Electric Power Systems Research 106 (2014) 232–240

Contents lists available at ScienceDirect

Electric Power Systems Research

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

A fuzzy-based approach for optimal allocation and sizing of capacitor

banks
Husam A. Ramadan ∗ , Mohamed A.A. Wahab, Abou-Hashema M. El-Sayed,
Mohamed M. Hamada
Electrical Engineering Department, Faculty of Engineering, Minia University, Minia 61111, Egypt

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

Article history: This paper proposes a fuzzy set optimization approach for capacitor allocation in radial distribution
Received 26 April 2009 system. In this approach, a membership function for voltage profile constraint has been used. Moreover,
Received in revised form 1 June 2013 another membership function incorporating feeder section active power losses and total power losses
Accepted 22 August 2013
constraints has been proposed. This membership function indirectly imposes thermal capability of the
feeder on the optimization process. The proposed approach has been applied to 9-bus and 34-bus radial
Keywords:
distribution systems. The results have been compared with those of two fuzzy approaches in literature.
Distribution systems
The comparison showed the effectiveness of the proposed approach for optimizing the sizes and locations
Capacitor allocation
Fuzzy sets theory
of the capacitor with running and total cost reduction.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction Published literature describing capacitor allocation algorithms are

Electric power system networks are typically composed of four
major parts: generation, transmission, distribution and loads. Dis-
tribution networks are crucial elements in electric power systems
since they link the generated power to the end user. Transmis-
sion and distribution networks share similar functionality; i.e.
both transfer electric energy at different levels from one point to
another; however, their network topologies and characteristics are
quite different.
Distribution networks are well-known for their high R/X ratio
and significant voltage drop that could cause substantial power
losses along the feeders. It is estimated that as much as 13% of the
total power generation is lost in the distribution networks [1]. Por-
tion of this loss is caused by the reactive current flowing in the
network. Voltage profiles throughout the network have to be kept
at acceptable levels to ensure service reliability among other issues.
Capacitor banks are commonly used in various parts of the electric
grid to maintain voltage levels within appropriate limits, mini-
mize the power losses and release the line capacity. With regard
to the power losses in the feeders, capacitor installations have
demonstrated their effectiveness in reducing the overall current by
canceling part of the reactive current supplied by the substation.
Capacitor installations have proven their economical impact by
significantly reducing the power losses and releasing line capacity.
∗ Corresponding author.
E-mail address: eng hosam ramadan@yahoo.com (H.A. Ramadan).
abundant. Grainger et al. pioneered the analytical methods [2].
In Ref. [2], fixed and switched capacitors are placed for opti-
mizing the net monetary savings associated with the reduction of
power and energy losses. Both the capacitor locations and sizes
are treated as continuous variables. A new voltage dependent
methodology for shunt capacitor compensation of primary distri-
bution feeders is presented in Ref. [3]. Ponnavaikko et al. [4] used
a numerical method called the method of local variations and fur-
ther expanded the problem to include the effects of load growth,
and switched capacitors for varying load.
Similarly, Baran et al. [5] formulated the capacitor placement
problem using mixed integer programming. The optimal selec-
tion and placement of capacitor banks using binary particle swarm
optimization (PSO) is integrated with the estimation of harmonic
levels in Ref. [6]. Methods based on heuristic search techniques are
introduced for distribution system loss reduction by reconfigura-
tion [7,8]. Abdel-Salam et al. [9] proposed a heuristic technique
based on the ideas from [7,8] to identify a section in the distribu-
tion system, with the highest losses due to reactive load currents,
then pinpoint the sensitive node in that section having the greatest
effect on the system loss reduction. Sizes of capacitors placed on
the sensitive nodes are determined by maximizing the power loss
reduction from capacitor compensation. Chis et al. [10] improved
the work of [9] by determining the sensitive nodes that have the
greatest impact on loss reduction for the entire distribution system
directly, by optimizing the capacitor sizes based on maximizing
the net economic savings from both energy and peak power loss
reductions.

0378-7796/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.epsr.2013.08.019
 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240 233

s 1 Ploss(1,2) 2 i i 1 Ploss(i 1,i 2) i 2 n 1 Ploss(n -1,n) n

Z1, 2 Zi+1, i+2
Zn-1, n

Q loss(1,2) Q loss(i 1,i 2) Q loss(n -1,n)

P1 , Q1 P2 , Q2 Pi , Q i Pi 1, Qi 1 Pi 2 , Q i 2
Pn -1 , Q n -1 Pn , Q n
Fig. 1. A single line diagram of a radial distribution (i) feeder.

