Journal of Engineering Science and Technology Vol. 14, No. 1 (2019) 087 - 107 © School of Engineering, Taylor’s University
87
OPTIMAL CONDUCTOR SELECTION IN RADIAL DISTRIBUTION SYSTEMS USING WHALE OPTIMIZATION ALGORITHM
SHERIF M. ISMAEL1, SHADY H. E. ABDEL ALEEM2,*, ALMOATAZ Y. ABDELAZIZ3
1Engineering for the Petroleum and Process Industries (Enppi), Cairo, Egypt 215th of May Higher Institute of Engineering, Mathematical and Physical Sciences,
Helwan, Cairo, Egypt 3Faculty of Engineering and Technology, Future University in Egypt, Cairo, Egypt
*Corresponding Author: [email protected]
Abstract
Nowadays, electrical power system networks are driven harder, and they are
required to deliver more energy. Electrical losses reduction is one of the most
important ways to conserve the generated energy, especially in the distribution
systems. In this regard, the optimal conductor selection can reduce the electrical
power losses, while enhancing the voltage profile in a cost-effective manner. In
this paper, a novel approach based on a recent meta-heuristic algorithm, known
as whale optimization (WO) algorithm is proposed to solve the optimal conductor
selection problem of radial distribution networks. An updated practical
conductor’s library is introduced. Further, practical techno-economic aspects are
considered such as load growth considerations and payback period calculations.
The objective function is to minimize the combined cost of energy loss and
conductors’ investment cost. The considered constraints are the bus voltage limits
and the conductors’ current carrying capacities. The proposed approach is applied
to two different systems; the first one is a 16-bus small-scale system and the
second is a large-scale 85-bus system. The obtained results are compared with
other results available in the literature, and showed the effectiveness of the
proposed algorithm in reducing the network losses, maximizing the overall
saving, while maintaining the specified constraints over almost a five-year period
while taking into account high annual load growth rate.
Keywords: Optimal conductor selection; Whale optimization algorithm; Radial
distribution networks.
88 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
1. Introduction
Due to the gap between the increasing electrical demand and the limited power
generation resources, researchers focused on conserving the electric energy in line with
increasing the generated power to cater for this continuous growth in electrical
demands. Power loss reduction is considered as one of the most important ways of
conserving the generated energy. The significant power losses in the overall power
system networks are concentrated in the distribution systems because they are usually
characterized by their reduced voltage levels, high lines’ currents, and low lines’
resistances [1, 2]. Researchers have explored many methodologies for power losses
reduction in distribution networks such as shunt capacitors placement, network
reconfiguration, distributed generation (DG) allocation, and optimal conductor
selection. Optimal selection of the conductor sizes of a distribution network is a
complicated problem as it should take into account many constraints such as the
specified voltage limits for the different buses, current carrying capacity of the feeders,
loading profile alteration, and load growth, as well as other economic considerations
that concern with cost of losses and conductors, and interest and depreciation rates.
In the literature, the idea of selecting the optimal conductor sizes had been
tackled years ago [1-4]. Funkhouser and Huber [1], in 1955, introduced the idea of
the selection of the Aluminium Conductor Steel Reinforced (ACSR) sizes based on
economic considerations. Ponnavaikko and Rao [2] presented a mathematical
method for the selection of the optimal set of conductors considering a non-uniform
distribution of loads, the maximum permissible voltage drop, and load growth in
future years. Tram and Wall [3] completed the work of [2] by introducing a
computerized methodology for the selection of optimal conductor sizes. Ranjan et
al. [4] proposed an optimization algorithm for the optimal conductor sizes selection
using evolutionary programming (EP).
One can see that the methodologies presented in the literature for solving the
optimal conductor selection problem can be categorized into two main approaches;
the first one is based on conventional analytical algorithms [5-14] and the second
is based on new intelligent meta-heuristic algorithms [15-20].
The analytical approaches have been utilized effectively in the optimal
conductor selection. Islam and Ghani [5] proposed an analytical algorithm for
optimal selection of conductors in radial distribution networks. Wang et al. [6]
introduced a practical procedure for optimal conductor selection considering the
non-uniform loading. Sivanagaraju et al. [7] proposed an analytical algorithm based
on radial network load flow to select the optimal conductor size in order to
maximize saving in energy loss and conductor investment costs. Mandal and Pahwa
[8] presented a method for selection of the optimal set of conductors considering
both technical and economic constraints. Falaghi et al. [9] introduced an analytical
algorithm for optimal selection of conductors in radial distribution networks
considering many factors such as the capital cost of conductors, cost of energy
losses, bus voltage profile, current carrying capacity of conductors, and load growth
within a specific period. Satyanarayana et al. [10] presented a model for improving
the maximum allowable loading of radial distribution feeders for different types of
load models based on the optimal selection of the conductors and incorporating
load growth profile. Kaur and Sharma [11] developed a generalized model for
optimal conductor size selection considering the non-uniform loading pattern, load
growth, load factor and diversity in load peaks. Raju et al. [12] proposed an
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 89
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
analytical method for conductor selection based on the cost of energy losses and
the interest and depreciation rates on the investment. Abul’Wafa [13] presented a
generalized model for optimal conductor size selection based on branch wise
minimization (BWM) technique. Franco et al. [14] developed an optimization
algorithm based on a mixed-integer linear programming (MILP) model for optimal
conductor selection in radial distribution systems.
