Multi-Objective Optimization using Microgenetic Algorithm Applied

to the Placement of Remote and Manual Switches in Distribution

Networks

Helton do Nascimento Alves and Railson Severiano de Sousa

Department of Electrical Engineering, Federal Institute of Maranhão, São Luís, Brazil

Keywords: Costs, Power Distribution Networks, Load Importance, Multi-Objective Microgenetic Algorithm,

Reliability, Switch Placement.

Abstract: This paper presents a new formulation for placement of switches in distribution of electric power. An

approach for determination of the location of tie switches and section switches using a multi-objective

microgenetic algorithm is proposed. In the procedure, load importance, reliability index, remote or manual

controlled switch and investments costs are considered. The results are based on simulations in a 69-bus test

system presented and the results are compared to the solution given by others search techniques. This

comparison confirms the efficiency of the proposed method which makes it promising to solve complex

problems of tie switches and section switches placement in distribution feeders.

1 INTRODUCTION

In order to increase the reliability in overhead radial

electrical energy distribution systems, tie switches

and section switches are normally installed. Section

switches (SS) enable the isolation of failed

components and tie switches (TS) are used for

interconnection between feeders. In the planning

process of distribution of electric power, the

decision of the strategy to be adopted in the

allocation of switching devices is an important

aspect to be considered. Another important

characteristic of these devices should be considered

is their way of operation: manual controlled switch

(MCS) or remote controlled switch (RCS). When

MCS is used, a maintenance crew has to be

dispatched to the switch site to perform the fault

isolation and load restoration. RCS are usually

initiated at a control room, where the operation crew

performs the action of the switches. In this case, the

switching time considerably decreases, performing

the restoration of system capacity and reliability

with minimum outage and least expenditure of

manpower. Of course, RCS are more expensive and

they need a communication system to be activated.

However, the appearance of an increasing number of

new automation equipment and communication

technologies has provided economic viability to the

application of RCS in distribution networks

(Sperandio et. al., 2007). In last decades, the electric

utilities have introduced remote control schemes in

distribution networks to increase the reliability and

to have faster responses in contingencies (Allan and

Billinton, 1976), however, the cost involved in this

process is very high and the amount of investment

generally have budget constraints. So an alternative

that has been considered by electric utilities is the

gradual replacement at strategic points of MCS by

RCS.

The selection of the number and location of the

switches depends on factors such as reliability

indices, cost of switches, maintenance and operation

costs. Besides the cost and reliability, other factors

connected to the system can be taken into account to

define the allocation of switches, such as load

importance. The solution of this problem is

considered a very difficult task because it is a

combinatorial constrained problem described by a

nonlinear and nondifferential objective function.

Several intelligent algorithms have been used to

solve such a problem applying different heuristics.

Simulated Annealing (Billinton and Jonnavithula,

1996), Genetic algorithms (Levitin, Mazal-Tov and

Elmakis, 1995; Golestani and Tadayon, 2011;

Dezaki et. al., 2012), Fuzzy Logic (Teng and Lu,

2002), Ant Colony (Falaghi, Haghifam and Singh,

Alves, H. and Sousa, R..

Multi-Objective Optimization using Microgenetic Algorithm Applied to the Placement of Remote and Manual Switches in Distribution Networks.

In Proceedings of the 7th International Joint Conference on Computational Intelligence (IJCCI 2015) - Volume 1: ECTA, pages 269-278

ISBN: 978-989-758-157-1

269

2009; Tippachon and Rerkpreedapong, 2009),

Particle Swarm Optimization (Golestani and

Tadayon, 2011; Moradi and Fotuhi-Firuzabad,

2008; Ziari et. al., 2009), Immune Algorithm (Chen

et. al., 2006) and Tabu Search (Toune, 1998)

In this paper, a Multi-Objective Microgenetic

Algorithm (MGA) is proposed and employed for the

allocation of SS and TS, in order to assist the

decision-taking during the planning of the

distribution system. Investments costs, reliability,

load importance and the use of MCS and/or RCS are

considered in the solution.

2 PROBLEM FORMULATION

2.1 Expert Knowledge

In distribution networks planning, the decision of the

strategy to be adopted in switch allocation is an

important aspect to be considered. This decision is

based on expert knowledge and influenced by

several parameters that determine the importance of

certain consumers and circuits. Technical and

economic aspects must be considered, seeking a

balance among: safe operation of the system, desired

level of reliability and investments.

