Professional Documents
Culture Documents
Optimized Allocation of Fault Indicators For Improving The Recloser Fuse Coordination RG
Optimized Allocation of Fault Indicators For Improving The Recloser Fuse Coordination RG
Optimized Allocation of Fault Indicators For Improving The Recloser Fuse Coordination RG
net/publication/365811236
CITATIONS READS
2 44
4 authors, including:
All content following this page was uploaded by Fernanda C L Trindade on 08 December 2022.
Abstract—New technologies are currently delivering protective Tgk time of fast recloser attempt to detect a phase-to-
devices that can enhance the design of protection coordination ground fault at bus k.
in modern distribution systems. This paper proposes an integer
Lks binary matrix that indicates whether a bus k protected
linear programming (IP) model to find the optimal location of a
on
Fault Transmitter (FT) in the distribution system. The proposed by an upstream fault transmitter s or not.
IP model provides the optimal FT allocation considering the Dmax maximum distance between a fault transmitter s and
minimization of sustained interruption frequency and momentary the upstream recloser.
interruption duration. A study of the appropriate number of FT
Smax maximum number of fault transmitter that can be
reveals the best alternative in the proposed adaptive protection
allocated into the network.
si
scheme. A Monte Carlo simulation with a richer representation
of stochastic fault scenarios is used for further evaluation of the N number of network customers.
solutions obtained with the proposed IP model. The combined nj number of downstream customers of fuse j | ∀j ∈ Ωf .
strategies allow full-scale studies of the optimal allocation of
r
FT in distribution systems. The results of studies using two test
ds distance between a fault transmitter s and the recloser.
systems certify the benefits of the proposed strategy in fuse-saving Functions:
coordination and in reducing the total duration of the momentary
interruption, which can cause power quality problems.
Index Terms—Allocation of fault transmitters, adaptive pro-
tection, integer linear programming, Monte Carlo simulation.
RT
IF
xs
Ve
total protection operation time.
average interruption frequency of the network.
Binary Variables:
fault transmitter allocation with the bus s.
ed
zks protection of node k by a fault transmitter s.
N OTATION yksj fuse-blowing for a fault at bus k that is protected by a
Optimization Model fault transmitter s and a fuse j.
Sets and Indexes:
itt
TFgkj melting time of fuse j for a phase-to-ground fault at type of a symmetric or an asymmetric fault at bus k.
bus k. sf number of saved fuses.
TSks time of fault transmitter s to detect a three-phase fault bf number of blown fuses.
at bus k. nf number of fast recloser attempts.
TSgks time of fault transmitter s to detect a phase-to-ground nd number of delayed recloser attempts.
fault at bus k. nt number of total recloser attempts; nt = nf + nd.
Tk time of fast recloser attempt to detect a three-phase tf kj melting time of fuse j for a fault at bus k.
fault at bus k. tf k minimum of {tf k1 , · · ·, tf kj }.
tsks time of fault indicator s to detect a fault at bus k.
The authors kindly acknowledge the support of the São Paulo Research
Foundation (FAPESP), Grants No. 2020/13578-4, 2016/08645-9, 2020/09838- tsk minimum of {tsk1 , · · ·, tsks }.
0, and 2021/11380-5, and the CNPq, Grants No. 422765/2021-0 and t̂k fast recloser attempts to detect a fault at bus k.
304373/2020-6.
The authors are affiliated with the University of Campinas – UNI- t̃k delayed recloser attempts to detect a fault at bus k.
CAMP, Campinas, Brazil (e-mail: cdsantos@ieee.org; fernanda@ieee.org; R electrical resistance of phase-to-ground and a phase-
chrlyra@unicamp.br), and the Mount Sinai Hospital Mount Sinai Heart, New
York, United of States (e-mail: petra.bartmeyer@gmail.com).
to-phase-to-ground faults.
2
on
of the protection scheme, adapting the protection coordination,
rial problem, and the best candidate sites for fault indicators and using the set of equipment FT-FR proposed in [11] to
can be delivered by heuristic computing techniques [4]- [7], save fuses by improving the recloser-fuse coordination. An
evolutionary computing [8], and commercial solvers [9], [10]. optimal solution defines the places and the number of FT of a
These previous papers are based on permanent faults and they given distribution system. The FR is considered to be installed
si
showed that the optimal allocation depends on the location of together with the substation relay and every recloser of the
protection devices and fault rates. distribution feeder. It can lead to forcing or inhibiting their
An improvement of fault indicators is the Fault Transmitter operation based on information provided by the FTs and preset
r
(FT)-Fault Receiver (FR), which can help fault location, but logic rules. The IP model considers all protection coordination
also aid adaptive protection. The set of equipment FT-FR
can speed up the overcurrent protection scheme by inhibiting
or forcing the operation of a protection device; in case of
permanent faults, the set of FT can also provide the fault
location information and support the network restoration time.
Ve
constraints, relying on three-phase and phase-to-ground faults
to represent the fault-state operation of the distribution system.
