ticular focus is kept on preserving the coordinated operation of the protective directional overcurrent relays
(DOCRs) of the distribution network under any operating conditions. DOCRs coordination preservation is
merged into the FCLs allocation problem by a compulsory set of main-backup DOCRs pairs operation constraints
for both near-end and far-end faults at every possible network topology. Furthermore, to lessen the cost of FCLs,
a new hybrid approach is developed. Firstly, optimal re-adjusting of a selected small percentage of the DOCRs is
done for just one time at nearly zero cost. Then, the FCLs allocation problem is solved. Different engineering
strategies for maintaining DOCRs coordination are compared for a 33 kV meshed case-study system.

1. Introduction devices and a possible need for replacement of protection system

equipment at very high cost. Furthermore, the penetration of DGs in the
Distributed generators (DGs) are widely connected to modern power power system may block feeder protection, and cause miscoordination
systems as they can improve reliability, ecological conditions, and of relays. Losing of coordination between protective devices leads to
economics of the system. Radial structure of distribution networks is trip healthy feeders causing unnecessary supply interruptions. This re-
commonly used worldwide. This is because it has a simple design, less sults in reducing the supply reliability and increases the cost of energy
fault current level, less affected by DGs, easy to operate, and has less not served. Also, protection miscoordination can delay fault clearance
restrictions on DGs integration. Also, radial networks have a simple and resulting in abnormal high stresses on the system equipment reducing
efficient protection scheme. However, meshed networks have many its lifespan [1,2].This can magnify or urge the needed investments for
advantages over radial ones. Power loss reduction, improved voltage equipment replacement and upgrading. To overcome these technical
regulation, high level of flexibility, and high capability to handle load challenges and lessen the associated economic loss, it is essential to
growth establish the main merits of meshed systems. Also, they are prevent or at least reduce the resultant increase in short circuit current
proved to be capable of accommodating high penetrations of DGs re- levels. Thus, complicated and expensive devices like fault current lim-
ducing or postponing the need for big investments to build new lines iters (FCLs) must be employed in this situation. This paper is devoted to
and substations for supplying new customers. Hence, meshed networks optimizing these FCLs.
substitute radial ones in some countries [1, 2]. Directional overcurrent relay (DOCR) is the main protective device
However, DGs bring serious impacts on stability, power quality, and in the distribution system and a backup protection in the transmission
operation of protective devices. Particularly, integration of DGs in system [3]-[9]. To ensure that the protection system is reliable, simple,
meshed networks drastically changes the direction, and the magnitude secure, and fast, an appropriate coordination among main and backup
of fault currents. This leads to severe mal-operation of the protection DOCRs should be designed [4]. This ensures that, under fault

⁎
Elec. Eng. Dept., Mansoura University, Mansoura 35516, Egypt
E-mail addresses: kelmitwally@yahoo.co.uk (A. Elmitwally), kelmitwally@yahoo.co.uk (A. Amer).

https://doi.org/10.1016/j.epsr.2020.106417
Received 30 September 2019; Received in revised form 10 April 2020; Accepted 9 May 2020
Available online 01 June 2020
0378-7796/ © 2020 Elsevier B.V. All rights reserved.
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Nomenclature TDSi ith relay time dial setting

N number of DOCRs
FCL fault current limiter If,i short circuit current seen by the ith relay
DG distributed generator (tj,backup,far)k operation time of the backup relay of the jth primary-
DOCR directional overcurrent relay backup DOCR pair for a far-end fault at the kth network
L number of sites with FCLs configuration
ti operating time of the ith DOCR ti, fark operating time of the ith primary DOCR for a far-end fault
Mi current multiple of ith DOCR pickup current at the kth network configuration
th
(Ii, near, ADG)k near-end fault current seen by the i DOCR with DGs TDSi,min, TDSi,max minimum and maximum value of ith relay time
th
for the k network configuration dial setting, respectively
ti, neark operating time of the ith primary DOCR for a near-end (tj,backup,near)k operation time of the backup relay of the jth primary-
fault at the kth configuration backup DOCR pair for a near-end fault at the kth network
A, B, C characteristics constants of the overcurrent relay configuration
Ipi,min, Ipi,max minimum and maximum value of ith relay pickup (tj,primary,near)k operation time of the primary relay of the jth pri-
current setting, respectively mary-backup DOCR pair for a near-end fault at the kth
Xmin, Xmax minimum and maximum values of the FCL inductive network configuration
reactance, respectively (tj,primary,far)k operation time of the primary relay of the jth primary-
CTI positive coordination time interval between backup and backup DOCR pair for a far-end fault at the kth network
primary relays configuration
(Ii, near, BDG)k near-end fault current seen by the ith DOCR without (Ii, far, BDG)k far-end fault current seen by the ith DOCR without DGs
DGs for the kth configuration for the kth configuration
RFCLS, XFCLS size of the resistive and inductive FCL installed at the SICRS, SICxS size-independent cost related to the installation of a
sth site, respectively resistive and inductive FCL at the sth site, respectively
GSA gravitational search algorithm (Ii, far, ADG)k far-end fault current seen by the ith DOCR with DGs for
Pop initial population vector the kth configuration
N index of a particle in population Rmin, Rmax minimum and maximum values of the FCL resistance,
REP repository vector respectively
F1, F2 objective functions Vel velocities vector of PSO particles
Ipickup,i ith relay pickup current setting MOPSO multi-objective PSO
Kc number of network configurations SDCR,SDCX per-unit size-dependent cost of the resistive and in-
PSO particle swarm optimization ductive FCL, respectively

conditions, the backup DOCR should not start working before a pre- fault conditions. This lowers their assumed life span and increases their
specified time interval measured from the presumed operating instant failure rates under normal conditions degrading the power system re-
of its main relay [5]. DOCRs coordination is accomplished by si- liability and economics [15].
multaneous adjusting of the pick-up current (Ipickup) and time–dial Nowadays, FCLs are used to limit the power system fault currents to
setting (TDS) of each relay. Many methods are applied to solve the a much reduced level [16-23]. In normal operation, FCL is invisible for
DOCR coordination problem including curve fitting, graph theory, and the network and has negligible energy loss or voltage drop. When a
optimization. As it has voluminous constraints, the DOCRs coordination fault happens, FCL appears in the network with inserted impedance of
is a complex problem [8]. Optimization is considered as the most effi- definite value [18]. FCLs have three types: passive, solid-state, and
cient coordination method [5, 6]. hybrid [20]. In passive FCL, inductance is added to the circuit when a
The DOCR TDS and Ipickup are commonly determined based on a fault happens, based on the electromagnetic design of the inductor. In
fixed network topology [7-9]. But, the topology of the modern power solid-state type, the FCL impedance is added by using controlled power
system is not fixed in practice. It may change due to unplanned line electronics switches. The hybrid FCL uses mechanical, solid-state
outage, intentional line outage for maintenance, frequent strategic switches, and superconducting FCL element [20]. The impedance in-
network reconfiguration for power loss reduction and voltage im- serted in the circuit by the FCL at fault represents the FCL size. The
provement, and switching operations in the network after faults for greater the FCL size is, the more its cost becomes [22]. Hence, if FCLs
load restoration [10-12]. This leads to DOCRs miscoordination because are optimally applied, they can relive the negative effects of DGs pe-
the values of Ipickup and TDS become inappropriate for a new possible netration on power system by keeping the fault current level at almost
topology [10].The most probable network configurations must be its usual level before DGs connection. Thus, the need for equipment
considered in identifying TDS and Ipickup of DOCRs. They should replacement and protection system reconfiguration vanish [21]. Also,
guarantee coordination of DOCRs for every possible topology. This adds life span of main power system components is maintained due to re-
an extra challenge to the DOCRs coordination problem [11]. ducing the electromagnetic stresses under fault condition. Finally, FCL
DGs much heighten the levels of fault currents to a limit that may be improves system reliability by lowering the failure rates of main power
outside the handling capacity of circuit breakers and switches already system components [15].
installed in the network. The switchgear equipment can become un- In [13], an optimization technique is used to select the impedance
suitable and needs a costly replacement. Besides, the routes and di- type of FCL and its size to maximize the mitigation effect of FCL on
rection of fault currents can be changed due to DGs [13]. This can cause distribution network with DGs. But, the relay coordination problem was
serious fuse-recloser miscoordination and relay-relay miscoordination not studied in that paper. Also, the authors in [17] searched the optimal
[14]. Typically, these aforementioned problems are expensively solved locations and sizes of supportive FCLs in a meshed power system using
by protective equipment replacement, protection system reconfigura- iterative mixed integer nonlinear programming. Nonetheless, neither
tion, and disconnection of DGs at fault time [13]. Furthermore, the topology variation nor relay coordination was addressed. Solutions of
enlarged fault current causes extra high stresses on power system DOCRs miscoordination problem due to DGs are addressed in [16]-
equipment such as transformers, lines, switchgear, and cables under [28]. One solution is the use of adaptive relaying schemes where

