You are on page 1of 9

Downloaded from http://iranpaper.

ir
http://www.itrans24.com/landing1.html

Ain Shams Engineering Journal xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Ain Shams Engineering Journal


journal homepage: www.sciencedirect.com

Electrical Engineering

A multi-objective optimal sizing and siting of distributed generation


using ant lion optimization technique
Mohammad Jafar Hadidian-Moghaddam a, Saber Arabi-Nowdeh b, Mehdi Bigdeli c,⇑, Davood Azizian d
a
Young Researchers and Elite Club, Zanjan Branch, Islamic Azad University, Zanjan, Iran
b
Electrical Department, Payambar’azam Student Research Center, Aq’qala, Golestan, Iran
c
Department of Electrical Engineering, Zanjan Branch, Islamic Azad University, Zanjan, Iran
d
Department of Electrical Engineering, Abhar Branch, Islamic Azad University, Abhar, Iran

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

Article history: In this paper, a new optimization method is proposed to solve the optimal sizing and siting problem of
Received 10 August 2016 distributed generation (DG) in a distribution system (DS). The optimization problem is solved using a
Revised 2 January 2017 new ant lion optimizer (ALO) with considering different objectives. These objectives are reduction of pur-
Accepted 6 March 2017
chased energy cost from upstream network (due to DGs’ power generation), reliability improvement,
Available online xxxx
reduction of DGs’ application cost, reduction of DS losses and reduction of buses’ voltage deviation.
This problem is solved as a multi-objective optimization (MOO) in addition of a single-objective opti-
Keywords:
mization (SOO). The proposed DGs optimal sizing and siting is implemented on 33 and 69-bus IEEE net-
Optimal sizing and siting
Distributed generation
works. The obtained results show that the ALO has the best performance in extracting the solution of DGs
Ant lion optimizer optimal sizing and siting problem in comparison to particle swarm optimization (PSO) and genetic algo-
Reliability rithm (GA).
Ó 2017 Ain Shams University. Production and hosting by Elsevier B.V. This is an open access article under
the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction literature have studied the optimal sizing and placement of DGs
[5–7]. In order to maximize the voltage stability and reducing real
The rapid load growth, environmental pollution and global power losses in a distribution network, an invasive weed optimiza-
warming have led to a new trend of power generation locally at tion based approach has been proposed in [5]. Ref. [6] proposes a
the distribution voltage level by using distributed generation method for optimal placement and sizing of a DG at multi-load
(DG) [1,2]. DGs can be either connected or disconnected from the levels to minimize the active power losses. A method that uses pro-
main grid, and thereby can provide higher flexibility and expand- gramming technique and genetic algorithm (GA) has been sug-
ability. Accurately designed and operated DG units can lead to gested in [7]. In addition, multi-objective approaches have been
enhancement of power quality, reliability and security of the used for optimal sizing and siting of DG units in the literature
distribution networks. Moreover, system’s power losses and trans- [8–11]. A multi-objective approach based on the weight algorithm
mission congestion can be decreased by employing renewable- has been proposed in [8] for this purpose, considering DGs uncer-
energy resources [3,4]. In spite of the mentioned potential benefits, tainty and power quality.
without suitable placement and sizing, penetration of DG units in a In [9], a multi-objective algorithm that uses the double trade-off
distribution network can encounter with several challenges and method for optimal sizing and siting of multiple DGs was sug-
potential drawbacks such as unacceptable voltage profile, debilita- gested for minimization of network costs and improvement of
tion of voltage stability, inaccurate operation of the protection power quality. Two strategies based on Monte Carlo simulation
devices and etc. have been presented in [10] to achieve stochastic placement of
Optimal sizing and placement of DG units can be obtained multiple DG units in low and medium-voltage distribution grids.
through defining a proper objective function, which could be In [11], a multi-objective algorithm was proposed for DG place-
constructed of single or multi objectives. Considerable amount of ment in which objectives were presented as minimization of cost
index, technical risks, and economic risk using non-dominated GA.
In [12], the optimal placement of DG units in radial DS has been
Peer review under responsibility of Ain Shams University.
investigated via bacterial foraging optimization algorithm to min-
⇑ Corresponding author.
imize the power losses, operational costs and the voltage profile
E-mail addresses: m.bigdeli@srbiau.ac.ir, bigdeli.mehdi@gmail.com (M. Bigdeli).

http://dx.doi.org/10.1016/j.asej.2017.03.001
2090-4479/Ó 2017 Ain Shams University. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