In addition, the method in [10] also accounts for varying loads of 2. Problem formulation
the distribution system considered. Hamada et al. [11] introduced a
new strategy for capacitor allocation handling the reduction in the The purpose of placing compensating capacitors along the dis-
section losses by adding a new constraint to the well-known con- tribution feeders is to lower the total power loss and bring the bus
straint (allowed voltage violation constraint). The new constraint voltages within specified limits while minimizing the total power
is the sectional ohmic losses in each branch of the feeder. Ref. [12] cost. The total power loss PTloss is given by
presents a fuzzy-based approach for capacitor placement for the 9-
bus feeder. Two membership functions for total real power losses 
n−1
PTloss = Ploss(i,i+1) i = 1, 2, . . ., n (1)
and voltage sensitivity have been defined to reduce the effort of
i=1
finding the optimal locations. The whole problem has been pre-
sented as a fuzzy-set optimization problem to minimize the total where, i is the bus number and n is the total number of buses
real losses and capacitor cost with voltage limit constraints. They as shown in Fig. 1. Considering investment cost, there is a finite
used the intersection principle in fuzzy as the fuzzy decision to find number of standard capacitor sizes that are integer multiples of
the capacitor location, then a variational method has been used to Q0c . The cost per kVAr varies from one size to another. Generally,
find capacitor sizes to attain minimum cost without violating the large capacitor sizes are cheaper than smaller ones. The available
voltage constraints. capacitor size is usually limited to [13]:
In Ref. [13], exactly the same procedures using the same feeder c
have been implemented, but with two different membership func- Qmax = L × Q0c (2)
tions. In fact, their membership functions for real power losses and where Q0c is the smallest capacitor size in Table A1 and L is an
voltage are the fundamental part of the membership functions that

integer. Therefore 
for each installation  are L capaci-
location, there 
have been used in [12]. However, they have relatively achieved tor sizes Q0c , 2Q0c , . . ., LQ0 c to choose from. Let K1c , K2c , . . ., KLc
better results by introducing a certain constant in the real losses be their corresponding equivalent annual cost per kVAr. The objec-
membership function depending on their experiences. In [14], the tive (Cost) function can be expressed as
authors used membership functions forms of [12], but replaced the
real losses by reactive losses and the intersection decision (using 
J

min operator) by product decision. They used the product fuzzy
decision to determine the location of the capacitors. To find the
capacitor sizes, they used their analytical method that has been
explained in [15]. This analytical method is based on differentiat-
ing a well-defined net saving function of power and energy losses
with respect to capacitor size, thus obtaining the optimum capac-
itor size. Their method has been applied on the 34-bus feeder
[15].
Ref. [16] presented fast power loss computation using super-
visory control and data acquisition system (SCADA), which a
fuzzy-based decision maker used to compute the suitable shunt
capacitor required to improve the power factor according to the
measured parameters. Most of these studies consider the loss
reduction for the capacitor allocation problem, but these studies
take in consideration the reduction of total losses not the reduction
in the individual sectional losses. The general capacitor placement
problem in distribution feeders consists of determining the optimal
location, type (fixed or switched), and size of capacitors; such that
power and energy losses are minimized while taking the cost of the
capacitor into account. For simplifying the problem fixed capacitors
are only considered in this paper.
This paper presents a fuzzy-based approach for capacitor alloca-
tion in radial distribution systems. Two membership functions are
defined in this paper, one for the voltage sensitivity and another
proposed one for the real sectional ohmic loss constraint. This con-
straint has been initially introduced by the authors in a previous
published paper, Ref. [11]. The problem is formulated as a fuzzy-set
optimization problem to minimize the real power loss and capaci-
tor cost with sectional loss and voltage limit constraints. Moreover,
this paper presents a comparison between the proposed approach
and previous works to show the validity of the proposed approach.
Cost = Cp × PTloss + Kjc Qjc (3)
j=1
where Cp is the cost per power loss ($/kW) [13] and j = 1,2,...,j repre-
sents the selected buses for compensation. The objective function
Eq. (3) is to be minimized subjected to two constraints:
The first constraint is the bus voltage constraint.
Vmin ≤ Vi ≤ Vmax , i = 1, 2, . . ., n (4)
The second constraint is the sectional ohmic loss constraint (pro-
posed in the author’s previously published paper).
(K+1) (K)
PSec,loss ≤ PSec.loss (5)
where (K + 1) is the case after the capacitor placement and (K) is the
case before the capacitor placement.
3. Application of fuzzy set theory
In the conventional methods, the large number of combinations
in the solution space makes the solution searching process time
consuming. The problem is formulated as a fuzzy reasoning opti-
mization model to minimize the cost and power loss subjected to
voltage limit and sectional losses constraints. Minimum operation
of fuzzy sets is employed to find the optimal locations and sizes of
capacitors.
3.1. The proposed method
Two membership functions are defined in this method, one for
the voltage sensitivity and another for the active sectional ohmic
loss. This method suggests a new membership function for the
234 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240

P 3.1.3. Decision making

1.0 By determining values of p and V , a membership function of
the placement decision S is determined, which is the intersection
0.8
of fuzzy sets of P and V . The candidate bus is the one that has
0.6 lowest value of S . Eq. (8) defines the membership function S .
 