The benefits of increasing the number of FT are evaluated
using the Pareto frontier defined by the trade-offs between
the minimum interrupting time of the protection scheme and
As in the fault indicator allocation problem, the choice of sites the number of FT allocated into the distribution system. The
ed
to allocate the FT devices strongly affects the performance minimum interruption frequency is based on the number of
of the protection schemes, power quality indexes, and eco- affected customers, as in the SAIFI index. The proposed
nomic aspects. methodology can also be adapted to consider other decision
Few studies investigate the use of the set of equipment FT- criteria such as line segment type (overhead, cable), terrain,
itt
FR to improve the operation time of overcurrent protection weather, age of equipment, and chemical pollutants. The 85-
in recloser-fuse coordination. The authors in [11] use FT- bus and 1080-bus test systems from [14] are used to verify
FR combined with a protection device (for instance, recloser the performance of the approach. The case studies and a
or relay) to detect and send fault information using high- Monte Carlo simulation with a richer representation of the
bm
speed wireless communication. The information from FT- stochastic nature of the faults certify the benefits of the
FR equipment, combined with predefined settings, drives the proposed strategy in fuse-saving coordination and in reducing
recloser in choosing the best response to clear a fault. As the total duration of the momentary interruption, which can
detailed in [11], the set of equipment FT-FR can indicate that cause power quality problems.
a fault is on an unfused line section; therefore, the recloser The main contributions of this paper are as follows: a) the
Su
allows a faster trip because there is no need to coordinate the design of an integer linear programming model that can reveal
recloser operation with the fuse operation. These aspects can optimized locations for a set of equipment FT; b) the proposal
make FT intuitive allocations but not as efficient as desired. of a multi-objective analysis for the FT allocation problem
Because of that, the FT allocation problem is not addressed in to assess decisions about the maximum number of FT; and
[11], as the installation buses of the equipment are defined by c) the evaluation of the set of equipment FT-FR technologies
hand, without guarantee of the best efficiency for the combined as alternatives to improve the coordination between recloser
FT-FR system and the protection devices. and fuses in a smart grid perspective. The combined strategies
The authors in [12] proposed an adaptive protection method- allow full-scale studies of the optimal allocation of FT in
ology based on optimization techniques. The proposal con- modern distribution systems.
siders the set of equipment FT-FR without optimize their
allocation into the network. In [12], the FT allocation problem
II. A N OVERVIEW OF P ROTECTION C OORDINATION
addresses the lateral branch fuses as installation bus candidates
and provides a decision-making based on the most usual con- At the planning stage, the protection scheme is designed to
figuration for energy flow. A two-stage optimization strategy satisfy coordination and selectivity between a set of protection
is proposed: a mixed-integer linear programming model is devices to improve the network reliability and reduce the
3
on
melt; a fault current greater than the imax may follow the fuse
blow before or during the recloser attempts [21]. Figure 1–c
illustrates these ideas.
The integration of the distributed energy resource (DER)
si
into the network can contribute to the recloser-fuse miscoor-
dination [22]– [23]. The additional fault current contribution
from DER (i2 in Fig. 1–b) can make i greater than imax ,
r
Figure 1. Recloser-fuse coordination: (a) without DER; (b) with DER; and which may cause the fuse to blow before the recloser opens
(i1 ≤ imax ≤ i) [24]– [25]. Figure 1–b and Fig. 1–c illustrate
(c) time-current curve of the recloser-fuse coordination.
t0 t Time [ms]
Δ0 Δ1 Δ2 Δ3
ts off
ts off
ts off
Fault Cleared*
On
FT-FR
Off
On
Rcl.2
Off
t0 t* Time [ms]
Δ*0 Δ*1 Δ*2 Δ*3
Figure 2. Example of improving the protection coordination (the crosshatched areas represent the time required for the FT to communicate with the FR and
operate the recloser — ∆∗0 ).
on
Table I The IP model uses information about three-phase and phase-
C OMPARISON OF OPERATION TIMES . to-ground faults to provide the optimal topology for a set
of FT. The proposed model is an approximation for the real
Expected
Adapted behavior as it excludes the stochastic nature of the; it is only
Attempt Operation Times Faster
si
Response — Rcl.2
Rcl.1 FT-FR considered the worst-case scenario. Thus, it is unnecessary to
consider all the information about faults (for instance, phase-
1 ∆0 ∆∗0 FT-FR ∆∗0
∆∗0 + tsoff ∆∗1 = ∆1 to-phase and phase-to-phase-to-ground faults) as it entails
r
2 ∆1 Rcl.1
3 ∆2 ∆∗0 + tsoff FT-FR ∆∗2 = ∆∗0 + tsoff an extra computational burden without translating into better
∆∗0 + tsoff ∆∗3 = ∆∗0 + tsoff
4 ∆3 FT-FR
before completing the short-circuit current detection, and the changes when considering temporary reconfiguration. In case
∆1 and ∆∗1 times are equal. It is worth remarking that the set of having a new permanent configuration of a given network,
of equipment FT-FR forces the recloser to open only when ∆∗0 the IP model can deliver new optimal locations for reinstalling
or ∆∗0 + tsoff is shorter than the coincident attempt time of the the set of equipment FT considering that FT can be easily
bm
III. T HE IP M ODEL FOR THE FT A LLOCATION P ROBLEM radial distribution system. The sites of essential customers and
The proposed model assumes a greenfield perspective in the facilities with low reliability can point to allocation priorities;
planning of the adaptive protection coordination and supposes however, this information is not always available.