2 M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx

enhancement of system simultaneously. A simple and fast tech- Max F 1 ¼ ðF 11 þ F 12 Þ=F 13 ð1Þ
nique has been presented in [13] to determine appropriate location
where F11 represents a reduction in the cost of purchased energy
and size of DG units with consideration of a voltage stability indi-
and F12 illustrates a reduction in the cost of energy that is not sup-
cator (VSI). In VSI consideration, vulnerable buses of the DS have
plied (resulted from applying optimal DGs allocation on the net-
been arranged (rank-wise) to form a priority list for allocation of
work). The objective of DGs optimization is maximizing F 11 and
DG units. In [14], an interactive satisfying method has been pro-
F 12 . Also F 13 refers to DGs’ installation cost and must be minimized
posed, which was based on hybrid modified shuffled frog leaping
in this study.
algorithm. In this DG optimization problem, the energy losses,
the energy cost and total pollutant emissions should be minimized
as the objectives. In [15], a differential evolution algorithm (DEA) 2.1.1. Reduction of purchased energy cost from upstream network
was presented for DG optimization in a radial DS to minimize The main advantage of power generation through DG units
the real power losses and improve the voltage profile within the within the DS is that the reduction of the injected power from
frame work of system operation and security constraints. the upstream network. In addition, since this study is scheduled
It was observed that most research works in DG optimization for a 20 years period, inflation and interest rates should be applied
problems have been focused on economic aspects while the relia- in the calculation of electricity prices [4,16,17]. Therefore, the
bility indices have not been sufficiently addressed yet. Thus in this profit stemming from installing DG units into the distribution net-
study, different objectives such as reliability indices (as the cost of work can be determined as follow:
energy not supplied (CENS)) were applied to the DG optimization. F 11 ¼ C Purchase  C Purchase ð2Þ
NODG DG
For this purpose, a multi-objective optimization (MOO) method
based on the ant lion optimizer (ALO) is proposed to solve the
X
N DG  i1
1 þ InfR
DG optimal sizing and siting problem in DS with reliability C Purchase ¼ PiDG  q ¼ PiDG  q0  ð3Þ
considerations. The proposed method is implemented on 33 and
DG
i¼1
1 þ IntR
69-buses IEEE distribution test systems and the optimization
results are compared to other optimization methods. where PiDG is the DG generated power, q is the price of energy, N DG is
Consequently, some important contributions of this work can the number of DGs, i is the study time period (year), q0 is the price
be listed as follows: of energy in the first year ($/MWh) and InfR and IntR refer to infla-
tion and interest rates, respectively.
 Application of ALO as a robust optimization method for sizing
and siting of DG units in DS 2.1.2. Reliability improvement
 Comparison of the introduced DG optimization results to the Reliability is another advantage of DG’s optimal application in
solutions gathered from two well-known intelligent algorithms DS. Due to fault occurrence in DS branches, some customers can
(PSO and GA) be affected by long-term interruptions during the fault location.
 Considering reliability in optimal sizing problem Also, some of the customers in unfaulted sections of DS that are
 Considering variable load (three level load as light, normal and fed in islanding state, may automatically be separated from the
heavy load) instead of constant load faulted section. In this condition, DG is capable to restore the
 Applying different factors in objective function such as losses power to the downstream section of faulted branch and improve
minimization, voltage profile improvement, and load supply the reliability of DS. So, DGs optimal application can reduce the
reliability. cost of DS due to the customer’s interruption. In this study the reli-
ability is modeled as the CENS of DS. The CENS is defined as follow
The rest of paper is organized as follows. The problem formula- [6,10,11]:
tion is presented in Section 2. The proposed MOO method and the
F 12 ¼ C ENS ENS
NODG  C DG ð4Þ
introduced algorithm for solving the optimization problem are pre-
sented in Sections 3 and 4, respectively. The Simulation results and  i1
NX
branch X
Nl
discussions are presented in Section 5. Finally, in Section 6 the 1 þ InfR
C ENS
DG ¼ ki  Li  qint
0   t i  Dj ð5Þ
obtained results are concluded. i¼1 j¼1
1 þ IntR

where CENS is the cost of energy that is not supplied (ENS) per year,
2. Problem formulation Nl is the number of grid’s branches, Li is the lengths of the branch, ti
is the time that takes to remove the fault, qint0 is the cost of energy
The main objective of this study is to maximize the advantages that is not supplied in first year.
or profits of DG application in DS considering CENS. When DGs are
integrated into the DSs, several performance factors should be con- 2.1.3. Reduction of DGs application cost
sidered in optimal sizing and siting. The most important factors vs
The DG application cost (DGAC) includes the investment (C In
DG ),
that describe the system performance include the profit derived
from the sale of active power to upstream market, load supply reli- operation (C Oper Main
DG ) and maintenance costs (C DG ). The DGAC is

