Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Engineering Science and Technology,

an International Journal
journal homepage: www.elsevier.com/locate/jestch

Full Length Article

A novel approach based on crow search algorithm for optimal selection

of conductor size in radial distribution networks
Almoataz Y. Abdelaziz a, Ahmed Fathy b,⇑
a
Electric Power & Machine Dept., Faculty of Engineering, Ain-shams University, Cairo, Egypt
b
Electric Power & Machine Dept., Faculty of Engineering, Zagazig University, Zagazig, Egypt

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

Article history: The selection process of conductor size in radial distribution network is very essential issue to improve
Received 10 December 2016 the performance of the network. The true conductor size selection leads to less power loss, achieve
Revised 11 February 2017 improvement of the bus voltage profile and obtain reduction in the annual operating cost of the system.
Accepted 21 February 2017
This paper proposes a novel approach based on crow search algorithm (CSA) for optimal selection of the
Available online xxxx
conductors in a radial distribution network. The CSA is a recent meta-heuristic algorithm which is based
on the intelligent behavior of crows. The objective function presented in our work is the sum of conductor
Keywords:
capital cost and the conductor energy loss cost. The proposed constraints are bus voltage limits and the
Optimal conductor size
Crow search algorithm
current capacity of the conductor. The independent variables are the type and the size of conductor such
Distribution network that minimizing the proposed objective function. The proposed approach is applied on two different net-
work topologies, the first one is 16-bus system and the second is large scale system of 85-bus system. A
sensitivity analysis of the CSA controlling parameters are also studied for 16-bus system. The obtained
results via CSA are compared to previous works’ results; the CSA results encourage the usage of the pro-
posed approach in optimal selecting the conductor type and size in the distribution network.
Ó 2017 Karabuk University. Publishing services 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 [14], partial enumeration technique [15], harmony search algo-

rithm with a differential operator [16].
The challenges faced the distribution networks have been raised Thenepalle [2] introduced two comparative approaches of opti-
in the last decades as the main requirement of operating the radial mal selection for conductors in radial network such that minimiz-
network is its reliability. It is known that the number of resources ing the power losses, maximizing the cost saving and maintaining
in such networks is limited and the disconnection of any feeder the bus voltages in their corresponding limits. The first approach is
will prevent supplying the load. Therefore, it is important to select analytical one based on performing load flow study while the sec-
proper types of conductors able to carrying the required energy to ond is based on genetic algorithm. Sivanagaraju et al. [3] presented
load. an analytical approach based on radial network load flow to select
Many researchers studied the optimal selection of the conduc- the suitable conductor size to maximize saving in conductor capi-
tors in radial networks. A review study of optimal selecting the tal and energy loss costs. Satyanarayana et al. [4] presents a model
conductor size in radial network approaches has been presented of overloading to improve the loading of radial feeder based on the
in [1]. These literature methodologies can be classified into two optimal selection of the conductor. Legha et al. [5] used GA to
main categories; the first one is conventional approaches based select the conductor size to minimize the system overall annual
on analytical analysis [2–4] and the second is based on meta- cost, a backward-forward sweep method has been used to perform
heuristics as genetic algorithm [2,5–7], evolutionary strategies the load flow study of radial network. In [6] two actions are taken
[8,9], modified differential evolution [10], imperialist competitive via GA; the capacitor optimal placement and the optimal selection
algorithm [11], bacterial foraging algorithm [12], particle swarm of conductors in radial network to reduce the system losses and
optimization (PSO) [13], discrete particle swarm optimization enhance the voltage profile. Sharma et al. [7] presented an
approach based on GA to minimize the network losses and maxi-
mize the saving cost via selecting the networks’ conductors opti-
⇑ Corresponding author. mally. Mendoza et al. [8] presented an evolutionary strategy (ES)
E-mail address: [email protected] (A. Fathy). to solve the problem of selecting the network feeders while Ranjan
Peer review under responsibility of Karabuk University.

http://dx.doi.org/10.1016/j.jestch.2017.02.004
2215-0986/Ó 2017 Karabuk University. Publishing services 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: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
2 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

