I.J.

Intelligent Systems and Applications, 2013, 07, 42-49

Published Online June 2013 in MECS (http://www.mecs-press.org/)
DOI: 10.5815/ijisa.2013.07.06

Optimal Placement and Sizing of Capacitor and

Distributed Generation with Harmonic and
Resonance Considerations Using Discrete
Particle Swarm Optimization
M. Heydari
E-mail: [email protected]

S.M. Hosseini
Faculty of electrical engineering, Babol University, Iran
E-mail: [email protected]

S.A. Gholamian
Faculty of electrical engineering, Babol University, Iran
E-mail: [email protected]

Abstract— Presence of distributed generation (DG) in distribution system is greater than in transmission
distribution systems has significant impacts on the system. Most of loads connected to distribution systems
operational characteristics of thes e systems, also using are induction loads which make the system power factor
capacitor for reactive co mpensation and loss reduction will be lag and the voltage drop to a lower level even
is so common. Injected harmonic currents fro m non- lower than the acceptable range. Any components
linear loads into distribution system distort all of connected to power system for working properly and
voltages and currents and must be considered when safely must receive voltage in the defined range, any
placing the capacitor banks so that the resonance will voltage out of this boundary can damage these
not occur. In this paper d iscrete particle swarm components. In power system, operator is obligated to
optimization (DPSO) approach is used for the optimal maintain voltage level o f each costumer bus within the
placement and sizing of d istributed generations and required limit. A most acceptable voltage variation is
capacitors in distribution systems for simu ltaneous within the range of [1].
voltage profile imp rovement, loss and total harmonic
As mentioned above, drop of voltage is unavoidable,
distortion (THD) reduction. There is a term in the
so this drop should be compensating somehow. One of
objective function which prevents harmonic resonance
the most widely method is using shunt capacitors. In
between capacitor reactance and system reactance.
addition imp roving voltage profile, shunt capacitors can
Constraints include voltage limit, voltage THD,
reduce loss, enhance power factor and voltage stability
number/ size of capacitors and generators. For
of the system [2]. Another method is using distributed
evaluating the proposed algorith m, the IEEE 33-bus test
generation (DG).DG is play ing an increasing role in
system is modified and employed.
distribution system. Improving voltage profile, reducing
system losses, reinforcing grid and improving system
reliability are some advantages of using DG[3]. A ll of
Index Terms— Capacitor, Distributed Generat ion,
these advantages will be achieved only on the condition,
Optimal Placement, Harmonic, Resonance, Discrete
these (DG & capacitor) p laced in proper buses. Any
Particle Swarm Optimization
improper placing may adverse system condition. There
have been some studies for optimal p lacing such as:
tabu search [4], analytical approaches [5, 6], second-
I. Introduction order algorith m [7], meta-heuristic
approaches[8,4],fu zzy-GA algorith m[10],etc. Most of
Centralized Power p lants deliver the electricity to the
these techniques suffer fro m the inability to escape local
end-user via transmission system. Distribution system optimal and burden heavy computational
makes a link between the high voltage transmission time.Pso(particle swarm optimization)is a good method
system and consumers. In distribution system, the which is fast and have ability to escape local
voltage levels are low but current levels are high in the optimal.Pso is a self-adaptive search optimization was
compare o f transmission system, so the loss , in introduced by Kennedy and Ebhart[11]. Pso has been

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
 Optimal Placement and Sizing of Capacitor and Distributed Generation with 43
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

applied to many optimization problems such as: 2.3 Capacitor and Inductor:
dynamic systems [12], optimal capacitor placement in
Capacitor and inductor can be modeled similarly and
distribution system [13], constrained optimization [14],
their impedance at harmonic frequency can be
mu lti-objective optimization problems [15], etc. So me
expressed by:
works have used pso for optimal placing of capacitors,
it was considered as a continuous problem but as we
know capacitors are co mmercially available in discrete ⁄
sizes, so in this paper a modified version of pso which
is suitable for discrete problems and is called discrete
particle swarm optimizat ion (DPSO) or b inary partic le
swarm optimizat ion is used .Most of these works
consider all of loads are linear and exclude the effect of where f is the main frequency of the system.
harmonic in the objective function. But in pract ice, a
portion of loads are none linear because of wide speared
use of electronic devices, so it can lead to imp roper
placing because of probation of resonance at some III. Problem Formulation
harmonic frequencies. In this paper, there is a term in In this paper the following assumptions are made:
the objective function wh ich prevents harmonic
resonance.  Capacitors are fixed type.