Mendoza et al. [15], on the other side, the intelligent optimization techniques
have inspired many researchers to employ it for solving the optimal conductor
selection problem, such as evolutionary strategy (ES), discrete particle swarm
optimization (DPSO) [16], genetic (GA) [17], harmony search (HS) with a
differential operator [18], analytic and GA [19], Grasshopper Optimization
algorithm (GOA) [20], Crow search algorithm (CSA) [21] and sine-cosine
optimization algorithm (SCA) [22]. The practical selection of the optimal
conductors set should consider present loads in addition to the expected load growth
for a certain time span [23-25].
Optimal conductor selection can allow additional penetration of DGs into
distribution networks. Nowadays, the maximum amount of DG units that can be
integrated into the distribution system, without violating the system operational
performance limits, is called the hosting capacity (HC). One of the approaches used
by network operators to face the challenges of continuous load growth and high
DG penetrations is to reinforce the network. It is believed that network
reinforcement and optimal conductor selection are considered as effective
techniques for HC enhancement that will play a significant role in future power
systems and smart grids [26, 27].
The WO algorithm is proposed in this work due to the intelligent behaviour of
crows in storing their excess food, hiding their food place from others, and bringing
their food back when they need [28]. The WO algorithm has multiple advantages
such as having few parameters to be set, faster convergence capability and higher
sensitivity when compared to the widely known meta-heuristic algorithms. Due to
these advantages, it has been recently employed to solve many engineering
problems in the literature [29, 30].
In this paper, a novel approach based on a recent meta-heuristic algorithm, whale
optimization (WO) algorithm [28] is proposed to solve the optimal conductor selection
problem of radial distribution networks. From the literature survey, it is clear that
application of WO has not been discussed so far to solve the optimal conductor selection
problem in radial distribution systems. This encourages utilizing this algorithm for the
problem. A practical conductor’s library containing twenty conductor types is
introduced based on actual manufacturer data that comply with the BS 50182 [31]. This
rich library explores a wider search space and ensures finding the most optimal set of
conductors that satisfy the economic objective functions while complying with the pre-
set constraints. A constrained objective function is employed to minimize overall cost,
and comply with the system voltage limits and the conductors’ current carrying
capacities. Practical aspects are considered such as available market conductors
utilization, load growth considerations, and payback period calculations. The proposed
WO algorithm is applied on two different test systems, 16-bus system, and 85-bus
system. The achieved results are compared with other methods available in the literature
and showed the effectiveness of the proposed algorithm in reducing the network losses,
90 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
maximizing the overall saving, while maintaining the specified constraints over a five-
year period taking into account high annual load growth rate.
2. Problem formulation
2.1. Objective function
In optimal conductor selection problem, the goal is to select conductor size and type
from a set of available inventory such that the total cost is minimized while
satisfying nonlinear constraints on voltage limits of the buses and maximum
ampacities of the conductors. The cost consists of the annual cost of energy losses
and the capital investment cost of the conductor which is formulated as follows:
(1) , , , Cost j c CL j c CC j c
(2), , loss p eCL j c P j c k k LSF T
The loss factor which is defined as the ratio of energy loss in the system during
a given time period to the energy loss that could result if the system peak loss had
persisted throughout that period. LSF is expressed in terms of the load factor, the
average load to the peak demand in a given time period, as follows [16, 18]:
2
0.2 0 (3).8 LSF LF LF
Thus, the total annual cost of the energy loss can be expressed as follows:
( , 4)
b ntotal
loss p e
j c
CL P j c k k LSF T
The capital investment cost of the conductor is defined in the terms of the annual
depreciation of the capital cost of the jth branch with the cth conductor, thus:
, ( ) 5CC j c IDF l j A c IC c
so that,
( 1)
( (6)
1) 1
F
F
i iIDF
i
The objective function (OF) for the optimal selection of types and sizes of the
conductors can be expressed as:
OF min (7)total totalCL CC
2.2. Constraints
The considered constraints can be expressed, as follows:
Bus voltage constraint
, ) 8, ( min maxU m U m c U m m k c n
where Umin(m) and Umax(m) are considered as 0.9 p.u. and 1.1 p.u., respectively.
Conductor current carrying capacity constraint
, , 9 ( )maxI j c I c j b c n
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 91
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
2.3. Additional techno-economic considerations
Load growth
Practically, optimal conductor sizes selection should consider the existing loads in
addition to possible load growth for a certain planning period [10]. This period is
usually determined by the feeder ability to accept load growth without violating the
node voltage or the branch's current constraints. During the ‘pre-determined’
planning period, the load growth can be considered as an annual growth rate in
proportion to the connected loads in the base year (1st year). Mathematically, the
active and reactive power loads at the Nth year is given by:
1 for = 1, 2, 3,... (10)
1 for = 1,...
1 for = 1, 2, 3,...
1
1
(11) 1 1 for =
1
1,...