et al. [9] used evolutionary programming (EP) algorithm to solve X

nb

the problem. Kalesar et al. [10] presented a modified differential C Et loss ¼ ½P loss ðiÞ  ðK P þ K E  T  LSFÞ ð2Þ
i¼1
evolution (MDE) to solve the problem of selecting the conductors
in radial network taken the conductor capital cost, power loss where Ploss(i) is the power loss of branch i, KP is the annual cost of
and energy loss cost as objective function. Legha et al. [11–13] the losses (Rs/kW), KE is the cost of energy loss (Rs/kWh), T is the
introduced the imperialist competitive algorithm (ICA), bacterial period in hours (8760 h.), LSF is the loss factor and nb is the number
foraging algorithm (BFA) and particle swarm optimization (PSO) of branches.
to solve the optimal selection of network feeders. Sivanagaraju The annual capital investment cost for a branch i of type j is
et al. [14] considered the discrete particle swarm optimization given as follows:
(DPSO) is suitable for selecting the conductors in radial network.
C c ði; jÞ ¼ a  AðjÞ  C  LðiÞ ð3Þ
Kaur et al. [15] concerned with the power cost and peal load diver-
sity during a certain period in determining the optimal conductor where a is the carrying charge rate, A (j) is the cross section area of
type in a distribution network using partial enumeration tech- branch of type j (mm2), C is the cost of the line (Rs/mm2.km) and L
nique. Rao et al. [16] combined the harmony search algorithm with (i) is the length of conductor i (km).
a differential operator for optimal selecting the network conduc- Finally, one can write the objective function as follows:
tors to minimize the conductor investment and energy loss costs
taking the bus voltages and conductors’ maximum current capacity X
nb X
nc
F ¼ C Et loss þ C ct ¼ ½Ploss ðiÞ  ðK P þ K E  T  LSFÞ
as constraints. i¼1 j¼1
In this work; a novel approach based on a recent meta-heuristic
þ a  AðjÞ  C  LðiÞ ð4Þ
algorithm, crow search algorithm (CSA), is proposed to solve the
problem of optimal selection of the distribution network conduc- where Cct is the total capital investment cost of all branches and nc is
tors’ types. A constrained objective function combined the conduc- the number of allowable conductors’ types.
tor capital and the conductor energy loss costs are presented. The
bus voltage limits and the max current carrying capacity of con- B. Constraints
ductor are considered as constraints. The proposed CSA is applied
on two different network topologies, 16-bus system and 85-bus The proposed constraints can be described as follows:
system. A sensitivity analysis of the CSA controlling parameters,
maximum iteration, flight length and awareness probability, are  Bus voltage constraints
studied for 16-bus system and the best set of these parameters
are taken in analyzing 85-bus system. The obtained results V minðjÞ 6 jVðjÞj  V maxðjÞ 8j 2 nbus ð5Þ
via the CSA are compared to others in previous works; the
results ensure the superiority of the proposed algorithm in solving  Maximum current carrying capacity
the problem of optimal selection of the radial network conductor
type. jIði; cÞj 6 ImaxðcÞ 8i 2 nb ð6Þ

where Vmin(j) and Vmax(j) is the minimum and maximum voltage at

bus j, nbus is the number of buses, I(i, c) is the flow in conductor i
2. Optimization problem formulation of type c and Imax(c) is the maximum allowable current flows in con-
ductor c.
In this section the objective function and the proposed con-
straints are presented.
3. Crow search algorithm
A. Objective function
At recent days, Askarzadeh [18] proposed a novel metaheuris-
tics optimization algorithm based on the social behavior of the
The main object of this work is to select a suitable conductor in
crow and named Crow Search Algorithm (CSA). The idea of CSA is
radial distribution network that minimizes the sum of the conduc-
motivated from the storing process of the excess food in hiding
tor capital cost and the conductor energy loss cost with keeping
places then restoring it in the necessary time. It is known that
acceptable bus voltage profile and achieving acceptable power flow
the crow is very intelligent bird that observes the others hide their
in the network branches. Assuming that the conductor to be
food and steal it once they leave. After committing the theft, it
selected is between two buses m and n as shown in Fig. 1, the cur-
hides to avoid being a victim in the future. It is assumed that a flock
rent Imn is flowing from bus m to bus n to supply the load con-
of N crows, the crow number i has position at iteration k is xki . The
nected to bus n. The conductor power loss can be calculated as
hiding place of the food followed by crow i is memorized. Crow
follows:
moves in the search plane and tries to find the best food source
ðP2n þ Q 2n Þ which is defined as mki . The searching approach in CSA has two
Ploss ¼ jImn j2 Rmn ¼ Rmn ð1Þ probable scenarios; the first one is that the owner crow j of food
jV m j2
source mkj doesn’t know the thief crow i follows it therefore the
The total annual cost of the energy loss can be written as thief crow reaches to the hide place of owner crow. The updating
follows [17]: process of the crow thief position is done by
k
V m ∠δm V n∠ δ n xkþ1
i ¼ xki þ ri  fli  ðmkj  xki Þ ð7Þ
Imn
flki
where ri is a random number in range [0,1], is the flight length of
crow i at iteration k.
Rmn+jXmn The second probable scenario is that the owner crow j know
m n Pn+jQn that the thief crow i follows it therefore, the owner crow will
deceive crow i by going to any another position of search space.
Fig. 1. Electrical equivalent of a conductor between two buses. The position of crow i is updated by a random position.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx 3