that the set of equipment FT-FR can improve the time response The solution space of the allocation of the FT problem is
of the recloser-fuse selectivity [11]– [13]. The parameters of represented by all downstream buses of the recloser (Ωs ),
the set of protection devices (recloser and fuses) are defined where the benefit to allocate a FT is evaluated. Note that
for the first attempt of three-phase and phase-to-ground faults the cardinality of the set Ωb is the same as Ωs , and the
before integrating the DER and, consequently, the time current greater the number of buses in the protection zone of the
curves of this equipment set have preset values. Another recloser, the higher computational burden entailed. The FT
assumption is that the DER contributes to the fault current protection zones depend on the presence of DER. In radial
if this event occurs in any bus of the set Ωb . It is also worth networks without DER, the set of buses downstream from the
remarking that the proposed IP model aims to provide optimal FT define the protection zone. For radial networks with DER,
sites for FT allocations on a given topology of the distribution the protection zone is given by Ωs because of the possibility
system; the FR location is considered fixed with the substation of direct and reverse current flows. Both cases can be tackled
recloser. by commercially available equipment FT [27].
5
on
conflicting; therefore, a given solution that reduces the total , (8)
X
operation time also reduces the number of blown fuses. The xs ≤ Smax
(9)
use of RT and IF in (1) aims to a broader evaluation of the s∈Ω
FT locations. zks ≤ Lks , ∀k ∈ Ωb , s ∈ Ωs (10)
si
min {IF + RT} (1) zks , yksj , xs ∈ {0, 1}, ∀k ∈ Ωb , ∀s ∈ Ωs , ∀j ∈ Ωf (11)
The average interruption frequency of the distribution sys- Constraint sets (4)-(5) compute the operation of the protec-
tem (IF) is computed according to the blown fuses (yksj ) tion coordination for phase-to-ground and three-phase faults,
r
and the number of customers in the whole network (N ) and respectively. In both equations, the fuse j blows if its melting
each downstream bus of the fuses j (nj ). In (2), the binary
variable yksj indicates that a given fault (three-phase or phase-
to-ground) at bus k is seen by the fault transmitter s and
blows the fuse j; note that the number of customers nj is
computed for all nodes that are protected by an upstream
Ve
time for a fault at bus k is shorter than the fast recloser
attempts and the communication time between FT and FR,
s, that detects the same fault (yksj = 1); otherwise, the fault
is cleared and the distribution system returns to the steady-
state operation (yksj = 0). Note that in (4)-(5) the time of the
ed
fuse j. The IEEE Standard 1366-2012 [29] defines the SAIFI recloser is only computed when there is no FT s which detects
as the average number of customers disconnected from the the same fault at bus k (zks = 0); otherwise, the time of the
network for a given period (greater than 5 minutes). The use recloser is not considered (zks = 1). Additionally, M is a large
of the SAIFI enables the IP model to consider the effects of a constant value that assures the inequalities (4)-(5) whenever
given fault without considering the time required to reconnect yksj = 1 (for instance M = max{TSgks , Tgk }. Equation (6)
itt
the affected customers; it is assumed that fuse replacement limits to one the number of FT that can protect the bus k. This
requires more than 5 minutes. Although the reconnecting time constraint set secures that the recloser only be activated by the
could be included in the formulation, it would add much FT with the smallest detection time regarding the presence of
complexity to a model that already requires a large number
bm
DER in the network. Constraint set (7) uses the variable zks
of details. to relate the protected bus k to the location of FT (xs ). In (8),
1 X X X the maximum wireless coverage distance (Dmax ) to allocate
IF = nj yksj (2)
N a fault indicator is imposed. Constraint (9) limits the number
k∈Ωb s∈Ωs j∈Ωf
of allocated FT (xs ) in the network — the study described
Equation (3) computes the total interrupting time, as the in Sec. III-E shows the appropriate value of Smax . Constraint
Su
time attempts of the recloser for the three-phase (Tk ) and (10) imposes the network configuration, i.e., that a fault at bus
phase-to-ground (Tgk ) faults and each allocated FT also con- k can only be detected by a fault indicator allocated at bus s if
sidering the time to detect three-phase (TSks ) and phase- Lks = 1. In (11) the binary nature of the variables zks , yksj ,
to-ground (TSgks ) faults. All information about the electrical and xi are imposed.
characteristic of the short-circuit current magnitude is com-
puted during the preprocessing step, the times Tk and Tgk are
D. Proposed IP Model
computed from the IEC 60255 Standards [30], and the times
TSks and TSgks are provided as proposed in [11]. The RT To sum up, the FT allocation optimization problem that
component of the objective function comprises two terms: the reveals the best location for FT is given by (12).
first computes the operation time of the adaptive protection
scheme (zks = 1), while the second term considers the full min (1)
recloser operation time (zks = 0).
subject to: (4) − (11) (12)
Commercial solvers for integer linear programming prob-
X X
RT = [(TSks +TSgks )zks +(Tk +Tgk ) (1 − zks )] (3)
k∈Ωb s∈Ωs
lems can deliver optimal solutions for (12).