0.4 S (i) = min V (i), P (i) (8)

0.2 The optimal capacitor size, to be installed at the candidate bus,

is automatically determined during choosing the candidate buses;
0.0 since the scenario herein is to put a capacitor at the candidate bus.
0.04 0.02 0 0.02 0.04 Ploss(i) Then, compute P , V and S at this case. If S decision leads to
the same bus as the candidate bus; then continue increasing the
Fig. 2. The proposed membership function P for sectional ohmic loss. capacitor size, at this bus, by integer steps until S decision leads
to another candidate bus. Consequently, the problem of choosing
the optimal allocation and size by the proposed method; became a
V
matter of one fuzzy optimization problem. Therefore, the proposed
1.0 method can solve the optimal capacitor allocation and sizing with-
out the need of other method as given in [12,13]. Therefore, the
0.8
proposed method will greatly reduce the time and effort spent in
0.6 searching for the optimal locations and sizes.

0.4 3.1.4. The steps of the proposed method

(1) Perform the load flow program to calculate bus voltages and
0.2 power losses in each section (original case).
(2) Find the lowest voltage bus, and then put a capacitor at this
0.0 bus, its value equal to the smallest one in Table A2.
0.0 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 VP .U . (3) Perform the load flow program and calculate bus voltages and
power losses in each section (case K).
Fig. 3. Membership function v for voltage sensitivity. (4) From results of steps (1 and 3) calculate Ploss , hence calculate
the membership function P
(5) From results of step (3) calculate the membership function V .
sectional ohmic loss. This membership function guarantees the (6) Calculate the membership function S , and then identify the
achievement of sectional ohmic loss constraint that is defined in candidate location (optimal bus) as the bus with lowest mem-
Ref. [11]. It is worthily mentioning that the capacitor allocation bership function S .
problem in Ref. [11] was treated with the conventional optimiza- (7) Install a capacitor at this bus, beginning form the smallest one
tion methods. The membership functions are defined as follows: in Table A2.
(8) Perform the load flow program and calculate bus voltages and
3.1.1. Membership function for the real sectional loss power losses in each section (case (K + 1)).
According to the constraint in Eq. (5), a sectional ohmic loss in (9) Calculate the membership functions P and V as in steps (4
any branch of the feeder is not allowed to increase during capacitor and 5).
placement scenario. To guarantee the achievement of this con- (10) Repeat steps (4, 5, 6, and 8) until satisfying the voltage con-
straint a new membership function p is proposed as shown in straint, and the cost function is still decreasing.
Fig. 2.
  P 2 If Ploss values have any positive value, this means that the
loss (i) losses in certain section is increased than it was before putting this
p (i) = exp −w , i = 1, 2, . . ., n − 1 (6)
PTloss capacitor. Therefore, remove the capacitor causing this violation
K+1 K (i, i + 1), and try with the bus, which following it in S value. If the decrease
where Ploss (i) = Ploss (i, i + 1) − Ploss PTloss =
in the cost function is steady, the optimal solution is reached.
K+1 K
PTloss − PTloss ,
(K + 1) is the case after the capacitor placement, The following two sections will shortly demonstrate methods in
K is the case before the capacitor placement, W is a weighting [12,13] that are used for the sake of comparison with the proposed
factor whose value is determined from experiences. method.
In the definition of this membership function, the bus with
low membership value is the bus at which the loss is significantly 3.2. Method (1) [12]
decreased. A bus of small loss decreasing is given a high member-
ship function value. In this method, membership functions are defined for the volt-
age sensitivity and real power loss. The two membership functions
3.1.2. Membership function for the bus voltage are defined to find the optimal location of shunt capacitors, then
For the voltage sensitivity, the same membership function for the local variational method is used to find the capacitor size. The
voltage given in Ref. [13] is chosen. This membership function is objective function in Eq. (3) is used to minimize the real power
illustrated as given in Fig. 3, and is defined as follows: loss and capacitor cost with voltage limits constraints Eq. (4). This
method has been applied to 9-bus feeder in [12]. The membership
1
V (i) =
2 , i = 1, 2, . . ., n (7) functions are defined in the following:
1 + (V (i) − 1)/(Vmax − Vmin )
3.2.1. Membership function for the power loss
where V(i) is the bus i voltage, Vmax is the high voltage limit, Vmin To minimize the real power loss, a membership function given
is the low voltage limit. in Fig. 4 is defined. A high loss branch gives a low value (P ); on the
 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240 235

P P

1 .0 1 .0

0 .8

0 .6

0 .4

0 .2 0 .0
LP Ploss ( i , i 1)
0 .0
0 .0 0 .1 0 .3 0 .4 0 .5 0 .6 Ploss (i, i 1) / PTloss Fig. 6. Membership function P for real power loss for method (2) [13].

