Optimal Placement and Sizing of DG Considering Power and Energy Loss Minimization in Distribution System

International Journal on Electrical Engineering and Informatics - Volume 12, Number 3, September 2020
Optimal Placement and Sizing of DG considering Power and Energy Loss

Minimization in Distribution System
Arnab Pal, Ajoy Kumar Chakraborty, and Arup Ratan Bhowmik
Department of Electrical Engineering

National Institute of Technology Agartala
Agartala, India
[email protected], [email protected], [email protected]
Abstract: Optimal placement of Distributed Generation (DG) in the distribution network is a

significant solution to resolve the power loss issue. Selection of the optimal locations and sizes
of the DG is a practical problem and challenging job to achieve the minimum loss. In this
work, the optimal location and appropriate size of the DGs have been found out to minimize
the total active power loss of the distribution system with less penetration level of the DG
considering of 24 hours duration. A modified search algorithm has been proposed in this paper
which is capable to obtain the best solutions. Moreover, several reputed soft computing
techniques have been applied to solve the problem and validate the optimal results. Technical
aspects like energy losses have been minimized instead of power loss and practical factors like
hourly basis solar output, wind generation and load variation have been considered to make
this work more realistic. The solution techniques have been tested on the IEEE 33, IEEE 69
and 118 radial distribution test bus systems and results have been compared with recent
literature. The quality of the results establishes the efficiency of the proposed modified
algorithm to solve the optimal location and sizing problem with the effect of PV and wind
power variation along with the variable load.
Keywords: Distributed generation; optimal DG placement; renewable generation; energy loss

minimization; optimization techniques.
1. Introduction
DG is a small scale electrical power generator typically located at near to the load in the
distribution system. DG plays a very important role to minimize the power loss and voltage
improvement of the system, its size is around 50 kW to 100 MW [1]. The energy loss can be
reduced and the voltage profile can be improved of the distribution network by placing DGs
optimally. The DG also helps to improve the reliability of the power system. DG is beneficiary
for the consumer and utility both, particularly where the central generation or utility is
unfeasible or transmission system is weak to transmit sufficient power to fulfill the load
demand. Since DG is connected close to the load at distribution network, the loss for the power
travel through transmission and distribution line can also be avoided. Most importantly, it can
supply the power to the consumer as well as to the grid if generation is more than the local
demand.
But DG should be placed at such locations so that minimum loss can be achieved
considering the constraints of the system. DG placement at non-optimal place can cause
maximization of power losses, over voltage or under voltage and may be the cause of violation
of the line capacities. Also DG penetration limit is a very important issue to the researchers
because there is some limitation to inject power into the distribution system. The system may
not be healthy or stable by injecting more and more power by DG as high penetration can
create high node voltages, high fault current and impact on grid stability [2]. Many
optimization techniques have been proposed and applied so far in the literature to get the
optimal location as well as the size of DG. Most of the earlier proposed algorithms were
mainly based on classical mathematical programming methods. These techniques could not
Received: February 8th, 2019. Accepted: September 11st, 2020

DOI: 10.15676/ijeei.2020.12.3.12
624
Arnab Pal, et al.
explain the complex objective functions which are not differentiable, mostly with complicated
constraints, as today’s DG placement problem is not a mathematically convex problem.
In [3], H.L.Willis has discussed zero point analysis migrating zero point of DG and an
analytical 2/3 method has been used for the optimal placement and sizing of DGs. However,
this method is not suitable for non-uniform loads. Wang et. al [4] have presented analytical
approaches for the optimal placement of DG with unity power factor in power systems. DGs
have been allocated in IEEE 33 bus system to minimize the power loss [5]. Optimal load flow
with second order algorithm method [6] has been used to determine the optimal location of DG
for minimum power losses. In [7], particle swarm optimization has been used for the optimal
placement of DG and the results have been compared to the analytical approach. The
approaches like prime dual interior point method [8], mixed integer nonlinear programming
[9], trade-off method [10], genetic algorithm [11] also have been employed to solve the DG
allocation in distribution systems. In [12], Ant Lion Optimization (ALO) has been imposed to
allocate DG on different distribution network at optimal places with optimal sizes which are
mainly renewable generation. Active power loss minimization has been done in [13] by placing
DG at optimal locations by applying Genetic Algorithm. In [14], authors have used a new
metaheuristic technique and hybrid GWO to allocate DGs optimally in different radial bus
systems such as IEEE 33, 69 and Indian 85 for loss minimization and bus voltage
improvement. The PSO method has been implemented for DG allocation to get the benefit of
voltage improvement and reduction of transmission line losses in [15]. In [16], placement of
multiple DGs has been done in a microgrid. In [17], DG allocation at optimal locations using a
combination tool of genetic algorithm and simulated annealing. The optimal results have been
compared with using simple genetic algorithm and tabu search in [18]. To find out DG’s
optimal location in distribution (radial) system an improved multi-objective harmony search
(HS) has been used in [19]. The nonlinear constraints have been considered for placing and
sizing of DG in the radial bus system in [20]. Article [21] has been presented as system loss
minimization considering yearly load variation in an existing distribution system and optimal
DG size and location have been found out by cuckoo search algorithm. In [22] Mukul Dixit et
al. have utilized Power Loss Index (PLI) approach and Index Vector Method (IVM) to
determine the appropriate allocation of DGs and shunt capacitors. Ankit Uniyala and Ashwani
Kumar have considered three objective functions and solved with NSGA-II with fuzzy
satisfying method to get optimal sizes and locations of DG in [23]. In [24] Imran Ahmad
Quadri et al. have used teaching learning-based optimization to reduce the distribution system
loss, improve the voltage and annual energy saving. Asmaa H. Ali et al. in [25], have used
ALO to allocate DG in 33 and 69 RDN considering the renewable sources. In [26] Imen Ben
Hamida et al. have reconfigured the distribution to minimize the loss of the system with DG
integration considering the variation of load and DG generation. Nazari-Heris et al. have used
Q-PSO in [27]. In [28], R. Arulraj and N. Kumarappan have proposed the allocation of a single
DG and multiple DG in the distribution system by SLPSO. In [29] Coelho, F.C et al. have
planned a new metaheuristic technique named War Optimization to get the right places and
sizes for DG in order to achieve the minimum active power loss. In [30], Karar Mahmoud and
Yorino Naoto have shown a combination of analytical expressions and OPF to obtain the
optimal places, sizes and DG type. T. Yuvaraj et al. have used the Bat Algorithm to optimal
allocate DG on IEEE 33 RDS in [31]. In [32], Belkacem Mahdad and K. Srairi have allocated
multi-type DG in presence of SVC in the system by adaptive DS. In [33], E. S. Ali has applied
ALOA in only 69 distribution bus system for allocation of renewable DG. In [34] loss–
voltage–cost index method has been incorporate for optimization and placed capacitor bank
and DG in distribution system to achieve stable voltage, optimal annual operating cost and
minimum power loss. Optimal locations and sizing for renewable generation, capacitor bank
and energy sources have been done by an integrated planning framework in [35]. DGs have
been considered as solar PV source to get highest voltage stability in the system, and a new
power flow algorithm developed to avoid iteration for optimal allocation of DG considering
line sensitivity factor using HOMOR software in [36]. Transient performance of power system
625
has been enhanced with the help of DG in [37]. Energy storages and fuel cells have been
allocated as DG in distribution network in [38]. Locations and sizes of DG have been found out
optimally in [39] using a robust algorithm. Authors have minimized the power loss of radial
distribution network by optimal network reconfiguration in [40]. Coordination of protection
device has been done in presence of DG in [41]. Optimal reconfiguration has been presented
considering the DG in [42]. Authors in [43], have introduced an optimization method to
allocated DGs in distribution system. In [44] and [45], optimal allocation of renewable DG and
storage has been done in distribution network. Network reconfiguration of the distribution
network and DG allocation both have been done in [46 – 48]. An innovative algorithm has
been proposed in [49] to allocate DG in radial network and [50] has presented a hybrid
technique for the same. GWO has been used in [51] to allocate electric vehicle charging station
in IEEE 33 distribution network. [52 – 54] are also good work of allocation problem on
distribution network.
Mirjalili. S. et al. [55] have presented a bio inspired optimization technique named Salp
Swarm Algorithm in 2017 to solve engineering problems with a single objective or multi-
objective function. The outcomes on the numerical capacities demonstrate that the SSA
algorithm can enhance the underlying arbitrary arrangements viably and join towards the ideal.
The aftereffects of the genuine contextual investigations show the benefits of this algorithm
proposed in taking care of true issues with troublesome and obscure pursuit spaces. Whale
Optimization Algorithm [56] is a metaheuristic optimization technique inspired by nature. It is
copied from hunting behaviour of humpback whales i.e. bubble-net searching technique.
Mirjalili. S. et al. have tested with 29 mathematical and 6 design problems with this
optimization technique and the results are far better than other conventional techniques. In
[57], Mirjalili et al. have proposed an effective optimization algorithm to solve the real
problem called Moth Flame Optimization. The technique inspired by the transverse orientation
which is the navigation system of moths. By this technique, they can travel a long distance in a
straight line and trapped other small insects around the artificial light. MFO has been tested and
got the impressive result with 29 benchmark problem and 7 practical engineering problems.
Pal, A. et al. [58] in 2017 have introduced two new algorithms to determine optimal locations
for DG. Algorithm 1 is capable to find the best optimal locations by searching all the possible
combination of DG’s locations for a certain number of DG but it searches a large number of
combinations and takes huge time. Algorithm 2 finds the same optimal locations as algorithm 1
but it consume very less time. In 2018 Pal, A. et al. [59] have proposed an algorithm named
One by One Search Algorithm (OBOSA) to find out only optimal locations of DG but the sizes
were predefined. The obtained results are far better than other optimization techniques because
it searches all the possible location one by one and computing time is also very less, unlike
optimization techniques.
In this work, the OBOSA has been modified to make it robust to solve the sizing problem
along with the locations of the DG. Moreover, state of the art technique such as grey wolf
optimizer (GWO), salp swarm algorithm (SSA), whale optimization algorithm (WOA) and
moth-flame optimization (MFO) have been applied to solve the same problems. Therefore,
various optimization techniques and modified one by one search algorithm (MOBOSA) have
been used to determine validated results of the optimal nodes and respective sizes of the DG on
IEEE 33, IEEE 69 and 118 (appendix – A) distribution systems. Each technique has strength
and weakness depending on different problem. For some cases, one technique is performing
well but for other cases, it is not able to provide a good result that’s why in this paper many
techniques have been applied to find out optimal locations as well as the sizes of the DG. But
the most beauty of this paper is energy loss minimization because the distributed generators are
renewable based like solar power and wind power generation. These sources are non-
dispatchable energy source which totally depends upon the weather condition and human
doesn’t have any control of their output. Therefore, solar irradiance, wind speed and load are
changing with the time. Hence, which is an optimal location at a particular time it would not be
optimal at another time because by that time, the DG power output and load have changed. For
626
Arnab Pal, et al.
this reason, the energy loss minimization has been formulated in this work to minimize the
losses in 24 hours duration. An annual daily average of PV and wind power output on hourly
basis have been taken to make this research realistic. Finally, the DGs have been allocated at
optimal nodes for which the energy loss of the system will be minimum for the entire day
considering the hourly variation of PV and wind power along with the variable load. As R/X
ratio is high for the distribution network, the conventional load flow method is not suitable.
Therefore, the energy losses, node voltages and line current have been calculated by BIBC
BCBV based forward-backwards sweep (FBS) load flow.
2. Problem Formulation
A. Power Loss Minimization
The main objective is to find out the locations and sizes of the DGs for which it will offer
the minimum power/energy loss when the voltage of the nodes are within the acceptable limits.
The losses and voltages of the radial distribution network can be calculated with the help of
forward-backwards sweep load flow [60]. Let’s take a 6 bus distribution system as an example
shown in Figure 1.
Figure 1. Example of 6 test bus system
where B1 , B2 , B3 ,..., B6 are the branch currents, like B1 is the branch current in between bus number
1 and 2. I1 , I 2 , I 3 ,..., I 6 are the equivalent current injection at the respective bus. Now, branch
current can be calculated using the KCL.
B1 = I 2 + I 3 + I 4 + I 5 + I 6
B2 = I 3 + I 4 + I 5 + I 6 (1)
B=
3 I 4 + I5
B4 = I 5
B4 = I 6
Therefore, the relations matrix with branch currents and current injection can be developed
as
 B1  1 1 1 1 1  I 2 
  