ability, power losses and voltage profile. Each of these factors can defined as follows [16,17]:
be considered as an objective function or applied as a part of objec- Inv s
F 13 ¼ DGAC ¼ C DG þ C Oper Main
DG þ C DG ð6Þ
tive function. Then the MOO of sizing and siting of DGs are imple-
mented in DS using ALO, PSO and GA. Inv s
The C DG ($/MW h) will be paid initially and it is related to the pur-
The important factors applied in problem’s objective function chase and installation of each DG. It can be calculated by (7):
are described and formulated as follows.
Inv s
X
N DG
C DG ¼ C iinv estment ð7Þ
2.1. First objective function i¼1

The first objective function in optimal sizing and siting of DGs in where C iinv estment is the cost of purchasing and installation of i-th DG
DS can be defined as (1). units.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx 3

The C Oper where Li is the load in i-th year, L0 is the load at the first year and
DG ($/MWh), which is important regarding to DG units, is
the operational cost that can be defined as (8): Lgrowth is the load growth (Lgrowth is assumed to be 1% in this study).

X
N DG
C Oper
DG ¼ C operation
i ð8Þ 3. Multi-objective optimization
i¼1
MOO is an area with several decision indices. In this area, the
where, C operation
i is the operating cost of i-th DG and can be calculated optimization problem has more than one objective function and
as (9): it is optimized, simultaneously. MOO technique has been applied
 i1 in engineering science, economics and logistics, where the decision
1 þ InfR
C operation ¼ C operation  ð9Þ variables are needed to be determined optimally subjected to dif-
i 0
1 þ IntR
ferent optimization constraints. This paper uses the MOO method
The C Main
DG ($/MW h) is an annual cost that represents the interest
based on the ALO for optimal planning and sizing of DGs. In math-
and inflation rates and can be modeled as (10): ematical terms, a MOO problem can be formulated as [21,22]:

X
N DG min = max FðxÞ ¼ ½f 1 ðxÞ; f 2 ðxÞ; . . . ; f 3 ðxÞ
C Main
DG ¼ C maintenance
i ð10Þ hðxÞ ¼ 0 ð15Þ
i¼1
gðxÞ 6 0
where C maintenance
i is the maintenance cost of i-th DG and can be cal-
culated as (11). where F is an objective function that includes several single-
objective functions. x represents objective (Input) vector or an out-
 i1
1 þ InfR come. If an objective function is needed to be maximized, it is
C maintenance ¼ C meintenance  ð11Þ
i 0
1 þ IntR equivalent to minimize its negative (or inverse). h and g are set of
equal and unequal constariants, respectively.
MOO formulations can be expressed in two ways. First, as a
2.2. Second objective function multi-objective function with weights (the weighted sum method)
where the multi-objective formulation is converted to a single-
Power losses reduction is another advantage of DGs’ optimal objective function that is created from the weighted sum of indi-
application in DS. The DS losses are depend on the line resistance vidual objectives. In the second form, as a multi-objective index,
and the passing currents. Also, the losses are usually seen as ther- the multi-objective formulation is transformed into a single-
mal losses. In this study, the maximizing of lossess reduction is objective function using the goal programming method. In this
considerd as the second objective function. The DS losses reduction study, the first method is used. In mathematical terms, it can be
may be considered as maximizing the objective function that is formulated as (16) [21,22]:
shown in (12). X
min = max wi f i ðxÞ
Max F 2 ¼ LossDS DS
NODG  LossDG ð12Þ i
X ð16Þ
where LossDS and LossDS
refer to the losses without and with DG wi ¼ 1; 8i : wi P 0
NODG DG
i
application, respectively.
The main disadvantage of this method is to determine the accu-
2.3. Third objective function rate weights, especially when there is not enough information
about the problem. Consequently, any attained optimal point will
The DGs application in DS can reduce the voltage drop in system be a function of the coefficient that is used to combine the objec-
branches and also improve the buses’ voltage profile to 1 per unit. tives. Other limitations of this approach are comprehensively pre-
In this section, minimization of voltage deviation is considered as sented in [22]. To cope with this challenge, it can be converted into
the objective and so the third objective function may be defined a single-objective optimization (SOO) by fuzzy optimization strat-
as (13) [18–20]: egy and then be solved by an optimization algorithm.
With fuzzy based technique, the simultaneous optimization of
X
Nb
an objective function and constraints can be achieved [22]. The
Min F3 ¼ jðV i  1Þj ð13Þ
i¼1
MOO is transformed into a single-objective model of maximizing
the minimum degree of satisfaction among the membership func-
where Nb is the number of buses and Vi is the magnitude of voltage tions. In order to determine the optimal point, both objective func-
at bus i. tions and constraints must be characterized by membership
functions. In simultaneous MOO technique, the first step is to
2.4. Load modeling introduce the li for i-th fuzzy objective function. In other words,
for each function a l is defined which can be defined based on
Considering a constant load cannot provide accurate results for Eq. (17). For the multiple-objective functions, a suitable member-
the system utilization. Usually, the load of a DS varies from hour to ship function should be defined for each objective function based
hour. However, for a multi-year study, considering an hour to hour on Fig. 1.
load is not reasonable because it needs a huge amount of computa-
fi  fi
tions. One of the best solutions for modeling the time-varying load is li ¼ max
ð17Þ
fi max  f i min
using three different load levels, including light, normal and heavy
loads [6]. The load pattern, the price of electricity and the price of Therefore, it is necessary for each objective function to be
ENS have certain values in each level. According to a 20 years plan, solved to calculate the minimum value of the objective function.
the network load grows each year due to the population rise. This According to (17), each objective function that is closer to its
load growth for every year can be calculated as (14): maximum value has the minimum value for l. The mathematical
expression of this relationship is presented in (18) and (19)
Li ¼ L0  ð1 þ Lgrowth Þi1 ð14Þ
[21,22].

