Maximizing Reliability

R by Optimal Siting
S of
Distributed Generation and Protectivee Devices
Maurice Tristan G. Donalvo
Adonis Emmanuel DC. Tio
Wilbert Rey D. Tarnate
Eleectrical and Electronics Engineering Institute
University of the Philippines Diliman
Quezon City, Metro Manila, Philippines
[email protected]
[email protected]
[email protected]

Abstract— The placement of protective andd switching devices, Average Interruption Frequeency Index (SAIFI), System
as well as distributed generation (DG), affecct the reliability of Average Interruption Duration Index (SAIDI), and Momentary
distribution systems. Existing placement methodologies
m give Average Interruption Frequenccy Index (MAIFI) or by using
suboptimal solutions since device placem ment is optimized cost indices.
separately from DG placement, and vice verssa. In this paper, a
binary formulation for the System Average Interruption In these studies, differentt optimization algorithms have
Frequency Index (SAIFI) as a function of thee locations of fuses, been used. Some studies have used
u the deterministic method in
reclosers, and DGs is developed. Commercially-available obtaining the optimal locationn of these devices [2,3,5,6,7,9],
optimization software is used to minimize SAIFI to get the while some have used heuristtic method such as the Genetic
optimal location of fuses, reclosers, and DGs, subject to various Algorithm [4,8,11,12], ant-coloony algorithm [10], and binary
constraints. Using this methodology, the t simultaneous particle swarm algorithm [13]. We can see that studies
optimization of fuse, recloser and DG placemment is achieved for regarding finding the optimal location of protective devices
maximum reliability. have been done using either thhe deterministic or the heuristic
approach. On the other hand, studies regarding finding the
Keywords— Reliability; SAIFI; Protective Devices;
D Distributed optimal location of both protective devices and DGs have been
Generation done using only the heuristic method.
m
I. INTRODUCTION In this paper, a binary moodel for SAIFI is presented and
The goal of the distribution system iss to continuously used to determine the location of “fuse-blow” and “fuse-save”
supply power to their customers, achievinng high reliability. fuses, reclosers, and distributted generation that will yield
However, because of certain factors like eqquipment failures, optimum system reliability. Havving the binary model for SAIFI
vehicular accidents, weather conditions, and many more, will allow the use of determ ministic methods in finding the
customers experience interruptions [1]. These factors, optimal location of protective devices
d and DGs.
combined with the fact that the distribution system is radial in
nature, add up to low distribution system reliaability. II. OPERATION OF RAD
DIAL DISTRIBUTION SYSTEM

Protective devices and distributed genneration (DG) are A. Protective Devices

installed in the distribution system to help h improve the Protective and switching devices
d are used by distribution
reliability of the system. Protective devicees such as fuses, utilities to protect the system and
a the equipment when a fault
reclosers, switches and circuit breakers are installed
i to isolate occurs. These devices also isollate the source of the fault, thus
faults, thus decreasing the number of affectted customers. On making the system stable and ablea to serve other loads that are
the other hand, distributed generation can serrve as a temporary really not affected by the fault.
source of power to the customers downstrream of the fault,
leading to a decrease in the frequency and duration of
interruption.
Many studies have been made regarrding finding the
optimal location of protective devices and DGs
D to improve the
reliability of the distribution system. Somee have considered
only the optimal location of protective devvices [2-9], while
some have considered the optimal location of both protective
devices and DGs [10-13]. In these studdies, reliability is Fig. 1. A Simple Distribution
D System
evaluated either by using reliability indices such
s as the System