 B2  0 1 1 1 1   I 3  (2)
 B3  = 0 0 1 1 0  I 4 
    
 B4  0 0 0 1 0  I5 
 B  0 0 0 0 1   I 6 
 5 
Let’s consider this matrix as a general form, where BIBC is the matrix containing only 0
and 1.
[ B ] = [ BIBC ][ I ] (3)
where BIBC is ‘bus injection to branch current’. Using KVL, the relation between branch
current and bus voltage can be expressed as below.
V=
2 V1 − B1 Z12
V= V2 − B2 Z 23
3
(4)
V=
4 V3 − B3 Z 34
V=
5 V4 − B4 Z 45
V=
6 V3 − B5 Z 36
627
where V1 is the voltage at the substation, V2 , V3 ,..., V6 are the voltages of the respective bus and Z is
line impedance. Using equation (5) all the voltages can be replaced with V1 , as:
V6 =−
V1 B1 Z12 − B2 Z 23 − B5 Z 36 (5)
Therefore, the bus voltage can be presented only with the help of branch current and line
parameter as below
V1  V2   Z12 0 0 0 0   B1 
V  V    
 1   3   Z12 Z 23 0 0 0   B2 
(6)
V1  − V4  =
 Z12 Z 23 Z 34 0 0   B3 
      
 1   5   Z12
V V Z 23 Z 34 Z 45 0   B4 
V  V   Z Z 23 0 0 Z 36   B5 
 1   6   12
Equation (6) can be written in general form as
[ ∆V ] =
[ BCBV ][ B ] (7)
where BCBV is ‘branch current to bus voltage’ and bus voltage can be calculated by
[ ∆V ] =
[ BCBV ][ BIBC ][ I ] (8)
= [ DLF ][ I ]
where DLF is the ‘Distribution Load Flow’, which is the simple multiplication of BCBV and
BIBC matrixes. The voltages and currents can be updated using equation (9 – 11) by an
iterative process.
*
 P + jQ  (9)
I ik =I ir (Vi k ) + jI ii (Vi k ) = i k i 
 Vi 
[ DLF ]  I k 
 ∆V k +1  = (10)
V =
k +1
 V  +  ∆V 
0 k +1
(11)
k th
where I i is the injected current at k iteration at the bus number i and V 0 is the initial
r i
voltage. I i , I i are the real and imaginary part of the current respectively. The calculation for
the power loss has been presented below:
Pl i = ( IB i ) 2 * Zi (12)
Z=i R i + jX i (13)
where PL is total system loss, Pl is the branch loss, IB is branch current, Z
i i i
is the
impedance, R is the resistance and X is the reactance of the i branch.
i i th
Active Power Loss = Re( Pl i )

Reactive Power Loss = Im( Pl i )
When the objective is to minimize the total power loss, the objective function can be
written as
n
(14)
∑
min( f ) = Re{( IBi )2 *Zi }
i =1
Where n = number of branch.
B. Energy Loss Minimization

Objective Function has been presented below for energy loss minimization considering the
hourly variation of load, PV, wind power.
24 n
min( f ) = ∑∑ Re{( IBt i )2 * Zi } (15)
=t 1 =i 1
where t is the hour in a day and IBt i is the current flowing through the i th branch at time tth
hour.
628
Arnab Pal, et al.
C. Constraints
• Voltage constraint:
The allowable voltage limits should be maintained at the time of DG allocation. DG should
not create under or over voltage in the system.
Vmin ≤ Vt i ≤ Vmax where i=1 to nb and t = 1 to 24 (15.1)
where Vmin = 0.94 and Vmax = 1.06
When PV and wind generation have been considered as DG, the lower limit of the voltage
constraint have not been taken into account because PV/wind power output may be less or zero
at a particular time in a day, which depends on solar irradiance and wind speed. Therefore, it
may not maintain the desired voltage improvement.
• DG penetration limit constraint:
Total DG output power should be less than or equal to the total load (100% penetration
level). Over penetration may cause unstable and disturbance in the system [61, 62] because the
total DG output more than the total load can create over voltage, high fault current, and reverse
power flow.
nDG
∑ P DG ≤ 100% of the lotal load (15.2)

i
i =1
• Power balance constraint:

The summation of the total load and power losses should be supplied from the DG and
utility as follow
0
PU + PDG − Ploss − Pload = (15.3)
Where, PU = Power consumption from central generation, PDG = Total power delivered
from all DG, Ploss = Total power loss in the network, Pload = Total load in the network.
3. Methods and Tools

A. Salp Swarm Algorithm (SSA) [55]
The main objective of Salp Swarm Optimizer is to find out the global optimal point. The
leader salp goes at the food direction and follower salps follow that. So, the total salp chain
will travel in the direction of the food or the optimal point.
With random positions, initialize the multiple salps to start close to the global optimum
point SSA algorithm starts close to the global optimal. Then find out the best fitness salp by
calculating each salp’s fitness. As the target of the salp chain is to find out the food location,
we have to set the best salp to the variable F. Using equation (17) update the coefficient for
every dimension. Then need to update the position to go towards food and to get back the
outsider salps into the boundaries. Update the leader salp’s position using equation (16) and the
position update formula is equation (19) for the follower salps. These steps will continue until
it is not fulfilling the end condition. This updated technique has the potential to chase that food
source which is moved during the search.
 Fj + c1 ((ub j − lb j )c2 + lb j ) c3 ≥ 0
x1j =  (16)
 Fj − c1 ((ub j − lb j )c2 + lb j ) c3 < 0
Equation (16) is the updated formula for the leader salp even the food location has changed.
Where
th
x1j = Position of the leader salp (first salp in the chain) in the j dimension.
F j = Food location at the j th dimension.
c1 , c2 and c3 are the random numbers.
lb j = Lower bound of the j th dimension.
4l 2
−( )
c1 = 2e L (17)
629
In SSA, c1 is such a coefficient that helps to maintain the exploitation and exploration
balanced. Equation (17) is the expression for c1 . Where, L = Max iteration number, l =
Iteration number
1 2
x ij
= at + v0 t (18)
2
i
Equation (18) is for the position update formula for follower salps. Where x j = follower
v final
salps position in j th dimension and i ≥ 2 . v0 = Starting speed, t = Time. a= , where,
v0
x − x0
v= .
t
The mismatch between iteration and v0 is 1, so the equation can be expressed as follow:
1 i i −1
xij
= (x j + x j ) (19)
2
• Pseudocode for the SSA [55]:

Salp population Initialization xi ( i = 1, 2, ..., n) using ub and lb
while (not satisfied the stop condition)
fitness calculation for each salp
F=the best salp (search agent)
Update c1 by equation (17)
for each salp ( x i )
if ( i ==1)
Position update of the leading salp using equation (16)
else
Position update of the follower salp using equation (19)
end
end
Improve the salps according to the variables’ upper and lower bounds.
end
return F
B. Whale Optimization Algorithm [56]

WOA may be the global optimizer as it incorporates with exploitation and exploration skill.
It allocates a search space near the best solution and permits other search agents to utilize the
best point by using a hyper-cube mechanism. Only two parameters A and C need to adjust.
 
