Page 1 of 8
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 1980
DG Allocation in Primary Distribution Networks for
Loss Reduction and Voltage Profile Improvement
using Particle Swarm Optimization
RAJA KUMAR .R.B.R
M.Tech – Power System Engineering,
PRIST Deemed Univeristy, Madurai, Tamilnadu
Abstract – Distributed Generation (DG) plays a vital role in a
power system network. Renewable energy based DG units are
located close to the consumers or load centers in order to
improve voltage profile, reduce the network power losses and
improves substation capacity release etc. Thus, while allocating
DG units care has to be taken in order to maximize the benefits.
In this thesis, by installing DG, an optimal way of managing real
and reactive power and improving the nodal voltages in primary
distribution network explained. Optimal location of DG is
identified by using voltage stability index (VSI). The optimal
rating of DG is computed by using Particle Swarm Optimization
(PSO) technique to ensure reduction in power losses and to
attain better voltage regulation. To demonstrate the efficiency of
proposed techniques a clear and detailed analysis of performance
has been carried out on IEEE-33 & 69 bus systems.
I.INTRODUCTION
Overview:
An Electric power system comprises of a generating system, a
transmissionsystem and a distribution system. Generating station
convert fuel energy into electricity, transmission system
connects the generating stations and distribution substations and
distribution system distribute power to consumers. In view of
network structure, transmission and distribution systems are
different .Generally, transmission system will have loop
structure and distribution system tends to have a radial structure.
Distribution Systems:
Distribution systems are employed with radial structure in order
to obtain operational simplicity. By means of an interconnected
transmission network, primary distribution substation receives
power from generating stations. Radial Distribution System
(RDS) network is passive in nature and transfers power to
consumers from the substation. Thus, in RDS the power flow is
unidirectional. In case of distribution lines, due to high R /X
ratio, high voltage drops, large power loss will occur. Everyday
distribution networks are experiencing many changes in the load.
At most of the nodes, RDS experience a sudden collapse in the
voltage during critical load conditions because of low voltage
stability index. In this thesis, for RDS a voltage stability index
(VSI) is proposed for all the nodes. It is observed that node with
minimum VSI value is more sensitive and leads to collapse in
voltage. During past years, several techniques were implemented
by placing dispersed sources injecting reactive power like
capacitor banks in order to obtain improvement in voltage and to
reduction in power losses. Even through the implementation of
capacitor placing method which is promising in nature, the
voltage profile improvement obtained is below desired voltage
level (1.0 p.u.). As RDS is passive in nature, it is less reliable.
Many solutions are suggested recently by incorporating
electrical sources based on renewable energy technology to
overcome the passiveness of RDS and also to improve reliability
of the system and voltage profile. These embedded generations
in RDS are called as Distributed or Dispersed Generation (DG).
The objectives of the project are:
Developing an optimal way for managing the reactive
and real power and also improving the voltages of the
nodes in RDS with DG.
Implementation of voltage stability index technique for
optimal placing of DG units.
Implementing the PSO algorithm to obtain the optimal
size of the DG and thereby reducing the losses.
A detailed analysis of the performance of these methods
is to be carried out on IEEE-33 & 69 bus systems to
explain the effectiveness of the proposed techniques.
II.LOAD FLOW ANALYSIS
Real-time applications such as optimization of network,
switching, estimation of the state, and so on, requires an efficient
and standard power flow technique. Due to special features of
distribution systems such as Radial structure, high ratio of R/X
and wide-ranging reactance and resistance values. Newton
Page 2 of 8
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 1981
Raphson (NR) and Gauss Seidel (GS) techniques may become
ineffective. In particular, in standard fast-decoupled NR method,
the assumptions that are used for the simplifications are not valid
in RDS. This makes the transmission systems power flow
computation different from distribution systems. Hence, for
distribution networks, an efficient power flow algorithm is
desired. In order to carry out the unbalanced and balanced RDS
analysis various methods are proposed. Basically they are
divided into two types. The first type includes modification of
traditional techniques such as GS and NR. Second type is based
on forward and backward sweep process using Kirchhoff’s laws.