N
L
M
L
N
L
M
L
L
L
P g N M
P g N M F
Q g
P N
Q NN M
Q g N M F
Equations (8) and (9) of the proposed constraints shall be checked initially in
the base year, and then the load growth shall be calculated according to Eqs. (10)
and (11). Further, the loading pattern shall increase gradually till violation of any
of the constraints occurs. In this condition, the breaking year ‘M’ is reached and the
maximum duration that the feeder can handle the load growth without violating the
preset constraints is determined. Beyond the breaking year, the optimally selected
conductors can accept additional load growth until the end of their lifetime, but
with the support of external compensators such as the shunt capacitors, DGs, and
energy storage schemes. Additionally, Fig. 1 shows the load growth pattern during
the conductor’s lifetime. Table 1 gives the numerical values of the parameters used
in the formulation of the objective function and the constraints.
Fig. 1. Load growth pattern during conductor lifetime.
Table 1. Numerical values of the used parameters.
Parameter Value Parameter Value
kp (Rs/kW) 2500 i (%) 8
ke (Rs/ kWh) 0.5 F (years) 25
IC(c) (Rs/mm2/km) 500 IDF 0.1
LF 0.4 g (%) 10
LSF 0.2
Load
PL(N)
PL(1)
N ……
Load growth is possible with alternative solutions
Feeder cannot accept load growth due to
Constraint
violation
breaking
year
M1 year
base
year
planned
lifetime
F
Feeder can accept load growth
92 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Payback period
The optimal selection of distribution system conductors may result in purchasing of larger
conductor sizes which means additional conductors’ cost. This additional cost should be
justified to the planners of the distribution system (decision makers), as follows:
Optimally selected conductors will result in reduced overall network losses,
therefore allows for more rigid load growth plans.
Reducing the network losses will result in enhanced system performance and
will save annual energy consumption
The cumulative annual energy saving will equalize with the additional initial
conductors cost in a certain period known as the payback period. After this
period, the annual energy saving is purely gained by the network operators.
Considering that optimal conductors’ selection led to increasing the original
system conductor size from the base case area say Ax (j,c) to the optimized area Ay
(j,c). Accordingly, the procedure for calculating the payback period can be
presented as follows:
The total cost of the original system conductors can be formulated as follows:
, ( 2) 1
bor g
x
j
iC A j c l j IC c
The total cost of the optimized system conductors can be formulated as follows:
, (13)
j
bOpt
yC A j c l j IC c
The total additional cost due to optimal conductor selection is:
(14)total Opt origC C C
The cost of annual energy saving due to optimal conductor selection for all
network branches is:
(15)( )x x Optbase
as loss elossC P P T K
(16 )
total
as
CPP
C
3. Whale Optimization Algorithm (WO)
Nature-inspired meta-heuristic algorithms have shown surprisingly efficient results to
tackle difficult problems. Mirjalili and Lewis [28] developed n this domain, the WO,
which is a nature-inspired meta-heuristic optimization algorithm based on mimicking
the hunting behaviour of humpback whales. Whales are considered as highly intelligent
animals with emotions. Whales have common cells in their brains similar to those of
human called spindle cells. These cells are responsible for judgment, emotions, and
social behaviours in humans. The hunting technique of the humpback whales is called
bubble-net feeding technique. Humpback whales prefer to hunt small fiishes (preys)
close to the surface. When the humpback whale detects its prey, it dives around 12 m
down and then starts to create bubbles in a spiral shape around the prey. The prey fears
to cross these bubbles that appear as a trap. At that moment, the whale swims up to the
surface and collects his trapped prey as shown in Fig. 2.
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 93
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
(a) Illustrative scheme
for the technique
(b) Real scene of the
hunting process
Fig. 2. Bubble-net hunting technique of the humpback whales.
3.1. Mathematical model
The bubble-net feeding technique of the humpback whales is mathematically
modelled in three stages [28], as briefed below.
i. Encircling the prey
ii. Bubble-net hunting method (exploitation phase)
a. Shrink mechanism
b. Spiral update of the position
iii. Globalization of the search (exploration phase)
3.1.1. Encircling the prey
Humpback whales can recognize the location of prey and encircle them. Since the
position of the optimal design in the search space is not known in advance, the WO
algorithm assumes that the current search agent is the target prey or is close to the
optimum one. After the best search agent is defined, the other search agents will try
to update their positions towards the best search agent which is represented by the
following equations:
. ( ) ( 7)( 1)C X t X tD
( ) . ( 1) (18)X t X t A D
The vectors A and C are calculated as follows:
2 . (19)a r aA
2 . (20)rC
3.1.2. Bubble-net hunting technique
a. Shrink mechanism
This behaviour is achieved by decreasing the value of a in Eq. (19) from 2 to 0.
Accordingly, the value of A is also decreased by a random value in the interval [-
94 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
a, a]. Figure 3 shows the possible positions from the random whale's position at (X,
Y) toward the prey position at (X*,Y*) that is applied to (0 ≤ A ≤ 1).
b. Spiral update of the position
In this stage, the attacking mechanism is executed and the distance between the
whale located at (X, Y) and the prey located at (X*,Y*) is calculated. Then, a spiral
equation is created between the position of the whale and prey to mimic the helix-
shaped movement of the whales as shown in Fig. 4 and Eq. (21).