In CSA, the scenario is determined by the following if

expression: rj P pkj
ifr j P pkj update position by eqn:7 xkþ1
i ¼ xki ðrandpermðnc ; nb  nc ÞÞ i ¼ 1; 2; :::; nb
else ð8Þ else
update to random position xkþ1
i;j ¼ xki;j ðrandpermðnc ; 1ÞÞ i ¼ 1; 2; . . . ; nb & j ¼ 1; 2; . . . ; nf
end
where rj is a random number in range [0,1] and pki is the probability
of awareness of crow j at iteration k. ð10Þ
The parameter flki plays an important role in catching the global Where randperm is Matlab function used to randomize the rank
optimum solution as small value of flki leads to local minimum of the selected conductor. The final stage in the proposed steps is
while large value lead to global minimum, Fig. 2 shows the effect checking the constraints, if it is achieved then the optimal solution
of flki in searching process. The main steps followed in CSA are is obtained else repeat the CSA shown in Fig. 3.
shown in Fig. 3.
Referring to Fig. 3, the initial position of the flock is defined and
then the corresponding fitness function is calculated. The flock’s 5. Simulation results
position is updated at iteration k based on Eq. (8) and then the
new fitness function is calculated, the obtained fitness function In order to show the validity of the proposed methodology
at iteration k is compared with previous one and updating flock’s incorporated CSA; two radial distribution networks are studied,
position is placed. the first one is small scale network of 16-bus system and the sec-
ond one is large-scale network of 85-bus system. In this work; the
4. The proposed solution methodology parameters of computing the conductor energy loss annual cost are
used as given in [9], KP = 2500 Rs/kW, KE = 0.5 Rs/kW and LSF = 0.2.
The steps followed in solving the selection of conductor’s type is
shown in Fig. 4. The starting point in the proposed methodology is A. 16-bus distribution network
formulating a matrix of dimension (nf  nb) representing the initial
The system configuration is given in Fig. 5, the rated network
conductors’ types, where nf is the flock size and nb is the number of
of this network is 11 kV, the line and load data are obtained
branches as follows:
from [9]. The available conductors’ types are given in Table 1
2 3
c4 c5 c7 c3    c8 with their electrical specifications [16]. The controlling parame-
6 7 ters of the proposed CSA are given in Table 2. The bus voltage
6 c2 c7 c1 c3  c5 7
6 7 of the original network and that optimized via the CSA are
6  c7 7
xInitial ¼ 6 c8 c3 c5 c6 7 ð9Þ shown in Fig. 6.
6. .. .. .. .. .. 7
6. 7 The algorithm is applied on 16-bus system and the obtained
4. . . . . . 5
results are compared to those obtained via EP [9] and HSDE [16];
c6 c5 c8 c4 c2 c1
this comparison is given in Table 3. It is clear that, the network
where ci is the conductor of type i. In this work; the number of power loss optimized by the proposed CSA is 35.6553 kW which
conductor types used is eight, nc = 8, load flow analysis based on is less than the others; this means the power loss is reduced by
forward sweep algorithm described in [19,20] is performed and 33.32% than the original network loss. The optimal annual cost
the corresponding fitness function is calculated. The CSA steps obtained by the proposed methodology is 175207.4555 Rs./year
shown in Fig. 3 are performed and repeated until the problem which will save about 24.48% than the original network cost. The
constraints are achieved then the optimal solution is obtained. response of the CSA is given in Fig. 7. Fig. 8(a) shows the current
The main contribution of this paper is to adapt the CSA to be flow in branches and their maximum limits while Fig. 8(b) shows
applicable in solving the optimization problem under study. The the comparison between the branch power loss in the original
design variables are the conductor’s types therefore; the updating and optimized networks, it’s clear that the current flow doesn’t
process is performed via two stages, the first one is done using Eq. exceed the line capacity; additionally the power loss in each
(8) and then the flock position is updated by the following pro- branch in the optimized network is less than that of the original
posed equation. network.

Crow i Crow i Crow j

xik xik
Crow j
mjk
xik+1

mjk xik+1

Origin Origin

(a) If fl <1 (b) If fl >1

Fig. 2. The effect of the crow flight length on search process.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
4 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Start

Initialize the position of a flock of N crows (xi0)

k=1

i=1

Define fli k, pi k of all crows

Select randomly a crow (j)

No Yes
rj > p i k

Update the position of Update the position of

crow i randomly crow i using eqn. (7)

No
Last crow?
i=i+1

Yes
Calculate the fitness function Fk (xk)

No
k=k+1 Fk (xk) < Fk-1 (xk-1)

Yes
Update the memory of crows

End

Fig. 3. The flowchart of CSA.

Finally; one can derive that the proposed CSA is succeeding in tical parameters of the CSA response are tabulated in Table 4. The
finding the optimal conductor type incorporated in the radial net- highlighted case is the base one which gives the best power loss
work to minimize the total annual cost of the network operation. and standard division. It is clear that as the maximum iteration
is increased, the optimal cost is decreased. The responses of CSA
B. Sensitivity analysis of changing CSA controlling parameters for some cases are shown in Fig. 9. Referring to Fig. 9; one can
derive that for maximum iteration equal to 1000 and 2500; the
It is important to show how the controlling parameters of the optimal solution is obtained after 250 iterations while for the
proposed algorithm affect on the optimal solution obtained for two other cases, the algorithm consumes large time to obtain the
16-bus system therefore, in this section the effects of CSA control- optimum solution.
ling parameters on the optimal solution are studied, the studied
parameters are (maximum iteration, flight length, awareness pos- ii. The effect of flight length on optimal solution
sibility and population size).
The flight length has significant impact on the CSA performance
i. The effect of maximum iteration on optimal solution as it said before; less flight length leads to local optimum solution
while large flight length helps in obtaining the global optimum
It is assumed that the maximum iteration is changed from 500, solution. Different cases are studied and tabulated in Table 5. The
base case, to 5000 in step 500. The obtained results and the statis- results ensure that larger values of flight length lead to global

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx 5

Start

Input the data of

available conductors and
the network data

Input a vector of conductor types for

original network

Perform load flow

Update the vector Perform the CSA shown in Fig. 3

of conductor types

No
Constraints achieved?