Some iterations are committed for exploration ( A ≥ 1) and remaining for exploitation ( A < 1) by
decreasing the A for the smooth movement between exploitation and exploration.
The whale optimization starts directly with random solution, then they update their
positions with the best position acquired so far or random search agents. To assign exploration

and exploitation, decreased the parameter a from 2 to 0. When ( A > 1) , a search agent is

chosen randomly, and the best solution is chosen when ( A < 1) for position update of the
search agents. WOA is capable to shift between circular or spiral movement. When the stop
criterion is satisfied then the algorithm will terminate.
  * 
=D C. X (t ) − X (t ) (20)
   
1) X * (t ) − A.D
X (t + = (21)
630
Arnab Pal, et al.
Equation (20) and Equation (21) are the position update formula of the whale to catch the
victim. Where,

D = Distance calculation between whale and victim.
t = Iteration Number.
X * = Obtained best position.

X = Position Vector.
 
A and C = Coefficient vector, which can be calculated by equation (22) and equation
(23).
   
= A 2a.r − a (22)
 
C = 2.r (23)
 
Where a decreased the value 2 to 0, r = Random vector.
  
= X (t + 1) D′.ebl .cos(2π l ) + X * (t ) (24)
For the spiral position update using equation (24) to target the food by calculating the
distance between whale and victim. Where the formula to calculate distance is
 * 
D′ X (t ) − X (t )
= (25)
Where b = Constant, l = Random number in between -1 and 1.

  
  X * (t ) − A.D if p < 0.5
X (t + 1) =   bl  (26)
 D′.e .cos(2π l ) + X (t ) if p < 0.5
*
Here p = Random number in between 0 and 1.

   
=D C. X rand − X (27)
   
X (t + 1)
= X rand − A.D (28)
With the help of equation (27) &equation (28) find the location of the victim by varying the
 
vector A within -1 to 1. Where X rand = Random position of the whale.
• Pseudocode for the WOA [56]:

Whales population Initialization X i (i = 1, 2,3..., n)
fitness calculation of each whale agent
X * =the best search agent
while (t < max number of iterations)
for each agent
a, A, C , l and p update
if ( p < 0.5)
if (|A| < 1)
current search agent position update by the Eq. (20)
else if (|A|>1)
Select a random search agent ( X rand )
Update the position of the current search agent by the Eq. (28)
end
else if ( p ≥ 0.5)
Position update of the current search by the Eq. (24)
end
end
Take back those agents who have gone out of the search boundary.
Calculate every search agent’s fitness
If better solution obtained then Update X *
631
t=t+1
end
return X *
C. Moth-Flame Optimization (MFO) [57]

MFO is the moth populations based search, where moths are the search agents search the
optimal point by considering the moonlight as the reference with a fixed angle. By this way,
they can obtain long straight accurately and by changing the position vector, they can fly 1, 2, 3
and hyper dimension. At any point the best positions of moths are called flame, then each moth
follows the flame’s position in a spiral path to update theirs for a better position. MFO
algorithm mainly depends upon three functions.
i.e. MFO = ( I , P, T ) (29)
where I is a function that creates moths’ position randomly in search space and generates the
population matrix M which will be updated and calculate their fitness values.
I : ϕ → {M , OM } (30)
M describes the position of the moths and OM is their fitness.
 m1,1 m1,2  m1,d 
m m2,2  m2,d 
M =
2,1 (31)
     
 
 mn,1 mn,2  mn,d 
where d = number of variable, n = number of moths
Calculated fitness values from objective function for each corresponding population set will
be store in OM matrix.
 OM 1 
OM 
OM =
2
 (32)
  
 
OM n 
Flames are also playing a very important role to update the positions of moths. Matrix for
flames is similar to position matrix for moths.
 F1,1 F1,2  F1,d 
F F  F 
F=
2,1 2,2 2,d 
(33)
     
 
 Fn,1 Fn,2  Fn,d 
where d = number of variable, n = number of flames. The fitness values for the flame’s
position will be store in OF matrix as
 OF 
1
OF 
OF =   (34)
2
 
OF 
 
n
P: The function P generates the matrix when moths are fly in the search space. After that
update it according to the fitness function value.
P:M → M (35)
T: The function T is a simple true or false generator, when the stop condition will be
satisfied then it gives true, and false when it is not satisfied to stop the search.
T : M → [true, false] (36)
To update the position of the moths according to flame position use the following equation.
M i = S ( M i , Fj ) (37)
632
Arnab Pal, et al.
S ( M i , Fj ) Di * ebt cos(2π t ) + Fj
= (38)
where logarithmic spiral function is S and Di is the distance between i moth and j flame, t is
th th
a random number in between -1 and 1 and b is a constant with gives the shape to that spiral
path.
D=i Fj − M i (39)
where Fj and M i are the fitness values of flame and moth respectively.
Try to make an adaptive reduction of flames with the iteration process by using the
following equation
 N −1 
=flame no round  N − l *  (40)
 T 
where T = maximum iteration number, N = maximum flame number, l = current iteration
number.
D. Modified One By One Search Algorithm

This technique is not like other optimization. It searches all the possible locations one by
one and highlights the best solution. The major steps of the modified one-by-one search
algorithm (MOBOSA) are as follows:
Step 1: Set DGn = Number of DG, DGpen = total penetration by the DG, PLmin = Total
loss without DG. Initiate EL as a zero matrix.
Step 2: Start counting of the hour in a day, t = 1 to 24.
Step 3: i = 2 to number of bus, DG can be connected at ith bus. As bus number 1 is
considered as substation.
Step 4: Insert the network data without DG i.e. bus data, line data and loads at the tth hour as
the load is variable.
Step 5: Check the available power output from the solar PV and wind generator at the tth
hour as solar irradiance and wind speed is not fixed.
Step 6: Connect PV/Wind at the ith bus and check the total power loss of the system and
store the loss value at ELi,t= ELi,t+ loss when DG is connected at ith bus at tth hour.
Step 7: i = i+1 and go to step 4. Stop when i = total number of DG to be connected.
Step 8: t = t +1 and go to step 3, stop when t = 24.
Step 9: From EL matrix calculate the energy loss for each bus to find out the bus numbers
which are suitable for all over the day as well as year considering all constraints i.e. equation
15.1 to 15.3.
Step 10: Find the bus numbers from EL matrix where DGs provide minimum energy loss.
E. Grey wolf optimization (GWO) [63]

GWO is a strong metaheuristic optimization technique proposed by S. Mirjalili et al. in
2014 [63]. This method is based on the hunting tactic of the grey wolf. The implementation of
this algorithm to solve the proposed problem has been shown in Figure 2 using a flowchart.
633
Figure 2. Flow chart of GWO [64]
4. Results and Discussion

All programs have been written in MATLAB programming language on a computer having
2.27 GHz Intel core (TM) i3 CPU M 350 with 3-GB RAM.
Total active and reactive load and total active power loss and reactive power loss have been
shown in below Table 1.
Table 1. Total load and loss of different distribution system

Bus Real Power Reactive Power Real Power Reactive Power
System Load Load Loss Loss
IEEE33 3715 kW 2290 kvar 202.4733 kW 135.0311 kvar
IEEE69 3802.29 kW 2694.1 kvar 224.9827 kW 102.1584 kvar
118 22709.72 kW 17041.067 kvar 935.6724 kW 707.7794 kvar
A. Optimal Locations and Sizes of DG for Power Loss Minimization

GWO, SSA, WOA, MFO and MOBOSA techniques have been implemented to the DG
allocation problem to find out the optimal locations and as well as the sizes of the DG on IEEE
33, IEEE 69 and 118 test distribution systems. Tables 2 – 6, tables 7 – 11 and tables 12 – 16 are
for IEEE 33, IEEE 69 and 118 radial bus systems respectively to allocate 5 and 10 DGs using
different solving tool.
The results for the IEEE 33 distribution system have been presented in Table 2 – 6 with
optimal DG locations and respective sizes using different techniques along with their
634
Arnab Pal, et al.
computational time. The power losses, with and without DG also have been shown below with
the total DG penetration.
Table 2. Optimal allocation of DG on IEEE 33 bus system using GWO