Fig. 4. Membership function P for real power loss for method (1) [12].
a low loss branch gives a high value. The membership function
defined for the power loss is given as:
V
 Ploss (i, i + 1)
1.0 1− , i = 1, 2, . . ., n ; Ploss (i, i + 1) ≤ LP
P (i) = LP (11)
0.8 0, i = 1, 2, . . ., n ; Ploss (i, i + 1) > LP

0.6 where Lp is a constant whose value is determined from experiences.

0.4
3.3.2. Membership function for the bus voltage
0.2 For voltage magnitude V, a membership function V is defined as
shown in Fig. 3. A bus with high voltage deviation gives a low value
0.0
(V ). A bus with low deviation gives a high value. The membership
0.0 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 V(i) function is defined in Eq. (7). With P and v defined, a membership
function of the placement decision S is determined. An optimal
Fig. 5. Membership function v for voltage sensitivity for method (1) [12]. bus is the one that has lowest S . Eq. (8) defines the membership
function S . The optimal size of shunt capacitor selected for the
other hand, a low loss branch gives a high value. The membership installation is one that gives the lowest cost while satisfying the
function chosen for the real power loss is written as: voltage constraint. For comparison, this method is also applied to
 −wp  the two feeders mentioned before.
loss (i, i + 1)
P (i) = exp , i = 1, 2, . . ., n (9)
PTloss
4. Numerical examples and computational results
where Ploss (i,i + 1) is the real power loss between buses i and i + 1,
Ploss is the total real power loss, W is the weighting factor whose In order to test the proposed method and validate its results, the
value is determined from experiences. distribution systems with 9 and 34-buses respectively have been
considered as examples. The results of the proposed method are
3.2.2. Membership function for the voltage sensitivity compared with the second and third method. Constants and con-
For voltage regulation, membership function given in Fig. 5 is ditions have been considered for the three methods, are as follows
defined. A bus with high voltage deviation gives a low value (V ). A [13]:
bus with low deviation gives a high value. The membership function
chosen as follows: • Power losses cost Cp = 168 U.S.$/kW
  V (i) − 1 2 • Voltage limits:
V (i) = exp −w , i = 1, 2, . . ., n (10)
Vmax − Vmin
Vmin = 0.9 p.u.
where is the bus i voltage, Vmax
Vmax = 1.1 p.u.
=
=the high voltage limit,
Vmin =the low voltage limit, • Fixed capacitors are only considered.
W =weighting factor whose value is determined from experiences. • The possible capacitor sizes have been shown in Table A2 with
corresponding cost/kVAr.
With P and V defined; a membership function of the place-
ment decision S is determined, which is the intersection of fuzzy 4.1. The first system
sets of power loss and the voltage sensitivity. An optimal decision
is the one that has lowest S . Eq. (8) defines the membership func- The 9-bus radial distribution feeder in [3] is considered as the
tion S . Implementation of this method for the two well-known test feeder. The rated voltage is 23 kV. The single line diagram, the
test feeders, 9-bus and 34-bus distribution feeders. load data and feeder-line parameters for the system are shown in
Tables B1 and B2. By applying the load flow program on this feeder
3.3. Method (2) [13] before compensation, the value cost and the total power losses
are U.S.$ 131,675 and 783.8 kW, respectively. The maximum and
3.3.1. Membership function for the power loss the minimum bus voltage magnitudes were 0.9929 and 0.8375 p.u
For power loss Ploss , a membership function P is defined as respectively, where the voltage of the substation (bus number 0) is
shown in Fig. 6 A high loss branch gives a low value (P ). Whereas, assumed to be 1 p.u.
236 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240

1.15
Vmax
1.1
1.05
1
0.95
Vmin
Voltage in p.u.

0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
1 2 3 4 5 6 7 8 9
Bus No
Without Qc Method1
proposed method Method 2

Fig. 7. The voltage profile for the 9 bus feeder.