6
Algorithm 1: The Pareto Frontier for Smax A. Setting the Recloser Operation
input : Ωf , Ωb , and Ωs The first procedure simulates the recloser attempts by
output: Pareto Frontier defining the number of fast (nf) and delayed (nd) current
1 // Creating the Smax candidates curves, always considering fast curves first. The total number
2 define: W = {1, 2, · · · , |Ωs |} of recloser attempts (nt) is defined as nt = nf + nd, where
3 for w ∈ W do
nf ∈ {1, 2} and nd ∈ {0, 1, 2}. The parameters for the recloser
4 set: Smax = w
operations are based on protection standards and selected with
5 // Solving the proposed IP model
6 RTw ← solution of (3) from the IP model (12) a random uniform distribution.
7 save point (w, RTw ) of Pareto Frontier
B. Drawing the Location and Nature of the Fault
The second procedure draws the location (k), the type of
E. Searching for an Appropriate Value for Smax fault (k t ), and computes the fault current magnitude using
short-circuit analysis. Both approaches are random and drawn
The decision on the maximum number of FT (Smax ) de-
from given data: location k is selected from the options in the
pends on the available technology and should consider the
set Ωb , and the type of the fault k t is defined from the closed
on
coverage of the most significant number of customers in
interval [0%, 100%]. For the type of fault, the classification is
the network [9]. The conflict between the network coverage
based on the value of k t and accurate probabilities of occur-
and the number of FT allocation entails a complex decision-
rence of a phase-to-ground (pphg ), phase-to-phase-to-ground
making problem. The multi-objective approach proposed here
(p2phg ), phase-to-phase (p2ph ), and three-phase fault (p3ph ).
si
is an alternative to assess these problems; the main building
The probability values used here are provided as input to the
blocks are the total interrupting time of protection coordina-
method. To compute the fault current magnitude, the phase-
tion (RT).
to-ground and phase-to-phase-to-ground fault studies consider
r
The purpose is to obtain the trade-off between the mini- a random uniform distribution for fault resistance of up to
mization of RT and the number of FT to allocate into the
distribution systems (Smax ). The rationale behind the study
of the multi-objective approach is that the trade-off curve
provides information that would help utilities to decide on
the appropriate number of FT allocation to be allocated; this
Ve
40 ohms.
objective function to be minimized and the maximum number between nd and nf. The Monte Carlo study evaluates the
of allocated FT (Smax ) is limited in (12). cases according to the next two hypothesis: the first case
To obtain the Pareto frontier, the multi-objective optimiza- considers that the set of recloser attempts only comprises fast
tion process considers different scenarios for the number of operations (nt = nf and nd = 0); the second case considers
bm
FT Smax . The Smax value varies in the set W , composed by both the fast and the delayed operations (nt = nf + nd)
numbers between a single device (say, w = 1) and w = |Ωs | performed by the recloser to detect and, if possible, clear the
(a safe upper bound). For each value of Smax , the associated fault. Note that phase-to-phase and phase-to-phase-to-ground
interrupting time of protection scheme RTw is computed faults must have detection times shorter than those used by
with (3). Algorithm 1 summarizes the computational process the proposed IP model; phase-to-ground and three-phase faults
Su
to reveal the trade-off between RT and number of FT. may have detection times less than or equal to the detection
times addressed by the IP model.
In the first case, considering nf = 1, the fuse is saved if the
IV. E VALUATION WITH M ONTE C ARLO S IMULATION
minimum time to detect the fault between the fast attempt
The optimal solution for the FT allocation problem delivered of the recloser (t̂k ) and the communication time between
by the IP model is certified with a Monte Carlo simulation, FT and FR (tsk ) is lower than the melting time of the fuse
which considers a rich representation of stochastic scenarios (tf k ); otherwise, the fuse blows. To simulate recloser multiple
and uses three complementary procedures. The first procedure attempts, the circuit is disconnected at least once; therefore,
(Sec. IV-A) selects the sequence of fast and delay attempts the recovery time tsoff is added to the communication time
of the recloser. The second procedure (Sec. IV-B) defines the between FT and FR (Sec. II-B) and the contribution of the
nature of the faults. The third procedure (Sec. IV-C) computes DER to the fault is not considered for computing the following
the times of the adaptive protection operation with the set detection time because the anti-islanding protection is assumed
of equipment FT-FR to evaluate if the fuse is saved for the to disconnect the DER reclosing. In this case, the fuse is
designed scenario. The three procedures are detailed in the considered saved if the minimum time to detect the fault
next sections. between t̂k and tsk +tsoff is lower than tf k ; otherwise, the fuse
7
blows. Note that the times of fuse and the set of equipment FT- Algorithm 2: Monte Carlo Simulation
FR are computed as the minimum value of the sets of the of input : Ωf , Ωb , Ω∗s , sf ← 0, and bf ← 0
fuse times {tf k1 , · · ·, tf kj|j=|Ωf | } and the fault indicator times output: RT, sf, and bf
{tsk1 , · · ·, tsks|s=|Ωs | } for the given type of fault k t . 1 iter ← 0
The second case considers the fast and delayed operations 2 while iter < max_iter do
(nt = nf + nd) performed by the recloser to detect and 3 // First procedure
may clear the fault of the network. The fuse is considered 4 nf ← randomly uniform choose from {1, 2}
saved if the minimum time to detect the fault between the 5 nd ← randomly uniform choose from {0, 1, 2}