*' . . cos (( 1) (21)2 ) ( )szD e z tX Xt
*' ( ( 2) )( 2)i X t X tD
It is worth mentioning that humpback whales swim around the prey within a
shrinking circle and along a spiral-shaped path simultaneously. To model these
simultaneous movements; a 50% probability is assumed to choose between either
the shrinking or updating the spiral movements to the whales’ position, as follows:
*
' *
. if < 0.5 (23)
. . cos (2 ) ( ) ( 1)
( 1) if 0.5sz
X A D p
D e z
t
X pX tX t
t
Fig. 3. Bubble-net search technique.
Fig. 4. Spiral updating of the position.
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 95
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
3.1.3. Globalization of the search
The position of the search agent is updated according to a randomly chosen search
agent instead of the best search agent found so far. To consider this randomness;
Eqs. (17) and (18) are updated after replacing X* with Xrand, as follows:
. ( ) (2( 4) )randC X t X tD
( )( 1) ( . 25)rand t DX X At
The WO algorithm includes two internal parameters to be adjusted (A and C) depending
on the selection of the vector a . To sum up, the WO procedure is shown in Fig. 5.
Fig. 5. Flowchart for the WO algorithm.
96 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
4. Results and Discussion
In this work, twenty conductor types are used. The numerical values of the used
parameters are summarized in Table 1. Ismael et al. [22] and Leppert and Allen
[23] described, that the load flow analysis is based on a forward-sweep algorithm
is performed and the corresponding fitness function is calculated using the WO. In
order to verify the effectiveness of the proposed WO algorithm; two radial
distribution networks are examined in MATLAB platform, the first one is a small-
scale network of 16-bus system, the second one is a large-scale network of 85-bus
system.
4.1. 16-bus system
The 16-bus system configuration is shown in Fig. 6, the base voltage and apparent
power of this system is 11 kV and 100 MVA respectively. The substation bus
voltage (bus 1) is 1 p.u. and the line and load data for this system are obtained from
[4]. Practical conductor’s library is introduced based on actual manufacturer data
[33]. The available conductor types and their electrical specifications are presented
in Table 2. According to Thenepalle [19], the conductors’ library is presented in
Table 3 to be compared with the results obtained using the proposed algorithm.
Other published results for the 16 bus are excluded from the comparison because
of noncompliance with the considered constraints.
Fig. 6. Configuration of 16-bus radial distribution network.
Table 2. Electrical specifications of the used ACSR conductors.
Conductor
type A (mm2) R (Ω/km) X (Ω /km) Imax (A)
1 6.5 2.718 0.374 70
2 13 1.374 0.355 120
3 16 1.098 0.349 130 4 20 0.9116 0.345 150
5 25 0.6795 0.339 175
6 30 0.5449 0.335 200 7 40 0.4565 0.353 250
8 42 0.3977 0.327 270
9 45 0.3841 0.327 257 10 48 0.3656 0.329 260
11 50 0.3434 0.328 270
12 55 0.302 0.327 290 13 65 0.2745 0.315 305
14 80 0.2193 0.282 395
15 80 0.2214 0.268 380 16 80 0.2221 0.271 385
17 95 0.1844 0.266 425
18 110 0.1589 0.261 470 19 130 0.1375 0.256 510
20 140 0.1223 0.252 560
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 97
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Table 3. Electrical specifications of the used ACSR conductors in [19].
Conductor
type A (mm2) R (Ω/km) X (Ω /km) Imax (A)
A 12.90 1.3760 0.3896 115
B 15.91 1.098 0.3100 138
C 19.55 0.9108 0.3797 150
D 25.87 0.6795 0.2980 180
E 32.26 0.5441 0.3673 208
F 37.32 0.4565 0.2850 226
G 42.07 0.3841 0.2795 250
H 48.39 0.3657 0.3579 270
The used parameters of the proposed WO algorithm are presented in Table 4.
The obtained WO results are compared to those obtained using EP [4] and HSDE
[18] and the comparison is presented in Tables 5 and 6. It is noticed that the
network active power loss optimized by the proposed algorithm is reduced to
26.0841 kW, which is less than other published results; this means the power loss
is reduced by 51.275% compared to the original network loss. The optimal annual
cost obtained by the proposed algorithm is 180,167.364 Rs./year, which will save
about 22.34% of the original network cost with a payback period of 6.76 years.
The bus voltage values of the original network and that optimized using the
proposed algorithm are presented in Fig. 7. Figure 8 shows the current flow in
branches compared to their maximum limits. The response of the WO algorithm
is presented in Fig. 9. The load growth results are presented in Table 7. Voltage
profile and branch current flows of the 16-bus system over a five-year span are
presented in Figs. 10 and 11, respectively. From the presented figures, one can
notice that the proposed optimal conductor selection algorithm succeeded in
minimizing the overall system costs and keeping the bus voltages and branch
currents within the prescribed limits not only for the current load profile but also
for a five-year span with high annual load growth factor of 10%. As shown in
Table 7, the voltage of bus 16 begins to exceed its allowable limit (0.9 p.u.).
Therefore, other compensation techniques should be adopted to allow for more
load growth without violating the system limits.
A sensitivity analysis has been performed for selecting the optimum WO
algorithm parameters, mainly vector ā and n. A case study was examined to
calculate the losses’ cost of the 16-bus system considering various WO parameters
as presented in Table 8. Three independent values were assumed for each of ā and
n. It is assumed that all other parameters are kept constant. It was concluded that
the linear decrement of vector ā from 2 to 0 leads to improved results while other
values give worse results because they de-emphasize exploitation and exploitation
process. In addition, it was noticed that large number of search agents allow for
better reach to the global optimum.