In general, from the point of view of reliability, the

following criteria may be taken into account in the

switches allocation:

 Minimizing the number of consumers affected by

an outage in the distribution system;

 Restoration of service to critical loads;

 Preference must be given to the installation in:

o circuits with high incidence of permanent

faults;

o points of interconnection between different

feeders;

o along the main section of the feeder, by

dividing the load into blocks. It should be

considered the voltage drop and maximum

demand allowed in the restoration of each

load block by a tie switch;

o points near to the beginning of circuits with

high loading;

o before and after points where there are loads

priority, with high continuity demanded;

o places easily accessible.

2.2 Distribution Feeder Model

In general, the distribution feeder model is

represented by sections with its respectively length,

cable type, origin node, end node and active and

reactive load. In order to evaluate the switches

placement problem is needed that the feeder model

also contains line’s failure rate, mean time for

restoring by switching and mean outage time of a

fault in the feeder. Every possible solution defined

by proposed algorithm, identify a group of sections

where the SS are allocated at their beginning.

Besides that, it was considered that a TS is

somewhere allocated downstream of the line section

containing a SS. It can happen that the same TS is

downstream of more than one SS. This occurs when

the sections that contain SS belong to the same set of

line sections that start at the substation and end on a

terminal node. In the Fig. 1 is showed an example

with 5 SS allocated but only 2 TS are necessary to

be allocated downstream of SS.

Figure 1: Example of tie switches allocated downstream of

section switches.

2.3 Expected Unsupplied Energy Due

to Power Outages and Costs

Considering MCS and RCS

Simultaneous Placement

Switches Placement in distribution networks can

reduce down time by isolating the faulted part of the

circuit after protection operation. Hence, the

upstream and downstream sections of the faulted

section can be restored. In this case, the outage time

and the expected unsupplied energy due to power

outages are reduced.

Manual and remote switches perform the same

function, changing only the operation form. Outage

time could be reduced by replacing manual by

remote switches. This reduction is caused to increase

reliability and reduce outage cost in network.

Moreover, most devices in automation system

resulting in more cost to the system..Simultaneously

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

270

placement of manual and remote controlled switches

decreases the total cost in life time. Due to these

facts, it is essential to develop appropriate strategies

to consider the gains and costs produced with the use

of remote switches.

Expected unsupplied energy due to power

outages can be calculated using Eq. (1). The first

part of Eq. (1) is the unsupplied energy up to

detection of the fault and switching and the last part

is the unsupplied energy after switching.



(



)

=









(





)





+(



−



)(



()



)





(1)

where, EUE(S) is the expected unsupplied energy

due to power outages considering switches installed

according S [kWh]; S is the set of section switches

and tie switches installed; Ns is the number of

sections in the feeder; Ltot is the total load in the

feeder [kW]; Li(S) is the load interrupted in the

feeder by a fault in section i after service restoration

performed with the operation of switches installed

according S [kW]; λi is the annual failure rate of

section i; Tswitch is the mean time for restoring by

switching (hours); and Toutage is the mean outage

time of a fault in the feeder.

When all switches installed in the feeder are

MCS, the time Tswitch(MCS) is considered in the

calculation of EUE. Likewise, when all switches

installed in the feeder are RCS, the time

Tswitch(RCS) is considered in the calculation of

EUE. When MCS and RCS are simultaneous

placement, the calculation of EUE will depend on

the procedures used for operation and maintenance

teams. In this case, in order to define EUE, the

following switches are identified considering a fault

in line section i:

 CLU(i): the closest upstream SS of the line

section i that isolate the fault from main source

(Fig. 2 - SS

);

 CLD(i): the closest downstream SS of the line

section i that isolate the fault from an alternative

source. There may be more than one CLD

triggered to isolate distinct alternative sources

(Fig. 2 - SS

, SS

and SS

);

 TST(i): TS triggered to restore part of the feeder

load through alternative source considering a

fault in line section i. There may be more than

one TST triggered for interconnection with

distinct alternative sources (Fig. 2 - TS

, TS

and

);

 CLD&TST(i): CLD(i) and TST(i) triggered to

restore part of the feeder load through alternative

source (Fig. 2 -SS

&TS

, SS

&TS

and

&TS

). If both CLD(i) and TST(i) are RCS,

CLD&TST(i) is considered RCS, otherwise

CLD&TST(i) is considered MCS.

Figure 2: Feeder with switches installed considering a

fault in line section 7.