delayed attempt of the recloser (t̃k ) and the communication 6 // Second procedure
time between FT and FR (tsk +tsoff ) is lower than the melting 7 k ← randomly uniform choose from Ωb
time of fuse (tf k ); otherwise, the fuse blows. The worst case 8 kt ← randomly uniform choose from [0, 100%]
here is represented by the delayed attempt of the recloser and 9 if kt ≤ pphg then
10 R ← randomly uniform choose from [0, 40]
the communication time between FT and FR after the network 11 kt is a phase-to-ground fault
is out during the last attempt of the recloser. 12 else if pphg < kt ≤ (pphg + p2phg ) then
For both cases, when the fuse is saved, the protection 13 R ← randomly uniform choose from [0, 40]
kt is a phase-to-phase-to-ground fault
on
operation time is computed by the equations A1, A2, and A3 14
(Alg. 2 lines 28, 34−35, and 43−45, respectively), focusing on 15 else if (pphg +p2phg ) < kt ≤ (pphg + p2phg + p2ph ) then
evaluating the benefits to speed up the protection coordination 16 kt is a phase-to-phase fault
17 else
time for temporary faults. In contrast, the IF objective is only
18 kt is a three-phase fault
computed when the fuse is blown, disregarding the network
si
restoration time. Algorithm 2 summarizes the computation 19 Compute the fault current magnitude
process of the Monte Carlo simulation. 20 // Third procedure
21 tf k ← min{tf k1 , · · · , tf kj }
r
22 tsk ← min{tsk1 , · · · , tsks }
V. T ESTS AND R ESULTS 23 if nd = 0 then
else
sf ← sf + 1
RT ← RT + min{t̂k , tsk } + 2[ms] (A1)
16GB using Ubuntu 20.04. The solver Gurobi delivered an
ed
30 bf ← bf + 1
optimal solution within seconds for all the instances. The
31 else
case study considers the placement of FT in two different
32 if min{t̂k , tsk + tsoff } < tf k then
test systems: 85-bus and 1080-bus benchmark; all data of the 33 sf ← sf + 1
distribution systems are provided by [14]. Table II shows the RT ← RT + (min{t̂k , tsk }
itt
34
recloser parameters, the protection coordination settings, and 35 +(nf − 1) · min{t̂k , tsk + tsoff } + nt · 2[ms])
the electrical distance between the DER and the substation (A2)
for both test systems. For instance, the maximum electrical 36 else
37 bf ← bf + 1
distance in the 85-bus is 19.6940 Ohms (the distance from
bm
curve from the IEC 60255 [30]; the time multiplier settings are 44 +(nf − 1) · min{t̂k , tsk + tsoff }
defined as 0.05 for fast attempts and 0.15 for delayed attempts 45 +nd · min{t̃k , tsk + tsoff } + nt · 2[ms]) (A3)
(a safe coordination for the downstream fuses). The recloser 46 else
intervals (contacts open) are defined as 2 [ms] to compute the 47 bf ← bf + 1
total operation time. Table III shows the bounds of sets Ωf
48 iter ← iter + 1
and the number of fuses in each network. For the 85-bus test
system, a total of 7 fuses are used to cover 68% of the buses
on laterals. For the 1080-bus test system, a total of 40 fuses
are used to cover 71% of the buses on laterals. adopted from [11] tsoff = 50 [ms]. The value of M in equation
Having defined the recloser branch, the cardinalities of sets (4) and (5) is |Ωb | × |Ωs | × |Ωf |, as indicated in Sec. III-C.
is defined as |Ωb | = |Ωs | = 80 for the 85-bus test system and The probabilities of the event of faults are defined as stated
|Ωb | = |Ωs | = 1079 for the 1080-bus test system. The maximum in [35]; the values are as follows: pphg = 70%, p2ph = 15%,
distance between the recloser and a given FT (Dmax ) and the p2phg = 10%, and p3ph = 5%.
detection times of the communication time between FT and FR The following subsections present the four studies per-
(TSk and TSgk ) are computed as presented in [11]; it is also formed for each test system. The first study (Sec. V-A) applies
8
on
r si
1080-bus test system.
Table II
Ve
Figure 3. Solutions of the adaptive protection scheme with the set of equipment FT-FR and recloser-fuse coordination: a) 85-bus without DER; b) 1080-bus
without DER; c) 85-bus with DER; d) 1080-bus with DER; e) the Monte Carlo simulation for the 85-bus test system; f) the Monte Carlo simulation for the
S ET OF FUSES .
considered in the computation of the parameters used in the
Test System
Lower Upper
Number of Fuses
IP model, and the fuse-saving coordination is adopted in the
Bound of Ωf Bound of Ωf protection scheme for both test systems. Set W consists of all
85 15K 50K 7 candidates in Ωs for the 85-bus test system and a small portion
1080 15K 40K 40 of the Ωs for the 1080-bus test system; the last assumption
Su
on
can improve the selectivity time and save fuses. A small provides the best curve coverage rate when considering 85
number of FT provides greater benefits to save fuses in both and 1080-bus systems; therefore, this value is used in the
test systems, and from four devices, the original protection Monte Carlo simulations. These topologies of FT provide a
selectivity is established. The RT function has improvements 17.52% reduction in the RT and a 94.34% reduction in the
with fewer FT than in the first study case (without DER), IF objective compared to the original protection scheme of
si
showing that using the set of equipment FT-FR and recloser the 85-bus test system (without FT). In comparison, the 1080-
approach can handle the presence of the DER to provide the bus test system has a 29.10% reduction in the RT objective
quality places of the FT and improve the performance of the
r
and a 83.05% reduction in the IF objective compared to
protection scheme. Figure 3 – c and Fig. 3 – d illustrate these the original protection scheme (|Ω∗s | = |Ωb |). These results
results. As observed in the previous study, the number of FT
increases, and the location of the devices moves from regions
close the DER to positions closer to the lateral fuses. Ve
reinforce that the proposed methodology for the allocation of
the fault indicator is effective in reducing the total interrupting
time and frequency. Note from Tab. IV that the Planned Cases,
Smax = 11, have IF and RT similar to the Utopian Cases —
Smax = |Ωs |.