2
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

DOCRs can have dynamic setting to keep coordination under varying 4 The determined sizes and locations might not be optimal. Empirical
operating conditions as discussed in [19]. The adaptive relaying scheme approaches are used in many cases.
depends on a dedicated communication system that suffers from cyber 5 The assessment of the overall mitigation effect of FCLs on power
security and reliability issues. Also, they are costly and technically system is ignored. The DOCRs coordination is merely focused.
complicated. Remedial resetting of Ipickup and TDS of some or all the
DOCRs is another solution reported in [23]. Though it is relatively In this paper, FCLs are optimally allocated to mitigate the negative
simple and cheap, this later solution is tedious and time consuming task impacts of DGs on a variable-topology power distribution network. A
for power system operation staff in large networks [22]. Besides, a fault proposed multi-objective optimization scheme is applied to minimize
may occur during the resetting process that will be treated improperly the total cost of requested FCLs. Meanwhile, the optimal coordination
by the DOCRs leading to undesired consequences. Despite their relative of DOCRs must be preserved under all system topologies and operating
high cost, properly installed FCLs can restore and sustain the lost conditions. Hence, the DOCRs coordination constraints are integrated
DOCRs coordination due to DGs [18]. The authors in [17] studied the into the FCLs allocation problem. Besides, to achieve the optimal
use of FCLs for restoring the coordination of DOCRs under integration DOCRs coordination at lower FCLs cost, a combined re-setting-FCLs
of DGs. The locations and sizes of FCLs are determined by trial and approach is proposed. Before searching the optimal FCLs, a small per-
error. The obtained FCLs have much high sizes, and hence high cost. centage (in order of 10%) of critical DOCRs are selected for one time re-
Moreover, the overall mitigation effect of FCLs in terms of its capability adjustment at nearly zero cost.
to reduce fault current is ignored. In a previous paper [22], the authors The contributions of this paper are:
investigated the maintenance of overcurrent relays coordination under
DGs by optimal planning of FCLs of different types. Despite the obvious ❖ A method is proposed to maximize the mitigation effect of
reduction in sizes of FCLs, the required number and expected cost of FCLs on power systems with DGs at minimum cost based on
FCLs were still high. Also, the overall mitigation effect of FCLs, network multi-objective optimization.
topology variation, and effect of far-end fault are missing. In [23], the ❖ Constraints of the optimal DOCRs coordination are integrated
relay coordination under DGs is restored by a combined FCL-resetting into the FCLs allocation problem.
approach. This method empirically allocates FCLs to restore DOCRs ❖ Permanent re-setting of a selected small ratio of DOCRs is
coordination. Then, the effect of resetting only one of the relays is as- proposed to further diminish the FCLs cost.
sessed to reduce the size of FCLs. However, the determined sizes and ❖ The varying topology of the network is taken into account.
locations of FCLs are not optimal. Also, the overall mitigation effect of ❖ Both the near-end and far-end fault currents are considered.
FCLs is overlooked.
In [24], the authors proposed a technique to allocate FCLs to reduce 2. Formulation of the DOCRs coordination
the current surges through a two-stage approach. They combined the
hierarchical fuzzy logic decision method with hashing integrated gen- 2.1. DOCR characteristics
eric algorithm to find the optimal placements of FCLs. Then, PSO is
applied to determine the optimal FCL size. However, topology varia- The characteristic of the DOCR is mathematically defined in [29] as:
tion, and relay coordination were not analyzed. The authors in [25]
proposed a protection coordination index. Then, it is used to find the A Ifi
ti = TDSi ⎛⎜ + B ) withMi =
locations and sizes of the FCLs taking in consideration multiple DGs
⎝ Mi C − 1 Ipickup,i (1)
units. But, multiple system configurations and cost evaluation were
th
missing. An approach to identify the optimal size of FCL through bio- Where ti is the operating time of the i DOCR, Mi is the current
geography-based optimization is studied in [26]. The work estimated multiple of its pickup current, Ifi is the short circuit current seen by the
the reliability improvement raised by FCL due to fault current reduction relay i. The constants A, B, C define the type of the overcurrent relay
and fixing the fuse-recloser protection mismatch under DGs. Topology according to IEEE standard C37.112-1996 [29].
variation and relay-relay coordination were ignored. Simultaneous es- The ith relay pickup current setting (Ipickup,i) is kept between a
timation of the optimal location and size of DGs and the optimal size of minimum value (Ipi, min) and a maximum value (Ipi, max). The ith relay
FCLs is presented in [27]. A nondominated sorting genetic algorithm is time dial setting (TDSi) is kept between a minimum value TDSi,min and a
utilized to obtain optimal decision variables. However, FCLs are as- maximum value TDSi,max. Ipi, min and Ipi, max are chosen as 1.25 and 2
sumed only in series to DGs. Topology variation and relay coordination times the maximum load current seen by the relay, respectively.
were not considered. In [28], a method to simultaneously optimize the TDSi,min and TDSi,max are set as 0.1s and 1.1s, respectively [29].
power quality and protection coordination in a microgrid with DGs
under multiple modes is presented. A solid state FCL is used as an in- 2.2. Coordination problem
terface between the microgrid and the utility to get coordination be-
tween upstream and downstream networks. The settings of the over- An optimization problem is developed to find the value of TDS and
current relays and the characteristics of the FCL are determined by Ipickup of each DOCR. To obtain an accurate coordination, the objective
genetic algorithm. Nevertheless, few relays and one FCL with pre-as- function is formed to minimize the operating times of the primary
sumed location are coped. Also, economic evaluation and network re- DOCRs to achieve fast fault clearance. Meanwhile, coordinated opera-
configurations are absent. tion of each primary-backup DOCRs pair in terms of their relative op-
In general, the previous research works that addressed the appli- erating time is considered as a set of obligatory constrains for each
cation of FCLs for DOCRs coordination under DGs have the following possible network configuration [11]. Furthermore, near-end and far-
drawbacks: end faults are examined to ensure a universal and robust coordination.
This is mathematically expressed as:
1 The assurance of DOCRs coordination for far-end faults is over- Kc Kc N
N
looked. The coordination is verified for near-end faults but it may be min F = ∑ ∑ tk + ∑ ∑ tik,far
i = 1 i, near
lost for far-end fault. k=1 k=1 i=1
(2)
2 The cost of FCLs is not explicitly expressed. Sum of FCLs impedance
th
values are used as an implicit cost function. Where, ti, neark is the operating time of the i primary DOCR for a
3 The network topology variation is not considered in the problem three-phase near-end fault at the kth network configuration.ti, fark is the
formulation. operating time of the ith DOCR for a three-phase far-end fault at the kth