Table 4. Numerical values of the used parameters of the proposed WO algorithm.
Parameter Value
Number of search agents (n) 100
a Linearly decreases from 2 to 0
Maximum number of iterations 500
98 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Table 5. Comparison of conductor selection results for the 16-bus system.
Branch
no.
From
bus
To
bus
EP
[4]
HSDE
[18]
The proposed
WO
1 1 2 Raccoon H 17
2 2 3 Raccoon H 17
3 3 4 Raccoon H 14
4 4 5 Raccoon H 14
5 5 6 Raccoon H 12
6 6 7 Raccoon H 12
7 7 8 Raccoon H 12
8 8 9 Raccoon H 12
9 9 10 Raccoon G 12
10 10 11 Raccoon G 8
11 11 12 Rabbit H 6
12 12 13 Rabbit G 6
13 13 14 Rabbit F 5
14 14 15 Weasel E 3
15 15 16 Squirrel E 1
Table 6. Comparison of final results for the 16-bus system.
Variables Original
system
After optimal conductor selection
EP [4] HSDE [18] The proposed WO
Umin at node 16 0.8867 0.9153 0.9288 0.9300
Power Loss (kW) 53.47 37.36 36.17 26.0841
Loss reduction (%) - 30.14 32.36 51.275
Optimal cost (Rs) 231,990 208,796 200,783 180,167.36
Net saving (%) - 10 13.45 22.34
Payback period (years) - - - 6.76
Fig. 7. Bus voltages of the original and optimized 16-bus network.
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Vo
ltag
e (
pu
)
Bus number
Original network Optimized netwok by WO
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 99
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Fig. 8. Branch currents compared to their maximum limits.
Fig. 9. The variation of fitness function for the16-bus network.
Table 7. The load growth results obtained for 16-bus system.
Year Total active
power (kW)
Total power
losses (kW)
Umin at bus
16 (p.u.)
1 675.75 26.08 0.9300
2 743.33 31.95 0.9225
3 817.66 39.63 0.9141
4 899.42 48.15 0.9048
5 989.37 59.28 0.8943
Fig. 10. Voltage profile over a five-year span for the 16-bus system.
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cu
rren
t (A
)
Branch no.
Branch current flow Branch current capacity
0.88
0.9
0.92
0.94
0.96
0.98
1
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Vo
ltag
e (p
u)
Bus number
1st Year 2nd Year 3rd Year 4th Year 5th Year
100 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Fig. 11. Branch currents over a five-year span with 10% load growth factor.
Table 8. Sensitivity analysis for the WO algorithm parameters.
n= 10 n= 40 n= 100
ā = from 1 to 0 208,972.04 195,963.53 188,010.24
ā = from 2 to 0 193,021.91 185,919.58 180,167.36
ā = from 4 to 0 197,029.68 186,964.29 181,332.88
4.2. 85-bus system
The 85-bus system configuration is shown in Fig. 12, the base voltage and apparent
power of this system is 11 kV and 100 MVA respectively. The substation bus voltage
(bus 1) is 1 p.u. and the line and load data for this system are initiated [32] and
developed [18]. The obtained WO results are compared to those obtained using BWM
[13], HSDE [18], and CSA [21] and the comparison is presented in Tables 9 and 10.
It is noticed that the network active power losses optimized by the proposed algorithm
are reduced to 83.55 kW, which is less than other published results; this means the
power loss is reduced by 73.55% compared to the original network loss. The optimal
annual cost obtained by the proposed algorithm is 372,243.95 Rs./year, which will
save about 66.36 % of the original network cost with a payback period of 0.53 years.
The bus voltage values of the original network and that optimized using the proposed
algorithm are presented in Fig. 13. Figure 14 shows the current flow in branches
compared to their maximum limits.
Fig. 12. Configuration of 85-bus radial distribution network.
0
50
100
150
200
250
300
350
400
450
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cu
rren
t (A
)
Branch no.
1st Year 2nd Year 3rd Year
4th Year 5th Year Branch current capacity
43
40 41 42 44 45 46 47
37 38 39
23
25 26 27 28 29 30 31 32 33 34 35 36
22 21 20 19 18
24
78 85 48
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 49
50 56
51
55 54 53 52
57
16 17 58
24 23
59 55
77 80 81 82
60 61 62
83 84
63 64 65 66 76
67 68 69 70 71
72
79 75 73 74
SS
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 101
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Fig. 13. Bus voltages of the original and optimized 85-bus network.
Fig. 14. Branch currents compared to their capacities for the 85-bus system.
Voltage profile and branch current flows of the 85-bus system over a five-year
span are presented in Figs. 15 and 16, respectively. From the presented figures, one
can notice that the proposed optimal conductor selection algorithm succeeded in
minimizing the overall system costs and keeping the bus voltages and branch
currents within the prescribed limits not only for the current load profile but also
for a five-year span with high annual load growth factor of 10%. After the 5th year,
the voltage of bus 54 begins to reach its minimum allowable limit (0.9 p.u.).
Therefore, other compensation techniques should be adopted to allow for more load
growth without violating the system limits.