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

4 M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx

ALO fully discussed in [23–26]. So, this algorithm will be described


briefly here.

4.1. ALO algorithm operators

Ants’ random (walking) movement in the search space can be


defined as (21) [23–26]:

xðtÞ ¼ ½0; cumsumð2rðt1 Þ  1Þ; cumsumð2rðt2 Þ


 1Þ; . . . ; cumsumð2rðt n Þ  1Þ ð21Þ

where cumsum calculates cumulative set, n is maximum repetitions,


Fig. 1. l for i th objective function. t indicates random walks step and r(t) is a random function which is
defend as (22) [23]:

lD ðxÞ ¼ minðlf 1 ðxÞ; lf 2 ðxÞ; . . . ; lfm ðxÞÞ ð18Þ 1 if rand > 0:5
rðtÞ ¼ ð22Þ
0 if rand 6 0:5
max lD ðxÞ s:t: hðxÞ ¼ 0; gðxÞ 6 0 ð19Þ
where t indicates random walks phase (repetition) and rand is a
As mentioned before, in the fuzzy technique, the simultaneous random value that is produced with uniform distribution in [0 1]
optimization of an objective function and constraints can be [23].
achieved. In this regard, the constraints are checked after genera-
tion of variables and calculation of the objective function. If con-
straints are in the desired range, the value of 1 assigned to them, 4.2. Random walks of ants
otherwise a negative number assigned to them. The penalty factor
is defined by lc and it will be inserted into the lD. Based on this According to updating position of ants in each optimization
definition, the MOO can be written as follows [21,22]: walk, normalization is conducted in order to maintain random
walks inside the search space [23–26]:
max lD ðxÞ
ð20Þ
S:T lD ðxÞ ¼ minðlf 1 ðxÞ; lf 2 ðxÞ; . . . ; lc1 ðxÞ; lc2 ðxÞ; . . .Þ ðX ti  ai Þ  ðdi  cti Þ
X ti ¼ t þ ci ð23Þ
di  ai

4. Ant lion optimizer (ALO) where ai is the minimum random walk for variable i, bi is the
maximum random walk for variable i, cti is minimum of the i-th
ALO algorithm was presented by Mirjalili in 2015 [23]. This variable in repetition t, and dti indicates maximum i the variable
algorithm mimics the hunting mechanism of ant lions in nature. in repetition t.

Fig. 2. The IEEE 33-bus DS.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx 5