3
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

network configuration. N and Kc are the number of DOCRs and network Where, (I i, near , BDG )k is the three-phase near-end fault current seen by
configurations, respectively. the ith DOCR without DGs for the kth configuration. (I i, near , ADG )k is the
Subject to: the DOCRs setting constraints: three-phase near-end fault current seen by the ith DOCR with DGs for
the kth configuration. (I i, far , BDG )k is the three-phase far-end fault current
Ipi,min ≤ Ipickup,i ≤ Ipi, max (3)
seen by the ith DOCR without DGs for the kth configuration. (I i, far , ADG )k is
TDSi, min ≤ TDSi ≤ TDSi, max (4) the three-phase far-end fault current seen by the ith DOCR with DGs for
the kth configuration.
the DOCRs coordination constraints: The function F2 represents the total cost of installing and operating
Δt jk, near ≥ CTI FCLs. Each FCL can be made of a resistive part and an inductive part.
(5)
Each part has a size-independent cost that accounts for commissioning,
Δt jk, far ≥ CTI site-preparation, and installation. It also has a size-dependent cost that
(6)
accounts for its purchase and operating cost. An example of the oper-
Where, ating cost is the cooling system operation cost for superconducting FCL.
In (10), L is the number of sites with FCLs. SICRs is the size-independent
Δt jk, near = (t j, backup, near )k − (t j, primary, near )k (7) cost related to the installation of a resistive FCL at the site s. SDCR is
per-unit cost of the resistive FCL. RFCLs is the size of the resistive FCL
Δt jk, far = (t j, backup, far )k − (t j, primary, far )k (8)
installed at the site s. SICXs is the size-independent cost related to the
Where, (tj,backup,near)k is the operation time of the backup relay of installation of an inductive FCL at the site s. SDCX is the per-unit cost of
the jth primary-backup DOCR pair for a three-phase near-end fault at the inductive FCL. XFCLs is the size of the inductive FCL installed at the
the kth network configuration, (tj,primary,near)k is the operation time of site s.
the primary relay of the jth primary-backup DOCR pair for a three-phase Subject to: the DOCRs coordination constraints:
near-end fault at the kth network configuration. (tj,backup,far)k is the Δt jk, near ≥ CTI (11)
operation time of the backup relay of the jth primary-backup DOCR pair
for a three-phase far-end fault at the kth network configuration, Δt jk, far ≥ CTI (12)
(tj,primary,far)k is the operation time of the primary relay of the jth pri-
mary-backup DOCR pair for a three-phase far-end fault at the kth net- the FCL component size constraints:
work configuration. CTI is a positive coordination time interval be- RFCLmin ≤ RFCL ≤ RFCLmax (13)
tween backup and primary relays to ensure that the backup relay
operates, if the fault is not cleared, after the primary relay. This ac- XFCLmin ≤ XFCL ≤ XFCLmax (14)
counts for current transformer error, DC offset component of fault Where, Rmin, Rmax are the minimum and maximum values of the FCL
current, and safety margin for relay errors [4]. The number of con- resistance, respectively. Xmin, Xmax are the minimum and maximum
straints set expressed in (5) and (6) is huge for a large-scale network. values of the FCL inductive reactance, respectively. It is recalled that
constraints (11) and (12) are the same as (5) and (6) but after the
3. FCL allocation problem connection of DGs.

FCLs are installed to minimize the increase in short circuit currents 4. Solution Algorithm
due to the connection of DGs. This mitigates the negative impacts of
DGs on power system. Besides, loss of coordination among DOCRs can Mixed-integer nonlinear programming (MINLP), as a deterministic
be avoided for all possible network configurations. The less the short optimization method, was reported to solve the DOCRs coordination
circuit currents become, the better the mitigation effect of FCLs is. problem for the first time in 1988. Afterwards, linear programming (LP)
Meanwhile, it is required to minimize the cost of FCLs sustaining the technique gained good recognition to solve this problem, including
coordination between all main-backup DOCR pairs. So, there are two simplex, two-phase simplex and dual simplex methods. These LP tech-
contradictory objectives associated to the FCLs allocation problem. The niques make a simplifying assumption. They express the operational
first is to maximize its mitigation effect by minimizing the increase in time of relay as a linear function of its time multiplier setting.
short circuit currents due to DGs connection. The second is to minimize Moreover, the deterministic methods entrap in a local optimal solution
the FCLs cost. The later depends on the FCL technology (mechanical, due to its dependency on a single guessed starting point [30]. Hence,
electronic, superconducting, etc.), the FCL type (resistive, inductive), biologically-inspired methods (BIM) are introduced to overcome the
the FCL size (impedance value), and the number of FCL units as it de- limitations of deterministic optimization methods by producing an in-
termines the FCLs fixed cost. Commissioning, housing, installation, itial population of feasible solutions [31]. In the last decade, BIM have
monitoring and control system, and cost of cooling system are com- become a widely recognized and extensively used tool to solve various
prised into the FCL fixed cost [16, 22]. The main constraint is to engineering problems including the relay coordination [30]. These
guarantee the coordinated operation between all main-backup DOCR techniques include genetic algorithm (GA), non-dominated sorting GA
pairs for every considered network configuration. In short, it is desired (NSGA II), modified particle swarm optimization (MPSO), teaching-
to maximize the benefit of the minimum-cost FCLs needed to guarantee learning based optimization (TLBO) and modified adaptive TLBO
coordinated operation of DOCRs. Mathematically, the problem is a (MATLBO). To reduce the search space and computational time, hybrid
constrained multi-objective optimization problem. methods are also utilized for solving the relay coordination problem,
The objective functions are expressed as: including GA-LP, GA-NLP and PSO-gravitational search algorithm [31].
In brief:
Min F1
Kc N
= ∑ ∑ abs {[(Ii,near ,BDG)k − (Ii,near ,ADG)k] + [(Ii,far ,BDG )k − (Ii,far ,ADG )k]} • The quality of the solution obtained by the deterministic optimiza-
k=1 i=1 tion methods greatly depend on the assumed initial (starting) point.
(9) So, it can be trapped in a local optimal solution that is far from the
desired global optimal solution. Also, it may suffer from numerical
L
instability. Additionally, some simplifying assumptions to the pro-
Min F2 = ∑ [[ SICRs + (SDCR*RFCL
s
)] + [ SICXs + (SDCX *XFCL
s
)]]
s=1 (10) blem might be needed.