0.84
0.86
0.88
0.9
0.92
0.94
0.96
0.98
1
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85
Vo
lta
ge (
pu
)
Bus number
Original network Optimized network via WO
0
100
200
300
400
500
600
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82
Cu
rren
t (A
)
Branch no.
Branch current flow Branch current capacity
102 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Table 9. Comparison of conductor selection results for the 85-bus system.
Bra
nch
no
.
From
bu
s
To b
us
BW
M
[13
]
HS
DE
[18
]
CS
A [
21
]
Pro
po
sed
WO
Bra
nch
no
.
From
bu
s
To b
us
BW
M
[13
]
HS
DE
[18
]
CS
A [
21
]
Pro
po
sed
WO
1 1 2 F H F 20 43 34 44 B H C 3
2 2 3 F H D 20 44 44 45 B D B 2
3 3 4 F H A 20 45 45 46 B G A 1 4 4 5 F H F 20 46 46 47 B D C 1
5 5 6 F H B 20 47 35 48 B E G 8
6 6 7 E H A 20 48 48 49 B A E 3 7 7 8 E H A 20 49 49 50 B D C 2
8 8 9 E H D 20 50 50 51 B D C 1
9 9 10 E G D 12 51 48 52 B D H 5 10 10 11 B H G 8 52 52 53 B B C 3
11 11 12 B E B 8 53 53 54 B A A 2
12 12 13 B C H 3 54 52 55 B G A 1 13 13 14 B A H 2 55 49 56 B D C 1
14 14 15 B A F 1 56 9 57 B H B 18
15 2 16 B H C 1 57 57 58 B H F 17 16 3 17 B G B 3 58 58 59 B A B 1
17 5 18 B E E 8 59 58 60 B E B 17
18 18 19 B F H 6 60 60 61 B D A 3 19 19 20 B A H 3 61 61 62 B E A 1
20 20 21 B B B 2 62 60 63 B H D 12
21 21 22 B A G 1 63 63 64 B G H 12 22 19 23 B C H 1 64 64 65 B H G 1
23 7 24 B C A 1 65 65 66 B A H 1
24 8 25 B G G 20 66 64 67 B G A 12 25 25 26 B H C 20 67 67 68 B D D 6
26 26 27 B G D 18 68 68 69 B E C 5
27 27 28 B H C 18 69 69 70 B E D 2 28 28 29 B H A 17 70 70 71 B A F 1
29 29 30 B G G 17 71 67 72 B H B 1
30 30 31 B H A 14 72 68 73 B B E 2 31 31 32 B H C 14 73 73 74 B H H 1
32 32 33 B E B 12 74 73 75 B A E 1
33 33 34 B H A 12 75 70 76 B C F 2 34 34 35 B F E 8 76 65 77 B A C 1
35 35 36 B H E 1 77 10 78 B C H 1
36 26 37 B A A 1 78 67 79 B A G 1 37 27 38 B C A 1 79 12 80 B D E 5
38 29 39 B G E 1 80 80 81 B B E 3 39 32 40 B B A 3 81 81 82 B G G 1
40 40 41 B B B 2 82 81 83 B C H 1
41 41 42 B F B 1 83 83 84 B F A 1 42 41 43 B B D 1 84 13 85 B A C 1
Table 10. Comparison of final results for the 85-bus system.
Variables Original
system
After optimal conductor selection
BWM [13] HSDE [18] CSA [21] Proposed
WO
Umin at node 54 0.8713 0.9458 0.9296 0.8900 0.9436
Power Loss (kW) 315.84 134.26 256.96 254.152 83.55
Loss reduction (%) - 57.51 18.64 19.53 73.55
Optimal cost (Rs) 1,106,732 485,040 929,035 927,786 372,243
Net saving (%) - 56.17 16.06 16.17 66.36
Payback period (years) - - - - 0.53
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 103
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
Fig. 15. Voltage profile over a five-year span for the 85-bus system.
Fig. 16. Branch currents over a five-year span for the 85-bus system.
5. Conclusions
In this work, the WO algorithm is proposed to solve the problem of optimal selection
of conductors of radial distribution networks. The WO algorithm is applied on two
different test systems, 16-bus system, and 85-bus system. The achieved results are
compared with other methods available in the literature and showed the superiority
of the proposed algorithm over the other published methods. One can notice that the
proposed set of conductors explored a wider search space that succeeded in finding
the most optimal set of conductors that satisfy the objective functions while
complying with the pre-set constraints. Finally, the proposed WO algorithm
succeeded reducing the network losses, maximizing the overall saving while
maintaining compliance with the specified constraints over five years with a
reasonable payback period, while taking into account high annual load growth rate.
Following highlights are observed through simulation case studies:
0.88
0.9
0.92
0.94
0.96
0.98
1
1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64 67 70 73 76 79 82 85
Volt
age (
Pu
)
Bus number
1st year 2nd year 3rd year 4th year 5th year 6th year
0
100
200
300
400
500
600
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83
Cu
rre
nt
(A)
Branch number
1st Year 2nd Year 3rd Year 4th Year 5th Year Branch capacity
104 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
The introduction of the proposed actual conductors library provided optimal
techno-economic results for the problem of conductors selection in radial
distribution networks.