4.3. Sliding ants toward ant lion RtA þ RtE


Antti ¼ ð27Þ
2
Following relations express mathematical modeling of trapped
where RtA is random walk around ant lion, which is selected by roul-
ant‘s sliding and its seeking to escape (since radius of hyper-
ette wheel, and RtE is random walk around elite.
sphere in ants random walks is reduced) [23–26]:
Pseudo code of ALO algorithm is presented as follows:
ct ¼ ct =I ð24Þ
Insert the ants and ant lions’ first population randomly
t t
d ¼ d =I ð25Þ Ants and ant lions fitness calculation
Determination of the best ant lions as the elite (optimal)
where I is a ratio, ct is minimum value of all variables in the repeti-
while the end index is unsatisfied
tion t, and dt indicates a vector containing maximum value of all
for every ant
variables in the repetition t. In the above relations, I is equal to
An ant lion is elected by roulette wheel
I ¼ 10w Tt in which t is repetition of current phase, T is maximum
Update parameter c and d (Eqs. (24) and (25))
number of repetitions and W is a constant value based on the cur-
Generation a random walk
rent repetition.
Normalization the random walk (Eqs. (21) and (23))
Update the ant position (Eq. (27))
4.4. Catching prey and re-building the pit end for
The fitness for all ants is calculated.
In order to improve the chance of ant lion’s to catch its prey, the If an ant lion is better than the ant in view of fitness value,
latest position of hunted ant can be defined as (26) [23]: replace it instead the ant (Eq. (26)
t
Antlionj ¼ Ant ti if
t
f ðAnt ti Þ > f ðAntlionj Þ ð26Þ If an ant lion is better than elite in fitness, update the elite
end while
where Antliontj
indicates position of ant lion j, and Antti is the posi- Return elite
tion of ant i in the current repetition.
Table 1
Commercial data of DGs.
4.5. Elitism
Parameter Unit Value (33-bus Value [6]
system) (69-bus system)
Elitism is a key feature in complementary algorithms and
DG investment cost $/each DG 150,000 318,000
allows them to keep the best responses obtained in each phase
DG operation cost $/MW h 29 29
of process. The best obtained ant lion so far must be saved in each DG maintenance cost $/MW h 7 7
repetition to be considered as an expert and it must be able to be Interest rate % 12.5 12.5
effective on movement of all of ants during the repetitions. It is Inflation rate % 9 9
assumed that each ant can walk around the selected ant lion simul- Planning period Year 20 20
(DG life time)
taneously by Roulette Wheel in random way as follows [23–26]:

Fig. 3. The IEEE 69-bus DS.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

6 M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx

Table 2
Technical and commercial data of load [6]. 5. Simulations results and discussions

System Percentage of Time Price of not Market price


The proposed method is implemented on IEEE 33 and 69-buses
condition peak load (%) duration supplied energy ($/MW h)
(h/year) ($/kW h) DS. The IEEE 33-buses test system is depicted in Fig. 2. The base
voltage of this system is 12.66 kV and the maximum capacity of
Low load 60 2190 2.68 35
Medium load 80 4745 3.76 49 the system is 10 MVA. The network has 3715 kW and 1790 kvar
Peak load 100 1825 4.92 70 load at peak times and at the first year. The lowest voltage within
20 years is 0.8758 pu [27]. In addition, the IEEE 69-buses system is
depicted in Fig. 3. The voltage of this system is 12.66 kV and the
maximum capacity is 10 kV A. This system has 3802.2 kW and
2694.1 kvar load at peak times and in the first year [28]. Commer-
Table 3
Results of F1 optimization (SOO) in 33-bus system. Table 6
Results of MOO in 33-bus system.
ALO PSO GA
ALO PSO GA
The location for installing DG DG1 18 18 18
DG2 – – – The location for installing DG DG1 16 16 16
Total capacity of DG DG1 500 495.72 489.63 DG2 33 32 33
DG2 0 0 0 Total capacity of DG DG1 500 497.40 460.93
The objective function value 1.495 1.225 1.317 DG2 479.61 468.34 500
The objective function value F1 1.27 1.24 1.23
F2 61,136 59,013 58,654
F3 9.32 9.48 9.51
MOO value F 0.0750 0.0733 0.0725
Table 4
Results of F2 optimization (SOO) in 33-bus system.
Table 7
ALO PSO GA Results of F1 optimization (SOO) in 69-bus system.
The location for installing DG DG1 33 32 32
ALO PSO GA
DG2 15 15 15
The location for installing DG DG1 54 54 54
Total capacity of DG DG1 487.18 500 491.45
DG2 – – –
DG2 498.3 500 500
Total capacity of DG DG1 1000 999.46 997.05
The objective function value 44,117 43,730 43,082
DG2 0 0 0
The objective function value 1.302 1.300 1.298

Table 5 Table 8
Results of F3 optimization (SOO) in 33-bus system. Results of F2 optimization (SOO) in 69-bus system.
ALO PSO GA ALO PSO GA
The location for installing DG DG1 17 16 17 The location for installing DG DG1 52 49 49
DG2 16 33 15 DG2 50 50 52
Total capacity of DG DG1 474.67 461.49 469.77 Total capacity of DG DG1 896.97 933.77 1000
DG2 493.86 485.33 470.93 DG2 975.33 1000 1000
The objective function value 8.25 8.57 8.71 The objective function value 22,632 22,493 22,419