Yes
Optimal solution is obtained

End

Fig. 4. The proposed solution methodology flowchart.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

SS

Fig. 5. Configuration of 16-bus radial distribution network.

Table 1
The electrical specifications of the available conductors.

Conductor type A (mm2) R (X/km) X (X/km) Imax (A)

Squirrel 12.9 1.376 0.3896 115
Gopher 15.91 1.098 0.31 138
Weasel 19.55 0.9108 0.3797 150
Ferret 25.87 0.6795 0.298 180
Rabbit 32.26 0.5441 0.3673 208
Mink 37.32 0.4565 0.285 226
Beaver 42.07 0.3841 0.2795 250
Raccoon 48.39 0.3657 0.3579 270

Table 2 leads to local optimum solution shown in the first three figures
The controlling parameters of CSA. shown in Fig. 10 while the last one ensures the large values of fl
Parameter Value
lead to global optimum solution.
Flock (population) size 50
Awareness probability 0.1
iii. The effect of awareness possibility on optimal solution
Flight length 2
Max. iteration 500 The awareness probability has important effect on the updating
process; it is assumed that this parameter is changed from 0.1 to
optimum cost. Additionally; the base case is the best due to less 0.7 in step 0.1 and the obtained results are tabulated in Table 6.
cost, power loss and standard division. The CSA responses for some It is clear that base case achieves less power loss with acceptable
cases are shown in Fig. 10. It is noted that the smaller values of fl operating cost.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
6 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

1
Original network
Optimized network via CA

0.98

0.96

Voltage (pu)
0.94

0.92

0.9

0.88
2 4 6 8 10 12 14 16
Bus number

Fig. 6. The bus voltages of the original 16-bus network and optimized one by CSA.

Table 3
The optimal results obtained for 16-bus system.

Branch number From bus To bus Load Length (km) Original network EP [9] HSDE [16] The proposed CSA
PL (kW) QL (kvar)
1 1 2 48.75 43 2.00 Raccoon Raccoon Raccoon Raccoon
2 2 3 37.5 33 1.60 Raccoon Raccoon Raccoon Raccoon
3 3 4 37.5 33 2.30 Raccoon Raccoon Raccoon Raccoon
4 4 5 75 66.1 2.90 Raccoon Raccoon Raccoon Raccoon
5 5 6 48.75 43 2.20 Rabbit Raccoon Raccoon Rabbit
6 6 7 37.5 33 1.57 Rabbit Raccoon Raccoon Raccoon
7 7 8 12 10.6 2.40 Rabbit Raccoon Raccoon Ferret
8 8 9 37.5 33 40 Rabbit Raccoon Raccoon Gopher
9 9 10 37.5 33 2.30 Weasel Raccoon Beaver Beaver
10 10 11 75 66.1 2.50 Weasel Raccoon Beaver Mink
11 11 12 18.75 16.5 2.70 Weasel Rabbit Raccoon Weasel
12 12 13 48.75 43 3.20 Squirrel Rabbit Beaver Squirrel
13 13 14 48.75 43 1.70 Squirrel Rabbit Mink Weasel
14 14 15 75 66.1 3.80 Squirrel Weasel Rabbit Squirrel
15 15 16 37.5 33 2.00 Squirrel Squirrel Rabbit Gopher
Power loss (kW) 53.4752 37.36 36.17 35.6553
Loss reduction – 30.14% 32.36% 33.32%
Optimal cost (Rs.) 231990 208796 200783 175207.4555
Net Saving – 10% 13.45% 24.48%

5
x 10
1.78

1.775

1.77
Fitness function (Rs.)

1.765

1.76

1.755

1.75
0 50 100 150 200 250 300 350 400 450 500
Iteration number

Fig. 7. The variation of fitness function with number of iterations for 16-bus network.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx 7

300 6
Branch current flow Original network
Branch capacity Optimized network via CA

250 5

200 4

Power loss (kW)

Current (A)

150 3

100 2

50 1

0 0
0 2 4 6 8 10 12 14 1 0 2 4 6 8 10 12 14 16
Bus number Bus number

(a) (b)

Fig. 8. (a) The branch current flow (b) the branch power loss.

Table 4
The effect of changing the maximum iteration on the optimal solution for 16-bus system.

Max. iterations Power loss (kW) Loss reduction (%) Optimal cost (Rs.) Mean (Rs.) Std. dev. (Rs.)
500 35.6553 33.32367 175207.4555 175,300 197.5
1000 39.4198 26.28396 172821.4429 172,900 476.1
1500 38.5467 27.91668 171859.7998 172,300 723.2
2000 38.1358 28.68507 172216.6583 172,800 532.4
2500 38.5467 27.91668 171645.7796 171,800 651.8
3000 40.5941 24.08799 172005.8296 172,200 497.8
3500 37.3550 30.14519 172201.5046 172,500 485.4
4000 36.7637 31.25094 172159.5833 172,500 461
4500 38.1261 28.70321 171964.2946 172,400 590
5000 38.6662 27.69321 172173.9796 172,400 440.7

5 5
x 10 x 10
1.78 1.8
Optimal annual cost (RS.)