4
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

• Metaheuristic methods produce an initial population of feasible 4.2. FCLs allocation

solutions. Therefore, they generally reach at least a sub-optimal
solution that is very close to the global optimal one. The multi-objective PSO (MOPSO) is introduced in [32, 33]. It is
• Metaheuristic methods' computation time is generally comparable employed to solve the FCLs allocation problem to decide the optimal
to that of deterministic ones. locations and sizes of FCLs as formulated in Section 3. The MOPSO-
• Metaheuristic methods are widely recognized and extensively used based solution algorithm is described below.
in recent literature for proficiently solving power system problems
and relay coordination problem in particular with proved accuracy. 1 Generate an initial population vector (pop) with n particles and
initial velocities vector (vel).
2 For each particle, insert the associated FCLs in the power system
4.1. DOCRs coordination with DGs connected and calculate the fault currents values for each
bus in the system. Repeat for each topology.
The combined PSO-GSA method (PSOGSA) is described with enough 3 Estimate the values of objective functions F1, F2 in (9) and (10) for
details in [31]. It is used herein to solve the DOCRs coordination pro- all particles.
blem discussed in Section 2 and formulated in (2)-(8) to get the optimal 4 Check constrains, and penalize particles having violations [33].
DOCRs Ipickup and TDS settings. The flowchart of the PSOGSA is il- 5 Create hypercubes for the search space as a reference system for the
lustrated in Fig.1. particles to locate them.

Input network data

Calculate the relay near-end and far-end

fault currents for each topology

Initiate counters for iteration& particle

Generate initial population of N

particles & their velocities

Compute operation time of each DOCR,

for each topology

Estimate objective functions F for particles

Next Examine constraints &

iteration
penalize violating particles Next
particle

Identify pbest, and gbest values

Find mass and acceleration of each particle

Modify weight, velocity and position of particles

n ≤ nmax
Yes
Yes No
it ≤ itmax
No
Save
results
Fig. 1. Flowchart of DOCRs coordination by the PSOGSA

5
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Read network data

Calculate the relay near-end

fault current without DG for each topology

Initiate counters for iteration t& particle

Generate randomly n particles &

their velocities

Connect DGs& FCLs , Compute fault current

of every DOCR, for each topology

Evaluate the objective functions F1, F2

Check constraints and

penalize failing particles Next
particle
Increase counter t by 1, & Get new weight,
velocity and position of particles

Save the nondominated particles in a set

Obtain the global nondominated set G

Yes
Size of G ≤ threshold

No
Reduce size of G

Make the external Pareto set equals G

t=
t+1
Calculate distances of members in G

Obtain the global best set

Yes
n ≤ nmax
No
Yes
t ≤ tmax
No
Display
solution
Fig. 2. Flowchart of identifying optimal FCLs by MOPSO

6
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

6 Store the particle position of the non-dominated solution in the Table 1

repository vector (REP). PSOGSA algorithm data
7 Initiate the iteration counter at t=1. Parameter Value
8 Update: the value of inertia weight, values of velocity for each size of the swarm "no of particles" 120
particle, and the position of each particle as in [32].
maximum number of iterations 100
9 Update the vector REP with new particles by inserting the non-
dimension of test functions 58
dominated solutions and remove the dominated solutions [33]. gravitational constant G0 1
10 Apply the pareto dominance to decide if the previous position of the weighting factor c1 0.5
particle is eliminated or stay [33]. weighting factor c2 1.5
11 Check if the termination criterion is verified, and stop. Otherwise, descending coefficient 20

increase iteration counter by 1 then go to step 8.

Table 2
Fig. 2 represents the flowchart of MOPSO algorithm used to obtain
MOPSO parameters
optimal FCLs. Figure 2Figure 2
Parameter Value
number of particles 100

5. Results and discussion population size 100

repository size 100
The proposed approach is applied to a part of the IEEE 30 bus test maximum number of iterations 100
system shown in Fig.3 [34]. In this work, 7 configurations are con- inertia weight 0.7298
inertia weight damping ratio 1
sidered for the system in Fig.3. The first configuration is the base case as
personal learning coefficient 1.49
depicted by Fig.3. The other 6 configurations are assumed by the most global learning coefficient 1.49
probable single line opening. The lines to open are 16-17, 12-16, 12-15, grid inflation parameter 0.10
15-18, 15- 23, and 27-30, respectively. This system has 29 DOCRs lo- number of grids per each dimension 10
leader selection pressure parameter 4
cated as shown in Fig.3. The DOCRs are assumed to be identical with
repository member selection pressure 2
moderately inverse characteristics with the constants A, B, and C are pareto front 100
0.0515, 0.114, and 0.02, respectively [29]. The value of CTI is selected
to be 0.2s [16]. The full solution algorithm is coded in Matlab m-file.
The PSOGSA algorithm data are given in Table 1. Table 2 displays the 5.1. Relays coordination before adding DGs
MOPSO algorithm parameters. The upper and lower limits of variables
involved in the constraints are delivered in Table 3. The cost data of the Each DOCR is a main relay that has one or more backup DOCRs
FCL at the 33 kV voltage level are provided in Table 4 [16]. based on the system topology. The main-backup DOCR pairs for the
base configuration of the study system shown in Fig.3 are given in
Table 5.

Fig. 3. Meshed system under study

7
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Table 3 5.3.1. Resetting all the DOCRs

Limits of variables After adding the DGs in the system, the PSOGSA-based solution
Variable minimum value maximum value technique described in subsection 4.1 above is applied to get the new
optimal DOCRs setting parameters. Table 9 shows the obtained values
TDS, sec. 0.1 1.1 of Ipickup and TDS for each relay. This DOCRs setting problem has 535
Ipickup 1.25 of normal current 2 of normal current
constraints which are fulfilled under DGs for all considered configura-
XFCL, p.u 0 10
RFCL, p.u 0 10
tions. This confirms the coordinated operation of all main-backup
DOCR pairs in presence of DGs for all topologies. The obtained optimal
value of the objective function F in (2) is 131.1s. However, the fault
Table 4 current values in the system branches after DGs connection become
Fixed and incremental cost for a typical superconducting FCL very high as can be noted in Fig.4. The fault current of many branches is
almost doubled due to DGs to reach a high value of about 24kA. This
Voltage level FCL type Fixed Cost (million $)
prohibitively high fault current will impose high electromagnetic stress
33KV Resistive 0.39 0.15 on system components affecting its life span. Also, it can exceed the
Inductive 0.42 0.12 fault handling capability of some protective devices making it necessary
to replace at a very high cost. It is emphasized that resetting DOCRs
only, without FCLs, did retain successful coordinated operation of re-
Table 5 lays under DGs. But, it can neither alleviate escalated electromagnetic
Main-backup DOCR pairs for the base configuration of the study system
stresses on system components nor prevent necessary replacement of
Main relay Backup relay Main relay Backup relay protective devices at a high cost.
1 19,20,23 15 13
2 15,20,23 16 18 5.3.2. Connect FCLs in series to DGs
3 15,19,23 17 10 In this scenario, FCLs are to be placed only in series to DGs. This is
4 15,19,20 18 2
5 9,12 19 17
thought as a practical solution to avoid unwanted impacts of DGs [13].
6 8,12 20 4,23 Each installed DG should be connected to the system through a series
7 8,9 21 3 FCL. The MOPSO-based solution algorithm presented in subsection 4.2
8 6,16,22 22 4,21,25 is applied to determine the optimal sizes of these FCLs. Assuming in-
9 5,16,22 23 11,25
ductive FCLs, Table 10 shows the optimal reactance values of the FCLs.
10 5,6,22 24 4,11,21
11 5,6,16 25 28,29 Fig.4 shows the near-end three-phase fault current levels seen by the
12 14 26 24,28 DOCRs for this case. FCLs cause the fault current in branches to be very
13 7 27 24,29 close to its original value before DGs. This confirms the overall miti-
14 1 28 26 gation effect of FCLs in addition to maintaining the DOCRs
29 27
coordination.Figure 4