GWO-33 Bus 3 DG 5 DG
Optimal DG Location 13 24 30 31 14 5 26 24
Optimal DG Size (kW) 850 1104 1101 647 603 72 949 962
Convergence Time (sec) 241.33 267.20
Total Loss (kW) 73.06 66.18
Total Loss Without DG (kW) 202.4733 202.4733
Total DG Penetration (kW) 3055 < 3715 3233 < 3715
Table 3. Optimal allocation of DG on IEEE 33 bus system using SSA

SSA-33 Bus 3 DG 5 DG
Total Loss (kW) 71.9386 67.6754
Table 4. Optimal allocation of DG on IEEE 33 bus system using WOA

WOA-33 Bus 3 DG 5 DG
Total Loss (kW) 72.5409 65.3723
Table 5. Optimal allocation of DG on IEEE 33 bus system using MFO

MFO-33 Bus 3 DG 5 DG
Total Loss (kW) 71.2696 65.511
Table 6. Optimal allocation of DG on IEEE 33 bus system using MOBOSA

MOBOSA-33 Bus 3 DG 5 DG
From Table 2 to table 6, the loss reduction with DG allocation can be observed. The
allocation of 5 DG creates more benefit in term of minimization of loss. The total DG
penetrations are within the limit for all the cases. It also can be seen that to allocate 3 DG on
635
IEEE 33 bus system the MFO technique gives the minimum loss and for 5 DG allocation WOA
gives the best result among others. By using the proposed MOBOSA method nearly same
results have been achieved with very less converging time. To optimal allocate 3 DG on 33 bus
system MFO consumes 447.61 seconds and MOBOSA requires only 1.64 seconds for the
same.
Table 7 – 11 shows the results of optimal DG allocation on IEEE 69 bus system using
different solution techniques. The optimal nodes and respective DG sizes have been presented
below with system power loss and DG penetration.
Table 7. Optimal allocation of DG on IEEE 69 bus system using GWO

Optimal DG Size (kW) 541.5 406.39 1751.43 1624 329 427 101 274
Total DG Penetration (kW) 2699.3 < 3802.3 2750 < 3802.3
Table 8. Optimal allocation of DG on IEEE 69 bus system using SSA

Total Loss (kW) 72.8715 70.2786
Total DG Penetration (kW) 2340 < 3802.3 3678 < 3802.3
Table 9. Optimal allocation of DG on IEEE 69 bus system using WOA

Total Loss (kW) 71.9555 71.7266
Table 10. Optimal allocation of DG on IEEE 69 bus system using MFO

Total Loss (kW) 69.4109 67.747
In table 7 – 11, it can be noticed that the losses have been reduced after DG allocation and
more number of DG helps to reduce more loss. MFO gives the best loss minimization among
all the applied techniques for optimal DG allocation on IEEE 69 bus system for 3 and 5 DG
allocation both. MOBOSA gives a better result than SSA where the total DG penetration is
2340 kW by SSA but MOBOSA achieved less loss with 1740 kW penetration even with very
less computation time consumption. All the DG penetration levels are within the limits.
636
Arnab Pal, et al.
Table 11. Optimal allocation of DG on IEEE 69 bus system using MOBOSA

The optimal results for the DG allocation on the 118 bus distribution system have been
presented below using different techniques in Table 12 – 16. The results have been shown with
optimal locations of the 3 and 5 DGs with respective optimal sizes. The total losses of the
system with and without DG also have been shown in the tables.
Table 12. Optimal allocation of DG on 118 bus system using GWO

Total Loss (kW) 727.3143 593.77
Table 13. Optimal allocation of DG on 118 bus system using SSA

Total Loss (kW) 719.8513 633.6502
Total DG Penetration (kW) 10056.466 < 22709.72 13228.6 < 22709.72
Table 14. Optimal allocation of DG on 118 bus system using WOA

Total Loss (kW) 671.6666 486.1729
Table 15. Optimal allocation of DG on 118 bus system using MFO

Total Loss (kW) 500.6186 433.5778
637
Table 16. Optimal allocation of DG on 118 bus system using MOBOSA

Total Loss (kW) 493.32 433.42
From Table 12 - 16, proposed MOBOSA provides the best result with the minimum loss for
the 3 and 5 DGs allocation both with less penetration and less computational time. For all the
cases, The loss has been minimized after allocating the DGs and the total DG penetration levels
are within the limits. It has also been observed that the more number of DG can reduce more
power loss.
It can be seen from the above tables (Table 2 – 16), most of the cases MFO provides good
results. SSA performs well when the number of nodes is less, but when search space is large
such as in the case of 69 and 118 bus systems, it is not efficient. In the opposite hand, WOA
performs moderately in the case of 69 and 118 bus systems. The proposed MOBOSA gives the
best result for the 118 bus system which is a larger system. By this proposed algorithm, DGs
have been allocated with equal optimal sizes in a particular distribution system. For 33 and 69,
MOBOSA has performed moderate but the advantage is that it provides the minimum loss with
less penetration of DG power compare to other methods. Consequently, equal DG sizes and
less penetration level could be huge cost beneficiary and easily maintainable.
It has been found that the voltage improvements of the systems are remarkable with the
results of MFO. Therefore, Figure 3 – 8 show the voltage profiles of the 33/69/118 bus systems
for the placement of 5 DGs using MFO technique. Figure 3 presents the voltage profile of
IEEE 33 distribution system without any DG. The improvement of the voltage profile of the 33
bus system can be noticed in Figure 4 when 5 DGs have been allocated optimally with optimal
size. Likewise, the improved voltage profile of IEEE 69 bus system has been presented in
Figure 6 with 5 DGs, where Figure 5 is for without DG. Figure 7 and 8 are for 118 bus
distribution system where Figure 7 shows the voltage profile without DG and Figure 8 displays
the improved voltage profile after allocating DGs in the system.
Figure 3. Bus voltages without DG of IEEE 33 bus system
638
Arnab Pal, et al.
Figure 4. Bus voltages with 5 DG of IEEE 33 bus system using MFO
Figure 5. Bus voltages without DG of IEEE 69 bus system
Figure 6. Bus voltages with 5 DG of IEEE 69 bus system using MFO
Figure 7. Bus voltages without DG of 118 bus system
639
Figure 8. Bus voltages with 5 DG of 118 bus system using MFO
Table 17. Comparison table for optimal locations and sizes of DGs on IEEE 33 bus System
Method DG Location DG size (kW) Loss (kW) Method DG Location DG size (kW) Loss (kW)
32 1200 13 754
GA/PSO
16 863 103.4 GWO[52] 24 850 73.06
[65]
11 925 30 1109
11 1500 25 1101
GA [65] 29 422.8 106.3 WOA 12 912 72.5409
30 1071.4 30 1034
13 981.6 24 912.68779
PSO [65] 32 981.6 105.3 SSA 13 867.04192 71.9386
8 981.6 30 1008.7476
12 1182.6 24 1100
TLBO
28 1191.3 124.7 MFO 14 754 71.2696
[66]
30 1186.3 30 1071
13 1083.4 30 973
QOTLBO
26 1187.6 103.4 MOBOSA 12 973 72.2
[66]
30 1199.2 24 973
Table 18. Comparison table for optimal locations and sizes of DGs in 69 bus System
Method IEEE Bus System DG locations DG sizes Loss
GWO [52] 69 11, 18, 61 541, 406, 1751 69.51
MFO 69 11, 61, 18 527, 1719, 380 69.41
The comparisons with existing literature have been presented in Table 17 for IEEE 33 bus
system. All the applied techniques in this work have performed better, and MFO has given the
best result among all. Table 18 presents the comparison between MFO and GWO (from
existing literature), where both have been used to allocate 3 DGs on IEEE 69 bus system. In
this case, the performance of MFO is better than GWO. Figure 9 - 11 have been shown to
describe the performances of each technique by the percentage of loss reduction for applied
distribution systems. Figure 9, 10 and 11 are for IEEE 33, IEEE 69 and 118 distribution system
respectively. It can be seen that the more number of DG helps to reduce more loss.
640
Arnab Pal, et al.
Figure 9. Percentage of loss reduction of IEEE 33 bus system
B. Optimal Locations of DG for Energy Loss Minimization

The energy loss minimization has been performed using the proposed MOBOSA
considering the variations of DG output and load. This has been tested on IEEE 33, IEEE 69
and 118 bus system. 3 and 5 DGs have been allocated in the 33 system. But, as the 69 and 118
bus systems are large, therefore, more number of DG has been incorporated. For 69
distribution system 5 and 10 DGs, and for 118 system 10 and 20 number of DGs have been
allocated. For each distribution system different PV, wind and load profile of 24 hours have
been considered. The input data and respective results for different test system have been
discussed below.
• DG Allocation on IEEE 33 Radial Network for Energy Loss Minimization:

The annual average variation of the PV power generation throughout the day for 33 bus
system has been shown in Figure 12. Figure 13 shows the average wind variation of 24 hours
in per unit for the same system. The load variation of 33 bus system also has been taken into
account which has been presented in Figure 14.
641
Figure 12. Annual daily average PV variation for IEEE 33 bus system
Figure 13. Annual daily average wind variation for IEEE 33 bus system
Figure 14. Annual daily average load variation for IEEE 33 bus system
Considering the above factors, the DGs have been allocated in 33 bus distribution system.
Table 19 presents the optimal DG locations on IEEE 33 bus system with 3 DGs as well as 5
DGs and minimum/maximum bus voltages have been shown. It is found that the energy loss is
minimized with the DG allocation and 5 number of DGs provide more loss reduction.
Table 19. DG allocation on 33 bus system for energy loss minimization

Number of Optimal Energy Loss Min bus Max bus
Type
DG Locations (kWh) Voltage (pu) Voltage (pu)
Without DG … ….. 3427.2 0.9151 1
PV 32,30,25 3022.3 0.9151 1
3 DG
Wind 32,30,25 2848.1 0.9412 1
PV 32,30,25,8,31 2843.4 0.9398 1
5 DG
Wind 32,30,25,8,31 2584.8 0.9423 1
This solution is for energy loss minimization which provides the optimal locations for
minimization of power losses of 24 hours. The voltage profiles of each hour of IEEE 33 bus
system have been presented in Figure 15 when no DG is connected in the network. The
642
Arnab Pal, et al.
differences between hourly voltage profiles can be seen due to hourly load variation in the
system. Figure 16 shows the 24 hours voltage profile when 5 DGs are connected at the optimal
nodes. The remarkable voltage improvement in each hour can be noticed in Figure 16
compared to Figure 15.
Figure 15. Bus voltages of 24 hours of IEEE 33 bus system without DG
Figure 16. Bus voltages of 24 hours of IEEE 33 bus system with 5 DG (wind generation)
• DG Allocation on IEEE 69 Radial Network for Energy Loss Minimization:

The PV and wind variations of 24 hours have been shown in Figure 17 and Figure 18
respectively, which have been considered for the IEEE 69 system to allocate the DGs. The
daily load variation for the same system has been taken as Figure 19 on hourly basis.
Figure 17. Annual daily average PV variation for IEEE 69 bus system
643
Figure 18. Annual daily average wind variation for IEEE 69 bus system
Figure 19. Annual daily average load variation for IEEE 69 bus system
Table 20 presents the optimal locations of the DG in IEEE 69 bus system considering the
variations of DG power and loads. 5 and 10 DGs have been allocated and it has been found that
the energy loss is significantly less for any cases when DGs are connected. The energy loss is
lesser with 10 DGs compare to 5 DG allocation and the minimum bus voltage of the system
has also improved.

Min bus Max bus
Number Energy Loss
Type Optimal Locations Voltage Voltage
of DG (kWh)
(pu) (pu)
Without
… …….. 3597.80 0.9091 1
DG
PV 61,64,59,65,21 2330.90 0.9091 1
5 DG
Wind 61,64,59,65,21 2304.50 0.9495 1
PV 61,64,59,65,21,12,11,62,18,17 2241.50 0.9429 1
10 DG
Wind 61,64,59,65,21,12,11,62,18,17 2188.70 0.95 1
The hourly voltage improvement of IEEE 69 bus system can be compared between Figure
20 and Figure 21, where Figure 20 represents the voltage profiles without DG and Figure 21 is
when 10 DGs are allocated optimally in the network. In Figure 21, the node voltages of 24
hours duration have improved upto a remarkable level with the optimally allocated DGs.
644
Arnab Pal, et al.
Figure 20. Bus voltages of 24 hours of IEEE 69 bus system without DG
Figure 21. Bus voltages of 24 hours of IEEE 69 bus system with 10 DG (wind generation)
• DG Allocation on 118 Radial Network for Energy Loss Minimization:

Figure 22 – 24 have been considered as input of DG power and load for the 118 bus system.
Where Figure 22 shows the 24 hours variation of PV power in pu and Figure 23 present the
hourly variation of wind power. The load variation of 24 hours duration has been shown in
Figure 24.
Figure 22. Annual daily average PV variation for 118 bus system
645
Figure 23. Annual daily average wind variation for 118 bus system
Figure 24. Annual daily average load variation for 118 bus system
The optimal allocations of 10 and 20 DGs in the 118 bus system have been presented in
Table 21 considering the above variation. It has been found that without DG the energy loss of
the system is high and it gets reduced with the allocation of DGs. The loss reduction is more
with 20 DGs compare to 10 DGs. The minimum bus voltage of the system also has been
improved by the DGs.
Energy Min bus Max bus
Number
Type Optimal Locations Loss Voltage Voltage
of DG
(kWh) (pu) (pu)
Without
… …….. 12892.00 0.9053 1
DG
PV 74,111,71,70,50,109,107,112,54,96 10992.00 0.9053 1
10 DG
Wind 74,111,71,70,50,109,107,112,54,96 7764.40 0.9598 1
74,111,71,70,50,109,107,112,54,96,76,97,
PV 10502.00 0.9338 1
42,108,110,72,80,32,66,31
20 DG
74,111,71,70,50,109,107,112,54,96,76,97,
Wind 6542.90 0.9654 1
42,108,110,72,80,32,66,31
Figure 25 shows the hourly voltage profiles of the 118 bus system without DG and Figure
26 presents the same with 20 DGs which are allocated optimally. The improvement of the
voltage profile for every hour can be noticed easily by comparing both the figures.
646
Arnab Pal, et al.
Figure 25. Bus voltages of 24 hours of 118 bus system without DG
Figure 26. Bus voltages of 24 hours of 118 bus system with 20 DG (wind generation)
Therefore, Figure 15 - 16, 20 - 21, and 25 – 26 have been presented for the of bus voltages
during 24 hours for IEEE 33, IEEE 69 and 118 bus systems respectively. The voltage profiles
have been shown with 24 different colors for 24 hours. From those figures, the improvement of
bus voltages can be noticed by comparing the voltage profiles between DG and without DG.
The improvement of voltage profiles has been shown considering the wind generator as DGs.
Voltage profile improvements due to PV have not been shown here because, PV generation is
zero at night, hence, the node voltages would be the same as without DG.
5. Conclusion and Future Work

In this work, the DGs have been placed at optimal nodes considering the variation of
PV/wind power generation and load variation in the distribution network. The proper locations
of the DG with optimal sizes have been found out such a way for which the energy loss in a
day will be minimum which is more realistic and sustainable.
Newly developed soft computing techniques (GWO, SSA, WOA, and MFO) and proposed
algorithm (MOBOSA) have been used to solve the problems. It has been found that no specific
technique is good for all types of problems. However, Modified OBOSA has achieved
minimum loss with optimal DG’s power injection to the system and it also takes the least
computational time to solve the problems. This technique arises to a situation where the equal
DG capacity is being installed in a particular system and the total penetration of DGs is less to
achieve the same loss minimization compare to others. Moreover, the identical DG sizes in the
system can lead to cost benefit and easy maintenance. All the techniques have been
implemented on IEEE 33, IEEE 69 and 118 radial distribution systems.
For future work, artificial intelligence, machine learning and mixed-integer linear
programming can be accommodated for more variability. Uncertainty of renewable energy
sources is present and forecasting may be required for this study. Hybrid power generations
647
like wind, PV hybrid and battery energy storage can be used to fulfil the power demand at any
time. The optimal number of DG can be found out by considering cost optimization.
6. References
[1]. F. Gonzalez-Longatt, and C. Fortoul, “Review of the distributed generation concept:
Attempt of unification,” In International Conference on Renewable Energies and Power
Quality (ICREPQ 05), España, pp. 16-18, 2005.
[2]. H. Kuang, S. Li and Z. Wu, "Discussion on advantages and disadvantages of distributed
generation connected to the grid," 2011 International Conference on Electrical and
Control Engineering, Yichang, pp. 170-173, 2011.
[3]. H. L. Willis, “Analytical methods and rules of thumb for modeling DG-distribution
interaction.” In Power Engineering Society Summer Meeting, IEEE, vol. 3, pp. 1643-
1644, 2000.
[4]. C. Wang, and M.H. Nehrir, “Analytical approaches for optimal placement of distributed
generation sources in power systems,” IEEE Transactions on Power systems, vol. 19(4),
pp. 2068-2076, 2004.
[5]. K. Mahesh, P. A. Nallagownden, and I. A. Elamvazuthi, “Optimal placement and sizing
of DG in distribution system using accelerated PSO for power loss minimization,” IEEE
Conference on Energy Conversion (CENCON), pp. 193-198, 2015.
[6]. N. S. Rau, and Y. H. Wan, “Optimum location of resources in distributed
planning,” IEEE Transactions on Power systems, vol. 9(4), pp. 2014-2020, 1994.
[7]. S. Kansal, V. Kumar, and B. Tyagi, “Optimal placement of different type of DG sources
in distribution networks,” International Journal of Electrical Power & Energy
Systems, vol. 53, pp. 752-760, 2013.
[8]. D. K. Khatod, V. Pant, and J. Sharma, “Evolutionary programming based optimal
placement of renewable distributed generators,” IEEE Transactions on Power
systems, vol. 28(2), pp. 83-695, 2013.
[9]. Y. M. Atwa, and E. F. El-Saadany, “Probabilistic approach for optimal allocation of
wind-based distributed generation in distribution systems,” IET Renewable Power
Generation, vol. 5(1), pp. 79-88, 2011.
[10]. L. F. Ochoa, and G. P. Harrison, “Minimizing energy losses: Optimal accommodation
and smart operation of renewable distributed generation,” IEEE Transactions on Power
Systems, vol. 26(1), pp. 198-205, 2011.
[11]. G. Celli, E. Ghiani, S. Mocci, and F. Pilo, “A multiobjective evolutionary algorithm for
the sizing and siting of distributed generation,” IEEE Transactions on power
systems, vol. 20(2), pp. 750-757, 2005.
[12]. E. S. Ali, S. M. Abd Elazim, and A. Y. Abdelaziz, “Ant Lion Optimization Algorithm for
optimal location and sizing of renewable distributed generations,” Renewable Energy,
vol. 101, pp. 1311-1324, 2017.
[13]. T. Shukla, S. Singh, and K. Naik, “Allocation of optimal distributed generation using GA
for minimum system losses in radial distribution networks,” International Journal of
Engineering, Science and Technology, vol. 2(3), pp. 94-106, 2010.
[14]. R. Sanjay, T. Jayabarathi, T. Raghunathan, V. Ramesh, and N. Mithulananthan, “Optimal
Allocation of Distributed Generation Using Hybrid Grey Wolf Optimizer,” IEEE Access,
vol. 5, pp. 14807-14818, 2017.
[15]. M. H. Albadi, H. M. Soliman, E. F. El-Saadany, M. A. Thani, A. AlAlawi, S. Al Ismaili,
and H. Baalawi, “Optimal allocation of PV systems in distribution networks using PSO,”
In Modeling, Simulation, and Applied Optimization (ICMSAO), 7th International
Conference on IEEE, pp. 1-5, 2017.
[16]. W. Prommee, and W. Ongsakul, “Optimal multiple distributed generation placement in
microgrid system by improved reinitialized social structures particle swarm
optimization,” Eur. Trans. Electr. Power, vol. 21(1), pp. 489–504, 2011.
648
Arnab Pal, et al.
[17]. M. Gandomkar, M. Vakilian, and M. Ehsan, “A combination of genetic algorithm and