4.2. The second system

The 34-bus radial distribution test system has been used. This
test system has a main feeder and four laterals (sub feeders). The
data of the feeder is presented in [13]. The single line diagram, the
load data and feeder line parameters for the system are shown in
Tables B1 and B2. The system voltage is 11 kV. Before compen-
sation, the cost is U.S.$ 327,212, this is based on the previously
defined cost function, the active and reactive losses are 221.5 kW
and 65.04 kVAr, respectively. The maximum and the minimum bus
voltage magnitudes were 1 and 0.9417 p.u respectively.
5. Discussion on results
Fig. 7 shows the voltage profile for the proposed method, meth-
ods (1) and (2) for the 9-bus system. The results show that without
capacitors the violation of minimum voltage limit starts from bus
no. 6 to bus no. 9. All capacitor allocation methods achieved volt-
age profile within the specified maximum and minimum voltage
limits. The voltage profile for the proposed method is higher than
that of the other two methods. However, the deviations between
the voltage profiles for the three methods are not large.
Fig. 8. The losses in each section for the 9 bus feeder.
Fig. 8 shows the active power losses in each section defined by
bus number for the 9-bus system for the three methods. From this
figure, it can be seen that the proposed method achieved reduc-
tion in all active power sectional losses when compared with those
obtained from the system without reactive compensation. Compar-
ing the active power sectional losses in the proposed method with
those obtained from methods (1) and (2) it can be seen that the
active power sectional losses increased in some section as bus no.
3 while these losses are decreased for most buses.
The total active power losses in the proposed method have been
decreased more than those obtained by the other two methods as
seen in Table 2. The imposed constraint on the sectional active
power losses serves in two issues: The first issue is reducing the
total active losses, and the second issue improving the thermal
capability of system feeder by prohibiting the increase in active
section losses than those of the uncompensated case.
The voltage profile for the 34-bus system is shown in Fig. 9.
From this figure it can be seen that the voltage profiles for the

1.15

Vmax
1.1
Voltage in p.u

1.05

0.95

Vmin
0.9

0.85
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
Bus No

Without Qc Proposed method Method 1 Method 2

Fig. 9. The voltage profile for the 34 bus feeder.

 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240 237

Table 1 Table 4
Capacitor values and locations for the 9-bus test system. Capacitor values and locations for the 34-bus test system.

Method no. The proposed method Method (1) [12] Method (2) [13] Method no. The proposed method Method (1) [12] Method (2) [13]
Bus no. Qc in MVAr Qc in MVAr Qc in MVAr
Bus no Qc in MVAr Qc in MVAr Qc in MVAr
1 0 0 0
1 0 0 0
2 0 0 0
2 0 0 0
3 0 0 0
3 3.15 0 1.05
4 0.6 2.1 0
4 1.5 2.1 1.05
5 0 1.05 0
5 0.9 2.5 1.95
6 1.35 0.3 0
6 0 0 0
7 0.15 0 0
7 1.5 0 0
8 0.15 0 0
8 0 0 0
9 0 0 0
9 0.45 0.9 0.9
10 0.15 0 0
Summation of MVAr 7.5 5.5 4.95
11 0 0 0
12 0 0 0
13 0 0 0
14 0 0 0
proposed method and for method (1) are close. The voltage pro- 15 0 0 0
file for method (2) at some buses is lower than that obtained by the 16 0 0 0
17 0 0.45 0
proposed method and method (1). All the methods including the 18 0 0.15 0
uncompensated case produced voltage profiles within the specified 19 0 0.15 0
maximum and minimum voltage limits. 20 0 0.15 0.15
Therefore, it can be concluded that the effect of the proposed 21 0.3 0.15 0.15
22 0 0.15 0.15
active power losses membership function and constraints may
23 0.15 0 0
impose a noticeable effect on the capacitor locations and MVArs 24 0 0.15 0.15
choices in such systems. The sectional active power losses for the 25 0 0 0.15
34-bus system for all methods are shown in Fig. 10. From this figure 26 0 0 0
it can be seen that, method (2) increased the active power losses in 27 0.45 0 1.05
28 0.15 0 0
sections no. (10–11), (22–23), (23–24), (24–25), (25–26), (31–33) 29 0 0 0
and (32–33). However, this increase in active power losses has not 30 0 0 0
been obtained in the proposed method, which validates the proper 31 0 0 0
behavior of the proposed active power loss membership function 32 0 0 0
33 0 0 0
(Figs. B1 and B2).
34 0 0 0.45
A summary of the results obtained for the 9-bus system is given Summation of MVAr 3.45 4.8 2.25
in Tables 1–3. Table 1 shows the locations and MVAr for the three
methods and the total MVArs required in each case. The proposed
method requires larger MVAr values than the other two methods.
The locations for capacitors are not the same for the three methods. method. In the proposed method, the fixed cost is higher than the
Table 2 explores the active losses in each section for each method. other two methods.
Table 3 shows that the proposed method achieved lower total active For the 34-bus system, the summary of results is given in
power losses and higher reduction than the other two methods. Tables 4–6. Table 4 shows the locations and total MVArs for the
This has been reflected on the lowest running cost of the proposed system. It is shown that the total MVArs of the proposed method

Table 2
The losses in each section for the 9-bus feeder in MW.

