You are on page 1of 17

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245325885

Optimal Sizing of Distributed Generation Placed on Radial


Distribution Systems

Article  in  Electric Power Components and Systems · January 2010


DOI: 10.1080/15325000903273403

CITATIONS READS

110 446

4 authors, including:

T. N. Shukla Shiv Pujan Singh


Kamla Nehru Institute of Technology, Sultanpur Indian Institute of Technology (Banaras Hindu University) Varanasi
5 PUBLICATIONS   158 CITATIONS    101 PUBLICATIONS   830 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Design and Development of Control Stratigies for Smart Energy Systems View project

Wide Area Measurement in Power System View project

All content following this page was uploaded by T. N. Shukla on 29 September 2016.

The user has requested enhancement of the downloaded file.


This article was downloaded by: [Dalhousie University]
On: 13 September 2012, At: 16:30
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Electric Power Components and Systems


Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/uemp20

Optimal Sizing of Distributed Generation


Placed on Radial Distribution Systems
a b b c
T. N. Shukla , S. P. Singh , V. Srinivasarao & K. B. Naik
a
Department of Electrical Engineering, Kamla Nehru Institute of
Technology, Sultanpur, India
b
Department of Electrical Engineering, Institute of Technology,
Banaras Hindu University, Varanasi, India
c
College of Engineering Science and Technology, Lucknow, India

Version of record first published: 27 Jan 2010.

To cite this article: T. N. Shukla, S. P. Singh, V. Srinivasarao & K. B. Naik (2010): Optimal Sizing
of Distributed Generation Placed on Radial Distribution Systems, Electric Power Components and
Systems, 38:3, 260-274

To link to this article: http://dx.doi.org/10.1080/15325000903273403

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Electric Power Components and Systems, 38:260–274, 2010
Copyright © Taylor & Francis Group, LLC
ISSN: 1532-5008 print/1532-5016 online
DOI: 10.1080/15325000903273403

Optimal Sizing of Distributed Generation Placed on


Radial Distribution Systems

T. N. SHUKLA,1 S. P. SINGH,2 V. SRINIVASARAO,2 and


K. B. NAIK3
1
Department of Electrical Engineering, Kamla Nehru Institute of Technology,
Sultanpur, India
Downloaded by [Dalhousie University] at 16:30 13 September 2012

2
Department of Electrical Engineering, Institute of Technology,
Banaras Hindu University, Varanasi, India
3
College of Engineering Science and Technology, Lucknow, India

Abstract Distribution network planning identifies the least cost network investment
that satisfies load growth requirements without violating any system and operational
constraints. Power injections from distributed generation change network power flows,
modifying energy losses. Determining appropriate location and optimal size of dis-
tributed generation with respect to network configuration and load distribution in the
feeder is main challenge in the changing regulatory and economic scenarios. Among
the benefits of distributed generation is the reduction in active power losses, which
can improve the system performance; reliability, and efficiency. In this article, the
multi-location distributed generation placement problem aims to minimize the total
active power loss of radial distribution networks using a genetic algorithm based
solution algorithm. This technical benefit of energy savings due to the reduction in
active power loss can also be translated into economic benefits. The loss sensitivity
to the change in active power injection is used in selecting candidate location(s)
for installation of distributed generation devices. A comparison of the results for
loss reduction and savings with other reported methods show the effectiveness of the
proposed method.

Keywords distributed generation, active power injection, sensitivity, losses, radial


distribution network, multiple candidate location

1. Introduction
Electric power systems are evolving from today’s centralized bulk system, with generation
plants connected to the transmission network, to the future’s more decentralized system,
with smaller generating units connected directly to distribution networks (DNs) near
demand consumption. This type of generating unit is defined as distributed generation
(DG) [1]. Penetration of DG is a new challenge for traditional electric power systems,
as it changes network power flows, modifying energy losses. This causes an impact on

Received 16 December 2008; accepted 10 July 2009.


Address correspondence to Professor Shiv P. Singh, Department of Electrical Engineering,
Institute of Technology, Banaras Hindu University, Varanasi, 221005, India. E-mail: sps5957@
indiatimes.com