For distribution networks power flow analysis, backward and
forward sweep based techniques gained more popularity because
of its high computational efficiency, low memory requirements
and strong convergence characteristic. In this, load flow study is
carried by using backward and forward sweep method.
In this algorithm, bus-branch oriented data is the only input.
Solving the power flow for RDS directly and developing the
formulation that includes advantages of topological
characteristics of the distribution networks are the main goals of
this chapter.It means in the new method, forward/backward
substitute of Jacobian matrix and time consuming LU
decomposition are not performed as in the traditional NR and
GS methods. In this new method, to get the load flow solution,
BIBC, bus-injection to branch-current matrix and simple matrix
operations are performed. Compared to all conventional
methods, this method is very efficient and robust. The results
explain the validity and feasibility of the proposed method.
IV.VOLTAGE STABILITY INDEX (VSI)
Voltage stability index represents a numerical solution
to identify the sensitive node of the system. It also helps the
operator to check the system to prevent from voltage collapse by
initiating automatic remedial actions. Main aim of VSI is to find
the distance between the current working point and stable point.
Voltage collapse generally starts at
most sensitive node in the system and then passes to all other
nodes which are sensitive.The node with most sensitivity to
voltage collapse exhibits any one of the below
features:
1) Highest critical point
2) Low margin of reactive power.
3) More deficiency of reactive power
V.PLACING AND SIZING OF DG WITHOUT
IMPLEMENTING OPTIMISATION TECHNIQUES
In order find the optimum location for placing DG units a
number of methods are developed. In this thesis, a technique
based voltage stability index (VSI) has been developed for
optimal placing of DG units. A node with minimum value of
VSI is considered as the optimum location for placement of
DG.In order to obtain optimal DG size, below steps should be
followed:
1) A node with minimum VSI should be found first and then he
DG is placed at that node.
2) Assuming Distributed generation power factor as constant,
size should be varied in constant steps from a minimum value to
a maximum value (feeder loading capacity) till minimum losses
are obtained
3) The size of DG which produces minimum loss is considered
as optimal size.
The system considered here is an IEEE-33 bus system. It has an
voltage of 12.66kV and total real power demand of 3.715 MW
and reactive power demand 2.3 MVAr. In the first case load
flow without DG is analysed and bus voltages magnitudes and
total power loss of the network in RDS are computed. VSI at
various buses is also calculated. From the analysis, we found
that the bus 18 is having lowest VSI value of 0.66721.Hence,
bus 18 is optimal location for placing DG.
For analysis, the following two cases are considered:
Case 1: DG operating at 0.9 power factor lagging
Case 2: DG operating at unity power factor
For finding the optimal size of DG in both the cases, the DG size
is varied from 0.5 MVA to 4.0 MVA in step of 0.5 MVA. From
the test result we observe that power losses are non-linearly
varying with capacity of generator. First the power losses are
decreased up to some minimum values and then start increasing
with DG capacity increment. Hence from the test results, it is
observed that in the base case without DG, total respective real
and reactive power losses are 210.99 kW and 143.03 kVAr.
Whereas the losses after placing a DG with 1 MVA at a lagging
power factor of 0.9 produces more loss reduction when
compared with the DG size of 0.85 MVA at unity power factor.
Hence, optimal placing and sizing of DG reflects the power loss
reduction in radial distribution system. We can also note that
substation capacity release is more for case 1 compared to case
2. As DG injects real and reactive power to the load centres
locally, it helps in improving the bus voltages and also reduce
the losses at the load side. Comparison of the voltage variation
for different cases viz. base case (without DG), Case 1 and Case
2 proves that improvement in voltage for case 1 is more
compared to case 2. Similarly analysis is done for IEEE-69 bus
network. It has an voltage of 12.66 Kv and total real power
demand of 3.802 MW and reactive power demand 2.694 MVAr.
There was a problem loading this page. Retrying...