Feeder section no. Method no.

Without Qc The proposed method Method (1) [12] Method (2) [13]

1 0.047 0.04 0.04 0.04

2 0.004 0.003 0.003 0.003
3 0.177 0.153 0.151 0.149
4 0.114 0.092 0.099 0.093
5 0.19 0.168 0.185 0.187
6 0.048 0.043 0.042 0.043
7 0.076 0.068 0.068 0.069
8 0.089 0.078 0.08 0.081
9 0.039 0.035 0.04 0.04

Table 3
Summary of results for the 9-bus test system.

Method no. Without Qc The proposed method Method (1) [12] Method (2) [13]

Max voltage 1 1 1 1
Min voltage 0.8375 0.9 0.9 0.9
Total power loss in MW 0.783 0.680 0.707 0.705
Running cost $ 131,675 115,250 118,776 118,550
Fixed cost $ 0 1195.3 1006.8 1055.95
Total cost $ 131,675 115,535.3 119,782.8 119,595
Loss reduction in kW 0 103.8 76.8 78.8
Cost saving in $ 0 16250.7 11892.2 12,180.05
238 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240

Fig. 10. The losses in each section for 34 bus feeder.

are higher than that obtained by method (2). Method (2) and the losses in each section for each method. Table 6 shows that the
proposed method have the same voltage membership function. proposed method offers less running cost than the other two meth-
The effect of unincreased sectional active power losses appeared in ods. The fixed cost for the proposed method is higher than that
large MVArs for the proposed method. Table 5 explores the active of method (2) and is lower than that of method (1). The proposed

Table 5
The losses in each section for the 34-bus feeder in MW.

Feeder section no. Method no.

Without Qc The proposed method Method (1) [12] Method (2) [13]

1 0.032 0.022 0.026 0.023

2 0.026 0.019 0.022 0.019
3 0.036 0.026 0.032 0.026
4 0.000 0.000 0.000 0.000
5 0.029 0.021 0.021 0.021
6 0.026 0.019 0.019 0.019
7 0.006 0.005 0.006 0.005
8 0.013 0.010 0.010 0.010
9 0.003 0.002 0.003 0.002
10 0.000 0.000 0.000 0.000
11 0.003 0.002 0.003 0.002
12 0.001 0.001 0.001 0.001
13 0.000 0.000 0.000 0.000
14 0.000 0.000 0.000 0.000
15 0.000 0.000 0.000 0.000
16 0.000 0.000 0.000 0.000
17 0.000 0.000 0.000 0.000
18 0.000 0.000 0.000 0.000
19 0.010 0.008 0.008 0.008
20 0.010 0.008 0.008 0.008
21 0.007 0.005 0.006 0.006
22 0.006 0.004 0.004 0.005
23 0.006 0.004 0.005 0.005
24 0.004 0.003 0.003 0.004
25 0.003 0.002 0.002 0.004
26 0.001 0.001 0.001 0.002
27 0.000 0.000 0.000 0.001
28 0.000 0.000 0.000 0.001
29 0.000 0.000 0.000 0.000
30 0.000 0.000 0.000 0.000
31 0.000 0.000 0.000 0.000
32 0.000 0.000 0.000 0.000
33 0.000 0.000 0.000 0.000
34 0.032 0.022 0.026 0.023

Table 6
Summary of results for the 34-bus test system.

Method no. Without Qc The proposed method Method (1) [12] Method (2) [13]

Max voltage 1 1 1 1
Min voltage 0.9416 0.951 0.951 0.952
Total power loss in MW 0.2215 0.162 0.179 0.171
Running cost $ 37,212 27,216 30,072 28728
Fixed cost $ 0 1005.3 1277.85 795.45
Total cost 37,212 28,221.3 31,349.85 29,523.45
Loss reduction in kW 0 59.128 42.38 50
Cost saving in $ 0 8990.7 5862.15 7688.55
 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240 239

method achieved the higher total cost reduction than the other two Appendix B.
methods.
See Tables B1 and B2 and Figs. B1 and B2.
6. Conclusion
Table B1
A fuzzy-based approach for optimal capacitor allocation and The load data and the feeder data of the 9-bus test system [3]. Base impedance
sizing has been presented. A new membership function for active = 529 .
sectional and total active power losses has been proposed and con-
Bus no. Load Sectional parameters
stant active sectional power loss concept has been implemented.
A voltage membership function has been used. The application of P Q Bus no. Ri , i + 1 () Xi , i + 1 ()

the proposed approach on 9-bus and 34-bus systems led to the (kW) (kVAr) From To
following conclusions: 1 1840 460 0 1 0.123 0.4127
2 980 340 1 2 0.014 0.6051
(1) The proposed membership function proved its proper behavior. 3 1790 446 2 3 0.746 1.205
(2) The increase in fixed cost resulted from the proposed method is 4 1598 1840 3 4 0.698 0.6084
5 1610 600 4 5 1.983 1.7276
compensated by the decrease in the running cost and generally 6 780 110 5 6 0.905 0.7886
lower total cost has been obtained. 7 1150 60 6 7 2.055 1.164
(3) The proposed approach is characterized by simple capacitor 8 980 130 7 8 4.795 2.716
allocation and sizing technique. 9 1640 200 8 9 5.343 3.0264