Fig. 4. The convergence curve of different methods in MOO in 33-bus system.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx 7

cial data regarding the selected DG unit has been specified in values in F3 optimization are 8.25, 8.57 and 8.71 for mentioned
Table 1, for both test systems. The selected DG has a maximum algorithms respectively.
rating of 500 kW and 1000 kW capacity. The DG’s costs for 33 Furthermore, the Fuzzy technique is applied in MOO. In this
and 69-buses systems are 150,000$ and 318,000$ respectively. strategy, the losses reduction is a priority. Therefore, its maximum
The load of IEEE test systems is considered in three levels. The value is twice its minimum. The next major objective function is
load increases 10% per year. In this study, the outage rate of the reliability and selling the energy which its maximum value is
branches is 0.046 for each km in a year [11]. It is assumed that three times of its minimum. The last one is the voltage profile
the reliability of other devices in these systems is equal to 100%. which its maximum value is four times of its minimum. The con-
In the outage condition of a branch, the consumed time between vergence curve of MOO is depicted in Fig. 4. This figure shows that
outage and repair is 8 h. Thus, due to a fault occurrence, the time the proposed algorithm has the best performance in solving the
of downstream costumers’ interruption is 8 h [20]. Technical and problem and has the best objective function (fuzzy index) values
commercial data of the loads in three levels are presented in versus PSO and GA. As shown in Fig. 4, the convergence speed of
Table 2. ALO is better than PSO and GA; also, the deviation of the extracted
MOO for optimal sizing and siting of DGs are implemented results of the ALO is less than PSO and GA. Therefore, the proposed
using ALO, GA and PSO methods. For comparison purpose, the pop- algorithm is an intelligent algorithm with higher convergence
ulation of three methods has been considered as 50 and also their speed in solving the DG optimization problem. The results of
maximum iteration is 100 times. The maximum capacity for each MOO DGs optimal sizing and siting are presented in Table 6. As
DG in IEEE 33 and 69-buses systems are 500 kW and 1000 kW, shown as in this table, the ALO results the best reliability, losses
respectively. Also in this study, the decision variables are consid- and voltage profile versus the PSO and GA.
ered as optimal size and installation place of DG units in DS.

5.2. Results for IEEE 69-bus system


5.1. Results for IEEE 33-bus system
The results of optimal sizing and siting of DGs in IEEE 69-bus
At the first, the results of SOO in 33-buses IEEE test system are system is done as SOO and MOO. The results of SOO are illustrated
presented in Tables 3–5. According to the results of these tables, in Tables 7–9. The optimization based on F1 shows that the ALO,
the ALO method has shown the best performance. In F1 based opti- PSO and GA are selected the bus 54 for the DG installation. In this
mization, in all the three methods, the bus 18 is selected to allocate case, the ALO has the best results from viewpoint of the objective
a DG. However, the ALO results have higher F1 value rather than function and the reliability. The objective function value in F1
other methods. Also in F2 and F3 based optimizations, the proposed optimization using ALO, PSO and GA are obtained as 1.302, 1.300
algorithm has the best objective function values versus PSO and and 1.298 respectively, while the values of objective function in
GA. In F2 optimization the objective function in ALO, PSO and GA
are obtained 44,117, 43,730 and 43,082 respectively, while these Table 10
Results of MOO in 69-bus system.

ALO PSO GA
Table 9
The location for installing DG DG1 50 52 52
Results of F3 optimization (SOO) in 69-bus system.
DG2 54 54 54
ALO PSO GA
Total capacity of DG DG1 1000 996.02 958.99
The location for installing DG DG1 15 15 15 DG2 422.67 347.37 369.50
DG2 54 44 43 The objective function value F1 1.32 1.22 1.20
Total capacity of DG DG1 989.12 983.77 965.46 F2 29,491 27,389 26,927
DG2 953.47 933.82 1000 F3 8.297 8.504 8.650

The objective function value 2.126 2.284 2.302 MOO Value F 0.319 0.265 0.254

Fig. 5. The convergence curve of different methods in MOO in 69-bus system.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