978-1-4799-7993-6/15/$31.00 ©2015 IEEE

 The capability of protective devices to isolate faults makes B. Distributed Generation
them an asset in improving the reliability of the system. Distributed Generation help improve the reliability of the
Consider the simple distribution system shown in Fig. 1. A system. However, how it improves the reliability depends on
fault occurring at any part of the system will make all how they are connected on the system. Two modes of
customers experience interruption. Placing a recloser or a fuse operation of distributed generations are discussed in this paper
anywhere in the system will decrease the number of affected - as Grid-Parallel Mode or as backup capacity [15]. These two
customers when the same fault occurs, thus improving modes treat distributed generation as an additional backup
reliability. source, but they differ on how will they connect and the time
Some protective devices considered in the study include they will connect to the grid.
fuses and reclosers. 1) Grid-Parallel Mode
1) Fuse In this mode, the DG is connected parallel to the grid. Both
Fuses are overcurrent devices that trips when it detects a DG and utility source are serving the loads on the system.
current that is greater than its rating. When fuses trip, it does When a permanent fault occurs, the protective devices of the
not reclose. Thus, when a fuse trips due to either a temporary DG will respond to cut the current flow from the DG to the
or permanent fault, all customers downstream of it will location of the fault, resulting in interruptions experienced by
experience a sustained interruption. customers served by the DG. However, if reclosers or fuses
near the location of the fault can respond immediately and trip
Fuses can be classified based on how they are coordinated before the protective devices of the DG, then no customers
with the other protective devices present in the system. They served by it will be affected by the fault. This will result in a
can either be coordinated using the “fuse-blow” scheme or the decrease in the number of interruptions experienced by the
“fuse-save” scheme [7]. customers of the DG.
a) Fuse-Blow Scheme 2) Backup Capacity
In a fuse-blow scheme, the protective devices are In this mode, DG acts as stand-by generators that will
coordinated in such a way that the fuse will trip first before the supply power to the grid whenever a utility experiences a
upstream protective devices whenever a permanent or supply interruption. They will only connect once the fault is
temporary fault occurs. The advantage of this scheme is that isolated. DG connected in this mode decreases the amount of
the customers affected are lesser compared to the other scheme. interruption time its customers will experience.
However, the problem is that both permanent and temporary
faults will cause sustained interruptions. Also, the fuse-blow C. Distribution System Reliability Indices
scheme is costly, since there is a need to replace the fuse every Reliability indices are used to quantify the reliability in a
time it detects a fault. distribution system. Most utilities use SAIFI, SAIDI or MAIFI
to quantify the reliability of the system. In this project, SAIFI is
b) Fuse-Save Scheme the index being considered. This index gives the average
In a fuse-save scheme, the fuse will be saved by an number of permanent interruption a customer will experience
upstream protective device through proper protection per year. Table I shows the equation for SAIFI and other
coordination. In this scenario, the protective device upstream of common reliability index [14].
the “fuse-save” fuse will react to the fault first. If the protective
device directly upstream of the “fuse-save” fuse is a recloser, TABLE I. EQUATIONS FOR RELIABILITY INDICES
then the recloser will open. If the fault happens to be
temporary, then the customers affected will only experience
SAIFI
momentary interruptions due to the opening of the recloser.
The advantage of this scheme is that it prevents the fuse
from reacting to temporary faults, thus decreasing the cost of SAIDI
replacement fuses. However, the problem is that the number of
customers affected by the fault (specifically the temporary
fault) increases. MAIFI

2) Recloser where:
Reclosers are protective devices that have both fault- = permanent failure rate of section i
interrupting and reclosing capabilities. Reclosers will open
whenever they detect faults, thus cutting the flow of current = momentary failure rate of section i
from the source to the loads downstream of the device. = repair rate of section i
Because of their reclosing capability, recloser “recloses”. It
checks whether the fault is still there, since faults can be = number of customers at section i
momentary. If the fault still exists, then it opens again. After = total number of customers.
some preset number of times, the recloser will reach its lockout
state. At lockout state, the recloser will be permanently open
until an action is made.
 III. METHODOLOGY
Before formulating the binary model for SAIFI, a test
system was developed to represent a general distribution 1 4
system in the smallest size possible. A reliability calculator
based on the load-point method [14] was developed in order to
validate the results of the binary formulation.
In the optimization problem, the objective function is the 1
SAIFI of the system. The SAIFI formulation is a function of ,
5
the location of both the protective devices and DGs. From [7],
the binary formulation for SAIFI as a function of location of
,
protective devices is already determined.
To determine the SAIFI formulation considering the
location of both protective devices and DG, the following
1
assumptions were considered:
• DGs have unlimited capacity. Thus, cost will not be a
part of the objective function. 6
,
• DGs operate in the grid-parallel mode.
• DGs are non-intermittent. 1
• DGs have on-site recloser for overcurrent protection.
• DGs have zero failure rates.
• Voltage profile, line loss, and power quality are not 1
considered. , ,

