Electrical Power and Energy Systems 54 (2014) 235–243

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

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

Capacitor placement for net saving maximization and system stability

enhancement in distribution networks using artificial bee colony-based
approach
Attia A. El-Fergany a,⇑, Almoataz Y. Abdelaziz b
a
Department of Electrical Power & Machines, Faculty of Engineering, Zagazig University, P.O. Box 44519, Zagazig, Egypt
b
Department of Electrical Power & Machines, Faculty of Engineering, Ain Shams University, Cairo, Egypt

a r t i c l e i n f o a b s t r a c t

Article history: This manuscript introduces an approach to allocate static capacitors along radial distribution networks
Received 12 March 2013 using the artificial bee colony algorithm. In general practice the high potential buses for capacitor place-
Received in revised form 1 July 2013 ment are initially identified using loss sensitivity factors. However, that method has proven less than sat-
Accepted 13 July 2013
isfactory as loss sensitivity factors may not always indicate the appropriate placement. In the proposed
approach, the algorithm identifies optimal sizing and placement and takes the final decision for optimum
location within the number of buses nominated. The result is enhancement of the overall system stability
Keywords:
index and potential achievement of maximum net savings. The overall accuracy and reliability of the
Artificial bee colony
Capacitor allocations
approach have been validated and tested on radial distribution systems with differing topologies and
Loss reductions of varying sizes and complexities. In the manuscript the results are compared with those obtained using
Net saving maximizations recent heuristic methods and show that the proposed approach is capable of producing high-quality solu-
Voltage stability index tions with good performance of convergence, and demonstrated viability.
Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction
Reactive power addition can be beneficial only when correctly
applied. Correct application means choosing the correct position
and size of the reactive power support. It is not possible to achieve
zero losses in a power system, but it is possible to keep them to a
minimum [1–3] to reduce the system overall costs. The reactive
power support is one of the well-recognized methods for the
reduction of power losses together with other benefits; such as in-
creased utilization of equipment, unloading of overloaded system
components, and stopping the premature aging of the equipment.
However, other alternatives can be used as the network reconfigu-
ration, which can provide the same mentioned benefits. Bear in
mind, too many capacitors at the wrong points will increase losses
on the lines. However, the minimization of losses does not guaran-
tee the maximization of benefits unless the problem is well-
formulated.
Numerous methods for solving this problem with a view to
minimizing losses have been suggested in the literature based
on both traditional mathematical methods and more recent
heuristic approaches. A comprehensive survey of the literature
from the last decade focusing on the various heuristic optimiza-
⇑ Corresponding author. Tel.: +20 100 5705526.
E-mail addresses: el_fergany@ieee.org (A.A. El-Fergany), almoatazabdelaziz@
hotmail.com (A.Y. Abdelaziz).
tion techniques applied to determine the OCP and size is pre-
sented in [4]. Several heuristic methods have been developed
in the last decade such as tabu search [5], PSO [6,7], the har-
mony search algorithm [8], ant colony optimization-based algo-
rithm [9,10] and a simulated annealing technique [11], GA [12]
and a GA-fuzzy logic algorithm [13] to solve capacitor place-
ment optimization problems.
The bacterial foraging with a PSO algorithm used to deter-
mine the optimal placement of capacitors has been introduced
in [14], and PGSA has been used for capacitor placement in
[15]. More recently, an immune based optimization technique
[16], the integration of DE and PS [17], and Big Bang-Big Crunch
optimization [18] to obtain the optimum values of shunt capac-
itors in radial distribution networks have been utilized and
employed.
Algorithms for enhancing voltage stability of electrical systems
by OCP have been developed [19,20] and a relationship between
voltage stability and loss minimization and the concept of maxi-
mizing voltage stability through loss minimization were defined
and outlined [21,22].
In this article, an ABC-based algorithm is utilized to ascertain
the optimal size and select optimum locations of shunt capacitors.
High potential buses for capacitor placement are initially identified
by the observations of LSF with weak voltage buses. The proposed
method improves the voltage profile and reduces system losses in
addition to enhancing voltage stability. The method has been

0142-0615/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijepes.2013.07.015
236 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243