8 M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx

mentioned algorithms are obtained as 22,632, 22,493 and 22,419 evolution algorithm”, In International Conference on Swarm, Evolutionary,
and Memetic Computing, Springer, Berlin Heidelberg, pp. 133–142, 2012.
for F2 optimization. Also in F3 optimization the objective function
[16] Porkar S, Poure P, Abbaspour-Tehrani-fard A, Saadate S. A novel optimal
in ALO, PSO and GA is calculated as 2.126, 2.284 and 2.302 respec- distribution system planning framework implementing distributed generation
tively. In F2 and F3 based optimizations, the ALO determined the in a deregulated electricity market. Electr Power Syst Res 2010;80(7):828–37.
optimal sizes and generation capacities of DGs with the best losses [17] Hussein D. N., El-Sayed M. A. H., Attia H. A., ‘‘Optimal sizing and siting of
distributed generation”, In Power Systems Conference, MEPCON Eleventh
reduction and voltage profile rather than PSO and GA. Also the International Middle East, pp. 593–600, 2006.
problem is solved as MOO using fuzzy technique. The convergence [18] Ghosh S, Ghoshal SP. Optimal sizing and placement of distributed generation
curve of all algorithms in MOO is depicted in Fig. 5. This figure in a network system. Int J Electr Power Energy Syst 2010;32(8):849–56.
[19] Al Abri RS, El-Saadany EF, Atwa YM. Optimal placement and sizing method to
shows a less deviation in final solutions of ALO with the best values improve the voltage stability margin in a distribution system using distributed
of objective function (fuzzy index) rather than PSO. The results of generation. IEEE Trans Power Syst 2013;28(1):326–34.
MOO are presented in Table 10. According to this table, the ALO [20] Singh RK, Goswami SK. Optimum allocation of distributed generations based
on nodal pricing for profit, loss reduction, and voltage improvement including
results the best reliability, losses and voltage profile in DG opti- voltage rise issue. Int J Electr Power Energy Syst 2010;32(6):637–44.
mization problem, versus the PSO and GA. [21] Bellman RE, Zadeh LA. Decision-making in a fuzzy environment. Manage Sci
1970;17(4):141–64.
[22] Carlos CA. A comprehensive survey of evolutionary-based multi-objective
optimization techniques. Knowledge Inform Syst An Int J 1999;1(3):269–308.
6. Conclusions [23] Mirjalili S. The ant lion optimizer. Adv Eng Softw 2015;83:80–98.
[24] Shin Mei R. N., Sulaiman M. H., Mustaffa Z., ‘‘Ant lion optimizer for optimal
In this paper, a new optimization method has been proposed to reactive power dispatch solution”, In International Conference on Advanced
Mechanics, Power and Energy, Kuala Lumpur, Malaysia, 2015.
determine the optimal location and size of DGs in DS considering [25] Raju M, Saikia LC, Sinha N. Automatic generation control of a multi-area
the reliability index as the cost of unsupplied energy. The opti- system using ant lion optimizer algorithm based PID plus second order
mization problem has been solved using ant lion optimizer algo- derivative controller. Int J Electr Power Energy Syst 2016;80:52–63.
[26] Kamboj V. K., Bhadoria A., Bath S. K., ‘‘Solution of non-convex economic load
rithm considering DGs’ installation cost reduction, DS losses dispatch problem for small-scale power systems using ant lion optimizer”,
reduction, voltage profile and reliability improvement as objective Neural Computing and Applications, Online Publication, 2016.
functions. The proposed method has been implemented on 33 and [27] Kayal P, Chanda CK. Placement of wind and solar based DGs in distribution
system for power loss minimization and voltage stability improvement. Int J
69-buses IEEE test systems as SOO and fuzzy MOO optimizations.
Electr Power Energy Syst 2013;53:795–809.
The obtained results show that ALO algorithm has the best results [28] Shyh-Jier H. An immune-based optimization method to capacitor placement in
in cost, loss reduction, voltage profile and reliability improvement a radial distribution system. IEEE Trans Power Deliv 2000;15(2):744–9.
versus PSO and GA in both test systems. Also in MOO problem, the
ALO has the best results from the viewpoint of objective function
value. Therefore, according to the good performance and the best Mohammad Jafar Hadidian-Moghaddam received his
B.Sc degree in Electrical Engineering from Azarbaijan
results rather than PSO and GA, the proposed method can be used Research Center, University of Applied Science and
as a reliable method in DG sizing and siting in DS. Technology, Tabriz, Iran in 2005. He received M.Sc.
degree from Islamic Azad University, Science and
Research Branch, in 2014. His research interests are in
References renewable energy, design optimization of power sys-
tems and migrogrids. Mr. Hadidian-Moghaddam is now