Without DGs, the PSOGSA-based solution technique described in 5.3.3. Combination of limited DOCRs re-setting and FCLs in series to DGs
subsection 4.1 above is applied to get the optimal DOCRs setting To reduce the cost of FCLs determined in subsection 5.3.2, resetting
parameters. Table 6 shows the obtained values of Ipickup and TDS for of a limited number of DOCRs suffering from mis-coordinated operation
each relay. It is worthy to note that this DOCRs setting problem has 535 due to DGs is carried out. This number of DOCRs to be re-set is about
constraints which are fully fulfilled. The obtained optimal value of the 10% of total DOCRs to avoid complexity and minimize risk during re-
objective function F in (2) is 118s. setting realization. The cost of DOCRs re-setting is minimal. As pro-
posed, 10% of the total 29 DOCRs will be re-set. So, R19, R16, and R11
which have the worst miscoordination conditions, least values of Δt
5.2. Effect of adding DGs on the DOCRs coordination after integration of DGs, are selected for resetting. A subset of con-
straints associated to those relays is defined and considered. Then, the
Three 10 MVA, 6.6 kV, 0.9 power factor synchronous DGs are added PSOGSA-based solution technique described in subsection 4.1 above is
to the test system in Fig.3 at buses 10, 12, and 19, respectively. Each DG applied to get the new optimal setting parameters of these 3 DOCRs.
has a 0.15 p.u transient reactance and connected to the system through Table 11 shows the obtained new values of Ipickup and TDS. Next, the
a 6.6/33 kV transformer of 0.05 p.u equivalent reactance [17]. Unlike
inverter-based DGs, synchronous DGs cause a large increase in the short Table 6
circuit current levels that severely impact the performance of DOCRs Ipickup and TDS settings of DOCRs
[21]. Due to the integration of these 3 DGs, many main-backup DOCR
Relay No. Ipickup TDS Relay No. Ipickup TDS
pairs deny coordination with Δt value less than 0.2s for one or more
possible configurations of the system. Table 7 indicates the mis- 1 3.7471 0.3588 16 6.1912 0.5269
coordinated main-backup DOCR pairs for the base configuration. 2 5.1739 0.2523 17 4.7935 0.1
3 3.678 0.2418 18 5.3648 0.1
Table 8 shows the miscoordinated main-backup DOCR pairs for the
4 4.6444 0.469 19 3.9182 0.4871
topology when line 16-17 is opened. 5 4.0701 0.439 20 3.3401 0.1
6 4.7654 0.5626 21 4.8791 0.1
7 4.6363 0.2675 22 2.841 0.1
5.3. Strategies for restoring DOCRs coordination under DGs 8 6.0988 0.5312 23 3.5703 0.7286
9 3.6318 0.3291 24 6.6123 0.6953
10 4.5882 0.2894 25 4.1396 0.6621
To restore the missed DOCRs coordination due to the connection of 11 4.5236 0.665 26 4.6381 0.2081
DGs as demonstrated in subsection 5.2 above, different restoration 12 4.9142 0.4811 27 3.5849 0.3503
strategies are employed. They include resetting of all DOCRs, applica- 13 7.4338 0.1 28 4.4444 0.1
14 6.4299 0.1 29 2.8004 0.2905
tion of FCLs in various scenarios, and hybridization of DOCRs resetting
15 5.7515 0.5081
and FCLs. These schemes are investigated in the following subsections.

8
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Table 7 5.3.4. Connect FCLs in unlimited locations

Miscoordinated main-backup DOCR pairs for the base configuration In this scenario, FCLs are placed in any possible location in the
Near end fault Far end fault system, in lines, next to generators, or next to DGs. The MOPSO-based
solution algorithm presented in subsection 4.2 is applied to determine
DOCR pair Δtj, near1 DOCR pair Δtj, far1 the optimal locations and sizes of FCLs. The number of FCLs is not
R1BR14 0.1769 R3BR21 0.1685
constrained. Assuming inductive FCLs, Table 13 shows the optimal lo-
R3BR21 0.1592 R9BR5 0.1872
R4BR24 0.1289 R11BR23 0.1407
cations and reactance values of the FCLs. Though the total size of FCLs
R5BR11 0.1179 R19BR4 0.068 is less than case 5.3.2, much bigger number of FCLs is suggested by the
R6BR11 0.1816 R27BR28 0.198 algorithm. This makes the total cost of FCLs to be much higher than
R7BR13 0.1838 R12BR6 0.161 before. So, the practical solution of connecting FCLs only in series to
R8BR6 0.1915 R16BR11 0.0944
DGs is also much better from the economic perspective. Fig.6 shows the
R9BR5 0.1312 R19BR4 0.1195
R10BR17 0.1907 R26BR28 0.1962 near-end three-phase fault current levels seen by the DOCRs for this
R11BR24 0.1696 case. Due to the big number of FCLs distributed in the network, the fault
current seen by most of DOCRs after FCLs in presence of DGs is less than
its initial value before DGs are connected. This reflects the prominent
Table 8 mitigating effect of FCLs on the system.
Miscoordinated main-backup DOCR pairs for the configuration number 2 (when
line 16-17 is open) 5.3.5. DOCRs re-setting and unlimited FCLs
Near end fault Far end fault To reduce the cost of FCLs determined in subsection5.3.4, resetting
of a limited number of DOCRs suffering from miscoordinated operation
DOCR pair Δtj, near2 DOCR pair Δtj, far2
due to DGs is carried out. Then, FCLs are searched. As proposed, 10% of
R3BR21 0.163 R3BR21 0.1725
R4BR24 0.1327 R11BR23 0.1279 the total 29 DOCRs will be re-set. So, R19, R16, and R11 which are
R5BR11 0.1271 R18BR16 0.176 related to the worst-violated main-backup coordination constraints, see
R6BR11 0.1881 R19BR4 0.0433 Tables 7 and 8, are chosen to be reset. To re-adjust those relays only, a
R8BR6 0.164 R24BR25 0.1797 sub-set of constraints associated only to these relays is defined and
R9BR5 0.1084 R27BR28 0.198
R10BR17 0.191 R19BR4 0.1091
considered. The PSOGSA-based solution technique described in sub-
R11BR24 0.1663 R26BR28 0.1963 section 4.1 above is applied to get the new optimal setting parameters
R16BR11 0.0909 of these 3 DOCRs. Table 11 shows the obtained new values of Ipickup and
TDS. Then, the MOPSO-based solution algorithm presented in subsec-
tion 4.2 is applied to determine the optimal locations and sizes of FCLs.
Table 9 Assuming inductive FCLs, Table 14 shows the optimal locations and
New Ipickup and TDS setting for DOCRs under DGs reactance values of the FCLs. The total number and sizes of FCLs are less
Relay No. Ipickup TDS Relay No. Ipickup TDS than the previous case in subsection5.3.4. This makes the total cost of
FCLs to be less than before. Fig.7 shows the near-end three-phase fault
1 5.4144 0.3618 16 6.0327 0.6957
current levels seen by the DOCRs for this case.
2 3.3633 0.2838 17 4.6553 0.1
3 5.1676 0.2492 18 4.4401 0.1
4 4.6675 0.4821 19 4.0412 0.6296 5.3.6. DOCRs re-setting and stipulated FCLs number
5 4.3898 0.5194 20 3.2702 0.1 This scenario is similar to that in subsection 5.3.5 but the number of
6 3.9599 0.6798 21 4.464 0.1 FCLs to be installed is restricted to 3. However, the locations and sizes
7 4.949 0.648616 22 3.8132 0.100001
of these FCLs are still to be optimally determined. R19, R16, and R11
8 5.9753 0.7877 23 4.0748 0.6723
9 3.6229 0.45 24 7.5415 0.6252 are re-set as provided in Table 11. Then, the MOPSO-based solution
10 3.961 0.3014 25 4.2255 0.5819 algorithm presented in subsection 4.2 is applied to determine the op-
11 4.4614 0.7124 26 5.474 0.2012 timal locations and sizes of the stipulated 3 FCLs. Assuming inductive
12 4.6267 0.6116 27 5.4203 0.2776
FCLs, Table 15 shows the optimal locations and reactance values of the
13 9.9659 0.5451 28 6.5438 0.1
14 9.5169 0.1 29 2.9294 0.1827
FCLs. The total cost of FCLs is noticeably less than before. Fig.8 depicts
15 5.8227 0.4774 the near-end three-phase fault current levels seen by the DOCRs for this
case.