Nomenclature

n total number of lines Sli actual line flow of line i

N total number of network buses Srated
li rated line transfer capacity
PLoss total network peak active loss PFmin lower limit of overall system power factor at substation
QLoss total network peak reactive loss (slack bus)
VSI(j) voltage stability index of bus j PFmax upper limit of overall system power factor at substation
Iij current of line i–j (slack bus)
Rij resistance of line i–j SN number of food sources/colony size
Xij reactance of line i–j D number of optimization parameters
|Vi| voltage magnitude of bus i fi cost value of ith solution
|Vj| voltage magnitude of bus j fiti modified fitness of ith solution
Pj total effective real power load fed through bus j xjmax upper bounds for the dimension j
Qj total effective reactive power fed through bus j xjmin lower bounds for the dimension j
Ce energy cost uij random number in the range [1, 1]
T time period NCN maximum cycle number
PLa total active power loss after compensation kVC penalty function for voltage limit constraint
PLb total active peak power loss before compensation kPFC penalty function for power factor constraint
CCi cost of installation kLFC penalty function for line flow constraint
CO capacitor operating cost kCC penalty function for maximum total compensation con-
CC cost of the capacitor purchase straint
NB number of candidate effective buses (that have compen-
sations with values >0) List of Abbreviations
a depreciation factor ABC artificial bees colony
lF magnifying factor DE differential evolution
nl number of load buses GA genetic algorithm
PSlack active power supplied from the slack bus LSF loss sensitivity factor
QSlack reactive power supplied from the slack bus OCP optimal capacitor placement
PD(i) active power demand of load at bus i PGSA plant growth simulation algorithm
QD(i) reactive power demand of load at bus i PSO particle swarm optimization
PL(i) active power loss at branch j PS pattern search
QL(i) reactive power loss at branch j VSI voltage stability index
QC(i) amount of reactive power of installed capacitors at bus i HS heuristic search
Vi,min lower permissible voltage limit at bus i EA evolutionary algorithm
Vi,max upper permissible voltage limit at bus i P.U. per unit
Q min
Ci lower reactive power limit of compensated bus i LF load flow
Q max
Ci upper reactive power limit of compensated bus i

tested and validated on a variety of radial distribution systems and
the detailed results are presented.
Different simplified methods of normal load distribution flow
and other special techniques have been proposed [23,24]; these
deal mainly with high ratio of R/X in distribution systems. As nei-
ther Newton–Raphson nor Gauss–Seidel methods have proven effi-
cient, have experienced difficulties and may not be convergent
[25–29]. A method which can find the LF solution of a radial distri-
bution system directly by using topological characteristic of distri-
bution network may overcome the limitations of Newton–Raphson
and Gauss–Seidel methods. The advantage of this technique is that
there is no problem of convergence with respect to radial networks
with high ratio of R/X. The distribution power flow suggested in
[24] is used in this study.
For stable the operation of the radial distribution networks, VSI
(j) P 0. The node, at which the value of the VSI has lower value, is
more sensitive to collapse. The node with the smallest VSI is the
weakest node and the voltage collapse phenomenon will start from
that node. Therefore, to avoid the possibilities of voltage collapse,
the VSI of nodes should be maximized.
3. Problem and objective function formulation
The objective of capacitor placement in the distribution system
is to maximize the peak active power loss reduction, reduce capac-
itor costs and enhance the system static stability subject to specific
operating constraints. The objective function is mathematically
formulated as,

2. Voltage stability index

Many different indices have been introduced to evaluate the

power systems security level from the point of voltage static stabil-
ity [30–33]. A new steady state VSI is proposed [33] for identifying
the node, which is most sensitive to voltage collapse and expressed
in Eq. (1) is utilized in the work. Fig. 1 shows the simple electrical
equivalent of the radial distribution system.

VSIðjÞ ¼ jV i j4  4fPj  X ij  Q j  Rij g2  4fPj  Rij þ Q j  X ij g:jV i j2 ð1Þ

Fig. 1. Line i–j power system model.
 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243 237

( " # )
X
NB X
N
where
Maximize C e  ðPLb  P La Þ  T  a: C Ci  N B þ C C  Q C ðiÞ  C O  N B þ lF  VSIðjÞ
i¼1 j¼2
" # 9
XN X
N
>
>
ð2Þ kVC ¼ WV  maxð0; hjV i j  V i;max iÞ þ maxð0; hV i;min  jV i jiÞ >
>
>
>
i¼1 i¼1 >
>
Subject to the satisfaction of the Active and reactive power flow >
>
kPFC ¼ WPF  ½maxð0; hPF ov erall  PF max iÞ þ maxð0; hPF min  PF ov erall iÞ >
>
>
balance equations and a set of inequality constraints. " # =
X n
rated
kLFC ¼ WL  maxð0; hSli  Sli iÞ >
>
3.1. Power balance constraints >
>
i¼1 >
>
" * +! # >
>
XNB nl
X >
>
Power balance (Active and Reactive) constraints, which are kCC ¼ WC  max 0; Q C ðiÞ  Q D ðjÞ >
>
>
;
equality constraints and include two nonlinear recursive power i¼1 j¼1
flow equations, can be formulated as follows, ð10Þ
nl
9
X X
n
>
>
PSlack ¼ PD ðiÞ þ PL ðjÞ >
>
WV, WPF, WL, and WC are the penalty function weights having large
>
= positive value.
i¼1 j¼1
ð3Þ
X
NB nl
X X
n >
>
Q Slack þ Q C ðiÞ ¼ Q D ðiÞ þ Q L ðjÞ >
>
>
; 4. Identification of potential buses using LSF
i¼1 i¼1 j¼1