The optimal conductor selection is considered as one of the important tools for
reducing the network losses in a cost effective manner.
The WO algorithm proved its efficiency and suitability for solving small scale
and large scale system applications.
Load growth considerations and payback period calculations are important
techno-economic aspects for the distribution system planners.
Finally, our study was limited to the constant loading profile in balanced and
sinusoidal electrical systems. Another factor that was beyond the framework of this
study, but will be included in future studies, is the consideration of a real-time
loading profile in unbalanced and non-sinusoidal distribution networks.
Nomenclatures
CA
, Coefficient vectors
A(c) Cross section area of the cth type conductor
a
Linearly decreasing factor to represent the shrinking bubbles
b Total number of branches in the system
Cas Cost of annual energy saving due to optimal conductor selection
CC Capital investment cost of the conductor
CCtotal Total capital investment cost of all branches
CL Annual cost of energy losses
CLtotal Annual cost of the energy loss OptC Total cost of the optimized system conductors origC Total cost of the original system conductors
c Branch type
iD distance between the ith whale to the prey
F Planned lifetime of the feeder
g Annual load growth rate
IC(c) Investment cost of the cth type conductor per unit area per unit
length
IDF Interest and depreciation factor
I(j,c) Current flowing in conductor j of type c
Imax(c) Maximum current carrying capacity of the cth type conductor
i Interest rate value
j Branch number
k Total number of buses in the system
kp Cost of peak demand power loss
ke Cost of energy losses
LSF Loss factor
LF Load factor
l(j) Length of branch j
M Breaking year
m Bus number
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 105
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
n Number of the available conductors’ types
OF Objective function
PL(1) Active load power in the 1st year
PL(N) Active load power in the Nth year
Ploss (j,c) Active power loss of the jth branch with a cth conductor under
peak load base
lossP Network losses of the original network
OptlossP Network losses of the optimized network
PP Payback period
p Random number in [0, 1]
QL(1) Reactive load power in the 1st year
QL(N) Reactive load power in the Nth year
r
Random vector in [0, 1].
s Constant for defining the shape of the logarithmic spiral
T Number of hours per year (8760 hour)
t Index for the current iteration
Umax (m) Maximum voltage at bus m
Umin(m) Minimum voltage at bus m
X Random position vector
X* Position vector of the best solution obtained so far
randX
Random position vector (a random whale)
z Random number in [−1, 1]
totalC Total additional cost due to optimal conductor selection
References
1. Funkhouser, A.W.; and Huber R.P. (1955). A method for determining ACSR
conductor sizes for distribution systems (include discussions). Transactions of
the American Institute of Electrical Engineers. Part III: Power Apparatus and
Systems, 74(3), 479-484.
2. Ponnavaikko, M.; and Rao, K.S.P. (1982). An approach to optimal distribution
system planning through conductor gradation. IEEE Transactions on Power
Apparatus and Systems, PAS-101(6), 1735-1742.
3. Tram, H.N.; and Wall, D.L. (1988). Optimal conductor selection in planning
radial distribution systems. IEEE Transactions on Power Systems, 3(1), 200-206.
4. Ranjan, R.; Chaturvedi, A.; Solanki, P.S.; and Das, D. (2003). Optimal
conductor selection of radial distribution feeders using evolutionary
programming. Proceedings of the Conference on Convergent Technologies for
the Asia-Pacific Region (TENCON). Bangalore, India, 456-459.
5. Islam S.J.; and Ghani M.R.A. (1999). Economical optimization of conductor
selection in planning radial distribution networks. Proceedings of the IEEE
Transmission and Distribution Conference. New Orleans, Los Angeles, United
States of America, 858-863.
6. Wang, Z.; Liu, H.; Yu, D.C.; Wang, X.; and Song, H. (2000). A practical
approach to the conductor size selection in planning radial distribution
systems. IEEE Transactions on Power Delivery, 15(1), 350-354.
106 S. M. Ismael et al.
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
7. Sivanagaraju, S.; Sreenivasulu, N., Vijayakumar, M.; and Ramana, T. (2002).
Optimal conductor selection for radial distribution systems. Electrical Power
System Research, 63(2), 95-103.
8. Mandal S.; and Pahwa, A. (2002). Optimal selection of conductors for
distribution feeders. Proceedings of the IEEE Power Engineering Society
Winter Meeting. New York, United States of America, 1323.
9. Falaghi, H.; Ramezani, M.; Haghifam, M.-R.; and Milani, K.R. (2005).
Optimal selection of conductors in radial distribution systems with time
varying load. Proceedings of the 18th International Conference and Exhibition
on Electricity Distribution (CIRED). Turin, Italy, 1-4.
10. Satyanarayana, S.; Ramana, T.; Rao, G.K.; and Sivanagaraju, S. (2006).
Improving the maximum loading by optimal conductor selection of radial
distribution systems. Electrical Power Components and Systems, 34(7), 747-757.
11. Kaur, D.; and Sharma, J. (2008). Optimal conductor sizing in radial
distribution systems planning. International Journal of Electrical Power and
Energy Systems, 30(4), 261-271.
12. Raju, M.R.; Murthy, K.V.S.R.; Ravindra, K.; and Rao, R.S. (2010). Optimal
conductor selection for agricultural distribution system - A case study.