Considering the possible situations that can occur

for the simultaneous operation of manual and remote

controlled switches, the expected unsupplied power

can assume the following values during switches

operation:

 Case 1: CLU(i) and/or some CLD&TST(i) are

RCS

o During time T

switch(RCS)

, L

tot

will be

disconnected by main breaker. In T

switch(RCS)

the RCS switches are triggered to restore L

through main source and/or alternative

source.

o Between T

switch(RCS)

and T

switch(MCS)

(when

MCS switches are triggered) the load L

1,i

tot

-L

will be disconnected for time

switch(MCS)

-T

switch(RCS)

. In T

switch(MCS)

the MCS

switches are triggered to restore L

o Between T

switch(MCS)

and T

outage

the load L

2,i

1,i

- L

will be disconnected for time T

outage

switch(MCS)

 Case 2: CLU(i) and all CLD&TST(i) are MCS

o During time T

switch(MCS)

, L

tot

will be

disconnected by main breaker. In T

switch(MCS)

theMCS switches are triggered to restore L

through main source and/or alternative

source.

o Between T

switch(MCS)

and T

outage

the load L

2,i

tot

-L

will be disconnected for time T

outage

switch(MCS)

Multi-Objective Optimization using Microgenetic Algorithm Applied to the Placement of Remote and Manual Switches in Distribution

Networks

271

 Case 3: CLU(i) and all CLD&TST(i) are RCS

o During time T

switch(RCS)

, L

tot

will be

disconnected by main breaker. In T

switch(RCS)

the RCS switches are triggered to restore L

through main source and/or alternative

source.

o Between T

switch(RCS)

and T

outage

the load L

1,i

tot

- L

will be disconnected for time T

outage

switch(RCS)

Based on these considerations, the expected

unsupplied energy due to power outages is

calculated in this work using Eq. (2):



(



)

=





(





∗



)





+

,

(



)

∗

(





−



)

∗









+(

,

(



)

∗(



−



)∗



)





(2)

Case 1: t1=Tswitch(RCS); LA,i= L1,i; LB,i=

L2,i; t2=Tswitch(MCS);

Case 2: t1=Tswitch(MCS); LA,i= 0; LB,i=L2,i;

t2=Tswitch(MCS);

Case 3: t1=Tswitch(RCS); LA,i=0; LB,i=L1,i;

t2=Tswitch(RCS);

where, Li(S) is the load interrupted in the feeder by a

fault in section i after service restoration performed

with the operation of switches installed according S

[kW]; Lx is the part of Ltot restored after RCS

switches are triggered; L1,i(S) is the load interrupted

in the feeder by a fault in section i after service

restoration performed with the operation of RCS

installed according S [kW]; Ly is the part of Ltot

restored after MCS switches are triggered; L2,i(S) is

the load interrupted in the feeder by a fault in section

i after service restoration performed with the

operation of MCS installed according S [kW];

Tswitch(MCS) is the mean time for restoring by

manual controlled switching (hours). It is considered

the same time for all line sections; and

Tswitch(RCS) is the mean time for restoring by

remote controlled switching (hours). It is considered

the same time for all line sections;

Costs associated with system expected outage to

customers due to supply outages and switches

placement can be calculated using Eq. (3).

()

)()(

1)()(

SCOSTSCOST

iSEUEkSCOST

MCSRCS

crese

++=



−

(3)

where, k

is the energy cost ($/kWh); A is the

planning horizon (years); i

cres

is the annual load

growth; COST

RCS

(S) and COST

MCS

(S) are the costs

of RCS and MCS installed according S, respectively,

all installed in the first year of planning. It includes

capital cost, installation cost and maintenance cost.

Many works only consider switches’ absolute

value in the cost evaluation (Assis et. al.,2012; Ziari

et. al., 2009; Villasanti et. al., 2008), although

utilities are already having costs with unsupplied

energy, so it should be included in cost equation.

Another difference on this formulation is related to

switch placement as a planning process, including

the planning horizon. This index is used as object

function in MGA in order to evaluate the costs.

2.4 Reliability Assessment

According to Billinton and Jonnavithula (1995),

reliability evaluation includes all the segments of an

electric power system in an overall assessment of

actual consumer load point reliability. The primary

reliability indices are the expected failure rate, the

average duration of failure and the annual

unavailability, at the customer load points.

Individual customer indices can also be aggregated

with the number of customers at each load point to

obtain system reliability indices. These indices are

the system average interruption frequency index

(SAIFI), the system average interruption duration

index (SAIDI), the customer average interruption

duration index (CAIDI) and the average service

availability index (ASAI). The most common

reliability indices used by electric utilities are SAIFI

and SAIDI. They are used to measure the impact of

power outages in terms of the number of

interruptions and interruption durations respectively

(Allan and Billinton, 1993). These indices can be

calculated for the overall system or for subsets of the

system depending on requirements for the

performance measures. These indices depend on the

circuit topology and location of switches. In order to

compare different switches allocation, is used

SAIDI, as follows:

tot

SEUE

SSAIDI

)(

)( =

(4)

where, SAIDI(S) is the system average

interruption duration index according S.