The estimation of these candidate nodes basically helps in sig-

3.2. Voltage limit constraint nificant reductions of the search space for the optimization proce-
dure. In this proposed work, LSF is utilized for this purpose [22]. It
The voltage magnitude at each bus must be maintained within is intuitive that a section in a distribution system with high losses
its limits and is expressed as, and lower voltage or VSI has higher priority for placement of
capacitors. Whereas, a low loss sections with good voltage are
V i;min 6 jV i j 6 V i;max ; i ¼ 1:::N ð4Þ not optimal for capacitor placements.
The LSF may be able to predict which bus will have the greatest
3.3. Reactive compensation limit loss reduction when reactive compensation is put in place. Con-
sider a distribution line connected between ‘i’ and ‘j’ buses as
Reactive power constraint in which injected reactive power at shown in Fig. 1.
each candidate bus must be within their permissible ranges. Active power loss in the ijth line between i–j buses is given by
3I2ij Rij and can be expressed as shown in the following equation,
Q min max
Ci 6 Q Ci 6 Q Ci ; i ¼ 1:::NB ð5Þ
ðP2j þ Q 2j Þ
Pijloss ¼ 3  :Rij ð11Þ
jV j j2
3.4. Line capacity limit
Thus, the sensitivity analysis factor is a derivative of the power
The apparent power flow through the line Sl is restricted by its loss with reactive power; Qj, as indicated in the following equation,
maximum rating limit as,
@Pijloss Qj
¼6  Rij ð12Þ
Sli 6 Srated
li ; i ¼ 1:::n ð6Þ @Q j jV j j2
The values are arranged in descending order for all the lines of
3.5. Maximum total compensation the given system. The descending order of the elements vector
will decide the sequence in which the buses (receiving-end of
From practical limitation, maximum compensation by using lines) are to be considered for compensation. Receiving-end buses
capacitor bank is limited to the total load reactive power demand. of the lines of higher LSF and lower VSI have a greater chance of
X
NB nl
X being identified as candidate locations for capacitor installations
Q C ðiÞ 6 Q D ðjÞ ð7Þ for the purposes of bi-objective (net saving and VSI) solution
i¼1 j¼1 and loss minimizations objective. However, in case of the objec-
tive is to maximize system VSI only, the buses with lower VSI
3.6. Overall system power factor are identified.