• Overlapping outages are not considered. 1

, ,
7
• Protective devices do not fail when isolating faults.
• Protective devices are properly coordinated. 1
• The paper presents a distribution system with a single , , ,
feeder only. However, the methodology can be easily
extended to systems with multiple feeders.
where:
A. Formulation of SAIFI
SAIFI can be computed using the formula: λi = permanent failure rate of section i, in failures per year
∑ γi = momentary failure rate of section i, in failures per year
1
Nt = total number of customers
The denominator, Nt, represents the total number of
customers the distribution system serves and is a constant Ni = number of customers of section i
value. The numerator is not constant, and its value is dependent Ntotal(i) = total customers downstream of section i
on the configuration of protective devices and DG. The terms
in the numerator can be divided into six and is shown in (2) to aji = section immediately downstream of section j in the
(7). path from j to i
xij = the decision variable in this paper. This variable takes
the values of either 0 or 1. A value of 0 means that there is a
2 protective device j or DG located at section i. A value of 1
means that there is no protective device or DG located at
section i. The variable j can take the value of 1 for fuse-blow
fuse, 2 for fuse-save fuse, 3 for reclosers, 4 for manually-
operated switch, and 5 for DG.

, s = set of all sections

3
p[1,i) = set of all sections in the path from the substation to
section i, excluding section i
,
 p(j,i] = set of all sections in the path from the section j to tells that in location i, either a device is installed or not.
section i, excluding section j and including section i
p(j,i) = set of all sections in the path from the section j to 6. Coordination Constraint: The equation
section i, excluding both sections j and i
1 (13)
d(i) = set of all sections downstream of section i, section i
not included
tells that device i in location k cannot be coordinated
dDG(i) = set of all DGs downstream of section i with device j in location l.
idu(j,i) = set of all sections that are both downstream of
section j and upstream of section i, sections j and i not included 7. Model Limitation Constraint: Assuming that i and j are
sections and that section j is located downstream of
idu[j,i) = set of all sections that are both downstream of section i, the equations
section j and upstream of section i, section j included and
section i not included
2 (14)
iDG,du[j,i) = set of all DGs that are both downstream of
section j and upstream of section i, including section j and 1 (15)
excluding section i.
Note that in the formulation, upstream sections of a given tells that a switch cannot be installed at the load side of
section pertain to the section itself plus all sections not the fuse and that a fuse-save fuse cannot be installed
included in the set of all downstream sections. Consider the downstream of a fuse-blow fuse.
developed test system in Figure 2. If we consider section 10,
then the sections downstream of node 10 are nodes 10, 8. Maximum Allowable Index Value: The equation
11,12,13,14, 15, 16, 17, and 18. Therefore, the upstream nodes
are the nodes 10, 9, 8, 7, 6, 5, 4, 3, 2, and 1. (16)
B. Constraints
tells that a maximum value for SAIFI is established.
Constraints used in this paper are used to model the
distribution system. These constraints are formulated using the
variables discussed. 9. Linearization Constraint: The following equations may
be used to linearize product terms
1. Fixed Devices: The equation 0 indicates that a
device j is fixed in location i. ∑ 1 (17)
2. Banned Devices: The equation 1 indicates that a
device j is not allowed in location i. 0 (18)
3. Number of Device Constraints: The equations
; 0,1
∑ , (8)

∑ , (9) C. Determining the Optimal Location of Protective Devices

and DG
∑ , , (10) The binary formulation for SAIFI obtained is a function of
the location of protective devices and DGs. Therefore, this
give the formulation when the maximum number of formulation will be used to obtain the optimal location of
devices and DG are given. protective devices and DG. Commercially-available software
such as MATLAB’s binary programming toolkit and MS
4. Device Required Constraints: The equation Excel’s Nonlinear Optimization Solver can be used to
determine the location of protective devices and DGs that will
3 (11) minimize the SAIFI.