260
Optimal DG Placement in Distribution Systems 261

Nomenclature
Ce energy cost per kWh
Cdg DG cost per kW, including purchase, installation, operation, and
maintenance costs
F sum of energy loss cost and DG cost
jIij j current flowing in section i -j
jIij jschedule current carrying capacity of line section i -j
k number of load levels
LN total number of line sections
n number of total nodes
Ng number of nodes where DGs are placed
PDGj real power generation by the DG placed at the j th bus during peak
Downloaded by [Dalhousie University] at 16:30 13 September 2012

period
PDi real power load at the i th bus
Pi net real power available at the i th bus
PL total real power loss in the system
PLwithout total real power loss without DG
PLk real power loss in section k
Qi net reactive power available at the i th bus
Tk duration of a particular load level k
Vi , ıi voltage magnitude and angle at i th bus, respectively
Vimin , Vimax acceptable voltage limit at bus i (i.e., 10%)

network operation and planning practices of distribution companies with both technical
and economic implications [2–4]. Determination of appropriate location and optimal size
of DG with respect to network configuration and load distribution in a radial distribution
feeder is a main challenge in the changing regulatory and economic scenarios.
Liew and Strbac [5] demonstrated that an active DN would allow considerably
greater penetration of embedded wind generation. They reported that the remarkable
and widely used loss-reduction method gives active power injection by placing optimal
DG resources at appropriate locations in existing DNs with minimum installation time.
Rau and Wan [6] proposed a methodology to identify the optimal locations of DG devices
in a meshed network to minimize losses, line loadings, and reactive power requirement
using the second-order method. In [7], El-Khattam et al. took into account the DG
variability, along with load curves, through a deterministic and stochastic analysis, aiming
the maximum insertion within specific penetration limits. Consequently, the results of
purely “static” snapshot approaches using specific loading and generation scenarios may
result in challenges or opportunities being neglected. Wang and Nehrir [8] suggested
an analytical approach for the best location of DG units with unity power factor for
minimum loss configuration using the two-thirds rule, and they analyzed a distribution
feeder for constant and time-varying uniformly, centrally, and increasing loads. In their
approach, DG size is not optimized and is line loading constraint is not considered.
Carpinelli et al. [9] reported a three-step procedure based on genetic algorithms (GAs)
and decision theory, which were applied to establish the best DG site and size on a
medium-voltage (MV) DN, considering technical constraints such as feeder capacity
limits, feeder voltage profile, and three-phase short-circuit currents in the network nodes.
262 T. N. Shukla et al.

They showed that the presence of DG significantly reduces the losses costs. A multi-
objective formulation of the siting and sizing of DG resources into existing DNs is
reported by Celli et al. [10], who proposed a GA and "-constrained methodology that may
permit the planner to decide the best compromise between the cost of network upgrading,
power losses, energy not supplied, and energy required by the served customers; however,
their methodology considered the case of a single location only. A methodology using
linear programming has been reported for the optimal allocation of embedded generation
by Keane and O’Malley [11], which explains the background of technical constraints to
accommodate increasing levels of DG in the existing networks, but there is no mention
of losses. To meet renewable energy targets in a cost-effective manner, it is suggested
that the best use of an existing network must be met. The under-voltage/over-voltage
problem may exist in the network because of load growth for a small period of hours
per day or per month or per year, but some buses of the network would always be
unaffected from such problems. Zhu et al. [12] suggested that DGs can be placed at such
Downloaded by [Dalhousie University] at 16:30 13 September 2012

locations.
Evolutionary techniques, such as the GA [13, 14] and particle swarm optimization
(PSO) [15], claim that global solutions are gaining popularity in solving the optimization
problems. Acharya et al. [16] suggested an analytical approach (heuristic method), derived
an expression based on the approximate loss formula from Elgerd [17], and calculated DG
size at various buses for minimum system loss to select the candidate location. Updating
loss coefficients ˛ and ˇ with the change in state variable voltage magnitudes and angles
at various nodes had been ignored, which may have led to inappropriate evaluation of
DG size. Most of the methods for different DG benefits suggested by various authors
using single-location DG have been reported in the literature.
In the present article, the GA-based methodology has been used to evaluate the
optimal size DG at multiple locations of radial DNs for a minimum loss configuration.
The loss sensitivity approach is used to decide the appropriate locations for placing
DG. The main focus of this article is to find and quantify the technical benefits (loss
reduction and voltage regulation), from the customer point-of-view, and overall economic
benefits, from the utility point-of-view. The study is performed on two widely used 33-
bus–32-branch and 69-bus–68-branch test systems. The results obtained by the proposed
method are compared with that of the heuristic method reported in [16] for base system
loads, and it is demonstrated that the proposed method outperformed. This idea is further
extended for variable load case loss reduction, and overall benefits are demonstrated for
a 15-year planning period.

2. Problem Formulation
Attention should be given to siting and sizing DG resources when planning a distribution
system and its impact. The inappropriate location and size of DG can result in very high
real power loss and reduced reliability. It is not advisable to install very high-capacity
DG in the network. The total capacity of DGs installed in the network should be such
that it is consumable within the DN boundary. Any attempt to install high-capacity DG
resources with the purpose of exporting power beyond the substation (reverse power flow
through substation) leads to a very high loss. Hence, in selecting the optimal size of DG
for minimum system loss, the network configuration and the loads on the distribution
system plays an important role.
For evaluating the technical impact of DG integration on the operation of the systems
considered for analysis, the daily load curves are approximated and divided into three
Optimal DG Placement in Distribution Systems 263

load levels in discrete form to include the effect of load variation. For each load level,
the active power loss and node voltage regulation were calculated. Natural gas micro-
turbines were selected, since they are more environmental friendly and offer a lower
operating cost.
In order to fulfill the above-mentioned goals, the problem is formulated to meet
two objectives. First, to select an appropriate location at which the system loss becomes
minimum; second, to compute the optimal DG size for minimum system loss. Neither
no-load losses nor non-technical or commercial losses depend on power flows; therefore,
they are not considered in this work.