The rest of this paper is organized as follows: in  Both linear and non-linear loads are presented in a
section 2 modeling of network co mponents at harmonic balanced 3-phase system.
frequencies is discussed. Problem formu lation for the
objective function for imp roving voltage profile and
minimizat ion of loss and THD is presented in 3.1 Constraints
section3.In section 4 the pso algorithm is described Voltage constraint will be defined as follows:
briefly. The results of DG and capacitor placement on
33-bus test system are presented and discussed in
section5. Finally, section 6 summarizes the main points √
and results of this paper.
where is lower , is upper bounds of rms
voltage and is harmonic order of voltage at bus i .
II. Model of Network Components at Harmonic In this paper =0.95 and =1.05.
Frequencies
In this paper for modeling of network co mponents at THD:
harmonic frequencies, some practical and
approximation of references [16,17] is used. Total harmonic d istortion of voltage (THD) should be
less than the maximu m of allowable

( )
2.1 Cable [√ ]
For very accurate model of cab le at high frequency ⁄
some modifications need to be applied to the resistance
and inductance of cable. But for practical harmon ic
( )
analysis a simple model is enough. For the
harmonic order the line impedance can be written as:

In this paper according to IEEE-519, =0.05.

2.2 Transformer
3.2 Indexing
Because of non-linear magnetizing properties of
transformer it is difficu lt to model it at harmon ic For forming of ob jective function some indexes are
frequencies and because of core saturation it produces defined.
harmonic itself. For harmon ic study in this paper we
assume transformer operates in normal condition. So
the impedance of transformer at harmonic frequency 3.2.1 Voltage deviation index
can be written as: This index is for imp roving of voltage profile and
defined as follows:

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
44 Optimal Placement and Sizing of Capacitor and Distributed Generation with
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

capacitors which make resonance will be o mitted

∑ automatically in the minimization progress of the
algorithm.

Where is the nominal of system voltage

( =1) , is the voltage at node i and n is nu mber of 3.3 Objective Function
busses.
By introducing above indexes objective function is
defined as follows:
3.2.2 Active and reactive power loss index (PL -I, QL-I)

where is the index weight, these weights indicate the

importance of each index in the p lacement problem.
They depend on the required analysis.
Where , are active and react ive power losses
after installation of DG and capacitor respectively and
, are active and reactive power loss before ∑ [ ]
installation.

In table 1 these weights is defined, for selecting these

3.2.3 THD index weights, the guides of [19,20] is used.

This index is for min imizing of total harmonic

distortion: T able 1: index weights

index Index weight value

∑ VD-I 0.3
PL-I 0.2
QL-I 0.2
Where is the total harmonic distortion of bus i. THD-I 0.15
RES-I 0.15

3.2.4 Resonance index (RES-I)

Many works concerning capacitor placement, IV. PSO Algorithm
consider all loads are linear, recently some wo rks take
Particle swarm optimizat ion is an algorith m
non-linear loads into account, so the capacitor
developed by Kennedy and Ebhart. This algorith m is
placement problem mixed with harmonic consideration,
based on social behaviors of bird flocking or fish
but as it is shown in [18] even in these works, capacitor
schooling and the methods which they use to find food
placement led to harmonic resonance at one or some
sources. In a simple way this algorith m is defined as
harmonic frequencies. In this paper there is a term in the
follows:
objective function wh ich prevents harmonic resonance.
So resonance index is defined as follow:  The search space is d-dimensional
 Particle: Each member is called particle and is
∑∑ presented by d-dimensional vector and described as :

[ ]
∑∑
Where is the (I-th) particle.
 Population: a set of n particles in the swarm is called
population and described as:
Where , are the voltage and current o f bus I at
harmonic o rder after installation and , [ ]
are before installation of DG and capacitor. When
resonance occurs, voltage or current at resonance  pbest: the best previous position for each particle is
frequency will increase and may go much more than the called particle best (pbest) and described as:
nominal amplitude of main frequency. So if resonance
occurs, the amplitude of this term will increase and

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
 Optimal Placement and Sizing of Capacitor and Distributed Generation with 45
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

[ ]
start
 gbest: the best position among all of particle is called
global best (gbest) and described as:
Getting the network information and
constant parameter
[ ]

 velocity: the rate of position change for each particle Randomly generates particle with discrete
is called particle velocity and described as: particle positions

[ ]
Power flow
 updating velocity: at iteration k the velocity for d-
dimension of i-particle is updated by: No
Harmonics?
( ) yes
( ) Harmonic power flow