pertains to the case where a device must be installed in

section i. IV. TEST RESULTS AND ANALYSIS
5. General Design Constraint: The equation Several case studies were performed to determine the effect
of simultaneously optimizing the location of protective devices
and DGs as compared to optimizing the location of either the
3 (12)
protective device or DGs with the restriction that the other one
is fixed. The test system used is the developed test system
 - Candidate Location for Protective Device

- Candidate Location for Distributed Generation

Fig. 2. Developed Test System

shown in Figure 2. The reliability parameters are shown in 5. Fuses cannot be installed in the main feeder.
Table II.
6. DGs can only be installed in the main feeder.
TABLE II. RELIABILITY PARAMETERS OF THE DEVELOPED TEST 7. Recloser is installed in section 1.
SYSTEM

Section λ γ r Customers
The three test case studies are as follows:
1 0.1 0.2 0.5 100
3 0.2 0.3 1 125 Case 1 - Fixed Location of Devices, Optimal Location of DG:
Case 1 represents the situation of finding the optimal
4 0.15 0.3 1.25 500
location of DG given that the locations of the protective
5 0.3 0.1 1.2 300 devices are fixed. Table III shows the assumed location of
8 0.6 0.3 3 50 protective devices for the purpose of discussion in this paper.
9 0.25 0.4 2.5 1000
11 0.4 0.3 3 400 TABLE III. FIXED LOCATION OF PROTECTIVE DEVICES FOR CASE 1