C. Considering an Appropriate Value of Smax
ed
Figure 4 illustrates the locations of the set of equipment
The first two studies showed the benefits of the adaptive FT-FR and the existing of protection devices (say, recloser
protection approach when considering three-phase and phase- and fuses) in the planned scenarios for the 85 and 1080-bus
to-ground faults. These results indicate the maximum number systems. The locations were drawn by NetworkX 2.6.2 — a
of FT (Smax ) to be allocated by the IP model. The values
itt
the feeder (Fig. 4–a and Fig. 4–c), which allows the detection
that the inclusion of an extra IF (w + 1) will not entail major of a great number of faults; therefore, more customers can be
improvements in the RT objective. The shadow sites in Fig. 3– protected, and a significant number of saved fuses is achieved.
a and Fig. 3– b represent Pareto efficient solutions, meaning A feeder with DER causes changes the locations of the set
that none of the solutions are superior in both objectives to of equipment FT from the previous scenarios; now, the new
the other solutions in the area. The values in the shadow area locations are mostly arranged close to the bus where the DER
Su
are used to set the upper bound of Smax values with the best is installed, as illustrated in Fig. 4–b and Fig. 4–d. In case of
CR for both test systems – 14 for the 85-bus and 15 for the serial location, the devices in series can detect the same fault,
1080-bus. Considering the presence of DER, the maximum but the FR will only process the FT signal closest to the fault.
benefits are stated with a small number of FT (see Fig. 3–c Also, note from Fig. 4–b and Fig. 4–d that the equipment FT
and Fig. 3–d). Combining the results with and without DER, far from the DER was correctly located in spots where it is
the Smax can vary from 4 to 14 for the 85-bus test system and possible to detect the contributions of the substation and DER
from 4 to 15 for the 1080-bus test system. The appropriate to the short-circuit current.
values of Smax for both distribution systems are verified by
the Monte Carlo simulation described in Sec. V-D.
E. Addressing the Configuration Changes
The feeder reconfiguration can change the direction and
D. Using the Monte Carlo Simulation magnitude of current flows and the downstream current flow
The Monte Carlo simulations consider 10 thousand (Fig. 3- magnitude; consequently, the overcurrent protection settings
e) and 100 thousand (Fig. 3-f) iterations to assess the quality for the most used network configuration may not always
of the topologies delivered by the IP model (12) to 85-bus provide a proper protection to the entire network. When
10
on
r si
Ve
ed
itt
bm
Figure 4. Solutions of the FT allocation problem based on the recloser and fuses locations: a) 85-bus without DER; b) 1080-bus without DER; c) 85-bus
with DER; d) 1080-bus with DER.
Su
considering the installation of equipment FT, the lower ac- The 85-bus test system is used in this study. A feeder portion
curacy of the adaptive protection scheme is a possibility (see Fig. 5) that represents an 11-bus feeder with 381.90 kVA
that must still be thoroughly observed. The study addressed of demand (10.6% of the 85-bus demand) is transferred to
here investigates how the reconfiguration affects the adaptive the distribution system with the adaptive protection scheme
protection approach. The computational tests were designed already set. Regarding the previous aspects, there are 40 fea-
as following: (i) the reconfiguration is evaluated by adding sible reconfiguration scenarios without DER and 38 with DER.
a feeder portion at a given network bus; (ii) the radiality For each feasible scenario, the Monte Carlo simulation is
of the distribution network is preserved after changing the used to verify the performance of the proposed adaptive
configuration; (iii) the minimum voltage magnitude of the protection scheme after changing the network configuration;
network, v min , must be within a safety operation range (say, the simulation parameters are the same stated in Sec. V-D
0.93 p.u. ≤ v min ≤ 1.03 p.u.); and (iv) the bus where the and the FT locations are illustrated in Fig. 4–a and Fig. 4–b,
feeder portion is connected must be protected by at least one with Smax = 11.
upstream FT. The case that an FT is transferred to another
Under the evaluation conditions, the simulation studies
circuit after reconfiguration is not addressed in this study.
show that the reconfiguration does not significantly affect the
11
x09 x08 x10 x11 The next steps in this research will address the modeling of
Feeder
anti-islanding protections and the islanding operation schemes.
x07 x04
Bus
ACKNOWLEDGMENTS
x x01 x02 x03 x05 x06
The authors are indebted to Eng. Paulo Lima from the
Figure 5. The feeder portion transferred. Schweitzer Engineering Laboratories for the helpful comments
and for allowing them to draw upon his expertise in the set
of equipment FT-FR technologies. The authors also kindly
performance of the proposed adaptive protection scheme. The acknowledge the suggestions of their colleagues of the Energy
fault detection times with and without the set of equipment Network Optimization Laboratory.