simulated annealing for optimal DG allocation in distribution networks,” IEEE Canadian
Conf. of Electrical and Computer Engineering, 2005.
[18]. N. Koichi, Y. Hayashi, K. Ikeda, and T. Ashizada, “Application of tabu search to optimal
placement of distributed generators,” IEEE Power Engineering Society Winter Meeting,
2001.
[19]. K. Nekooei, M. M. Farsangi, H. Nezamabadi-Pour, and K. Y. Lee, “An improved multi-
objective harmony search for optimal placement of DGs in distribution systems,” IEEE
Transactions on smart grid, vol. 4(1), pp. 557-567, 2013.
[20]. A. A. Hassan, F. H. Fahmy, A. E. S. A. Nafeh, and M. A. Abuelmagd, “Genetic single
objective optimisation for sizing and allocation of renewable DG systems,” International
Journal of Sustainable Energy, vol. 36(6), pp. 545-562, 2017.
[21]. M. Majidi, A. Ozdemir, and O. Ceylan, “Optimal DG allocation and sizing in radial
distribution networks by Cuckoo search algorithm,” IEEE 19th International Conference
on Intelligent System Application to Power Systems (ISAP), 2017.
[22]. M. Dixit, P. Kundu, and H. R. Jariwala, “Incorporation of distributed generation and
shunt capacitor in radial distribution system for techno-economic benefits,” Engineering
Science and Technology, an International Journal, vol. 20(2), pp. 482-493, 2017.
[23]. A. Uniyal, and A. Kumar, “Optimal Distributed Generation Placement with Multiple
Objectives Considering Probabilistic Load,” Procedia Computer Science, vol. 125, pp.
382-388, 2018.
[24]. I. A. Quadri, S. Bhowmick, and D. Joshi, “A comprehensive technique for optimal
allocation of distributed energy resources in radial distribution systems,” Applied
Energy, vol. 211, pp. 1245-1260, 2018.
[25]. A. H. Ali, A. R. Youssef, T. George, and S. Kamel, “Optimal DG allocation in
distribution systems using Ant lion optimizer,” In Innovative Trends in Computer
Engineering (ITCE), 2018 International Conference on IEEE, pp. 324-331, 2018.
[26]. I. B. Hamida, S. B. Salah, F. Msahli, and M. F. Mimouni, “Optimal network
reconfiguration and renewable DG integration considering time sequence variation in
load and DGs,” Renewable Energy, vol. 121, pp. 66-80, 2018.
[27]. M. Nazari-Heris, S. Madadi, M. P. Hajiabbas, and B. Mohammadi-Ivatloo, “Optimal
Distributed Generation Allocation Using Quantum Inspired Particle Swarm
Optimization,” In Quantum Computing: An Environment for Intelligent Large Scale Real
Application, Springer, Cham, pp. 419-432, 2018.
[28]. R. Arulraj, and N. Kumarappan, “Optimal Single and Multiple DG Installations in Radial
Distribution Network Using SLPSO Algorithm,” In Intelligent and Efficient Electrical
Systems. Springer, Singapore, pp. 89-96, 2018.
[29]. F. C. Coelho, I. C. da Silva Junior, B. H. Dias, and W.B. Peres, “Optimal Distributed
Generation Allocation Using a New Metaheuristic,” Journal of Control, Automation and
Electrical Systems, vol. 29(1), pp. 91-98, 2018.
[30]. K. Mahmoud, and Y. Naoto, “Optimal Siting and Sizing of Distributed Generations,”
In Electric Distribution Network Planning, Springer, Singapore, pp. 167-184, 2018.
[31]. T. Yuvaraj, K. R. Devabalaji, and K. Ravi, “Optimal Allocation of DG in the Radial
Distribution Network Using Bat Optimization Algorithm,” In Advances in Power Systems
and Energy Management, Springer, Singapore, pp. 563-569, 2018.
[32]. B. Mahdad, and K. Srairi, “Adaptive differential search algorithm for optimal location of
distributed generation in the presence of SVC for power loss reduction in distribution
system,” Engineering science and technology, an international journal, vol. 19(3), pp.
1266-1282, 2016.
[33]. E. S. Ali, S. A. Elazim, and A. Y. Abdelaziz, “Optimal allocation and sizing of renewable
distributed generation using ant lion optimization algorithm,” Electrical Engineering,
vol. 100(1), pp. 99-109, 2018.
649
[34]. M. A. Tolba, A. A. Z. Diab, V. N. Tulsky, and A. Y. Abdelaziz, “LVCI approach for