The overall power factor (PFoverall) is the value of power factor at
the main feeder of the primary substation; supplying the radial
distribution network. System power factor should be maintained
within desirable lower and upper limits.
PF min 6 PF ov erall 6 PF max
A penalty factor associated with each violated constraint is
added to the objective function in order to force the solution to
stay away from the infeasible solution space; to respect the
inequality constraints. Therefore, the optimal solution is estab-
lished when no constraints is violated or even with acceptable tol-
erance and the objective function is maximized.
The penalty function can be formulated as follows,
Penalities ¼ kVC þ kPFC þ kLFC þ kCC
5. Artificial bees colony algorithm
The ABC algorithm was proposed by Karaboga for optimizing
numerical problems in 2005 [34]. It simulates the intelligent forag-
ing behavior of honey bees warms. It is a very simple, robust and
ð8Þ
population based stochastic optimization algorithm. The perfor-
mance of the ABC algorithm has been compared with those of
other well-known modern heuristic algorithms such as GA, DE
and PSO on constrained and unconstrained problems [35,36]. Now-
adays, the ABC algorithm is one of the most popular approaches
used in optimization problems and requires less control parameter
to be tuned. The ABC algorithm has three phases: employed bee,
onlooker bee and scout bee. In the employed bee and the onlooker
bee phases, bees exploit the sources by local searches in the neigh-
ð9Þ
borhood of the solutions selected based on deterministic selection
238 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243
in the employed bee phase and the probabilistic selection in the
onlooker bee phase. In the scout bee phase which is an analogy
of abandoning exhausted food sources in the foraging process,
solutions that are no longer beneficial for search progress are aban-
doned, and new solutions are inserted instead of them to explore
new regions in the search space. The algorithm has a well-balanced
exploration and exploitation ability.
The main steps of the ABC algorithm in the form of Pseudo-code
are given below [35]
Step 1: Initialize the population of solutions xi,j, i = 1, . . ., SN,
j = 1, . . ., D
Step 2: Evaluate the population
Step 3: Cycle = 1
Step 4: Repeat
Step 5: Produce new solutions (food source positions) vi,j in the
neighborhood of xi,j for the employed bees using the
formula v i;j ¼ xi;j þ /ij ðxi;j  xk;j Þ (k is a solution in
the neighborhood of i) and evaluate them
Step 6: Apply the greedy selection process
Step 7: Calculate the probability values Pi for the solutions xi,j by
means of their fitness values using Eq. (13)
fit
pi ¼ PSN i
m¼1 fit m
In order to calculate the fitness values of solutions, the
following Eq. (14) is employed;
( The pre-identification and the estimation of high potential
1
1þfi
if f i P 0
fit i ¼
1 þ jfi j if f i < 0
Normalize Pi values into [0, 1]. fi is obtained separately
for each individual ith solution through Eq. (2).
Step 8: Produce the new solutions (new positions) vi,j for the
onlookers from the solutions xi,j selected depending on
Pi and evaluate them
Step 9: Apply the greedy selection process
Step 10: Determine the abandoned solution (source), if it exists,
and replace it with a new randomly produced solution
xi,j for the scout using Eq. (15)
xji ¼ xjmin þ randð0; 1Þ  ðxjmax  xjmin Þ
Step 11: Memorize the best food source position (solution)
achieved so far
Step 12: Cycle = Cycle + 1
Step 13: Until Cycle = MCN.
There are three control parameters used in the ABC-based
algorithm; the number of the food sources which is equal to
the number of employed or onlooker bees, the value of limit
and the MCN. In ABC, providing that a position cannot be im-
proved further through predetermined number of cycles, then
that food source is assumed to be abandoned. The value of pre-
determined number of cycles is an important control parameter
of the ABC algorithm; this is termed the ‘‘limit’’ for
abandonment.
The ABC algorithm employs four different selection processes:
(1) A global selection process used by the artificial onlooker
bees for discovering promising regions.
(2) A local selection process carried out in a region by the arti-
ficial employed bees and the onlookers depending on local
information for determining a neighbor food source around
the source in the memory.
(3) A local selection process called greedy selection process car-
ried out by all bees.
(4) A random selection process carried out by scouts.
The procedure of the ABC algorithm to solve OCP can be sum-
marized in the flow chart diagram of Fig. 2.
6. Test cases, numerical results and simulations
In order to test the effectiveness and performance of the pro-
posed ABC-based algorithm, it has been applied to several distribu-
tion radial test systems (10-bus, 22-bus, 28-bus, 30-bus, 33-bus,
34-bus, 69-bus radial distribution feeders and an actual Portuguese
radial distribution system with 94 nodes).
Due limitations of space supposed by submission guidelines,
only two radial distribution systems: the 34-bus and the actual
Portuguese radial distribution system with 94-node are selected
for reporting and demonstration in this article, to examine the
applicability of the proposed approach. In all calculations; for all
the test cases, the following constants are assumed and applied
as shown in Table 1.
The net savings are calculated using:
sav ings
ð13Þ Net ¼ fcos t of energy reductions  a
year
 ½cos t of installation þ cos t of purchase
 yearly operating cos tg ð16Þ
buses significantly help in the reduction of the search space for
ð14Þ
the optimization procedure. Setting the lower limit of capacitor
range to 0, will permit the proposed approach to select the opti-
mum locations within the range of bus nominations initially iden-
tified by the LSF method which has to be set manually by the user;
just to set the number of buses due for search. After intensive trials
with nominating 10–25% of total network number of buses after
ranking is guaranteeing the optimal or near to optimal solutions
(these percentage figures are obtained after many trials). For small
size networks, user may nominate/set the initial number of higher
potential buses to 20–25% of network buses (i.e. for 34-bus test
case, set to 9 as initial number of buses for capacitor allocations)
and for medium size, nominate 15–20% of network buses. How-
ð15Þ ever, for large scale radial networks, set the number of potential
buses for capacitor placement to 10–15% of network buses (i.e.
for 94-bus radial network is 15 buses might be nominated initially
for capacitor placements).
It is well-known that LSF observations, may not lead to the opti-
mum locations. Due to the fact the LSF calculations depend on the
network topology, configurations, loading, etc. and to tackle these
limitations, the algorithm will search the optimum number of
buses and select them for capacitor placements.
The magnifying factor (lF) is chosen to equal 500 for the two
case studies of 34-bus and 94-bus radial distribution systems.
The proposed method has been programmed and implemented
using MATLAB [37,38].
6.1. 34-Bus test system numerical results and simulations
This 34-bus test case has 4-lateral radial distribution system
which is shown in Fig. 3. The data of the system are obtained from
[2]. The total load of the system is (4636.5 + j2873.5) kVA.
Using base LF to candidate the potential buses for capacitor
placement and based on LSF values; {19, 22, 20, 21, 23, 24, 25,
26 & 27}. Parameters adopted for the ABC algorithm for the test
case of a 34-bus, and the required inequality constraints that
should be respected are given in Table 2.
 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243 239

Cycle=1
Read system input data
(busdata, line data,)
Generate new population for employed
bees, run LF and evaluate the objective
function.
Run base case LF and get the initial
values for voltages, losses and VSI
Apply the greedy selection process

Calculate the probability values for the

Estimate the LSF values for the buses solutions

Produce the new populations for the

Rank buses and propose higher potential
onlookers and evaluate them
buses for capacitor placement.

Apply the greedy selection process for

the onlookers
Set lower and upper bounds for the
constraints, set algorithm control
Determine the abandoned solution
parameters and MCN

Memorize the best solution achieved

Initialize the population within their

lower and upper bounds, run LF and Cycle=Cycle+1
evaluate the objective function

No
If Cycle< MCN
Initialize the population, for amount of
vars to be injected within their lower and
upper bounds

Get Optimal Solution

and report

Fig. 2. Flow chart of ABC Algorithm and capacitor allocations.