FT varied up to 1% of the feasible scenarios in comparison
to the original configurations (without and with DER). The R EFERENCES
protection scheme with equipment FT outperforms the tradi- [1] F. J. Muench and G. A. Wright, “Fault indicators: Types, strengths amp;
tional scheme (without FT) for all reconfiguration scenarios applications,” IEEE Trans. Power App. Syst., vol. PAS-103, no. 12, pp.
3688–3693, 1984.
evaluated in this investigation. The average reduction of these [2] J.-H. Teng, W.-H. Huang, and S.-W. Luan, “Automatic and fast faulted
new scenarios for the RT objective is 13.83% for the case
on
line-section location method for distribution systems based on fault
without DER and 17.28% for the case with DER; in the indicators,” IEEE Trans. Pow. Syst., vol. 29, no. 4, July 2014.
[3] A. Shahsavari, S. M. Mazhari, A. Fereidunian, and H. Lesani, “Fault in-
original network configuration, the cases without and with dicator deployment in distribution systems considering available control
DER provide reductions of 13.46% and 17.52%, respectively. and protection devices: A multi-objective formulation approach,” IEEE
For the IF objective, no changes were observed in its Trans. Pow. Syst., vol. 29, no. 5, 2014.
si
[4] C.-Y. Ho, T.-E. Lee, and C.-H. Lin, “Optimal placement of fault
results for the different configurations evaluated without DER; indicators using the immune algorithm,” IEEE Trans. Pow. Syst., vol. 26,
therefore, the allocated set of equipment FT has saved fuses no. 1, 2011.
for all fault scenarios that may be affected by short-circuit [5] M. de Almeida, F. Costa, S. X. de Souza, and F. Santana, “Optimal
r
placement of faulted circuit indicators in power distribution systems,”
currents. For the scenarios with DER, fewer fuses are saved Elec. Power Syst. Res., vol. 81, no. 2, pp. 699–706, 2011.
after changing the network configuration, which provide an
average of a 6.27% reduction in the IF objective compared
to the original protection scheme of the 85-bus test system
(without FT).
Ve
[6] H. O. Cruz and F. Bertequini Leão, “Optimal placement of fault
indicators using adaptive genetic algorithm,” in IEEE General Meeting,
July 2017, pp. 1–5.
[7] S. Shi, B. Zhu, A. Lei, and X. Dong, “Fault location for radial
distribution network via topology and reclosure-generating traveling
waves,” IEEE Trans. Smart Grid, vol. 10, no. 6, pp. 6404–6413, 2019 .
[8] W. F. Usida, D. V. Coury, R. A. Flauzino, and I. N. da Silva, “Efficient
ed
VI. C ONCLUSIONS placement of fault indicators in an actual distribution system using
evolutionary computing,” IEEE Trans. Pow. Syst., vol. 27, no. 4, 2012.
An integer linear programming (IP) model for solving the [9] J. S. Acosta, J. C. López, and M. J. Rider, “Optimal multi-scenario,
multi-objective allocation of fault indicators in electrical distribution
allocation of fault transmitters (FT) and fault receivers (FR) systems using a mixed-integer linear programming model,” IEEE Trans.
into distribution systems has been developed. The model Smart Grid, vol. 10, no. 4, pp. 4508–4519, 2019.
itt
considers the integration of a set of equipment FT-FR to [10] M. Farajollahi, M. Fotuhi-Firuzabad, and A. Safdarian, “Simultaneous
placement of fault indicator and sectionalizing switch in distribution
the traditional overcurrent equipment to adaptive protection networks,” IEEE Trans. Smart Grid, vol. 10, no. 2, pp. 2278–2287,
coordination, while the objective function minimizes the pro- 2019.
tection operation time and the average interruption frequency [11] K. Hao, S. V. Achanta, J. Fowler, and D. Keckalo, “Apply a wireless
m
Carlo simulation is applied to further improve the solution with [13] C. d. Santos, P. M. Bartmeyer, and C. Lyra, “Allocation of fault
indicators for adaptive protection schemes,” in IEEE Innovative Smart
a richer representation of the stochastic nature of the faults; it Grid Technologies Conference - Latin America, 2019, pp. 1–6.
also validates the adaptive protection coordination obtained by [14] LaPSEE, “Testing systems,” Laboratory of Electrical Power System
using the set of equipment FT-FR approach. The case studies Planning, www.feis.unesp.br/lapsee, Tech. Rep., 2021.
[15] R. Brown, Electric Power Distribution Reliability, ser. Power Engineer-
put the approach into perspective with two test systems for the ing (Willis). CRC Press, 2017.
allocation of FT. A complementary study shows that under the [16] N. Bayati, H. R. Baghaee, A. Hajizadeh, and M. Soltani, “A fuse saving
evaluated conditions the reconfiguration does not significantly scheme for dc microgrids with high penetration of renewable energy
resources,” IEEE Access, vol. 8, pp. 137 407–137 417, 2020.
affect the performance of the proposed adaptive protection [17] IEEE, “IEEE recommended practice for protection and coordination of
scheme. industrial and commercial power systems (ieee buff book),” IEEE Std
All pieces considered, the proposed approach is flexible, 242-2001 (Revision of IEEE Std 242-1986), pp. 1–710, Dec 2001.