It’s common that researchers use SAIDI as a portion

of the EUE, however most of them, like Ziari et. al.,

(2009), Billinton and Jonnavithula (1996) and Chen

et. al. (2006) don’t evaluate the unsupplied energy

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

272

before switch’s triggered to simplify the analysis. So

this work performed a more realistic reliability

analysis, using this index as objective function of the

MGA.

2.5 Load Importance

Generally, the electric utilities need to prioritize the

service to some consumers due to its special

characteristics, such as: critical loads, big power

consumers, loads with high continuity demanded,

etc. The proposed algorithm has defined a new

variable called load importance (LI), which defines

the importance for the electric utility of each

consumer connected to the feeder. It establishes a

ranking between 0 and 1 (inclusive) for consumers

defined by the electric utility

In order to give preference to the switches

installation in sections situated before and after

priority loads is defined the priority index as

follows:

tot

SLI

SPRIORI



)(

(5)

where, LI

tot

is the sum of all LI defined to the feeder;

(S) is the sum of LI of all consumers not

interrupted by a fault in section i after service

restoration performed with the operation of switches

installed according S; and PRIORI(S) is the priority

index considering the set S of section switches and

tie switches installed.

This index is used as object function in MGA in

order to evaluate the load importance. Most of the

paper in this field, e. g. Assis et. al. (2012) Falaghi,

Haghifam and Singh (2009) Tippachon and

Rerkpreedapong (2009), Golestani and Tadayon,

(2011) and Ziari et. al., (2009), don’t include this

evaluation on it solution.

3 SOLUTION

In order to find an optimal allocation of switching

devices in distribution networks based on load

priority, reliability, costs, and considering the

simultaneous allocation of MCS and RCS, a multi-

objective algorithm is proposed. MGA is used in the

solution. The simultaneous allocation of MCS and

RCS is considered to calculate reliability index and

costs. Load importance is defined by electric utility.

A weight function based on reliability index, costs

and load importance is used as object function.

3.1 Microgenetic Algorithms

Genetic algorithms are simple, robust, flexible, and

able to find the global optimal solution. They are

especially useful in finding solution to problems for

which other optimization techniques encounter

difficulties (Goldberg, 1989). A basic genetic

algorithm is constituted by a random creation of an

initial population and a cycle of three stages,

namely:

1. evaluation of each chromosome;

2. chromosomes selection for reproduction;

3. genetic manipulation to create a new population,

which includes crossover and mutation.

Each time, this cycle is completed, it is said that a

generation has occurred.

The disadvantage of genetic algorithms is the

high processing time associated. That is due to their

evolutionary concept, based on random processes

that make the algorithm quite slow. However,

different methods for reducing processing time have

already been proposed, such as more appropriate

choice of solution coding and reduction of search

space using the specialist knowledge. One

alternative method known as microgenetic

algorithms, whose processing time is considerably

smaller, is shown in (Delfanti et. al., 2000;

Chakravarty, Mittra and William, 2001).

According to Souza, Alves and Ferreira (2004), most

of the genetic algorithms produce poor results when

populations are small, because insufficient

information is processed about the problem and, as a

consequence, premature convergence to a local

optimum occurs. Population size generally varies

from 30 to 300 individuals. In contrast, microgenetic

algorithms explore the possibility to work with small

populations (from five to 20 individuals usually) in

order to reduce the processing time. From a genetic

point of view, it is known that frequent

reproductions inside a small population may

disseminate hereditary diseases rarely found in large

populations. Therefore, small populations can act as

natural laboratories where desirable genetic

characteristics quickly can emerge. In microgenetic

algorithms, mutations are unnecessary because after

a certain number of generations, the best

chromosome is maintained and the rest are

substituted by randomly generated ones. On the

other hand, it requires adoption of some preventive

strategy against loss of diversity in population.

Basically, two mechanisms are used to prevent

loss of diversity in population (Ongsakul and

Tippayachai, 2002). First, the individuals are

Multi-Objective Optimization using Microgenetic Algorithm Applied to the Placement of Remote and Manual Switches in Distribution

Networks

273

selected (only once) for a binary tournament. In this

way, not only do the most developed individuals

have an opportunity to participate in the

reproduction but all of them do. The second

mechanism is to insert new individuals each time the

population becomes homogeneous. The best

individual is kept and inserted into a new population

randomly created. When it occurs, a migration has

occurred. If the same individual is the best one along

a certain number of migrations, the algorithm stops

and this individual represents the solution.