Table 1 Table 2
Constants for the rates using a long with test cases. Control parameters adopted for the ABC algorithm and target setting for the
constraints.
SN Item Proposed rate
Item Proposed Setting
1 Average energy cost $0.06/kW h
2 Depreciation factor 20% Colony size (SN) 60
3 Purchase cost $25/kVAr Limit 30
4 Installation cost $1600/location MCN 100
5 Operating cost $300/year/location Bus voltage constraint 0:95 6 V i 6 1:05
6 Hours per year 8760 Power factor constraint 0:95 6 PF ov erall 6 0:99
Allowable capacitor range 0 kVAr to 1500 kVAr with step of 50 kVAr

14 15 16

13 28 29 30 31 32 33 34
950 kVAr and 900 kVAr, respectively. The CPU computational time
Slack needed is 10.08 s to accomplish this optimization process by the
1 2 3 4 5 6 7 8 9 10 11 12 proposed ABC-based method. The results of the proposed method
compared with the results of GA method [12], PSO method [6],
17
HS-based method [2], PGSA method [15] and EA method [39] for
18 24 25 26 27 the reactive compensation required and relevant bus allocations
are shown in Table 3.
19 20 21 22 23 For comparison purposes, the reported figures in [12,6,2,15,39]
of reactive power at specific buses are recycled to calculate the sys-
Fig. 3. Single line diagram of a 34-bus radial distribution network.
tem losses and the net savings (see Table 4) with the same rates
proposed in this article as shown in Table 1 and Eq. (16).
The system overall power factor is significantly corrected from
Set the number of initial higher potential buses estimated by 0.8557 lagging (base case) to 0.9798 lagging with capacitor alloca-
LSF to 9. After running the proposed optimization algorithm to tions, respectively. The VSI of a 34-bus radial distribution system
select the optimal locations and determine the capacitor optimal without and with compensations is depicted Fig. 4.
sizes, the outcome leads to only 2 locations for capacitor placement From the results illustrated and shown in table 4, the proposed
which are buses 19 and 24 with optimum capacitor ratings of algorithm yields to reduce peak losses to 167.99 kW with 1,850
240 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243

Table 3
Optimal location of capacitor placement and value of capacitor size in KVAr.

Method Proposed GA [12] PSO [6] HS [2] PGSA [15] EA [39]

(Location, Size in kVAr) (19, 950) (5, 300) (19, 781) (26, 1400) (19, 1200) (8, 1050)
(24, 900) (9, 300) (22, 803) (11, 750) (22, 639) (18, 750)
(12, 300) (20, 479) (17, 300) (20, 200) (25, 750)
(22, 600) (4, 250)
(26, 300)
Total KVAr placed 1850 1800 2063 2700 2039 2550

Table 4
Results and comparisons of a 34-Bus radial feeder test case without and with OCP showing different heuristic approaches.

Point of comparison Without OCP With OCP

Proposed approach GA [12] PSO [6] HS [2] PGSA [15] EA [39]
Vmin (P.U.)a 0.9416 0.9496 0.9478 0.9486 0.9522 0.9479 0.9501
Vmax (P.U.)a 0.9941 0.9949 0.9949 0.9950 0.9953 0.9950 0.9952
VSImin 0.786 0.81294 0.8071 0.8097 0.8219 0.8074 0.8149
VSImax 0.9765 0.9797 0.9796 0.9800 0.9811 0.9800 0.9808
P34
28.6211 30.1221 29.0894 29.1353 29.3214 29.1149 29.2675
j¼2 SVIðjÞ
Ploss (kW) 221.7373 167.99 164.9586 169.3592 168.4811 171.9643 161.2673
Reductions in Ploss% – 24.24% 25.61% 23.62% 24.02% 22.45% 27.27%
Qloss (kVAr) 65.2230 49.015 49.9643 47.1771 48.4489 48.6740 49.0518
Reductions in Qloss% – 24.85% 23.39% 27.67% 25.72% 25.37% 24.79
PFoverall 0.8557 0.9798 0.9825 0.9970 0.9989 0.9738 0.9837
P
Qc (kVAr) – 1850 1800 2063 2700 2039 2550
Net Savings/year – $17756.00 $15093.00 $15570.00 $12017.00 $15590.00 $17173.00
a
The reported values of Vmin and Vmax are shown excluding the slack bus # 1.

Table 5
kVAr installed at 2 locations only (buses 19 & 24), which are Summaries for the VSI and Ploss objectives.
optimally selected using the current methodology out of 9 initial
Item Power loss minimization VSI maximization
higher potential buses estimated by LSF calculations. The net sav-
ings gained are $15093.00, $15570.00, $12017.00, $15590.00, SVImin 0.8163 0.82376
$17173.00 and $17756.00 using GA [12], PSO [6], HS [2], PGSA SVImax 0.981 0.9807
P34 29.2815 30.3061
[15], EA [39] and the proposed algorithm, respectively. One may j¼2 SVIðjÞ
Ploss (kW) 161.087 169.92
note the superiority of the proposed approach compared to the
Qloss (kVAr) 47.1452 49.2659
others. Tables 3 and 4 conclude that the proposed ABC-based ap- PF 0.9978 0.9952
Poverall
proach possesses higher system stability index and higher annual Qc (kVAr) 2600 (3 locations) 2450 (8 locations)
net saving compared to other heuristic methods with lesser num- Net Savings/year $17,018.00 $8,785.00
ber of locations which realizes, an added value to the proposed
approach.
In case of the objective is to minimize the active power loss only the extracted summaries for the cases of VSI maximization and Ploss
or to maximize VSI only, the maximization of the cost savings is minimization as well. In the case of VSI maximization, the nomi-
not guaranteed. Set the number of initial higher potential buses nated buses for capacitor allocations are identified based on lowest
to 9 (for Ploss objective using ranked buses with higher LSF and buses VSI values. For Ploss objective, the algorithm selected 3 loca-
for VSI, utilize the buses with lower values of VSI). Table 5 depicts tions out of 9 nominated buses with higher LSF. However, for the