Page 3 of 8
European Journal of Business &
Social Sciences
Available at https://ejbss.org/
ISSN: 2235-767X
Volume 07 Issue 04
April 2019
Available online: https://ejbss.org/ P a g e | 1982
From the analysis, we found that the bus 65 is having lowest VSI
value of 0.68345. Hence, bus 65 is the optimal location for
placing DG.
VI.PARTICLE SWARM OPTIMISATION
Optimization technique is used to find the best solution
for any given circumstances. For example, in a company if it is
required to improve its rating, technological and managerial
plans have to be taken many times. Here, the goal of the
plans is to either maximize the profits or to minimize the
spending effort. Optimization is referred as both minimizing and
maximizing the tasks. Since the minimization of any function is
same as maximizing its additive inverse , the terminology
minimization and optimization can be used interchangeably [12].
Because of this reason, optimization became very important in
many fields. Basically, in order to solve the optimization
problems, PSO algorithm is inspired by the animal’s activity. In
PSO, swarm means population; particle represents each member
of the population. Each particle searches through the entire space
by randomly moving in different directions and remembers the
previous best solutions of that particle and also positions of its
neighbor particles. Particles of a swarm adjust their position and
velocity dynamically by communicating best positions of all the
particles with each other. In this way, finally all particles in the
swarm try to move towards better positions
until the swarm reaches an optimal solution Thus, due to its easy
implementation and its ability to obtain fast convergence, PSO
technique is becoming very popular. Moreover PSO uses only
basic mathematics and it does not involve any derivative or
gradient information.
PSO Algorithm Parameters
For any given optimization problem, some of the
parameters in PSO algorithm may affect its efficiency. Some of
these parameter’s values and their choices have major impact on
the performance of the PSO technique, and other parameters
have small or no effect [9]. The basic parameters of PSO are
1. size of the swarm
2. number of iterations
3. components of velocity, and
4. acceleration coefficient.
In addition to these parameters, PSO technique is also influenced
by inertia weight, velocity clamping, and velocity constriction.
Swarm size
Swarm size is defined as the number of agents n in swarm. A
huge swarm generates more particles and most of the search
space is to be covered per iteration. Number of iterations may be
reduced in order to achieve best optimization value using large
number of agents. But the computational complexity per
iteration will be increased by using more amounts of agents and
also it is more time consuming. From most of the studies, it is
identified that PSO use swarm size in the interval of n ε [20,60].
Iteration numbers
Obtaining a best result depends on number of iterations which in
turn depends on
problem. If the number of iterations is too low, then the search
process may stop
prematurely. If the number of iterations is large, it may add
computational complexity
and thereby consumes more time.
Velocity Components
While updating agent’s velocity, the velocity
components plays a vital role. There are three terms in agent’s
velocity. They are inertia component, cognitive component,
social component.
Acceleration coefficients
The stochastic influence of the social and cognitive
components of the agent’s velocity depends upon acceleration
coefficients c1 and c2, together with the randomly generated
numbers r1 and r2, respectively. The confidence that an agent
has in itself is represented by c1andthe confidence that an agent
has in its neighbors is represented by c2[10]. The properties of
c1 and c2 :
When c1 =c2 =0, until search space’s boundary is met, all the
agents will continue to move with their current speed.
Neighborhood Topologies
For each agent a neighborhood must be defined [7]. The
extent of social interaction within swarm is computed by the
neighborhood. When the size of neighborhoods in the swarm is
small, it leads to less interaction [10]. Even though the
convergence is slower, the quality of solutions will be improved
for small neighborhood. The risk involving earlier convergence
will be occurred in case of larger neighborhood [7]. In order to
solve this earlier convergence problem, search process should be
started with small size of neighborhoods and later over the time,
the size can be increased. As the agents move towards near to
optimum region, this technique ensures faster convergence [10].
In the swarm, the social interaction among the agents is dealt in
PSO technique. Each agent in the swarm exchanges the
information about their success with other agents through
communication. All agents tend to move towards the agent when
that agent find a better position. Agents neighborhood determine
the performance of the [10]. Different types of neighborhood
topology
There was a problem loading this page. Retrying...