Where w is the inertia weight, and are the Evaluate fitness value of each particle
acceleration constants, and , are two random values
in range [0,1]. The acceleration constants ( , )
control how far a particle will move in a single iterat ion. Store p-best & g-best
Typically these both are set to a value of 2. The inert ia
weight (w) is used to control the convergence of yes
Converged Print
behavior of pso. Small values of (w ) lead to mo re rapid result
convergence usually on suboptimal position, but large No
value may prevent divergence. In general the inert ia
Update particle's position & velocity
weight is set according to the following equation: vector End

( ) Discretize the position of each particle

 updating position: Preserve a feasible particle

The i-particle position is updated by:

Fig. 1: flowchart of pso algorithm

For binary d iscrete search space, Kennedy and Ebhart

[21] have adopted the pso to search in binary space by
applying a sig moid t ransformat ion to the velocity V. Simulation Results:
component given in (21) to squash the velocities into a The presented algorithm was imp lemented and coded
range [0,1],and force the component values of position in Matlab co mputing environ ment. In order to evaluate
to be 0 or 1. The equation for updating positions in (20) the proposed algorithm, the 12.6 kv 33-bus IEEE
then is replaced by (22) distribution system is modified and applied, such that
the objective function given in [16] is min imized. The
( ) single diagram of this system is shown in Fig.2. The
specification of this system is given in[22] .two none-
linear loads are replaced with loads in buses 5 and
26.This loads are two six-pulse converter with active
( ) and reactive power of 1MW and 0.75M VA R.the
harmonic current spectra of these converters is given in
table 2.

Fig.1 shows the flowchart of this algorithm

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
46 Optimal Placement and Sizing of Capacitor and Distributed Generation with
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

T able 4: DG sizes

DG
100 300 800 3000 5000
SIZES(KW)

In table 5, rms voltages and THD of all buses for the

base case (no installation) are demonstrated. As
observed in table 5, the voltage level of so me buses
drop to level lo wer than acceptable limit and THD of
some buses go to level higher than upper limit of
defined range by IEEE-519 standard.

T able 5: bus voltage and T HD before installation

Bus Bus
number number
Fig. 2: single line diagram of 33-bus IEEE distribution system
0 1 - 17 0.8693 7.83

1 0.9952 0.23 18 0.9947 0.23

T able 2: six-pulse converter harmonic spectra
2 0.9717 1.51 19 0.9911 0.23
Harmonic order Percentage of harmonic current
3 0.9573 2.50 20 0.9904 0.23
1 100
4 0.9429 3.55 21 0.9897 0.23
5 20
5 0.9077 7.50 22 0.9681 1.52
7 14
6 0.9040 7.53 23 0.9613 1.53
11 9
7 0.8989 7.57 24 0.9579 1.53
13 8
8 0.8924 7.63 25 0.9040 7.82
17 6
9 0.8863 7.68 26 0.8990 8.26
19 5
10 0.8854 7.69 27 0.8869 8.38
23 4
11 0.8838 7.70 28 0.8783 8.46
25 4
12 0.8774 7.76 29 0.8745 8.50

13 0.8750 7.78 30 0.8701 8.54

Capacitors and Distributed generations are
14 0.8735 7.80 31 0.8692 8.55
commercially available in discrete sizes. In table 3and 4
the size of capacitor bank and DG are given: 15 0.8721 7.81 32 0.8689 8.55

16 0.8699 7.83
T able 3: capacitor sizes

Capacitor Capacitor Capacitor The DPSO algorithm is applied in this system for
siz e (kvar) size(kvar) size(kvar) placement of capacitor and distributed generation for
150 1500 2850 improving voltage profile and reducing loss and THD.
In table 6 the locations and sizes of capacitors and
300 1650 3000 Distributed generations are given. In table 7, rms
450 1800 3150 voltages and THD of all buses after installation of
capacitor and DG are demonstrated, as observed in table
600 1950 3300 7, after placement of capacitor and DG by this
750 2100 3450 algorith m, no voltage and THD violat ion is observed in
any buses.
900 2250 3600

1050 2400 3750

1200 2550 3900

1350 2700 4050

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
 Optimal Placement and Sizing of Capacitor and Distributed Generation with 47
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

T ABLE 6: the locations of capacitors and DGs with proposed method

Bus number Capacitor(KVAR) DG(KW) Bus number Capacitor(KVAR) DG(KW)