Table 10 5.3.7. Effect of hybrid resistive-inductive FCLs

Locations and values of inductive FCLs Unlike scenario 5.3.6 above, the stipulated 3 FCLs are assumed to be
Location reactance value, p.u resistive-inductive ones. Each FCL is composed of series-connected re-
sistive and inductive parts. Table 16 reveals the obtained optimal lo-
DG at Bus 10 5.0686 cations and impedance values of the FCLs. Interestingly, no FCL has
DG at Bus 12 5.6564
both resistive and inductive parts. One FCL is found purely inductive.
DG at Bus 19 3.0771
Summation of FCLs sizes=13.8 p.u
The other two are purely resistive. However, the overall fault current
F1=4300 A suppression capability of the FCLs scheme is enhanced. The total cost of
Total FCLs cost = 2.9 million $ FCLs is 1.26M$ which is the least. Also, the value of F1 is reduced to
1620 A. Fig.9 depicts the near-end three-phase fault current levels seen
by the DOCRs for this case.
MOPSO-based solution algorithm presented in subsection 4.2 is applied
to determine the optimal sizes of FCLs connected in series to DGs. As- 5.4. Evaluation and comparison
suming inductive FCLs, Table 12 shows the optimal reactance values of
FCLs. The total size of FCLs is less than that in subsection 5.3.2. This 5.4.1. Comparison to other methods
makes the total cost of FCLs to be obviously less than before. Fig.5 For the FCLs allocation scenario discussed in subsection 5.3.2, the
shows the near-end three-phase fault current levels seen by the DOCRs results obtained by the proposed method using MOPSO as manifested in
for this case. Table 10 are compared to other two approaches. The first is the multi-

9
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Fig. 4. Fault current level seen by the DOCRs

Table 11 Table 13
New Ipickup and TDS for re-set relays Locations and reactance values of FCLs
Relay No. Ipickup TDS Location Reactance, p.u Location Reactance, p.u

11 5.1083 0.707 G at Bus 1 0.0888 Bus 10,17 0.0728

16 6.2461 0.491 DG at Bus 10 0.1441 Bus 10,20 0.4239
19 5.879 0.4198 DG at Bus 19 0.6402 Bus 10,21 0.1178
G at Bus 13 0.802 Bus 10,22 0.3851
Bus 2,4 0.2331 Bus 12,16 0.0857
Bus 3,4 0.3709 Bus 14,15 0.0951
Table 12
Bus 2,5 0.0024 Bus 15,18 0.1153
Reactance values of FCLs in series to DGs
Bus 2,6 0.807 Bus 19,20 0.2535
Location reactance value, p.u Bus 4,6 0.9965 Bus 22,24 0.6635
Bus 4,12 0.3392 Bus 15,23 0.1194
DG at Bus 10 2.0445 Bus 5,7 0.0196 Bus 23,24 0.3412
DG at Bus 12 2.6928 Bus 6,8 0.6836 Bus 27,30 0.1
DG at Bus 19 2.1176 Bus 6,10 0.2555 Bus 6,28 0.0504
Summation of FCLs sizes =6.85 p.u Bus 9,10 0.5406 Bus 8,28 0.9695
F1 =2802 A Summation of FCLs sizes =10.02 p.u
Total FCLs cost = 2.08 million $ F1=7900 A
Total FCLs cost = 13.38 million $

objective genetic algorithm (MOGA) as a metaheuristic method pre-

sented in [35]. The second is the interior point nonlinear programming worthy to mention that the run time to extract the optimal solution in
(IPNLP) adopted in [36] as a deterministic method. The both are coded FCLs allocation as a planning problem is not of great importance. Un-
as built-in Matlab optimization toolbox functions called "gamultiob" for like operation problems that are solved repeatedly every short period,
MOGA and "fmincon" for IPNLP. They are used to solve the FCLs allo- the coped problem in this paper will be solved occasionally to get
cation problem formulated in Section 3. The parameters of the MOGA planning decisions. So, it is desired to get the best results in terms of
and IPNLP algorithms are revealed in Table 17. The results are in- accuracy paying less attention to run time. Besides, as a deterministic
dicated in Table 18. The proposed method is remarked to produce method, the solution got by IPNLP is sensitive to the start point selec-
better results. However, the computation time of the MOPSO is noted to tion. Basically, the start point is selected at the mid-range of the control
be larger than the other two techniques. Though IPNLP takes the variables to be determined (the X values of FCLs). As three variables are
shortest run time, it produces the worst solution in this case. It is searched, the start point is set as (5, 5, 5 p.u) which produced the above

Fig. 5. Figure 5 Fault current level seen by the DOCRs

10
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Fig. 6. Fault current level seen by the DOCRs