7. List of symbols

Cp =is the cost per power loss ($/kW),

(K + 1) =is the case after the capacitor placement,
K =is the case before the capacitor placement, Table B2
W =a weighting factor whose value is determined from experiences, The load data and the feeder data of the 34-bus test system [13]. Base
Ploss (i,i + 1) =the real power loss between buses i and i+1, impedance = 121 .
PTloss =total real power loss,
Bus no. Load Sectional parameters Length
V(i) =the bus i voltage,
(km)
Vmax =the high voltage limit,
Vmin =the low voltage limit, P (kW) Q (kVAr) Bus no. Ri , i + 1 Xi , i + 1
Lp =a constant whose value is determined from experience,
v =a membership function for voltage magnitude, From To (/km) (/km)
P =a membership function for the real power loss, 1 320 142.5 0 1 0.195 0.08 0.6
S =a membership function for the intersection between P and v , 2 0 0 1 2 0.195 0.08 0.55
I =is the bus number, 3 230 142.5 2 3 0.299 0.083 0.55
N =the total number of buses of the distribution feeder. 4 230 142.5 3 4 0.299 0.083 0.5
5 0 0 4 5 0.299 0.083 0.5
6 0 0 5 6 0.524 0.09 0.6
Appendix A.
7 230 142.5 6 7 0.524 0.09 0.4
8 320 142.5 7 8 0.524 0.09 0.6
See Tables A1 and A2. 9 0 0 8 9 0.524 0.09 0.4
10 230 142.5 9 10 0.524 0.09 0.25
Table A1 11 137 84 10 11 0.524 0.09 0.2
Available three-phase capacitor size and cost [17]. 2 1 72 45 2 2 1 0.524 0.09 0.3
2 2 72 45 2 1 2 2 0.524 0.09 0.4
Size (kVAr) 150 300 450 600 900 1200 2 3 72 45 2 2 2 3 0.524 0.09 0.2
2 4 13.5 7.5 2 3 2 4 0.524 0.09 0.1
Cost ($) 750 975 1140 1320 1650 2040
5 1 230 142.5 5 5 1 0.299 0.083 0.6
5 2 230 142.5 5 1 5 2 0.299 0.083 0.55
5 3 230 142.5 5 2 5 3 0.378 0.086 0.55
5 4 230 142.5 5 3 5 4 0.378 0.086 0.5
5 5 230 142.5 5 4 5 5 0.378 0.086 0.5
Table A2
5 6 230 142.5 5 5 5 6 0.524 0.09 0.5
Possible choice of capacitor sizes and cost/kVAr [17].
5 7 230 142.5 5 6 5 7 0.524 0.09 0.5
J 5 8 230 142.5 5 7 5 8 0.524 0.09 0.6
5 9 230 142.5 5 8 5 9 0.524 0.09 0.4
1 2 3 4 5 6 5 10 230 142.5 5 9 5 10 0.524 0.09 0.25
5 11 137 85 5 10 5 11 0.524 0.09 0.2
Qc (kVAr) 150 300 450 600 750 900
6 1 75 48 6 6 1 0.524 0.09 0.3
$/kVAr 0.5 0.35 0.253 0.22 0.276 0.183
6 2 75 48 6 1 6 2 0.524 0.09 0.3
7 8 9 10 11 12 6 3 75 48 6 2 6 3 0.524 0.09 0.3
Qc (kVAr) 1050 1200 1350 1500 1650 1800 9 1 57 34.5 9 9 1 0.524 0.09 0.3
$/kVAr 0.228 0.17 0.207 0.201 0.193 0.187 9 2 57 34.5 9 1 9 2 0.524 0.09 0.4
9 3 57 34.5 9 2 9 3 0.524 0.09 0.3
13 14 15 16 17 18 9 4 57 34.5 9 3 9 4 0.524 0.09 0.2
Qc (kVAr) 1950 2100 2250 2400 2550 2700
$/kVAr 0.211 0.176 0.197 0.17 0.189 0.187
19 20 21 22 23 24
Qc (kVAr) 2850 3000 3150 3300 3450 3600 S 1 2 3 4 5 6 7 8 9
$/kVAr 0.183 0.18 0.195 0.174 0.188 0.17
25 26 27
Qc (kVAr) 3750 3900 4050
$/kVAr 0.183 0.182 0.179
Fig. B1. The single line diagram of the 9 bus test system [3].
240 H.A. Ramadan et al. / Electric Power Systems Research 106 (2014) 232–240

2 2 2 3 2 4 6 2 6 3

2 1 6 1
S 6 10

1 2 3 4 5 7 8 9 11
5 1 9 1

5 2
9 29 3 9 4
5 3

5 4 5 5 5 6 5 7 5 8 5 9 5 10 5 11
Fig. B2. The single line diagram of the 34 bus test system [13].