0.98
VSI with OCP
VSI without OCP
0.96

0.94
Bus Stability Voltage Index

0.92

0.9

0.88

0.86

0.84

0.82

0.8

0.78
5 10 15 20 25 30
Bus Number

Fig. 4. VSI values against bus number for a 34-bus radial distribution feeder with and without OCP (2 locations).
 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243 241

Table 6 Table 8
Control parameters adopted for the ABC algorithm and targets for the constraints Summaries and results for the case of a 94-node actual Portuguese radial distribution.
(case of a 94-node).
Point of Without With compensation With
Item Proposed setting comparison compensation (bi-objective) compensation
(pure VSI
Colony size (SN) 100
objective)
Limit 50
using proposed
MCN 100
approach
Bus voltage constraint 0:90 6 V i 6 1:10
Proposed DE-PS
Power factor constraint 0:95 6 PF ov erall 6 0:99
approach based [41]
Allowable capacitor range 0 kVAr to 1500 kVAr with step of 50 kVAr
Vmin (P.U.)a 0.84848 0.90721 0.9038 0.93726
Vmax (P.U.)a 0.99508 0.99699 – 0.99726
Table 7 VSImin 0.5183 0.6774 – 0.771687
Optimal locations and sizes for the 94-nodes test case. VSImax 0.9803 0.9879 – 0.988937
P94
62.2577 75.0565 – 79.6829
j¼2 SVIðjÞ
Bus/location 18 21 54 Total Ploss (kW) 362.8580 271.3590 271.9854 317.293
KVAr 600 450 1050 2100 Reductions in Ploss (%) – 25.23% 25.04% 12.56%
Qloss (kVAr) 504.0420 374.5060 376.0200 419.713
Reductions in Qloss (%) – 25.70% 25.34% 16.73%
VSI objective, the approach has selected 8 locations out of 9 buses PFoverall 0.8769 0.9931 0.9878 0.998867
P
nominated with lower VSI (refer to Table 5). Higher potential buses QC (kVAr) – 2100 1900 2400
ranked based on VSI values are {27, 26, 25, 24, 23, 22, 21, 20, 19, 34, (3 (8 (8
locations) locations) locations)
33, 12, 32, 11, 31, 10, 9, 18, . . .} that are used only in case of the
Net Savings/year – $35732.00 $33302.00 $5248.60
objective is to maximize the system VSI, ignoring the maximum
a
revenue of capacitor allocation. The reported values of Vmin and Vmax are shown excluding the slack bus # 1.

The proposed ABC-based approach can reduce peak real losses to

161.087 kW (i.e. percentage of reduction is 27.35%) with total reac-
tive compensation of 2600 kVAr allocated at buses of 8, 18 and 25
with ratings of 900 kVAr, 900 kVAr and 800 kVAr, respectively. The
net yearly saving is $17018.00 which is lesser than that obtained
in the case of bi-objective (both saving and VSI maximizations) as
indicated in Tables 4 and 5. However, the net savings is dramatically
reduced in case of a pure VSI maximization objective. On the other
hand, the total VSI has been significantly improved. The constraints
have been checked and found within acceptable limits.
6.2. 94-Bus test system results and simulations
This proposed approach has been applied to an actual Portu-
guese radial distribution system with 94 nodes, as shown in
Fig. 5. The network layout, including line data and load data, and
its physical characteristics are summarized and obtained from
[40]. This network consists of 22-lateral radial branches with total
loads of (4797 + j 2323.9) kVA.
The bus voltages constraints are (1 P.U. ± 10%) have been pro-
posed in this test case with 15 kV voltage level.
The most likely buses for capacitor placements as identified by
LSF indicators are {11, 10, 90, 18, 8, 21, 54, 52, 15, 9, 83, 20, 16, 24,
23, 25, 12, 13, 19, 17, . . .}. Table 6 shows the parameters adopted
for the ABC algorithm for the test case of a 94-node actual Portu-
guese radial distribution and the required constraints.
Set the number of initial higher buses range resulted by the LSF
observations to 15, to let the proposed algorithm to select the opti-
mal locations and amount of compensations required accordingly.
The approach has selected 3 buses for optimal capacitor allocations
with the relevant amount of reactive compensation required per
each location which is depicted in Table 7. The computational time
exhausted to complete this optimization process is 70.25 s on aver-
age. Once again, this proves the ability of the proposed approach to
allocate capacitors at a minimum number of locations. The sum-
maries and numerical results are tabulated and shown in Table 8.
The reductions in the peak active and reactive losses are 25.23%
and 25.70%, respectively. However, the overall power factor has

45 49 50 51

44 48 71 73

34 38 43 46 47 69 70 72 74
Slack
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

35 39 52

36 40 53 59 60 61 62 63

37 41 42 54 58 64

55 56 57 65 66

76 87 88 89 93 94

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33

75 77 83 84 85 86 90 91 92
78

79 80 81 82

Fig. 5. Single line diagram of an actual Portuguese 94-nodes radial distribution system.
242 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243

1
VSI with OCP
VSI without OCP
0.95

0.9

Bus Stability Voltage Index

0.85

0.8

0.75

0.7

0.65

0.6

0.55

0.5
10 20 30 40 50 60 70 80 90
Bus Number

Fig. 6. VSI values against bus number for a 94-bus radial distribution feeder with and without OCP (3 locations).