3.2 Multi-Objective Algorithm

Multi-objective formulations are realistic models for

many complex engineering optimization problems.

In many real-life problems, there are typically

multiple conflicting objectives that need to be

evaluated in making decisions. Optimizing a

particular solution with respect to a single objective

can result in unacceptable results with respect to the

other objectives. A reasonable solution to a multi-

objective problem is to investigate a set of solutions,

each of which satisfies the objectives at an

acceptable level without being dominated by any

other solution. Typically, there is not a unique

optimal solution for such problems and it is

necessary to use decision maker’s preferences to

differentiate between solutions.

There are two general approaches to multiple-

objective optimization (Deb, 2001). One is to

combine the individual objective functions into a

single composite function. Determination of a single

objective is possible with methods such as utility

theory and weighted sum method. The second

general approach is to determine an entire Pareto

optimal solution set or a representative subset. A

Pareto optimal set is a set of solutions that are

nondominated with respect to each other. In this

paper is used the weighted sum method due to its

simplicity and its characteristic of being a priori

approach since the user is expected to provide the

weighting factors. Assign weighting factors for each

criterion used will reflect its relative importance to

the decision. The weighted sum method combines

the weighting factors and scores for each criterion to

derive an overall value.

In this paper, the solution method will find an

alternative for locating switches devices based on a

set of criteria, in which load importance, reliability,

costs and simultaneous manual and remote

controlled switches are considered. Regarding the

criteria adopted in this paper can be affirmed:

 High values of PRIORI(S) increase the chances

of S to be chosen for allocation.

 High values of SAIDI(S) decrease the chances of

S to be chosen for allocation.

 High values of COST(S) decrease the chances of

S to be chosen for allocation.

For all criteria have the same behavior and the

objective function to be minimized, the portion of

the objective function relative to PRIORI(S) is

calculated as shown in (6).This formulation ensures

that the lowest value of PRIORI(S) corresponds to

the value 1 and other values of PRIORI(S)

correspond to values less than 1.The values for the

weighting factors employed are defined according to

(7) and they reflect the relative importance for each

criterion. The proposed algorithm normalizes each

criterion by its maximum value at a given

population. This procedure is performed for each

new generation.



(



)

=



1−



(



)

−()

()



+



()

()

+



()

()

(6)





+



+



(7)

where, w

is weighting factor for PRIORI; w

weighting factor for SAIDI; w

is weighting factor

for COST; and OF is the objective function.

3.3 Proposed Algorithm

From studies and experiments with several methods

reported in the literature, a MGA is proposed for

solving the tie switches and section switches

placement problem. The MGA uses load

importance, costs and a reliability index as criteria to

find the optimal solution. In order to calculate costs

and reliability index is considered the simultaneous

allocation of MCS and RCS. The proposed

algorithm consists of the following steps:

1) Define the number of section switches (ns) and

maximum number of tie switches (nt) that can

be used to allocation;

2) Define the number of RCS (nr) used to

allocation;

3) Define the weighting factorfor each criterion;

4) Load importance (LI) variable is defined by the

system manager for each consumer connected

to the feeder;

5) Adopt OF expressed in Eq. (6) as objective

function. The MGA is applied to minimize OF;

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

274

6) Randomly create a initial population p with

ns sectionalizers allocated in each

cromosome and go to step 8;

7) Randomly create an population p-1 and add

to it the best chromosome from the last

migration;

8) Determine the objective function of each

chromosome;

9) Choose chromosomes from the present

population using the tournament method

based on crossover rate c. Make crossover

operation using pairs of chromosomes from

this subgroup, determining new

chromosomes;

10) Calculate the objective function value of the

new chromosomes;

11) Replace the present population for a p size

new population compost of the best

chromosomes from the present population

and the new chromosomes;

12) Repeat steps 9 to 11 until the population

reaches an homogeneous degree h previously

chosen or for g generations;

13) Find the best chromosome, keep it, and

discard the others.

14) Repeat steps 7 to 13 until the best individual

does not change for m migrations.

The homogeneous degree may be adjusted

between 90% and 99%. For instance, if this highest

degree is chosen, it means that the population is

considered homogeneous when all individuals have

at least 99% of their genes identical to the genes of

the most adapted individual.

Numbers c, g, h, m and p are previously

specified. The tournament method is a process in

which a population subgroup is randomly formed

and from which the most well-adapted individual is

elected for crossover.