[18] B. Hussain, S. M. Sharkh, S. Hussain, and M. A. Abusara, “An adaptive
robust, and scalable. Flexible because it allows the design of relaying scheme for fuse saving in distribution networks with distributed
FT allocation plans for distribution systems with and without generation,” IEEE Trans. Power Del., vol. 28, no. 2, pp. 669–677, 2013.
distributed energy resources (DER); robust in the sense that it [19] M. Abdi-Khorsand and V. Vittal, “Modeling protection systems in time-
domain simulations: A new method to detect mis-operating relays for
always gives feasible optimized FT allocation plans; scalable unstable power swings,” IEEE Trans. Power Syst., vol. 32, no. 4, pp.
because it can address both small and large scale problems. 2790–2798, 2017.
12
[20] M. Izadi, A. Safdarian, M. Moeini-Aghtaie, and M. Lehtonen, “Opti- Fernanda C. L. Trindade (M’13, SM’22) received the Ph.D. degree in
mal placement of protective and controlling devices in electric power electrical engineering from the University of Campinas, Campinas, Brazil
distribution systems: A mip model,” IEEE Access, vol. 7, pp. 122 827– (2013), where she is currently an Associate Professor. Her research interests
122 837, 2019. include distribution system automation, monitoring, and protection. She has
[21] S. Santoso and T. A. Short, “Identification of fuse and recloser operations been Associate Editor of the IEEE Transactions on Power Delivery and IEEE
in a radial distribution system,” IEEE Trans. Power Del., vol. 22, no. 4, Power Engineering Letters since 2021.
pp. 2370–2377, 2007.
[22] R. C. Dugan and D. T. Rizy, “Electric distribution protection problems
associated with the interconnection of small, disperesed generation
devices,” IEEE Trans. Power App. Syst., no. 6, pp. 1121–1127, 1984.
[23] A. F. Naiem, Y. Hegazy, A. Y. Abdelaziz, and M. A. Elsharkawy,
“A classification technique for recloser-fuse coordination in distribution
systems with distributed generation,” IEEE Trans. Power Del., vol. 27,
no. 1, pp. 176–185, 2012.
[24] P. Barker and R. De Mello, “Determining the impact of distributed
generation on power systems.” in Power Engineering Society Summer
Meeting, vol. 3, 2000, pp. 1645–1656 vol. 3.
[25] A. Fazanehrafat, S. Javadian, S. Bathaee, and M.-R. Haghifamt, “Main-
taining the recloser-fuse coordination in distribution systems in presence
of dg by determining dg’s size,” in 9th International Conference on
on
Developments in Power System Protection, 2008, pp. 132–137.
[26] E. Purwar, S. P. Singh, and D. N. Vishwakarma, “A robust protection
scheme based on hybrid pick-up and optimal hierarchy selection of
relays in the variable dgs-distribution system,” IEEE Trans. Power Del.,
vol. 35, no. 1, pp. 150–159, 2020.
[27] “SEL wireless protection system: SEL-FT50 fault transmitter,
si
SEL-RP50 fault repeater, and SEL-FR12 fault receiver.”
https://selinc.com/products/FT50/, 2021.
[28] J. M. Gers and E. J. Holmes, Protection of electricity distribution
networks. IET, 2004, vol. 47.
r
[29] IEEE, “IEEE guide for electric power distribution reliability indices,”
IEEE Std 1366-2012 (Revision of IEEE Std 1366-2003), pp. 1–43, 2012.
[30] M. A. Dumitrescu, Network Protection and Automation Guide: Protec-
tive Relays, Measurements & Control. Alstom Grid, 2002.
[31] S. Ruzika and M. M. Wiecek, “Approximation methods in multiobjective
programming,” Journal of Optimization Theory and Applications, vol.
126, no. 3, pp. 473–501, Sep 2005.
[32] M. Ehrgott, Multicriteria optimization. Springer Science & Business
Media, 2005, vol. 491.
Ve
Christiano Lyra (SM’01) finished high school as an AFS exchange student
in Philadelphia, PA, USA, in 1970. He received a B.Sc. degree in Electrical
ed
[33] I. Dunning, J. Huchette, and M. Lubin, “Jump: A modeling language for Engineering from the Federal University of Pernambuco, Brazil, in 1975, and
mathematical optimization,” SIAM review, vol. 59, no. 2, pp. 295–320, an M.Sc. and a Ph.D. degree in Electrical Engineering from the University
2017. of Campinas (UNICAMP), São Paulo, Brazil, in 1979 and 1984, respectively.
[34] Gurobi Optimization, LLC, “Gurobi Optimizer Reference Manual,” Following a brief career at the Power Company of the São Francisco River, he
2021. [Online]. Available: https://www.gurobi.com joined the Faculty of UNICAMP in 1978, where he is a Professor of Electrical
[35] W. E. Company, Electrical Transmission and Distribution Reference Engineering and has served as the head of the Department of Systems
itt
Book. Central Station Engineers of The Westinghouse, 1964. Engineering, Director of the Graduate Program in Electrical Engineering,
[36] P. Bartmeyer, “Locations of protection devices and fault transmitters,” Dean of the School of Electrical and Computer Engineering, a councilman at
https://github.com/petrabartmeyer/fault_transmitter_receiver, 2021. UNICAMP Press, and a member of the Board of Trustees of the University.
[37] A. A. Hagberg, D. A. Schult, and P. J. Swart, “Exploring network He coordinates the Institute for Advanced Studies (IdEA) and the Laboratory
structure, dynamics, and function using networkx,” in Proceedings of for Energy Network Optimization (LABORE).
bm