Optimal annual cost (RS.)

1.77 1.78
1.76
1.76
1.75
1.74
1.74

1.73 1.72

1.72 1.7
0 200 400 600 800 1000 0 500 1000 1500 2000 2500
Iteration no. Iteration no.
5 5
x 10 x 10
1.77 1.77
Optimal annual cost (RS.)

Optimal annual cost (RS.)

1.76 1.76

1.75 1.75

1.74 1.74

1.73 1.73

1.72 1.72
0 500 1000 1500 2000 2500 3000 3500 4000 0 1000 2000 3000 4000 5000
Iteration no. Iteration no.

Fig. 9. The effect of changing the maximum iteration on CSA response.

Table 5
The effect of changing flight length on the optimal solution for 16-bus system.

fl Power loss (kW) Loss reduction (%) Optimal cost (Rs.) Mean (Rs.) Std. dev. (Rs.)
0.3 44.7409 16.33337 187878.0227 189,700 2295
0.6 44.7409 16.33337 184428.4427 186,800 3019
0.9 44.7409 16.33337 187092.2727 189,900 3068
2.0 35.6553 33.32367 175207.4555 175,300 197.5
4.0 39.7409 25.68349 175366.0477 184,100 5471
6.0 40.8409 23.62647 179893.3977 183,700 5147

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
8 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

5 5
x 10 fl=0.3 x 10 fl=0.6
2.05 2.05

Optimal annual cost (Rs.)

Optimal annual cost (Rs.)

2
2

1.95
1.95
1.9

1.9
1.85

1.85 1.8
0 100 200 300 400 500 0 100 200 300 400 500
Iteration no. Iteration no.
5 fl=0.9 5 fl=6
x 10 x 10
2.05 2.05
Optimal annual cost (Rs.)

Optimal annual cost (Rs.)

2
2
1.95

1.95 1.9

1.85
1.9
1.8

1.85 1.75
0 100 200 300 400 500 0 100 200 300 400 500
Iteration no. Iteration no.

Fig. 10. The effect of changing the flight length on CSA response.

Table 6
The effect of changing awareness probability on the optimal solution for 16-bus system.

P Power loss (kW) Loss reduction (%) Optimal cost (Rs.) Mean (Rs.) Std. dev. (Rs.)
0.1 35.6553 33.32367 175207.4555 175,300 197.5
0.2 39.8883 25.40785 174759.6808 175,800 1117
0.3 39.1482 26.79186 175184.9592 175,400 757.7
0.4 38.3292 28.32341 174924.9637 175,800 864.6
0.5 37.8389 29.24028 175253.192 175,600 797.7
0.6 39.5331 26.07209 175041.7628 175,500 868.3
0.7 40.1360 24.94465 174872.04 175,400 752.2

The responses of CSA at different values of awareness probabil- The conductor types of original network, optimized network via
ity are shown in Fig. 11, once the value of P is increased; the algo- HAS, HSDE [16] and the system optimized via the proposed CSA
rithm failed in catching the global optimum solution and falling in are tabulated in Table 7. Referring to the obtained results; the power
local optima. loss of the network optimized via CSA is 254.1520 kW which saves
Finally one can get the best power loss, the best standard divi- the power loss about 19.53% compared to the original network; addi-
sion; the best statistical parameters and acceptable optimal oper- tionally the total annual cost is 927786.3683 Rs./year which will
ating cost are obtained in base case. Therefore the analysis save about 16.17% than the original network cost. It is clear that
performed in 86-bus system is done using the controlling parame- the results obtained via the proposed CSA are better than those
ters given in Table 2 (base case). obtained via the previous ones. Fig. 13 shows the bus voltages of
the original and optimized networks. The bus voltage of the opti-
C. 85-bus distribution network mized network is improved than the original one. To ensure the
validity of the obtained results; the current flow of each branch is
The methodology is applied on a large-scale distribution net- compared to the maximum allowable capacity of it; this comparison
work of 85-bus, 11 kV. The data of the system is given in [21]; is given in Fig. 14. It is shown that the obtained results confirm the
the configuration of 85-bus radial network is shown in Fig. 12. current flow in each branch doesn’t exceed the allowable limit.

5
x 10
1.86
Pa=0.2
Pa=0.3
Pa=0.4
1.84 Pa=0.5
Pa=0.6
Pa=0.7
Optimal Annual cost (Rs.)

1.82

1.8

1.78

1.76

1.74
0 50 100 150 200 250 300 350 400 450 500
Iteration no.