12 0.4 0.6 4.5 200 Fuse with fuse- Fuse with fuse-
Recloser
13 0.5 0.55 2 250 blow scheme save scheme
16 0.3 0.4 5 100 1,2,7,12 5 3,11,13
17 0.4 0.5 5 50
18 0.7 0.9 4.7 40
Case 2 - Fixed Location of DG, Optimal Location of Devices:
Case 2 represents the case of finding the optimal location of
From the figure, the candidate locations for protective protective devices given that the location of the DG is fixed. In
devices are denoted by a square, while the candidate locations this case study, we will assume that a DG is installed in
for DG are denoted by a circle. In total, there are 18 candidate Location 26.
locations for protective devices and 12 candidate locations for
DG. TABLE IV. RESULTS OF THE THREE TEST CASES
In these case studies, the developed test system has the Fuse with Fuse with
Distributed
following constraints: Case Recloser fuse-blow fuse-save SAIFI
Generation
scheme scheme
1. Only four reclosers are available. 0 1,7,12,15 5 3,11,13 N/A 0.7277
2. Only four fuses are available. 1 1,2,7,12 5 3,11,13 20 0.6823
3. Only 1 DG is available. 2 1,7,10,12 5 3,11,13 26 0.5367
4. One fuse-blow fuse must be installed in the system. 3 1,4,7,12 6 11,13,15 29 0.5154
Case 3 - Optimal Location of Devices and DGs: system and (2) the optimal location of protective devices and
Case 3 represents the case of simultaneously finding the DGs are, indeed, dependent on each other and thus,
optimal location of protective devices and DG. simultaneously optimizing both of them will yield better
system reliability.
The result of the three test cases is shown in Table IV. Note
that Test Case 0 is a base test case. This test case shows the REFERENCES
optimal location of protective devices given that no DG is [1] R. Brown, “Electric Power Distribution Reliability”, CRC Press, Taylor
installed in the system. From the results, we can conclude the and Francis Group, 2009.
following: [2] F. Soudi and K. Tomsovic. “Optimized distribution protection using
binary programming,” IEEE Transactions on Power Delivery, Vol. 13,
No. 1. 1998. Page(s) 218-224.
1. The location of protective devices affects the optimal [3] E. Zambon, D. Bossois, B. Garcia, E. Azeredo. “A novel nonlinear
location of DG. This can be seen on Test Cases 1 and programming model for distribution protection optimization”, IEEE
3. In Test Case 1, the location of the protective devices Transactions on Power Delivery, Vol. 24, No. 4, 2009, Page(s) 1951-
is fixed, leading to a solution where the optimal 1958
location of DG is on location 20. However, from the [4] L. da Silva, R. Pereira, J. Mantovani, “Allocaton of protective devices in
result of Test Case 3, a different set of location of distribution circuits using nonlinear programming models and genetic
algorithms,” Electric Power Systems Research, Vol. 69, 2004, Page(s)
protective devices have led to a different optimal 77-84.
location of DG. This time, the optimal location of DG [5] R. Bupasiri, N. Wattanapongsakorn, J. Hokierti and D. Coit.“Optimal
is on location 29. electric power distribution system reliability indices using binary
programming,” Annual Reliability and Maintainability Symposium,
2. The location of DG affects the optimal location of 2003. Page(s) 556-561.
protective devices. This can be seen on Test Cases 2 [6] F. Soudi and K. Tomsovic. “Optimal trade-offs in distribution protection
and 3. In Test Case 2, the location of the DG is fixed at design,” IEEE Transactions on Power Delivery, Vol. 16, No. 2, 2001.
location 26, leading to the optimal location of Page(s) 292-296.
protective devices when the DG is installed at location [7] A.E.D.C. Tio, I.B.N.C. Cruz, B.M. Malquisto, R.D. del Mundo. “A
26. However, from test case 3, a different location of binary programming model for reliability optimization considering fuse-
DG has led to a different optimal solution. blow and fuse-save schemes.” TENCON 2012 – 2012 IEEE Region 10
Conference, 2012, Page(s) 1-6.
3. Since the location of DG affects the optimal location of [8] W.R.D. Tarnate, I.B.N.C. Cruz, B.M. Malquisto, R.D. del Mundo.
protective devices, and on the other hand, the location “Maximizing service restoration in reliability optimization of radial
of protective devices affects the optimal location of distribution systems.” TENCON 2012 – 2012 IEEE Region 10
Conference, 2012, Page(s) 1-6.
DGs, then to improve further the reliability of the
[9] A.E.D.C. Tio, I.B.N.C. Cruz. “A binary formulation of SAIDI for the
system, DGs and protective devices must be optimized predictive reliability assessment of radial distribution systems with tie
simultaneously. The effect of it is seen on the third lines.” Page(s) 1-5.
case in Table IV. [10] L. Wang, C. Singh. “Reliability-constrained optimum placement of
reclosers and distributed generators in distribution networks using an ant
V. CONCLUSION colony system algorithm.” IEEE Transactions on Systems, Man, and
Cybernetics, Vol. 38, No. 6, 2008, Page(s) 757-764
In this paper, a binary formulation for SAIFI as a function
[11] Pregelj, M. Begovic, A. Rohatgi. “Recloser allocation for improved
of fuses, reclosers, and DGs was presented. It was used to reliability of DG-enhanced distribution networks.” IEEE Transactions
determine the optimal location of these devices to obtain on Power Systems, Vol. 21, No.3, 2006, Page(s) 1442-1449.
optimal system reliability. The developed binary formulation [12] Weiwei, W. Yafeng, L. Wenxia. “Optimal allocation of switches in DG
was validated by replicating published test results and by enhanced distribution systems.” Power and Energy Engineering
replicating results using a reliability calculator based on the Conference (APPEEC), 2011 Asia-Pacific, 2011, Page(s) 1-5.
load-point method. [13] W. Prommee, N. Pongprapunt, W. Ongsakul. “Improved reliability
model and optimal protective device placement in micro grid by
The binary formulation was then used to determine the improved binary particle swarm optimization,” 8th International
optimal location of fuses, reclosers, and DGs of a sample Conference on Power Electronics – ECCE Asia. 2011. Page(s) 1514-
feeder. Commercially-available software such as MS Excel and 1519
MATLAB were used in the optimization process. [14] R. Billinton, R. Allan, Reliability Evaluation of Power Systems, 2nd
edition, Plenum Press, New York, 1996.
Test cases were developed to demonstrate the effect of [15] Sustainability Victoria, “Guide to Connecting a Distributed Generator in
simultaneously optimizing the location of protective devices Victoria,” www.sustainability.vic.gov.au, Accessed September 5, 2013
and DGs. It was demonstrated that (1) installing DGs in the [16] L.Willis. “Power Distribution Planning Reference Book.”Marcel Dekker
distribution system can definitely improve the reliability of the Inc., 2004.