In this work, the chromosome is a vector divided

in two parts:

1. Section Switches Group (SSG): the first ns

positions are the sections in the feeder where SS

are installed. Each position is associated to one

of the sections in the feeder.

2. Tie Switch Group (TSG): The last nt positions

are the sections where tie switches are installed.

These sections are somewhere downstream of the

sections appointed by first nt positions. They are

defined based on SSG.

For instance, considering a feeder with Ns

sections, ns=4 and nt=2, a possible chromosome is

shown in Fig. 3. The sections i, j, k and l are

randomly chosen to receive SS, and the section m

and n are defined based on sections i, j, k and l to

receive TS. Only the genes from SSG are randomly

chosen and copied from parents to their offspring.

TSG is always defined based on SSG.

Figure 3: Genetics information stored in chromosome is

pointed to section array.

The sections represented in a chromosome that

contain remote controlled switch are identified by

negative sign. For instance, considering the example

shown in Fig. 3 and nr=2 (number of RCS), any of

the chromosomes below represent one possible

solution.

Figure 4: Example of 3 different chromosomes with the

same sections.

4 APPLICATION

The proposed algorithm is implemented using

MATLAB® on a 1GHz AMD Dual core personal

computer. In this paper, the test system selected to

illustrate the performance of the algorithm is a 69-

node radial distribution system which includes one

main feeder and seven laterals as shown in Fig. 4.

The system and load data can be referred to Baran

and Wu (1989). The line voltage in substation is

12.66 kV. The mean time for restoring by RCS used

is 3 minutes and by MCS is 48.8 minutes. The mean

outage time of a fault in the feeder is 153 minutes. It

is considered the cost of $ 5,000.00 for MCS and $

10,000.00 for RCS. The annual load growth used is

5% for a ten-year planning horizon. The cost of

energy considered is $ 0.14 / kWh. In the proposed

algorithm, PRIORI is calculated based on LI

established by the system manager. In order to

evaluate the effect of this variable in the solution, it

is considered that all consumers have LI = 0.1,

except a consumer randomly chosen with LI = 1.0

(section 20 is chosen). Table 1 shows the annual

Multi-Objective Optimization using Microgenetic Algorithm Applied to the Placement of Remote and Manual Switches in Distribution

Networks

275

failure rate per section adopted in this application.

The weighting factor of each criteria used in this

application is defined to four different groups: equal

weighting factor for each criteria (group 1 - w1=w2=

w3=1/3), a greater weighting factor for PRIORI

(group 2 - w1=0.8,w2=0.1 and w3=0.1), a greater

weighting factor for SAIDI (group 3 - w1=0.1,

w2=0.8 and w3=0.1) and a greater weighting factor

for COSTS (group 4 - w1=0.1, w2=0.1 and w3=0.8).

Figure 5: The diagram of a 69-bus test system array.

The MGA uses p=20, c=80%, g=200, m=5 and

h=95%. It is considered the allocation of 10-

sectionalizers (ns) and a maximum of 10 tie switches

(nt). Regarding the amount of remote controlled

switching allocated are considered three

configurations: all switches allocated are MCS (C1),

at least 50% of the switches allocated are RCS (C2)

and all switches allocated are RCS (C3).

The 30 executions’ means of the proposed

algorithm are presented in Table 2. The execution

time of MGA is about five minutes. In solutions 1, 2

and 3 the weighting factors used are identical (group

1) but the number of RCS installed is modified

(configurations C1, C2 and C3) in order to analyze

the effect on some parameters related to the feeder

when these changes occur. In solutions 3, 4, 5 and 6

all switches installed are RCS (C3) but the

weighting factors used are different in order to

analyze the effect on some parameters related to the

feeder when these changes occur.

Comparing the results of solutions 1, 2 and 3

verify that the set of sections selected by MGA

differs in all them. This shows that the type of

switch used (RCS or MCS) influences the choice of

the section for allocation. As expected, the larger the

number of RCS installed, the better are the values

obtained for SAIDI. Comparing solutions 1 and 3

(all MCS and all RCS respectively) the costs are

close for a ten-year planning horizon although the

investment cost for RCS is double of investment

cost adopted in this work for MCS. This occurs

because the larger the number of RCS installed, the

lower is the value of EUE. Solution 2 has the lowest

cost because it has 7-RCS installed which reduces

the value of EUE compared to solution 1 and 8-MCS

installed which reduces the investment cost

compared to solution 3.

Comparing solutions 3, 4, 5 and 6 verify that the

set of sections selected by MGA differs in all them.