1
VSI with OCP
0.95 VSI without OCP

0.9
Bus Stability Voltage Index

0.85

0.8

0.75

0.7

0.65

0.6

0.55

0.5
10 20 30 40 50 60 70 80 90
Bus Number

Fig. 7. VSI values against bus number with and without OCP (8 locations) for a 94-bus radial distribution feeder with a pure objective of VSI.

been enhanced from 0.8769 lagging to 0.9931 lagging. In addition,
a significant improvement has been observed regarding the system
stability aspects. Fig. 6 shows the improvement in the bus stability
indices before and after capacitor installations.
Compared to DE-PS method [41], the obtained net year savings
is higher and in less number of locations.
In case of the objective is to maximize the system SVI, the
ranked buses based on VSI values are {92, 91, 90, 33, 32, 31, 30,
94, 29, 28, 93, 27, 26, 89, 25, . . .}. Fig. 7 depicts the trend of VSI
against buses with and without compensations and outcome val-
ues are tabulated in Table 8 (column 4). The total system VSI and
VSImin are considerably improved to 79.6829 and 0.7717, respec-
tively. On the other hand, the yearly net savings has been dramat-
ically reduced. One may note that both net saving maximizations
and system VSI are conflicting objectives.
7. Conclusions
The application of the ABC-based optimization approach for
solving the problem of capacitor allocations (sizing and location)
to maximize the net benefits and to improve system static voltage
stability has been depicted and investigated. The OCP problem has
been solved by two step method, 1) the candidate locations for
compensation are pre-identified using LSF values based on the ini-
tial case of LF run, and 2) the final optimal locations and size of
capacitors being obtained using the ABC-based approach. The
numerical results of the simulation indicate a considerable
improvement in active and reactive power losses reductions, volt-
age stability enhancements, and power factor corrections while
maximizing the net savings. The proposed ABC-based approach
outperforms other methods showcased in the recent state-of-the
art literature in the area of OCP in terms of the quality of the solu-
tion and the computational efficacy. The main advantage of the
ABC-based algorithm is that it does not require expending more ef-
fort in tuning the control parameters, as in the case of GA, DE, PSO
and other EAs. This feature marks the proposed ABC-based algo-
rithm as being advantageous for implementation.
References
[1] Haque MH. Capacitor placement in radial distribution systems for loss
reduction. IEE Proc Gener Transm Distrib 1999;146(5):501–5.
[2] Chis M, Salama MMA, Jayaram S. Capacitor placement in distribution system
using heuristic search strategies. IEE Proc Gener Transm Distrib
1997;144(3):225–30.
[3] Short TA. Electric power distribution equipment and systems capacitor
application. 1st ed. CRC Press, Taylor & Francis Group; 2005. ISBN-
10.0849395763.
[4] Sirjani R, Azah M, Shareef H. Heuristic optimization techniques to determine
optimal capacitor placement and sizing in radial distribution networks: a
comprehensive review. PRZEGLA˛D ELEKTROTECHNICZNY (Electr Rev)
2012;88(7a):1–7.
[5] Gallego RA, Monticelli AJ, Romero R. Optimal capacitor placement in radial
distribution networks using tabu search. IEEE Trans Power Syst
2001;16(4):630–7.
 A.A. El-Fergany, A.Y. Abdelaziz / Electrical Power and Energy Systems 54 (2014) 235–243
[6] Prakash K, Sydulu M. Particle swarm optimization based capacitor placement
on radial distribution systems. In: IEEE PES general meeting, Tampa, FL; 2007.
p. 1–5.
[7] Singh SP, Rao AR. Optimal allocation of capacitors in distribution systems using
particle swarm optimization. Int J Electr Power Energy Syst 2012;43(1):
1267–75.
[8] Sirjani R, Mohamed A, Shareef H. Optimal capacitor placement in a radial
distribution system using harmony search algorithm. J Appl Sci 2010;10(23).
2996-2996.
[9] Chang CF. Reconfiguration and capacitor placement for loss reduction of
distribution systems by ant colony search algorithm. IEEE Trans Power Syst
2008;23(4):1747–55.
[10] Annaluru R, Das S, Pahwa A. Multi-level ant colony algorithm for optimal
placement of capacitors in distribution systems. Congress on Evol Comput, CEC
2004;2:932–7.
[11] Chiang HD, Wang JC, Cockings O, Shin HD. Optimal capacitor placements in
distribution systems: part 1: a new formulation and the overall problem. IEEE
Trans Power Deliv 1990;5(2):634–42.