[1] Ackermann T, Andersson G, Soder L. Distributed generation: a definition. Electr a lecturer in Zanjan Branch, Islamic Azad University,
Power Syst Res 2001;57:195–204. Zanjan, Iran.
[2] Acharya N, Mahat P, Mithulananthan N. An analytical approach for DG
allocation in primary distribution network. Int J Electr Power Energy Syst
2006;28:669–78.
[3] Koeppel G. Distributed Generation-Literature Review and Outline of the Swiss
Station, Internal Report. ETH Zurich; 2003.
[4] Viral R, Khatod DK. Optimal planning of distributed generation systems in Saber Arabi-Nowdeh received his B.Sc degree and M.Sc
distribution system: a review. Renew Sustain Energy Rev 2012;16(7):5146–65.
degree in Electrical Power Engineering from Urmia
[5] Rama Prabha D, Jayabarathi T. Optimal placement and sizing of multiple
University, Iran in 2008 and 2010, respectively. He
distributed generating units in distribution networks by invasive weed
received Ph.D. degree in Electrical Power Engineering
optimization algorithm. Ain Shams Eng J 2016;7(2):683–94.
[6] Khalesi N, Rezaei N, Haghifam MR. DG allocation with application of dynamic from Tehran University, Iran in 2014. He works in
programming for loss reduction and reliability improvement. Int J Electr Power Golestan Technical and Vocational Training Center,
Energy Syst 2011;33(2):288–95. Gorgan, Iran. His research interests are in renewable
[7] Moradi MH, Abedini M. A combination of genetic algorithm and particle energy, distribution system planning and reliability
swarm optimization for optimal DG location and sizing in distribution assessment of power system.
systems. Int J Electr Power Energy Syst 2012;34(1):66–74.
[8] Ugranl F, Karatepe E. Multiple-distributed generation planning under load
uncertainty and different penetration levels. Int J Electr Power Energy Syst
2013;46:132–44.
[9] Falaghi H, Haghifam MR. ACO Based Algorithm for Distributed Generation
Sources Allocation and Sizing in Distribution Systems. In Power Tech, IEEE
Lausanne; 2007. p. 555–60.
Mehdi Bigdeli received the B.Sc. degree in electrical
[10] Hadian A., Haghifam M. R., Zohrevand J., Akhavan-Rezai E., ‘‘Probabilistic
approach for renewable DG placement in distribution systems with uncertain engineering from Iran University of Science and Tech-
and time varying loads” In IEEE Power & Energy Society General Meeting, pp. nology (IUST) in 2004 and M.Sc. and Ph.D. degrees from
1–8, 2009. Faculty of Engineering of Zanjan University and Islamic
[11] Haghifam MR, Falaghi H, Malik OP. Risk-based distributed generation Azad University, Sciences and Research Branch, in 2006
placement. IET Gener Trans Distrib 2008;2(2):252–60. and 2012, respectively. His research interests are in
[12] Devabalaji K. R., Ravi K., ‘‘Optimal size and siting of multiple DG and fault detection, transient modeling and application of
DSTATCOM in radial distribution system using Bacterial Foraging nature-inspired algorithms in power systems analysis.
Optimization Algorithm”, Ain Shams Engineering Journal, in press, 2016. Dr. Bigdeli is is now an Assistant Professor in Zanjan
[13] Kayal P, Chanda CK. A simple and fast approach for allocation and size Branch, Islamic Azad University, Zanjan, Iran.
evaluation of distributed generation. Int J Energy Environ Eng 2013;4(1):1–9.
[14] Doagou-Mojarrad H, Gharehpetian GB, Rastegar H, Olamaei J. Optimal
placement and sizing of DG (distributed generation) units in distribution
networks by novel hybrid evolutionary algorithm. Energy 2013;54:129–38.
[15] Nayak M. R., Dash S. K., Rout, P. K., ‘‘Optimal placement and sizing of
distributed generation in radial distribution system using differential

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001
Downloaded from http://iranpaper.ir
http://www.itrans24.com/landing1.html

M.J. Hadidian-Moghaddam et al. / Ain Shams Engineering Journal xxx (2017) xxx–xxx 9

Davood Azizian was born in 1981 in Rasht, Iran. He


received the B.Sc. (2003) and M.Sc. (2005) degrees in
electrical engineering from the Zanjan University and
Ph.D. degree (2012) in electrical engineering from Isla-
mic Azad University, Sciences and Research Branch
respectively. He has work in Iran Transformer Research
Institute, R&D Department and Dry-Type Technical
Office of Iran Transfo Company, and Abhar branch of
Islamic Azad University. His fields of interests include
application of heuristic algorithms in power systems
optimization and dry-type transformers. Dr. Azizian is
now an Assistant Professor in Abhar Branch, Islamic
Azad University, Abhar, Iran.

Please cite this article in press as: Hadidian-Moghaddam MJ et al. A multi-objective optimal sizing and siting of distributed generation using ant lion opti-
mization technique. Ain Shams Eng J (2017), http://dx.doi.org/10.1016/j.asej.2017.03.001

You might also like