Proceedings of the IEEE International Conference on Intelligent and Advanced
Systems (ICIAS). Manila, Philippines, 1-6.
13. Abul’Wafa, A.R. (2016). Multi-conductor feeder design for radial distribution
networks. Electrical Power System Research, 140, 184-192.
14. Franco, J.F.; Rider, M.J.; Lavorato, M.; and Romero, R. (2013). Optimal
conductor size selection and reconductoring in radial distribution systems using
a mixed-integer LP approach. IEEE Transactions on Power Systems, 28, 10-20.
15. Mendoza, F.; Requena, D.; Bemal-agustin J.L.; and Dominguez-Navarro, J.A.
(2006). Optimal conductor size selection in radial power distribution systems using
evolutionary strategies. Proceedings of the IEEE/PES Transmission and
Distribution Conference and Exposition: Latin America. Caracas, Venezuela, 1-5.
16. Sivanagaraju S.; Rao J.V. (2009). Optimal conductor selection in radial
distribution system using discrete particle swarm optimization. World Journal of
Modelling and Simulation, 5(2), 96-104.
17. Devi A.L.; and Shereen A. (2009). Optimal conductor selection for radial
distribution networks using genetic algorithm in SPDCL, AP - A case study.
Journal of Theoretical and Applied Information Technology, 5(6), 674-685.
18. Rao R.S.; Satish K.; and Narasimham S.V.L. (2011). Optimal conductor size
selection in distribution systems using the harmony search algorithm with a
differential operator. Electrical Power Components and Systems, 40(1), 41-56.
19. Thenepalle, M. (2011). A comparative study on optimal conductor selection
for radial distribution network using conventional and genetic algorithm
approach. International Journal of Computer Applications, 17(2), 6-13.
20. Ismael S.M.; Aleem S.H.E.A.; Abdelaziz, A.Y.; and Zobaa, A.F. (2018).
Optimal conductor selection of radial distribution feeders: An overview and new
application using grasshopper optimization algorithm. Classical and Recent
Aspects of Power System Optimization, 185-217.
Optimal Conductor Selection in Radial Distribution Systems using Whale . . . . 107
Journal of Engineering Science and Technology February 2019, Vol. 14(1)
21. Abdelaziz A.Y.; and Fathy A. (2017). A novel approach based on crow search
algorithm for optimal selection of conductor size in radial distribution networks.
Engineering Science and Technology, an International Journal, 20(2), 391-402.
22. Ismael, S.M.; Aleem, S.H.E.A.; and Abdelaziz, A.Y. (2017). Optimal selection
of conductors in Egyptian radial distribution systems using sine-cosine
optimization algorithm. Proceedings of the Nineteenth International Middle
East Power Systems Conference (MEPCON). Cairo, Egypt, 103-107.
23. Leppert S.M.; and Allen, A.D. (1995). Conductor life cycle cost analysis.
Proceedings of the 39th Rural Electric Power Conference. Nashville,
Tennessee, United States of America, C2/1-C2/8.
24. Vahid, M.; Manouchehr, N.; Jamaleddin, A.; and Hossein, S.D. (2009).
Combination of optimal conductor selection and capacitor placement in radial
distribution systems for maximum loss reduction. Proceedings of the IEEE
International Conference on Industrial Technology. Gippsland, Australia, 1-5.
25. Mozaffari L.M; Ostovar F.; and Mohammad L.M. (2014). Combination of
optimal conductor selection and capacitor placement in radial distribution
systems using PSO method. Iraqi Journal for Electrical and Electronic
Engineering, 10, 33-41.
26. Ismael S.M.; Aleem S.H.E.A.; Abdelaziz A.Y.; and Zobaa A.F. (2019). State-
of-the-art of hosting capacity in modern power systems with distributed
generation. Renewable Energy, 130, 1002-1020.
27. Ismael S.M.; Aleem S.H.E.A.; Abdelaziz, A.Y.; and Zobaa A.F. (2018). Practical
considerations for optimal conductor reinforcement and hosting capacity
enhancement in radial distribution systems. IEEE Access, 6, 27268-27277.
28. Mirjalili, S.; and Lewis, A. (2016). The whale optimization algorithm.
Advances in Engineering Software, 95, 51-67.
29. Reddy P.D.P.; Reddy, V.C.V.; and Manohar, G.T. (2017). Whale optimization
algorithm for optimal sizing of renewable resources for loss reduction in
distribution systems. Renewables: Wind, Water, and Solar Journal, 4(3), 1-13
30. Prakash, D.B.; and Lakshminarayana, C. (2016). Optimal siting of capacitors
in radial distribution network using whale optimization algorithm. Alexandria
Engineering Journal, 56(4), 449-504.
31. NSAI Standards (2001). Conductors for overhead lines - round wire concentric
lay stranded conductors. Irish Standard, I.S. EN 50182:2001.
32. Das, D.; Kothari, D.P.; and Kalam, A. (1995). Simple and efficient method for
load flow solution of radial distribution networks. International Journal of
Electrical Power and Energy Systems, 17(5), 335-346.
33. Oman Cables Industry (SAOG) (2019). Overhead line conductors. Retrieved
February 25, 2019, from http://omancables.com/wp-content/uploads/
2017/01/Overhead-Line-Conductor.pdf.