optimal allocation of distributed generations and capacitor banks in distribution grids
based on moth–flame optimization algorithm,” Electrical Engineering, pp. 1-26, 2018.
[35]. O. D. M. Dominguez, M. P. Kasmaei, M. Lavorato, and J. R. S. Mantovani,“Optimal
siting and sizing of renewable energy sources, storage devices, and reactive support
devices to obtain a sustainable electrical distribution systems,” Energy Systems, pp. 1-22,
2017.
[36]. W. Kou, S. H. Jung, and S. Y. Park, “Optimal location strategy for distributed generation
to maximize system voltage stability based on line sensitivity factors,” Energy
Systems, vol. 9(3), pp. 511-528, 2018.
[37]. A. Chatterjee, S. P. Ghoshal, and V. Mukherjee, “Transient performance improvement of
thermal system connected to grid using distributed generation and capacitive energy
storage unit,” International Journal on Electrical Engineering and Informatics, vol. 2(3),
pp. 159, 2010.
[38]. M. Tabasi, and P. Asgharian, “Optimal Operation of Energy Storage Units in Distributed
System Using Social Spider Optimization Algorithm,” International Journal on
Electrical Engineering & Informatics, vol. 11(3), 2019.
[39]. O. Mohamed, M. Mohamed, and A. Kansab, “Optimal Placement and Sizing of
Distributed Generation Sources in Distribution Networks Using SPEA
Algorithm,” International Journal on Electrical Engineering & Informatics, vol. 11(2),
2019.
[40]. P. Khetrapal, “Distribution Network Reconfiguration of Radial Distribution Systems for
Power Loss Minimization Using Improved Harmony Search Algorithm,” International
Journal on Electrical Engineering & Informatics, vol. 12(2), 2020.
[41]. A. Tjahjono, D. O. Anggriawan, A. K. Faizin, A. Priyadi, M. Pujiantara, and M. H.
Purnomo, “Optimal coordination of overcurrent relays in radial system with distributed
generation using modified firefly algorithm,” International Journal on Electrical
Engineering and Informatics, vol. 7(4), pp. 691, 2015.
[42]. Y. Sabri, P. O. Hadi, and N. Hariyanto, “Decision area of distributed generation
investment as deferral option in industrial distribution system using real option
valuation,” International Journal on Electrical Engineering and Informatics, vol. 8( 1),
pp. 1, 2016.
[43]. A. Selim, S. Kamel, and F. Jurado, “Efficient optimization technique for multiple DG
allocation in distribution networks,” Applied Soft Computing, vol. 86, pp. 105938, 2020.
[44]. J. M. Home-Ortiz, M. Pourakbari-Kasmaei, M. Lehtonen, and J. R. Sanches Mantovani,
“Optimal location-allocation of storage devices and renewable-based DG in distribution
systems,” Electric Power Systems Research, vol. 172, pp. 11-21, 2019.
[45]. B. Mukhopadhyay, and D. Das, “Multi-objective dynamic and static reconfiguration with
optimized allocation of PV-DG and battery energy storage system,” Renewable and
Sustainable Energy Reviews, vol. 124, pp. 109777, 2020.
[46]. S. Nikkhah, and A. Rabiee, “Multi-objective stochastic model for joint optimal allocation
of DG units and network reconfiguration from DG owner’s and DisCo’s
perspectives,” Renewable energy, vol. 132, pp. 471-485, 2019.
[47]. U. Raut, and S. Mishra, “An improved sine-cosine algorithm for simultaneous network
reconfiguration and DG allocation in power distribution systems,” Applied Soft
Computing, pp. 106293, 2020.
[48]. K. S. Sambaiah, and T. Jayabarathi, “Optimal reconfiguration and renewable distributed
generation allocation in electric distribution systems,” International Journal of Ambient
Energy, pp. 1-14, 2019.
[49]. S. Saha, and V. Mukherjee, “A novel multiobjective chaotic symbiotic organisms search
algorithm to solve optimal DG allocation problem in radial distribution
system,” International Transactions on Electrical Energy Systems, vol. 29(5), pp. e2839,
2019.
650
Arnab Pal, et al.
[50]. H. Abdel-mawgoud, S. Kamel, M. Ebeed, and M. M. Aly, “An efficient hybrid approach
for optimal allocation of DG in radial distribution networks,” in 2018 International
Conference on Innovative Trends in Computer Engineering (ITCE), Aswan, pp. 311-316,
Feb. 2018.
[51]. A. Pal, A. Bhattacharya, and A. K. Chakraborty, “Allocation of EV Fast Charging Station
with V2G Facility in Distribution Network,” In 2019 8th International Conference on
Power Systems (ICPS), IEEE, pp. 1-6, , 2019.
[52]. A. Sobieh, M. Mandour, E. M. Saied, and M. M. Salama, “Optimal number size and
location of distributed generation units in radial distribution systems using Grey Wolf
optimizer,” Int. Electr. Eng. J, vol. 7(9), pp. 2367-2376, 2017.
[53]. M. R. Nayak, “Optimal Feeder Reconfiguration of Distribution System with Distributed
Generation Units using HC-ACO,” International Journal on Electrical Engineering &
Informatics, vol. 6(1), 2014.
[54]. T. C. Subramanyam, S. T. Ram, and J. B. V. Subrahmanyam, “Optimal Location for
Fixing Fuel Cells in a Distributed Generation Environment using Hybrid Technique,”
International Journal on Electrical Engineering and Informatics, vol. 8(3), pp. 567, 2016.
[55]. S. Mirjalili, A. H. Gandomi, S. Z. Mirjalili, S. Saremi, H. Faris, and S. M. Mirjalili, “Salp
Swarm Algorithm: A bio-inspired optimizer for engineering design problems,” Advances
in Engineering Software, vol. 114, pp. 163-191, 2017.
[56]. S. Mirjalili, and A. Lewis, “The whale optimization algorithm,” Advances in Engineering
Software, vol. 95, pp. 51-67, 2016.
[57]. S. Mirjalili, “Moth-flame optimization algorithm: A novel nature-inspired heuristic
paradigm,” Knowledge-Based Systems, vol. 89, pp. 228-249, 2015.
[58]. A. Pal, A. K. Chakraborty, A. R. Bhowmik, and B. Bhattacharya, “New algorithms for
DG allocation with less execution time to minimize the power loss,” In Electrical,
Electronics, Communication, Computer, and Optimization Techniques (ICEECCOT),
International Conference on IEEE, pp. 495-500, 2017.
[59]. A. Pal, A. K. Chakraborty, A. R. Bhowmik, and B. Bhattacharya,“Optimal DG allocation
for minimizing active power loss with better computational speed and high accuracy,” In
4th International Conference on Recent Advances in Information Technology (RAIT)
IEEE, pp. 1-6, 2018.
[60]. Y. A. Awoke, T. F. Agajie, and E. A. Hailu, “Distribution Network Expansion Planning
Considering DG-Penetration Limit Using a Metaheuristic Optimization Technique: A
Case Study at Debre Markos Distribution Network,” International Journal on Electrical
Engineering and Informatics, vol. 12, no. 2: pp. 326-340, 2020.
[61]. A. Bhowmik, A. Maitra, S. M. Halpin and J. E. Schatz, “Determination of allowable
penetration levels of distributed generation resources based on harmonic limit
considerations,” in IEEE Transactions on Power Delivery, vol. 18(2), pp. 619-624, April
2003.
[62]. Sharma, Sharmistha, Subhadeep Bhattacharjee, and Aniruddha Bhattacharya. “Quasi-
Oppositional Swine Influenza Model Based Optimization with Quarantine for optimal
allocation of DG in radial distribution network,” International Journal of Electrical
Power & Energy Systems, vol. 74, pp. 348-373, 2016.
[63]. S. Mirjalili, S. M. Mirjalili, and A. Lewis, “Grey wolf optimizer,” Advances in
engineering software, vol. 69, pp. 46-61, 2014.
[64]. U. Sultana, A. B. Khairuddin, A. S. Mokhtar, N. Zareen, and B. Sultana, “Grey wolf
optimizer based placement and sizing of multiple distributed generation in the distribution
system,” Energy, vol. 111, pp. 525-536, 2016.
[65]. M. H. Moradi, and M. Abedini, “A combination of genetic algorithm and particle swarm
optimization for optimal DG location and sizing in distribution systems,” International
Journal of Electrical Power & Energy Systems, vol. 34(1), pp. 66-74, 2012.
[66]. S. Sultana, and P K Roy, “Multi-objective quasi-oppositional teaching learning based
optimization for optimal location of distributed generator in radial distribution
651
systems,” International Journal of Electrical Power & Energy Systems, vol. 63, pp. 534-
545, 2014.
Appendix – A
From To Bus P Q From To Bus P Q
r(Ω) x(Ω) r(Ω) x(Ω)
bus bus No (kW) (kvar) bus bus No (kW) (kvar)
1 2 0.036 0.01296 1 0 0 60 61 0.207 0.0747 60 80.551 49.156
2 3 0.033 0.01188 2 133.84 101.14 61 62 0.247 0.8922 61 95.86 90.758
2 4 0.045 0.0162 3 16.214 11.292 1 63 0.028 0.0418 62 62.92 47.7
4 5 0.015 0.054 4 34.315 21.845 63 64 0.117 0.2016 63 478.8 463.74
5 6 0.015 0.054 5 73.016 63.602 64 65 0.255 0.0918 64 120.94 52.006
6 7 0.015 0.0125 6 144.2 68.604 65 66 0.21 0.0759 65 139.11 100.34
7 8 0.018 0.014 7 104.47 61.725 66 67 0.383 0.138 66 391.78 193.5
8 9 0.021 0.063 8 28.547 11.503 67 68 0.504 0.3303 67 27.741 26.713
2 10 0.166 0.1344 9 87.56 51.073 68 69 0.406 0.1461 68 52.814 25.257
10 11 0.112 0.0789 10 198.2 106.77 69 70 0.962 0.761 69 66.89 38.713
11 12 0.187 0.313 11 146.8 75.995 70 71 0.165 0.06 70 467.5 395.14
12 13 0.142 0.1512 12 26.04 18.687 71 72 0.303 0.1092 71 594.85 239.74
13 14 0.18 0.118 13 52.1 23.22 72 73 0.303 0.1092 72 132.5 84.363
14 15 0.15 0.045 14 141.9 117.5 73 74 0.206 0.144 73 52.699 22.482
15 16 0.16 0.18 15 21.87 28.79 74 75 0.233 0.084 74 869.79 614.775
16 17 0.157 0.171 16 33.37 26.45 75 76 0.591 0.1773 75 31.349 29.817
11 18 0.218 0.285 17 32.43 25.23 76 77 0.126 0.0453 76 192.39 122.43
18 19 0.118 0.185 18 20.234 11.906 64 78 0.559 0.3687 77 65.75 45.37
19 20 0.16 0.196 19 156.94 78.523 78 79 0.186 0.1227 78 238.15 223.22
20 21 0.12 0.189 20 546.29 351.4 79 80 0.186 0.1227 79 294.55 162.47
21 22 0.12 0.0789 21 180.31 164.2 80 81 0.26 0.139 80 485.57 437.92
22 23 1.41 0.723 22 93.167 54.594 81 82 0.154 0.148 81 243.53 183.03
23 24 0.293 0.1348 23 85.18 39.65 82 83 0.23 0.128 82 243.53 183.03
24 25 0.133 0.104 24 168.1 95.178 83 84 0.252 0.106 83 134.25 119.29
25 26 0.178 0.134 25 125.11 150.22 84 85 0.18 0.148 84 22.71 27.96
26 27 0.178 0.134 26 16.03 24.62 79 86 0.16 0.182 85 49.513 26.515
4 28 0.015 0.0296 27 26.03 24.62 86 87 0.2 0.23 86 383.78 257.16
28 29 0.012 0.0276 28 594.56 522.62 87 88 0.16 0.393 87 49.64 20.6
29 30 0.12 0.2766 29 120.62 59.117 65 89 0.669 0.2412 88 22.473 11.806
30 31 0.21 0.243 30 102.38 99.554 89 90 0.266 0.1227 89 62.93 42.96
31 32 0.12 0.054 31 513.4 318.5 90 91 0.266 0.1227 90 30.67 34.93
32 33 0.178 0.234 32 475.25 456.14 91 92 0.266 0.1227 91 62.53 66.79
33 34 0.178 0.234 33 151.43 136.79 92 93 0.266 0.1227 92 114.57 81.748
34 35 0.154 0.162 34 205.38 83.302 93 94 0.233 0.115 93 81.292 66.526
30 36 0.187 0.261 35 131.6 93.082 94 95 0.496 0.138 94 31.733 15.96
36 37 0.133 0.099 36 448.4 369.79 91 96 0.196 0.18 95 33.32 60.48
29 38 0.33 0.194 37 440.52 321.64 96 97 0.196 0.18 96 531.28 224.85
38 39 0.31 0.194 38 112.54 55.134 97 98 0.1866 0.122 97 507.03 367.42
39 40 0.13 0.194 39 53.963 38.998 98 99 0.0746 0.318 98 26.39 11.7
40 41 0.28 0.15 40 393.05 342.6 1 100 0.0625 0.0265 99 45.99 30.392
41 42 1.18 0.85 41 326.74 278.56 100 101 0.1501 0.234 100 100.66 47.572
42 43 0.42 0.2436 42 536.26 240.24 101 102 0.1347 0.0888 101 456.48 350.3
43 44 0.27 0.0972 43 76.247 66.562 102 103 0.2307 0.1203 102 522.56 449.29
44 45 0.339 0.1221 44 53.52 39.76 103 104 0.447 0.1608 103 408.43 168.46
45 46 0.27 0.1779 45 40.328 31.964 104 105 0.1632 0.0588 104 141.48 134.25
35 47 0.21 0.1383 46 39.653 20.758 105 106 0.33 0.099 105 104.43 66.024
47 48 0.12 0.0789 47 66.195 42.361 106 107 0.156 0.0561 106 96.793 83.647
652
Arnab Pal, et al.
48 49 0.15 0.0987 48 73.904 51.653 107 108 0.3819 0.1374 107 493.92 419.34
49 50 0.15 0.0987 49 114.77 57.965 108 109 0.1626 0.0585 108 225.38 135.88
50 51 0.24 0.1581 50 918.37 1205.1 109 110 0.3819 0.1374 109 509.21 387.21
51 52 0.12 0.0789 51 210.3 146.66 110 111 0.2445 0.0879 110 188.5 173.46
52 53 0.405 0.1458 52 66.68 56.608 110 112 0.2088 0.0753 111 918.03 898.55
52 54 0.405 0.1458 53 42.207 40.184 112 113 0.2301 0.0828 112 305.08 215.37
29 55 0.391 0.141 54 433.74 283.41 100 114 0.6102 0.2196 113 54.38 40.97
55 56 0.406 0.1461 55 62.1 26.86 114 115 0.1866 0.127 114 211.14 192.9
56 57 0.406 0.1461 56 92.46 88.38 115 116 0.3732 0.246 115 67.009 53.336
57 58 0.706 0.5461 57 85.188 55.436 116 117 0.405 0.367 116 162.07 90.321
58 59 0.338 0.1218 58 345.3 332.4 117 118 0.489 0.438 117 48.785 29.156
59 60 0.338 0.1218 59 22.5 16.83 118 33.9 18.98
Arnab Pal has received the B.Tech degree in Electrical Engineering from the
MAKAUT, WB, India in 2016 and M.Tech degree in Power System
specialization from the NIT Agartala, India in 2018. He is now a PhD
research scholar in the Department of Electrical Engineering at NIT Agartala.
His area of interest is Distribution Network, Distributed Generation,
Microgrid, Electric Vehicle and Application of Optimization Techniques.
Ajoy Kumar Chakraborty has completed his B.E.E (Hons) from Jadavpur
University, M. Tech in Power System from IIT Kharagpur and Ph.D in
Electrical Engineering from Jadavpur University. He has more than 100
publications in national and international journals and conferences. Author is
currently serving as Professor in NIT Agartala. His area of interest is Power
System, Distributed Generation, Smart Grid, Electric Vehicle and
Optimization.
Arup Ratan Bhowmik has obtained his B.Tech in Electronics and