References
[1] H.N. Ng, M.M.A. Salama, A.Y. Chikhani, Classification of capacitor allo-
cation techniques, IEEE Trans. Power Delivery 15 (1) (2000) 387–392,
Jan.
[2] S.H. Lee, J.J. Grainger, Optimum placement of fixed and switched capacitors
on primary distribution feeders, IEEE Trans. Power Apparatus and Systems
PAS–100 (1) (1981) 345–352, Jan.
[3] J.J. Grainger, S.H. Lee, Capacity release by shunt capacitor placement on distri-
bution feeders: a new voltage-dependent model, IEEE Trans. Power Apparatus
and Systems PAS-101 (5) (1982) 1236–1244, May.
[4] M. Ponnavaikko, K.S. Prakasa Rao, Optimal choice of fixed and switched shunt
capacitors on radial distributors by the method of local variations, IEEE Trans.
Power Apparatus Syst. PAS-102 (6) (1983) 1607–1615, June.
[5] M.E. Baran, F.F. Wu, Optimal capacitor placement on radial distribution sys-
tems, IEEE Trans. Power Delivery 4 (1) (1989) 725–734.
[6] T.M. Khalil, H.K.M. Yousssef, M.M. Abdel Aziz, Optimal capacitor placement on
radial distribution feeders in presence of nonlinear loads using binary particle
swarm optimization, in: 19th International Conference on Electricity Distribu-
tion, Vienna, 2007, pp. 119–122.
[7] S. Civanlar, J.J. Grainger, H. Yin, S.H. Lee, Distribution feeder reconfigura-
tion for loss reduction, IEEE Trans. Power Delivery 3 (3) (1988) 1217–1223,
July.
[8] T. Taylor, D. Lubkeman, Implementation of heuristic search strategies for distri-
bution feeder reconfiguration, IEEE Trans. Power Delivery 5 (1) (1990) 239–246,
Jan.
[9] T.S. Abdel-Salam, A.Y. Chikhani, R. Hackam, A new technique for loss reduction
using compensating capacitors applied to distribution systems with varying
load condition, IEEE Trans. Power Delivery 9 (2) (1994) 819–827.
[10] M. Chis, M.M.A. Salama, S. Jayaram, Capacitor placement in distribution systems
using heuristic search strategies, in: Proc Inst. Elect. Eng. Gen. Transm. Dist., vol.
144, No. 2, 1997, pp. 225–230.
[11] M.M. Hamada, M.A.A. Wahab, A.M. El-Sayed, H.A. Ramadan, A proposed strat-
egy for capacitor allocation in radial distribution feeders, in: Proc. 12th Middle
East Power Systems (MEPCON) Conf, 2008, pp. 146–151.
[12] H. Chin, W. Lin, Capacitor placement for distribution systems with fuzzy algo-
rithm, in: Proc. IEEE Region 10‘s nineth Annu. Int. Conf.,Theme, Frontiers of
Computer Technology, vol. 2, 1994, pp. 1025–1029.
[13] C. Su, C. Tsai, A new fuzzy-reasoning approach to optimum capacitor allocation
for primary distribution system, in: Proc IEEE Int. conf. Industrial Technology,
1996, pp. 237–241.
[14] H.N. Ng, M.M.A. Salama, A.Y. Chikhani, Capacitor placement in distribution
systems using fuzzy technique, in: Proc Can. Conf. Electrical and Computer
Engineering, vol. 2, 1996, pp. 790–793.
[15] M.M.A. Salama, A.Y. Chikhani, A simplified network approach to VAr control
problem for radial distribution systems, IEEE Trans. Power Delivery 8 (1993)
1529–1535.
[16] W.F. Mohammad, N. Tawalbeh, K.M. Al-Aubidy, Fast power loss computation
and shunt capacitor insertion using fuzzy logic technique, American Journal of
Applied Sciences 4 (1) (2007) 37–41.
[17] Y. Baghzouz, S. Ertem, Shunt capacitor sizing for radial distribution feeders with
distorted substation voltages, IEEE Trans. Power Delivery 5 (1990) 650–657.