0 - - 17 - -
1 1350 - 18 - 5000
2 - - 19 - -
3 750 - 20 1050 -
4 - - 21 - -
5 1500 3000 22 900 -
6 0 - 23 - -
7 0 - 24 - -
8 2700 - 25 - -
9 - 800 26 - -
10 - - 27 - -
11 - - 28 - -
12 - - 29 - -
13 - - 30 900 -
14 - - 31 1050 -
15 - - 32 - -
16 - -

T able 7: bus voltage and T HD after installation of capacitor and T able 8: comparison results
DG
Before
RESULTS Afte r installation
Bus Bus installation
number number
578 349
0 1 - 17 1.029 2.31
8.55 2.33
1 1.003 0.08 18 1.009 0.08
1 1.04
2 1.004 0.51 19 1.014 0.08

3 1.008 0.83 20 1.016 0.08 0.86 0.99

4 1.010 1.16 21 1.015 0.08

5 1.022 2.33 22 1.003 0.51

6 1.030 2.31 23 0.996 0.51

7 1.033 2.31 24 0.993 0.52

8 1.044 2.28 25 1.020 2.17

9 1.044 2.28 26 1.017 1.96

10 1.043 2.28 27 1.018 1.96

11 1.042 2.29 28 1.018 1.96

12 1.036 2.30 29 1.018 1.96

13 1.034 2.30 30 1.026 1.94

14 1.033 2.30 31 1.027 1.94

15 1.032 2.31 32 1.027 1.94

16 1.030 2.31

In table 8 and Fig.3 and 4 the results of two cases

(before installation and after installation) are co mpared.
As observed in table 8, after optimization by proposed Fig. 3: voltage comparison
method, power loss was decreased by 39.61% and
maximu m THD was decreased by 72.74%.

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
48 Optimal Placement and Sizing of Capacitor and Distributed Generation with
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

[6] N. Acharya, P. Mahat, N. M ithulananthan, "An

analytical approach for DG allocation in primary
distribution network, Int. J. Electr. Po wer Energy
Syst. 2006, PP.669-678.
[7] K.H. Kim, Y.J. Lee, S.B. Rhee, S.K. Lee, S. -K.
You, "Dispersed generator placement using fuzzy-
GA in distribution systems", in: IEEE PES
Summer Meeting, vol. 3 Ju ly, 2002, pp. 1148–
1153.
[8] Y.A. Katsigiannis, P.S. Georgilakis," Optimal
sizing of s mall isolated hybrid power systems
using tabu search", Journal of Optoelectronics and
Advanced Materials 10 (5) (2008) 1241–1245.

Fig. 4: T HD comparison [9] J.O. Kim, S.W. Nam, S.K. Park, C. Singh,
"Dispersed generation planning using improved
Hereford Ranch algorith m", Electr. Power Syst.
1998.
VI. Conclusion:
[10] J. Kennedy, R. Eberhart, "Particle swarm
In this paper a modified version of pso algorith m optimization", IEEE . Conf. Neural Netwo rks,
(DPSO) wh ich is suitable for d iscrete problems is Perth, Australia, November 1995, pp. 1942–1948.
applied for optimal p lacing and sizing of capacitor and
DG in distribution system. In some works in spite of [11] K.E. Parsopoulos, M.N. Vrahatis," On the
taking harmonic into account like [17] resonance occurs computation of all min imizers through particle
at some frequencies, in this work inclusion of resonance swarm optimization, IEEE Trans. Evolut.
index in the objective function prevents occurring Comput,2004,PP.211-224.
resonance. Simu lat ion was done on the modified 33-bus [12] X. Yu , X. Xiong, Y. Wu," A PSO-based approach
IEEE test system and results are printed. The results to optimal capacitor placement with harmonic
show a significant reduction of power losses and THD distortion consideration", Electr. Power Syst.
and improvement of voltage profile in the system after 2004,227-233.
optimal placement and sizing of capacitor and DG by
proposed method. [13] R.A. Krohling, L.S. Coelho," Co-evolutionary
particle swarm optimizat ion using Gaussian
distribution for solving constrained optimization
problems" IEEE Trans. Syst,2006,PP. 1407–1416.
References
[14] U. Bau mgartner, C. Magele, W. Renhart, "Pareto
[1] M.Wang,J.Zhong," a novel method for distributed optimality and particle swarm optimizat ion", IEEE
generation and capacitor optimal placement Trans. 2004,PP.1172–1175.
considering voltage profile",IEEE,2011.
[15] ‖Modeling and Simulation of the Propagation of
[2] K.Hghdar,H.A.Shayanfar," optimal placement and Harmonics in Electric Power Network Part I:
sizing of DG and capacitor for the loss reduction Concepts, Models and Simulation Techniques,‖
via methods of generalized pattern search and Task Force on Harmonics Modeling and
genetic algorithm",IEEE.2010. Simu lation, IEEE Trans. on PWRD, Vo l. 11, No. 1,
[3] A.Mady,I.Tawfeek,"optimal sizing of capacitor Jan. 1996 pp. 452-465.
banks and distributed generation in distorted [16] ‖Modeling and Simulation of the Propagation of
distribution networks by genetic Harmonics in Electric Power Netwo rk Part II:
algorith ms",CIRED,20th international conference Sample Systems and Examp les Task Force on
on electricity distribution,2009. Harmonics Modeling and Simu lation", IEEE Trans
[4] K. Nara, Y. Hayashi, K. Ikeda, T. Ashizawa," on Power Delivery, Vo l. 11, No. 1, pp. 466-474,
Application of tabu search to optimal placement of Jan. 1996.
distributed generators", IEEE PES Winter Meeting, [17] S.A.taher,S.A.Hosseni,"a novel technique for
vol. 2, 2001, pp. 918–923. optimal capacitor p lacement and sizing in
[5] H.L. Willis," Analytical methods and rules of distribution systems with non-linear loads based on
thumb for modeling DG d istribution harmonic source identification",IUST,2008,pp.99-
interaction",IEEE PES Su mmer Meeting, vol. 3, 106.
Seattle,WA, Jul 2000. [18] D. Singh, D., K.S. Verma, "Multiobjective
optimization for DG p lanning with load models",