Fig. 7. Fault current level seen by the DOCRs

Table 14 Table 18. It is evident that these conditions lead to the best solution.
Locations and reactance values of FCLs in scenario 5.3.5 This is deemed to be the global optimal solution. However, compared to
Location Reactance, p.u Location Reactance, p.u
the MOPSO solution, the reduction in FCLs cost (F2) is about 3% and
the reduction in the value of F1 (absolute current difference index) is
G at Bus 2 0.7989 Bus 2,6 0.064 about 55%. The large reduction in the value of F1 may reflect the im-
DG at Bus 10 0.9454 Bus 6,9 0.4385 proved mitigation effect of the FCLs on the system. More importantly,
DG at Bus 12 0.2455 Bus 12,16 0.1713
DG at Bus 19 0.4983 Bus 19,20 0.331
the DOCRs coordination is maintained for all methods. So, although the
G at Bus 8 0.9 Bus 22,24 0.8527 MOPSO method got a solution that is sub-optimal from the mathema-
G at Bus 13 0.917 Bus 15,23 0.1446 tical perspective, it is practically as good as the global solution. This
Bus 2,4 0.8559 Bus 23,24 0.0053 confirms the efficacy and adequacy of the proposed MOPSO-based
Bus 3,4 0.9058 Bus 24,25 0.1821
method.
Bus 9,11 0.9
Summation of FCLs sizes =9.15 p.u
F1=6960 A 5.4.2. Evaluation of FCLs scenarios
Total FCLs cost = 8.2 million $ Table 19 summarizes the results of the FCLs-based solutions for
restoring DOCRs coordination obtained in subsections 5.3.2 to 5.3.7.
These solutions are compared in terms of the values of F1 (absolute
Table 15 current difference index) and F2 (total FCLs cost). It is noticed that
Locations and reactance values of FCLs in scenario 5.3.6 installing a restricted number of FCLs to 3 only, as in subsections 5.3.6
Location Reactance, p.u and 5.3.7, gives better results for both F1 and F2. Optimal placement
and sizing of these 3 FCLs achieves superior performance compared to
Generator at Bus 2 0.1224
only connecting the FCL in series to a DG as in subsections 5.3.2 and
DG at Bus 10 1.1266
Line 2-6 0.1063 5.3.3. This indicates that there are more effective locations for FCLs
Sum of X_FCL =2.0851 p.u than next to DGs. Generally, resetting of selected 3 DOCRs, about 10%
F1=1980A of the total DOCRs in the network, remarkably reduces the required cost
F2= 1.51 M$ of FCLs. Moreover, installing hybrid FCLs of both resistive and in-
ductive parts reduces the FCLs cost and enhances its current limiting
capability compared to only inductive FCLs. Thus, the scenario pre-
discussed results indicated in the 3rd column of Table 18. Trying other
sented in subsection 5.3.7, the bottom row in Table 19, is deemed to be
start points in the neighboring space yielded similar results. None-
the optimal solution to maintain DOCRs coordination in presence of
theless, when the final result attained by MOPSO is inputted as a start
synchronous machine-based DGs. On the other hand, increasing the
point to the IPNLP, the results are manifested in the fourth column of
number of installed FCLs leads to a large increase of the cost. This

11
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Fig. 8. Fault current level after using stipulated 3 inductive FCLs

Table 16 Table 17
Locations and values of hybrid FCLs Parameters of MOGA and IPNLP methods
Location RFCL, p.u XFCL, p.u MOGA IPNLP

DG at Bus 12 0 0.1220 Population size: 100 Number of iterations: 100

Line 1-2 0.2321 0 Tournament size:2 X tolerance:1e-10
Line 27-30 0.0863 0 Crossover fraction:0.8 Function tolerance: 1e-6
F1 1620 A Crossover ratio:1 Constraint tolerance:1e-6
F2 1.26 M$ Mutation function rate:0.01 Unboundedness threshold:-1e-20
Pareto front population fraction:0.35
Number of iterations=100
Function tolerance: 0.0001
deteriorates the economic feasibility of FCLs. So, it is recommended to
Constraint tolerance: 0.001
restrain the number of FCLs to the minimum.

Table 18
6. Conclusion
Comparison between PSO, GA and IPNLP in solving FCLs allocation problem

In this paper, FCLs are optimally allocated to mitigate the negative Location FCL reactance (X), p.u

impacts of DGs on a variable-topology distribution network. A proposed MOPSO MOGA IPNLP MOPSO-guided
multi-objective PSO-based optimization scheme is applied to maximize IPNLP
the mitigation effect of FCLs at minimum cost. A particular focus is kept DG at Bus 10 5.0686 6.1 6.777 4.14
on assuring optimal DOCRs coordination with DGs and for all network DG at Bus 12 5.6564 5.002 8.066 3.38
DG at Bus 19 3.0771 7.871 6.777 5. 41
topologies. So, the coordinated operation constraints of main and
Summation of X of FCLs 13.8 p.u 18.973 p.u 21.6 p.u 12.9 p.u
backup relays for each possible topology and fault location are imposed F1 (absolute current 4300 A 5120 A 5700 A 1900A
into the FCLs allocation problem. Besides, to achieve the DOCRs co- difference index)
ordination at minimum FCLs cost, it is proposed to re-adjust a selected F2 (total FCLs cost) 2.9 M$ 3.77 M$ 3.85 M$ 2.81 M$
small percentage (in order of 10%) of the DOCRs in the protection Computation time 75 min. 56 min. 43 min. 43 min.

system for one time at nearly zero cost. Two scenarios are analyzed for
placement of FCLs. The first is to insert FCLs in-series to DGs only. The
objective functions values, the current difference index (F1), and the
second is to insert the FCLs in any branch in the network. For the case
FCLs cost (F2). However, when the proposed hybrid limited resetting-
study system assuming inductive type FCLs, it is found that inserting
FCLs approach is applied, the results of both FCLs placement scenarios
FCLs only in-series to DGs shows better results in terms of the two

Fig. 9. Fault current level after using stipulated 3 hybrid FCLs

12
A. Elmitwally, et al. Electric Power Systems Research 186 (2020) 106417

Table 19
Performance comparison of DOCRs coordination maintenance strategies by FCLs
Case FCL Location Type of FCL Is number of FCLs restricted? Resetting No. of FCLs F1, A F2, M$

5.3.2 In series to DGs inductive no no 3 4300 2.9

5.3.3 In series to DGs inductive no yes 3 2802 2.08
5.3.4 unlimited inductive no no 29 7900 13.3
5.3.5 unlimited inductive no yes 17 6960 8.2
5.3.6 unlimited inductive yes yes 3 1980 1.51
5.3.7 unlimited hybrid yes yes 3 1620 1.26