Fig. 11. The effect of changing the awareness probability on CSA response.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx 9

1
2
3 16
4 17
5

18 6
19
7
20 23
21 24 8
22 9 25
57 10 26
58
59 11 78 37 27
60
12 38
61 28
62 13 29
85 14 80 39 30
63
15 81
31
64 83 82 32
84 40
65 33
41
43 42 34
66 77 67 44
72 68 79 45 35
46
73 69 47 36 48
74 70 52 49
76 56 50
75 71
55 53 51
54
Fig. 12. Configuration of 85-bus radial distribution network.

Table 7
The optimal results obtained for 85-bus system.

Branch number From bus To bus Load Length (km) Original network HSA [16] HSDE [16] The proposed CSA
PL (kW) QL (kvar)
1 1 2 0 0 0.29 Raccoon Raccoon Raccoon Mink
2 2 3 0 0 0.43 Raccoon Raccoon Raccoon Ferret
3 3 4 56 54.8912 0.58 Raccoon Raccoon Raccoon Squirrel
4 4 5 0 0 0.29 Raccoon Raccoon Raccoon Mink
5 5 6 35.28 34.58145 1.16 Raccoon Raccoon Raccoon Gopher
6 6 7 0 0 0.73 Raccoon Raccoon Raccoon Squirrel
7 7 8 35.28 34.58145 3.19 Raccoon Raccoon Raccoon Squirrel
8 8 9 0 0 0.29 Squirrel Raccoon Raccoon Ferret
9 9 10 56 54.8912 0.39 Squirrel Rabbit Beaver Ferret
10 10 11 0 0 0.39 Squirrel Rabbit Raccoon Beaver
11 11 12 0 0 0.42 Squirrel Rabbit Rabbit Gopher
12 12 13 35.28 34.58145 0.19 Squirrel Ferret Weasel Raccoon
13 13 14 35.28 34.58145 0.23 Squirrel Ferret Squirrel Raccoon
14 14 15 35.28 34.58145 0.52 Squirrel Squirrel Squirrel Mink
15 2 16 112 109.7824 0.32 Squirrel Rabbit Raccoon Weasel
16 3 17 56 54.8912 0.58 Squirrel Ferret Beaver Gopher
17 5 18 56 54.8912 0.45 Squirrel Mink Rabbit Rabbit
18 18 19 35.28 34.58145 0.32 Squirrel Squirrel Mink Raccoon
19 19 20 35.28 34.58145 0.58 Squirrel Weasel Squirrel Raccoon
20 20 21 35.28 34.58145 1.1 Squirrel Rabbit Gopher Gopher

(continued on next page)

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
10 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

Table 7 (continued)