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49
 Optimal Placement and Sizing of Capacitor and Distributed Generation with 49
Harmonic and Resonance Considerations Using Discrete Particle Swarm Optimization

IEEE Transactions on Power Systems 2009,pp.  Power converters for electric machines
427–436.
 Simu lation techniques such as finite elements
[19] L.F. Ochoa, A. Padilha-Felt rin, G.P. Harrison, analysis and MATLAB
"Evaluating distributed generation impacts with a
 Novel electric machines for different applications
mu ltiobjective index", IEEE Transactions on
Power Delivery 2006,pp.1452-1458.  Renewable Energy
[20] J. Kennedy and R. C. Eberhart, ―A Discrete
Binary Version of the Particle Swarm A lgorith m
Proc. of the conference on Systems, Man,1997,
How to cite this paper: M . Heydari, S.M . Hosseini, S.A.
pp.4104-4109.
Gholamian,"Optimal Placement and Sizing of Capacitor and
[21] S.K. Goswami, S.K. Basu, A new algorith m for the Distributed Generation with Harmonic and Resonance
reconfiguration of distribution feeders for loss Considerations Using Discrete Particle Swarm Optimization",
minimizat ion, IEEE Trans,1992,1484-1491. International Journal of Intelligent Systems and
Applications(IJISA), vol.5, no.7, pp.42-49, 2013. DOI:
[22] Abou-Ghazala,A.,"optimal capacitor placement in 10.5815/ijisa.2013.07.06
Distribution Systems Feeding non linear loads",
IEEE Bolgona Power Teach conference,2003..

Authors’ Profiles
Mohamadreze Heydari born in sari,mazandaran,Iran
in 1985.recieved B.S degree fro m babol university of
technology in 2010. is a graduate student of babol
university of technology

Seyed Mehdi Hosseini received

B.S degree fro m university of
mazandaran in 2000,and received
M.S degree fro m Iran university of
science and technology in
2002,recieved PhD degree fro m
Iran university of science and
technology in 2009,faculty
member of babol university of
technology, babol, Iran. His
research interests are Reliability of Distribution
Systems, Distributed Generation, FACTS Devices

Seyed Asghar Ghol amian

received B.S degree fro m
khaje nasiredin toosi
university of technology in
1999,and received M.S degree
university of mazandaran in
2001,recieved PhD degree
fro m khaje nasiredin toosi
university of technology
2009,facu lty member of babol
university of technology, babol, Iran. His research
interests are:
 Design, simulation, modeling and control of electrical
machines
 Condition monitoring and fault diagnosis of electric
machinery
 Electric and Hybrid Electric Vehicles motor

Copyright © 2013 MECS I.J. Intelligent Systems and Applications, 2013, 07, 42-49