got better in terms of F1 and F2 values. But, the in-series to DGs only Fault Current Limiter in Electrical Distribution Grid, IEEE Transactions on Applied
scenario is still superior. Nevertheless, when the number of FCLs to be Superconductivity (July 2007).
[16] P Yu, B Venkatesh, A Yazdani, BN Singh, Optimal location and sizing of fault cur-
installed is restricted to only 3 like the number of DGs in the system, the rent limiters in mesh networks using iterative mixed integer nonlinear program-
scenario of inserting FCLs in any branch became the better scenario as ming, IEEE Transactions on Power System 31 (6) (2016) 4776–4783.
F1 and F2 values got drastically lower. In addition, when hybrid re- [17] W. El Khattam, T. Sidhu, Restoration of Directional Overcurrent Relay Coordination
in Distributed Generation Systems Utilizing Fault Current Limiter, IEEE
sistive-inductive FCLs are used, the performance of inserting FCLs in Transactions on Power Delivery 23 (2) (April 2008).
any branch scenario is further improved. It is emphasized that resetting [18] R. Chabanloo H. Abyaneh A. Agheli H. Rastegar, Overcurrent Relays Coordination
DOCRs only, without FCLs, could retain successful coordinated opera- Considering Transient Behavior of Fault Current Limiter and Distributed Generation
in Distribution Power Network, IET Gen. Trans. Dis. 5 (9) (2011) 903–911.
tion of relays under DGs. But, it can neither alleviate escalated elec- [19] H. Javadi, S. Mousavi, M. Khederzadeh, A Novel Approach to Increase FCL
tromagnetic stresses on system components nor prevent necessary re- Application in Preservation of Overcurrent Relays Coordination in Presence of
placement of protective devices at very high cost. As a future work, the Asynchronous DGs, Electrical Power and Energy Systems 44 (2013) 810–815.
[20] A. Elmitwally, Hybrid Superconducting Fault Current Limiter, International Journal
approach presented in this paper can be modified to include both dis-
of Electrical Power and Energy Systems 31 (10) (November 2009) 619–625.
tance and overcurrent relays in protection coordination. [21] S. Hemmati, J Sadeh, Applying Superconductive Fault Current Limiter to Minimize
the Impacts of Distributed Generation on the Distribution Protection Systems, 11th
References International Conference on Environment and Electrical Engineering (EEEIC’11),
2012.
[22] A. Elmitwally, S.Eladawy E.Gouda, Optimal Allocation of Fault Current Limiters for
[1] V. Pandi, H. Zeineldin, Weidong Xiao, Determining Optimal Location and Size of Sustaining Overcurrent Relays Coordination in a Power System with Distributed
Distributed Generation Resources Considering Harmonic and Protection Generation, Alexandria Engineering Journal (2015).
Coordination Limits, IEEE Transactions on Power Systems, Vol 28 (2) (2013) [23] D. Ibrahim, E. Abo El Zahab, S. Mostafa, New Coordination Approach to Minimize
1245–1254. the Number of Re-Adjusted Relays When Adding DGs in Interconnected Power
[2] J. Momoh, Y. Xia, G. Boswell, An Approach to Determine Distributed Generation Systems with a Minimum Value of Fault Current Limiter, Electrical Power and
(DG) Benefits in Power Networks, " 40 th North American Power Symposium (2008). Energy Systems 85 (2017) 32–41.
[3] M. Hussain, I. Musirin, A. Abidin, S. Rahim, "Directional Overcurrent Relay [24] HT Yang, WJ Tang, P Lubicki, Placement of Fault Current Limiters in a Power
Coordination Problem Using Modified Swarm Firefly Algorithm Considering the System Through a Two-Stage Optimization Approach, IEEE Transactions on Power
Effect of Population Size," 2014 IEEE 8th International Power Engineering and Systems Vol.33 (1) (2018) 110–131.
Optimization Conference (PEOCO2014), Langkawi, The Jewel of Kedah, Malaysia, [25] L. Huchel, H. Zeineldin, E.F El-Saadany,’’ Protection Coordination Index
24-25, March 2014. Enhancement Considering Multiple, DG Locations Using FCL,’’, IEEE Transactions
[4] A. Urdaneta, R. Nadira, L. Jimenez, Optimal Coordination Of Directional on Power Delivery Vol.32 (1) (2017) 344–350.
Overcurrent Relays In Interconnected Power Systems, IEEE Transactions on Power [26] S Ghaemi, M Abapour, Effect of Fault Current Limiter (FCL) on reliability and
Delivery 3 (3) (1988). protection coordination of distribution system, 24th Iranian Conference on
[5] R. Perveen, N. Kishor, R. Mohanty, Fault Detection and Optimal Coordination of Electrical Engineering (ICEE), Shiraz (2016) 727–731.
Overcurrent Relay in Offshore Wind Farm Connected to Onshore Grid with [27] M.E. Hamidi, R.M. Chabanloo, Optimal Allocation of Distributed Generation With
VSC–HVDC, International Transactions On Electrical Energy Systems (2015). Optimal Sizing of Fault Current Limiter to Reduce the Impact on Distribution
[6] A. Srivastava, J. Tripathi, Optimal Overcurrent Relay Coordination with Distributed Networks Using NSGA-II, IEEE Systems Journal 13 (2) (June 2019) 1714–1724.
Generation Using Hybrid Particle Swarm Optimization–Gravitational Search [28] A. Esmaeili Dahej, S. Esmaeili, H. Hojabri, Co-Optimization of Protection
Algorithm, Electric Power Components and Systems (2016) 1–12. Coordination and Power Quality in Microgrids Using Unidirectional Fault Current
[7] DS Alkaran, MR Vatani, MJ Sanjari, Optimal Overcurrent Relay Coordination in Limiters, in, IEEE Transactions on Smart Grid 9 (5) (Sept. 2018) 5080–5091.
Interconnected Networks by Using Fuzzy-Based GA Method, IEEE Transactions on [29] STANDARD IEEE, Inverse Time Characteristic Equations for Overcurrent Relays,
Smart Grid (2018). IEEE Transactions on Power Delivery 14 (3) (July 1999).
[8] A. Srivastava, J. Tripathi, Optimal Coordination of Overcurrent Relays using [30] H.H. Zeineldin, E.F. El-Saadany, M.M.A. Salama, Optimal coordination of over-
Gravitational Search Algorithm with DG Penetration, IEEE Transactions on Industry current relays using a modified particle swarm optimization, Electric Power
Applications (2014). Systems Research 76 (2006) 988–995.
[9] M. Singh, Protection Coordination in Grid-Connected & Islanded Modes of Micro- [31] Seyedali Mirjalili, Siti Zaiton, A New Hybrid PSOGSA Algorithm for Function
Grid Operations, IEEE Innovative Smart Grid Technologies Asia, Bangalore Optimization, International Conference on Computer and Information Application
(November 2013) 10–13. (ICCIA), 2010.
[10] A. Urdaneta, L. Perez, H. Restrepo, Optimal Coordination of Directional [32] Y. del Valle, G. Venayagamoorthy, S. Mohagheghi, J. Hernandez, Particle Swarm
Overcurrent Relays Considering Dynamic Changes in The Network Topology, IEEE Optimization: Basic Concepts, Variants, and Applications in Power Systems, IEEE
Transactions on Power Delivery 12 (4) (Oct 1997). Transactions On Evolutionary Computation 12 (2) (April 2008).
[11] A. Noghabi, J. Sadeh, H. Mashhadi, Considering Different Network Topologies in [33] M. Reyes sierra, C. Coello, Multi Objective Particle Swarm Optimizers: A Survey of
Optimal Overcurrent Relay Coordination Using a Hybrid GA, IEEE Transactions On the State of the Art, International Journal of Computational Intelligence Research 2
Power Delivery 24 (4) (2009). (3) (2006) 287–308.
[12] M. Yang, A. Liu, Applying Hybrid PSO to Optimize Directional Overcurrent Relay [34] Univ. Washington, "Power Systems Test Case Archive," Seattle, Mar. 2006.[Online].
Coordination in Variable Network Topologies, Journal of Applied Mathematics Available: http://www.ee.washington.edu/research/pstca.
(2013). [35] Kalyanmoy Deb, Multi-Objective Optimization Using Evolutionary Algorithms,
[13] H. Zeineldin, W. Xiao, Optimal Fault Current Limiter Sizing for Distribution Systems England: John Wiley & Sons, Chichester, 2001.
with DG, IEEE Power and Energy Society General Meeting (2011). [36] R.H. Byrd, Mary E. Hribar, Jorge Nocedal, “An Interior Point Algorithm for Large-
[14] G. Khan, C. Fatma, G. Bagriyanik, M. Bagriyanik, The effect of fault current limiters Scale Nonlinear Programming, SIAM Journal on Optimization 9 (4) (1999)
on distribution systems with wind turbine generators, Int. J. Renew. Energy Res. 3 877–900.
(1) (2013).
[15] L. Ye, M. Majoros, T. Coombs, A. Campbell, System Studies of the Superconducting

13