This is due to the different weighting factors

adopted. Only solution 3(in sections 9 and 47) and

solution 4 (in sections 13, 25 and 47) allocate

switches in the lateral that is section 20 (the highest

LI in the feeder). This is due to the considerable

value of w1used in these solutions in relation to the

others weighting factors. Solution 4 (the highest

value of w1) allocates SS before and after section 20

(sections 13 and 25) where there are loads priority,

Table 1: Switches allocated by the proposed algorithm.

No. Configuration Section Switches Tie Switches

1 Group 1 – C

3;4;15;18;27;

31;39;53;59;60

21;34;63;68

2 Group 1 – C

-3;-6;-26;-27;

50;51;53;-54;

55;-57

-63;68

3 Group 1 – C

-6;-9;-11;-26;

-35;-37;-39;-55;

-56;-57

21;-47;-63;-68

4 Group 2 – C

-3;-6;-13;-25;

-27;-31;-39;-54;

-55;-57

-47;-63;-68

5 Group 3 – C

-9;-11;-14;-21;

-31;-32;-39;-55;

-56;-57

21;-45;-63;-68

6 Group 4 – C

-3;-10;-26;-32;

-35;-44;-54;

-55;-57;-62

-63;-68

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

276

demanding high continuity. As expected, solution 5

(the highest value of w2) shows the best value for

SAIDI and solution 6 (the highest value of w3) the

best value for the costs.

These results confirm that the multi-objective

algorithm proposed achieves efficient solutions

considering simultaneously a different number of

RCS for allocation and different weighting factors

for criteria adopted.

Table 2: Reliability and Costs’ results.

No. Configuration

SAIDI

(h)

EUE

(MWh)

COSTS

(10

1 Group 1 – C

12.78 48.59 0.15

2 Group 1 – C

4.51 17.14 0.13

3 Group 1 – C

2.79 10.61 0.16

4 Group 2 – C

3.27 12.44 0.15

5 Group 3 – C

2.69 10.24 0.16

6 Group 4 – C

3.36 12.77 0.14

5 CONCLUSIONS

This study presented a method for allocating of

section switches and tie switches in radial

distribution networks based on a multi-objective

microgenetic algorithm. This algorithm is applied to

determine the advantage of having a switch installed

in a particular section or not considering load

importance, reliability index, remote or manual

controlled switch and investments costs. The main

stages and characteristics of MGA and its

application in the proposed problem were described.

In order to illustrate the performance of the

algorithm, several experiments using a real

distribution system were conducted. This work did

not consider the possible additional costs for the

electric utility associated with the interruption of an

important load. Solutions 1, 2 and 3 indicated a

considerable effect of number of RCS installed on

reliability index and investments costs. Solutions 3,

4, 5 and 6 indicated a considerable effect of

weighting factors on the problem objectives. These

results showed complete adaptation of algorithm to

different requirements that are determined by the

planner who can adapt the value of weighting factors

according to the technical and economic conditions.

The proposed algorithm has shown excellent results

making this tool a great potential to assist in

planning of distribution networks and also to make

improvements in existing networks.

ACKNOWLEDGEMENTS

The authors thank the IFMA (Federal Institute of

Marannhão) and FAPEMA (Foundation for

Research and Scientific and Technological

Development of Maranhão) for the support to the

development of this article.

REFERENCES

Allan R, Billinton RN (1993) Power system reliability and

its assessment: part III, distribution systems and

economic considerations, Power Eng J 7: 185–192.

Allan RN, Billinton RN, Oliveira MF (1976) An efficient

algorithm for deducing the minimal cuts & reliability

indices of a generalized network configuration. IEEE

Trans Reliab 25: 226-233.

Assis, L, González, JF, Usberti, F, Lyra, Cavellucci, C,

Von Zuben, F (2012) Algoritmo Memetico para

Alocação ótima de chaves manuais e telecomandadas

em um sistema de distribuição de energia elétrica.

Congreso Latino-Iberoamericano de investigación

Operativa. CLAIO.

Baran ME, Wu FF (1989) Optimal capacitor placement on

radial distribution system. IEEE Trans Power Deliv 4:

725-734.

Billinton RN, Jonnavithula S (1995) A test system for

teaching overall power system reliability assessment.

IEEE Trans Power Syst 11: 1670-1676.

Billinton RN, Jonnavithula S (1996) Optimal sectionalizer

allocation in electric distribution systems, IEEE Trans

Power Deliv 11: 1646-1651.