Communication Engineering from West Bengal University of Technology,
M.Tech degree in Electrical Engineering from Tripura University (A Central
University), India in 2011 and Ph.D in Engg. from NIT Agartala in 2015
respectively. His areas of interest include Application of Soft Computing
techniques to different Power System problems, Power Quality, FACTS,
Optimal Power flow, Economic load dispatch and Power system Operation &
Control. He has published more than 30 research papers in various national
and international journals and conferences. He is reviewer of different reputed journals
including IEEE Transaction of Power System, IEEE Transaction on Industrial Application.
653

Optimal Placement and Sizing of DG Considering Power and Energy Loss Minimization in Distribution System

Uploaded by

Copyright:

Available Formats

Optimal Placement and Sizing of DG Considering Power and Energy Loss Minimization in Distribution System

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Optimal Placement and Sizing of DG Considering Power and Energy Loss Minimization in Distribution System

Uploaded by

Copyright:

Available Formats

International Journal on Electrical Engineering and Informatics - Volume 12, Number 3, September 2020

Optimal Placement and Sizing of DG considering Power and Energy Loss

Department of Electrical Engineering

Abstract: Optimal placement of Distributed Generation (DG) in the distribution network is a

Keywords: Distributed generation; optimal DG placement; renewable generation; energy loss

Received: February 8th, 2019. Accepted: September 11st, 2020

Figure 1. Example of 6 test bus system

Active Power Loss = Re( Pl i )

B. Energy Loss Minimization

∑ P DG ≤ 100% of the lotal load (15.2)

• Power balance constraint:

3. Methods and Tools

• Pseudocode for the SSA [55]:

B. Whale Optimization Algorithm [56]

Where b = Constant, l = Random number in between -1 and 1.

Here p = Random number in between 0 and 1.

• Pseudocode for the WOA [56]:

C. Moth-Flame Optimization (MFO) [57]

D. Modified One By One Search Algorithm

E. Grey wolf optimization (GWO) [63]

Figure 2. Flow chart of GWO [64]

4. Results and Discussion

Table 1. Total load and loss of different distribution system

A. Optimal Locations and Sizes of DG for Power Loss Minimization

Table 2. Optimal allocation of DG on IEEE 33 bus system using GWO

Table 3. Optimal allocation of DG on IEEE 33 bus system using SSA

Table 4. Optimal allocation of DG on IEEE 33 bus system using WOA

Table 5. Optimal allocation of DG on IEEE 33 bus system using MFO

Table 6. Optimal allocation of DG on IEEE 33 bus system using MOBOSA

Table 7. Optimal allocation of DG on IEEE 69 bus system using GWO

Table 8. Optimal allocation of DG on IEEE 69 bus system using SSA

Table 9. Optimal allocation of DG on IEEE 69 bus system using WOA

Table 10. Optimal allocation of DG on IEEE 69 bus system using MFO

Table 11. Optimal allocation of DG on IEEE 69 bus system using MOBOSA

Table 12. Optimal allocation of DG on 118 bus system using GWO

Table 13. Optimal allocation of DG on 118 bus system using SSA

Table 14. Optimal allocation of DG on 118 bus system using WOA

Table 15. Optimal allocation of DG on 118 bus system using MFO

Table 16. Optimal allocation of DG on 118 bus system using MOBOSA

Figure 3. Bus voltages without DG of IEEE 33 bus system

Figure 4. Bus voltages with 5 DG of IEEE 33 bus system using MFO

Figure 5. Bus voltages without DG of IEEE 69 bus system

Figure 6. Bus voltages with 5 DG of IEEE 69 bus system using MFO

Figure 7. Bus voltages without DG of 118 bus system

Figure 8. Bus voltages with 5 DG of 118 bus system using MFO

Figure 9. Percentage of loss reduction of IEEE 33 bus system

Figure 10. Percentage of loss reduction of IEEE 69 bus system

Figure 11. Percentage of loss reduction of IEEE 118 bus system

B. Optimal Locations of DG for Energy Loss Minimization

• DG Allocation on IEEE 33 Radial Network for Energy Loss Minimization:

Table 19. DG allocation on 33 bus system for energy loss minimization

Figure 15. Bus voltages of 24 hours of IEEE 33 bus system without DG

• DG Allocation on IEEE 69 Radial Network for Energy Loss Minimization:

Table 20. DG allocation on 69 bus system for energy loss minimization

Figure 20. Bus voltages of 24 hours of IEEE 69 bus system without DG

• DG Allocation on 118 Radial Network for Energy Loss Minimization:

Figure 25. Bus voltages of 24 hours of 118 bus system without DG

5. Conclusion and Future Work

[17]. M. Gandomkar, M. Vakilian, and M. Ehsan, “A combination of genetic algorithm and