Branch number From bus To bus Load Length (km) Original network HSA [16] HSDE [16] The proposed CSA
PL (kW) QL (kvar)
21 21 22 56 54.8912 0.13 Squirrel Beaver Squirrel Beaver
22 19 23 35.28 34.58145 0.64 Squirrel Ferret Weasel Raccoon
23 7 24 35.28 34.58145 0.32 Squirrel Ferret Weasel Squirrel
24 8 25 56 54.8912 0.26 Squirrel Raccoon Beaver Beaver
25 25 26 0 0 0.39 Squirrel Beaver Raccoon Weasel
26 26 27 56 54.8912 0.19 Squirrel Mink Beaver Ferret
27 27 28 0 0 0.39 Squirrel Mink Raccoon Weasel
28 28 29 35.28 34.58145 0.39 Squirrel Beaver Raccoon Squirrel
29 29 30 35.28 34.58145 0.19 Squirrel Beaver Beaver Beaver
30 30 31 0 0 0.13 Squirrel Ferret Raccoon Squirrel
31 31 32 14 13.7228 0.13 Squirrel Raccoon Raccoon Weasel
32 32 33 0 0 0.58 Squirrel Raccoon Rabbit Gopher
33 33 34 0 0 0.45 Squirrel Beaver Raccoon Squirrel
34 34 35 35.28 34.58145 0.13 Squirrel Mink Mink Rabbit
35 35 36 56 54.8912 0.26 Squirrel Mink Raccoon Rabbit
36 26 37 56 54.8912 0.71 Squirrel Ferret Squirrel Squirrel
37 27 38 56 54.8912 0.39 Squirrel Gopher Weasel Squirrel
38 29 39 35.28 34.58145 0.32 Squirrel Squirrel Beaver Rabbit
39 32 40 0 0 0.71 Squirrel Squirrel Gopher Squirrel
40 40 41 35.28 34.58145 0.19 Squirrel Squirrel Gopher Gopher
41 41 42 35.28 34.58145 0.32 Squirrel Ferret Mink Gopher
42 41 43 35.28 34.58145 0.71 Squirrel Mink Gopher Ferret
43 34 44 35.28 34.58145 0.65 Squirrel Ferret Raccoon Weasel
44 44 45 35.28 34.58145 0.65 Squirrel Beaver Ferret Gopher
45 45 46 14 13.7228 0.39 Squirrel Weasel Beaver Squirrel
46 46 47 0 0 0.45 Squirrel Gopher Ferret Weasel
47 35 48 0 0 0.13 Squirrel Ferret Rabbit Beaver
48 48 49 36.28 35.56165 0.26 Squirrel Weasel Squirrel Rabbit
49 49 50 56 54.8912 0.32 Squirrel Beaver Ferret Weasel
50 50 51 0 0 0.97 Squirrel Weasel Ferret Weasel
51 48 52 35.28 34.58145 0.32 Squirrel Gopher Ferret Raccoon
52 52 53 56 54.8912 0.39 Squirrel Mink Gopher Weasel
53 53 54 56 54.8912 0.39 Squirrel Gopher Squirrel Squirrel
54 52 55 14 13.7228 0.39 Squirrel Rabbit Beaver Squirrel
55 49 56 56 54.8912 0.19 Squirrel Beaver Ferret Weasel
56 9 57 0 0 0.58 Squirrel Mink Raccoon Gopher
57 57 58 56 54.8912 0.13 Squirrel Rabbit Raccoon Mink
58 58 59 56 54.8912 0.39 Squirrel Ferret Squirrel Gopher
59 58 60 56 54.8912 0.52 Squirrel Rabbit Rabbit Gopher
60 60 61 56 54.8912 0.71 Squirrel Ferret Ferret Squirrel
61 61 62 14 13.7228 0.13 Squirrel Gopher Rabbit Squirrel
62 60 63 0 0 0.52 Squirrel Weasel Raccoon Ferret
63 63 64 0 0 0.13 Squirrel Ferret Beaver Raccoon
64 64 65 56 54.8912 0.13 Squirrel Gopher Raccoon Beaver
65 65 66 0 0 0.32 Squirrel Mink Squirrel Raccoon
66 64 67 0 0 0.64 Squirrel Rabbit Beaver Squirrel
67 67 68 56 54.8912 0.77 Squirrel Rabbit Ferret Ferret
68 68 69 0 0 0.32 Squirrel Rabbit Rabbit Weasel
69 69 70 35.28 34.58145 0.39 Squirrel Gopher Rabbit Ferret
70 70 71 56 54.8912 0.13 Squirrel Weasel Squirrel Mink
71 67 72 0 0 0.84 Squirrel Ferret Raccoon Gopher
72 68 73 56 54.8912 0.19 Squirrel Squirrel Gopher Rabbit
73 73 74 35.28 34.58145 0.71 Squirrel Beaver Raccoon Raccoon
74 73 75 56 54.8912 0.39 Squirrel Squirrel Squirrel Rabbit
75 70 76 14 13.7228 0.06 Squirrel Rabbit Weasel Mink
76 65 77 56 54.8912 0.45 Squirrel Ferret Squirrel Weasel
77 10 78 35.28 34.58145 0.39 Squirrel Rabbit Weasel Raccoon
78 67 79 56 54.8912 0.52 Squirrel Squirrel Squirrel Beaver
79 12 80 56 54.8912 0.39 Squirrel Weasel Ferret Rabbit
80 80 81 0 0 0.26 Squirrel Squirrel Gopher Rabbit
81 81 82 56 54.8912 0.06 Squirrel Rabbit Beaver Beaver
82 81 83 35.28 34.58145 0.77 Squirrel Beaver Weasel Raccoon
83 83 84 14 13.7228 0.71 Squirrel Weasel Mink Squirrel
84 13 85 35.28 34.58145 0.58 Squirrel Weasel Squirrel Weasel
Power loss (kW) 315.84 265.22 256.96 254.1520
Loss reduction – 16.03% 18.64% 19.53%
Optimal cost (Rs.) 1,106,732 940,693 929,035 927786.3683
Net saving 15.00% 16.06% 16.17%

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx 11

1
Original Network
Optimized network via CA

0.98

0.96

Voltag (pu)
0.94

0.92

0.9

0.88

0.86
0 10 20 30 40 50 60 70 80 90
Bus number

Fig. 13. The bus voltages of the original 85-bus network and optimized one by CSA.

300
Branch current flow
Branch cpapcity

250

200
Current (A)

150

100

50

0
0 10 20 30 40 50 60 70 80
Bus number

Fig. 14. The branch current and the maximum allowable branch capacity for 85-bus system.

The proposed methodology based on CSA is valid, efficient and one the power loss is saved by about 19.53% and the total cost is
superior in evaluating the optimal conductors in the radial distri- saved by 16.17%. The proposed CSA are compared to the previous
bution networks so as to minimize the summing of the conductor algorithms and the obtained results show the superiority and effi-
capital and power loss costs in both small-scale and large-scale ciency of the proposed algorithm in solving the problem of optimal
networks. conductor selection in distribution networks.