Chakravarty S, Mittra R, Williams NR (2001) On the

application of the microgenetic algorithm to the design

of broad-band microwave absorbers comprising

frequency-selective surfaces embedded in multilayered

dielectric media. IEEE Trans Microwave Theory Tech

49: 1050–1059.

Chen CS, Lin CH, Chuang HJ, Li CS, Huang MY, Huang

CW (2006) Optimal placement of line switches for

distribution automation systems using immune

algorithm. IEEE Trans Power Syst 21: 1209–1217.

Deb K. Multi-objective Optimization Using Evolutionary

Algorithms, Wiley, 2001.

Delfanti M, Granelli PG, Marannino P, Montagna M

(2000) Optimal capacitor placement using

deterministic and genetic algorithms. IEEE Trans

Power Syst, 15: 1041–1046.

Dezaki HH, Abyaneh HA, Agheli A, Mazlumi K (2012)

Optimized switch allocation to improve the restoration

energy in distribution systems. J Electrical Eng 63: 47-

52.

Falaghi H, Haghifam MR, Singh C (2009) Ant colony

optimization-based method for placement of

sectionalizing switches in distribution networks using

a fuzzy multiobjective approach. IEEE Trans Power

Deliv 24: 268–276.

Multi-Objective Optimization using Microgenetic Algorithm Applied to the Placement of Remote and Manual Switches in Distribution

Networks

277

Goldberg DE (1989) Genetic algorithms in search,

optimization and machine learning. Reading, MA:

Addison-Wesley.

Golestani S, Tadayon M (2011) Optimal switch placement

in distribution power system using linear fragmented

particle swarm optimization algorithm preprocessed

by GA. In: Energy Market (EEM), 8th International

Conference on the European, Zagreb, Republic of

Croatia, pp 537 – 542.

Levitin G, Mazal-Tov S, Elmakis D (1995) Genetic

algorithm for optimal sectionalizing in radial

distribution systems with alternative supply. Elsevier

Electr Power Syst Res 35: 149-155.

Moradi A, Fotuhi-Firuzabad M (2008) Optimal switch

placement in distribution systems using trinary particle

swarm optimization algorithm. IEEE Trans Power

Deliv 23: 271–279.

Ongsakul W, Tippayachai J (2002) Micro genetic

algorithm based on migration and merit order loading

solutions to the constrained economic dispatch

problems. Int J Electr Power & Energy Syst, 24: 223–

231.

Rural Utilities Service (2012) Design guide for

sectionalizing distribution lines. BULLETIN 1724E-

102 2012. http://www.rurdev.usda.gov/Support

Documents/UEP_Bulletin_1724E-102.pdf.. Accessed

26 september 2013.

Silva LGW, Pereira RAF, Mantovani JRS (2004)

Optimized allocation of sectionalizing switches and

control and protection devices for reliability indices

improvement in distribution systems. In Proc.

IEEE/PES &D Latin America, São Paulo, Brazil, pp

51-56.

Souza BA, Alves HN, Ferreira HA (2004) Microgenetic

algorithms and fuzzy logic applied to the optimal

placement of capacitor banks in distribution networks.

IEEE Trans Power Deliv 19:942-947.

Sperandio M, Aranha Neto EAC, Sica ET, Trevisan F,

Camargo CCB, Coelho J, Ramos R (2007)

Automation planning of loop controlled distribution

feeders. In: 2nd International Conference on Electrical

Engineering CEE’07, Coimbra, Portugal.

Teng J, Lu C (2002) Feeder switch relocation for customer

interruption costs minimization. IEEE Trans Power

Deliv 17: 254-259.

Tippachon W, Rerkpreedapong D (2009) Multiobjective

optimal placement of switches and protective devices

in electric power distribution systems using ant colony

optimization. Elsevier Electr Power Syst Res 79:

1171-1178.

Toune S, Fudo H, Gengi T, Fukuyama Y, Nakanishi Y

(1998) A reactive Tabu Search for service restoration

in electric power distribution systems. In: IEEE

International Conference on Evolutionary Computa-

tion, Anchorage-Alaska, pp 1-7.

Villasanti, A.; Baran, B.; Gardel, P. (2008) Multiobjective

allocation of remotely controlled switches in an

electric distribution power system. In: IEEE

Transmission and Distribution Conference and

Exposition: Latin America, 2008, IEEE/PES, pp.1-7.

Ziari I, Ledwich G, Wishart M, Ghosh A, Dewadasa M

(2009) Optimal allocation of a cross-connection and

sectionalizers in distribution systems. TENCON IEEE

Region 10 Conference, pp:1-5.

ECTA 2015 - 7th International Conference on Evolutionary Computation Theory and Applications

278