[12] Swarup KS. Genetic algorithm for optimal capacitor allocation in radial
distribution systems. In: Proc the 6th WSEAS Int. Conf Evolut Comput, Lisbon,
Portugal, June 16–18; 2005. p. 152–59.
[13] Hsiao YT, Chen CH, Chien CC. Optimal capacitor placement in distribution
systems using a combination fuzzy-GA method. Int J Electr Power Energy Syst
2004;26(3):501–8.
[14] Tabatabaei SM, Vahidi B. Bacterial foraging solution based fuzzy logic decision
for optimal capacitor allocation in radial distribution system. Electr Power Syst
Res 2011;81(4):1045–50.
[15] Rao RS, Narasimham SVL, Ramalingaraju M. Optimal capacitor placement in a
radial distribution system using plant growth simulation algorithm. Int J Electr
Power Energy Syst 2011;33(5):1133–9.
[16] Huang S. An immune-based optimization method to capacitor placement in a
radial distribution system. IEEE Trans Power Deliv 2000;15(2):744–9.
[17] Attia El-Fergany. Optimal capacitor allocations using evolutionary algorithms.
IET Gener Transm Distrib 2013;7(6):593–601.
[18] Sedighizadeh M, Arzaghi-haris D. Optimal allocation and sizing of capacitors to
minimize the distribution line loss and to improve the voltage profile using big
bang-big crunch optimization. Int Rev Electr Eng (IREE) 2011;6(4):2013–9.
[19] El Arini M. Optimal capacitor placement incorporating voltage stability and
system security. Eur Trans Electr Power (ETEP) 2000;10(5):319–25.
[20] Satpathy PK, Das D, Gupta PB. Critical switching of capacitors to prevent
voltage collapse. Electr Power Syst Res 2004;41(1):11–20.
[21] Jasmon B, Lee LHCC. Maximising voltage stability in distribution networks via
loss minimization. Int J Electr Power Energy Syst 1991;13(3):148–52.
243
[22] Ault Graham W, McDonald James R. Planning for distributed generation within
distribution networks in restructured electricity markets. IEEE Power Eng Rev
2000;20(2):52–4.
[23] Teng JH. A direct approach for distribution system load flow solutions. IEEE
Trans Power Deliv 2003;18(3):882–7.
[24] Gözel T, Eminoglu U, Hocaoglu MH. A tool for voltage stability and
optimization in radial distribution systems using MATLAB GUI. Simul Model
Pract Theory 2008;16(5):505–18.
[25] Abdellatif Hamouda, Khaled Zehar. Improved algorithm for radial distribution
networks load flow solution. Int J Electr Power Energy Syst 2011;33(3):
508–14.
[26] Iwamoto S, Tamura Y. A load flow calculation method for ill-conditioned
power systems. IEEE Trans Power Apparat Syst 1981;100(4):1736–43.
[27] Ghosh S, Das D. Method for load flow solution of radial distribution network.
IEE Proc Gener Transm Distrib 1999;146(6):641–8.
[28] Singh Sachin, Ghose T. Improved radial load flow method. Int J Electr Power
Energy Syst 2013;44(1):721–7.
[29] Augugliaro A, Dusonchet L, Favuzza S, Ippolito MG, Riva Sanseverino E. A
backward sweep method for power flow solution in distribution networks. Int
J Electr Power Energy Syst 2010;32(4):271–80.
[30] Moghavemmi M, Omar FM. Technique for contingency monitoring and voltage
collapse prediction. IEE Proc Gener Transm Distrib 1998;145:634–40.
[31] Musirin I, Rahman TKA. Estimating maximum loadability for weak bus
identification using FVSI. IEEE Power Eng Rev 2002;22:50–2.
[32] Anhit S, Ndarajah M, Kwang S. A maximum loading margin method for static
voltage stability in power systems. IEEE Trans Power Syst 2006;21(2):965–72.
[33] Charkravorty M, Das D. Voltage stability analysis of radial distribution
networks. Int J Electr Power Energy Syst 2001;23(2):129–35.
[34] Karaboga D. An idea based on honey bee swarm for numerical optimization.
Technical report TR06, Computer Engineering Department, Erciyes University,
Turkey; 2005.
[35] Karaboga D, Basturk B. On the performance of artificial bee colony algorithm.
Appl Soft Comput 2008;8:687–97.
[36] Karaboga D, Bakay B. A comparative study of artificial bee colony algorithm. J
Appl Math Comput 2009;214:108–32.
[37] http://www.mathworks.com.
[38] http://mf.erciyes.edu.tr/abc/.
[39] Ali E, Boudour M, Rabah G. New evolutionary technique for optimization shunt
capacitors in distribution networks. J Electr Eng 2011;62(3):163–7.
[40] Pires DF, Antunes CH, António GM. An NSGA-II approach with local search for a
VAR planning multi-objective problem. Inst Syst Eng Comput 2009;8. ISSN:
1645-2631.
[41] El-Fergany A. Optimal capacitor allocations using integrated evolutionary
algorithms. Int Rev Model Simul 2012;6:2590–9.