References
6. Conclusion
[1] A.S. Chambiya, N. KumarSadana, S. Mahajan, A method of selection of optimal
This work aims the optimal design of the radial distribution net- size of conductor in radial distribution system – a review, Int. J. Adv. Technol.
work by optimal selection of the suitable conductor such that min- Eng. Sci. 2 (06) (2014) 190–298.
[2] M. Thenepalle, A comparative study on optimal conductor selection for radial
imizes the capital and energy loss costs. A novel solution distribution network using conventional and genetic algorithm approach, Int. J.
methodology based on crow search algorithm (CSA) is presented. Comp. Appl. 17 (2011) 6–13.
The updating process in the CSA is modified to be adaptive to solve [3] S. Sivanagaraju, N. Sreenivasulu, M. Vijayakumar, T. Ramana, Optimal
conductor selection for radial distribution systems, Electr. Power Syst. Res.
the problem under study. The objective function presented in this 63 (2002) 95–103.
work is the sum of conductor capital cost and the conductor energy [4] S. Satyanarayana, T. Ramana, G.K. Rao, S. Sivanagaraju, Improving the
loss cost. The proposed constraints are the bus voltage limits and maximum loading by optimal conductor selection of radial distribution
systems, Electr. Power Compon. Syst. 34 (2006) 747–757.
the max current carrying capacity of conductor. The analysis is per-
[5] Ma.M. Legha, H. Javaheri, M.Mo. Legha, Optimal conductor selection in radial
formed on two different topologies radial networks, the first net- distribution systems for productivity improvement using genetic algorithm,
work is 16-bus, 11 kV network and the second is 85-bus, 11 kV Iraq J. Electri. Electron. Eng. 9 (1) (2013) 29–35.
network. In the first one; the network power loss is saved by [6] Ma.M. Legha, H. Javaheri, Combination of optimal conductor selection and
capacitor placement in radial distribution systems for productivity
33.32% compared to the original network loss and the annual cost improvement using genetic algorithm, Int. Res. J. Appl. Basic Sci. 6 (10)
is saved by 24.48% compared to the original while in the second (2013) 1415–1422.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004
12 A.Y. Abdelaziz, A. Fathy / Engineering Science and Technology, an International Journal xxx (2017) xxx–xxx

[7] D. Sharma, P. Jha, S. Vidyasagar, Optimal conductor selection using genetic [14] S. Sivanagaraju, J. Viswanatha Rao, Optimal conductor selection in radial
algorithm, Int. J. Eng. Res. Technol. 2 (4) (2013) 660–664. distribution system using discrete particle swarm optimization, World J.
[8] F. Mendoza, D. Requena, J. L. Bernal-Agustin, J. A. Dominguez-Navarro, Optimal Model. Simul. 5 (2) (2009) 96–104.
conductor size selection in radial power distribution systems using [15] D. Kaur, J. Sharma, Optimal conductor sizing in radial distribution systems
evolutionary strategies, in: IEEE PES Transmission and Distribution planning, Electr. Power Energy Syst. 30 (2008) 261–271.
Conference and Exposition, Latin America, Venezuela, 2006, pp. 1–5. [16] R.S. Rao, K. Satish, S.V.L. Narasimham, Optimal conductor size selection in
[9] R. Ranjan, A. Chaturvedi, P. S. Solanki, D. Das, Optimal conductor selection of distribution systems using the harmony search algorithm with a differential
radial distribution feeders using evolutionary programming, in: Conference on operator, Electr. Power Compon. Syst. 40 (2012) 41–56.
Convergent Technologies for the Asia-Pacific Region, vol. 1, 2003, pp. 456–459. [17] R. Ranjan, B. Venkatesh, D. Das, A new algorithm for power distribution system
[10] B.M. Kalesar, Conductor selection optimization in radial distribution system planning, Electr. Power Syst. Res. 62 (2002) 55–65.
considering load growth using MDE algorithm, World J. Model. Simul. 10 (3) [18] A. Askarzadeh, A novel metaheuristic method for solving constrained
(2014) 175–184. engineering optimization problems: crow search algorithm, Comput. Struct.
[11] Ma.M. Legha, R.A. Sangrood, A.Z. Raeiszadeh, Mo.M. Legha, Conductor size 169 (2016) 1–12.
selection in planning of radial distribution systems for productivity [19] D. Das, D. Kothari, A. Kalam, Simple and efficient method for load flow solution
improvement using imperialist competitive algorithm, Int. J. Techn. Phys. of radial distribution network, Int. J. Electr. Power Energy Syst. 15 (5) (1995)
Prob. Eng. 5 (15) (2013) 65–69. 335–346.
[12] Ma.M. Legha, H. Noormohamadi, A. Barkhori, Optimal conductor selection in [20] U. Eminoglu, M.H. Hocaoglu, A new power flow method for radial distribution
radial distribution using bacterial foraging algorithm and comparison with ICA systems including voltage dependent load models, Electr. Power Syst. Res. 76
method, WALIA J. 31 (3) (2015) 37–43. (2005) 106–114.
[13] Ma.M. Legha, F. Ostovar, Mo.M. Legha, Combination of optimal conductor [21] M.M. Hamada, M.A.A. Wahab, N.G.A. Hemdan, Simple and efficient method for
selection and capacitor placement in radial distribution systems using PSO steady-state voltage stability assessment of radial distribution systems, Electr.
method, Iraq J. Electr. Electron. Eng. 10 (1) (2014) 33–41. Power Syst. Res. 80 (2010) 152–160.

Please cite this article in press as: A.Y. Abdelaziz, A. Fathy, A novel approach based on crow search algorithm for optimal selection of conductor size in
radial distribution networks, Eng. Sci. Tech., Int. J. (2017), http://dx.doi.org/10.1016/j.jestch.2017.02.004