2.1. Selection of Location


In order to reduce the efforts to select appropriate nodes for the placement of DG devices
Downloaded by [Dalhousie University] at 16:30 13 September 2012

a priori, the sensitiveness of nodes toward the change in active power loss with respect
to change in active power injection at various nodes are identified. The loss sensitivity
factors at different buses have been evaluated using a load-flow program [18] to select
appropriate nodes for DG planning. These sensitivity factors reflect how the feeder power
losses change if more real power is injected at a particular node and also allow the
obtained candidate nodes to locate DG.
Loss sensitivity factors are initially evaluated for the base case (initial condition). In
order to decide the next location(s), the computation of loss sensitivity has been repeated,
taking into account the previously allocated DG, which is termed successive sensitivity
analysis. This process is repeated until the optimum number of locations is ascertained.
The optimal number of locations is that number of buses that give maximum benefits
(savings) for optimal size DGs placed at selected locations. The expression for line losses
described in [17] has been used for this purpose.
The change in active power loss of the system due to change in active power injection
at a node is expressed as

n
@PL X
D2 .˛ij Pi ˇij Qi /; (1)
@Pi j D1

where

˛ij D rij Cos.ıi ıj /=Vi Vj ;

ˇij D rij Sin.ıi ıj /=Vi Vj ;

and

rij C jxij D Zij

is the ij th element of ŒZBUS  matrix with ŒZBUS  D ŒYBUS 1 .


For the present problem, only the real power injection with a fixed generator output
would be of concern. These factors are considered while evaluating coefficients ˛ and ˇ.
It is found that the change in real power injection affects real power losses and the loss
sensitivity to the generator output.
264 T. N. Shukla et al.

2.2. Maximization of Benefit


Among the benefits of DG, the reduction in active power losses is one that can improve the
system performance, reliability, and efficiency. This technical benefit of energy savings
because of the reduction in active power loss can also be translated into economic benefits
(e.g., in avoided power purchases from the wholesale electricity market and deferred
investments in reinforcing and upgrading existing networks).
An established method using loss coefficients ˛ and ˇ, popularly referred to as the
exact loss formula [17], is used for loss calculation and is represented in Eq. (2):
n X
X n
PL D ˛ij .Pi Pj C Qi Qj / C ˇij .Qi Pj Pi Qj /: (2)
i D1 j D1

For the DG project (selection of candidate locations and evaluation of optimal DG


Downloaded by [Dalhousie University] at 16:30 13 September 2012

size) with minimum active power loss, the problem is formulated as a cost function in
terms of loss cost and the cost involved toward installing DG. Thus, the objective function
can be expressed as

To maximize
Ng
LN
!
X X
F D Ce PLwithout Ce PLk Tk C Cdg PDGi (3)
kD1 i

subject to
Ng n
X X
PDGi  .PDi C PL /;
i D1 i D1

Vimin  Vi  Vimax ;

jIij j  jIij jschedule:

At minimum losses, energy loss reduction cost and installation, operation, and mainte-
nance costs of DGs are used for evaluation of savings.

3. GA
The GA simulates the biological processes that allow consecutive generations in a
population to adapt to their environment. GAs are unconstrained optimization methods
that model the evolutionary adaptation in nature. They work with a population of solutions
and create new generations (solutions) by appropriate genetic operators. The objective
to be met is represented by the appropriate fitness function. An in-depth description of
the method is not provided, as GA has been applied in several problems and excellent
texts are available [13, 14]. However, its implementation for the DG allocation problem
is delineated in next section.
The advantages of using GAs are that they require no knowledge of gradient in-
formation about the response surface, they are resistant to becoming trapped in local
optima, and they can be employed for a wide variety of optimization problems. For the
advantages of parallel searching, robust searching, and searching mechanisms based on
the principle of natural evolution, the GA has found applications in many areas and has
Optimal DG Placement in Distribution Systems 265

become one of the most successful optimization algorithms. On the other hand, it is very
difficult to achieve an analytical relationship between the sensitivity of simulated power
system and the parameter values to be optimized. Since GAs do not need this kind of
information, they are suitable in the present optimization task.

3.1. Coding Strategy


When applying GAs to optimize the DG allocation and sizing problem, an important
aspect is the coding of potential solutions. In a general way, the potential solution is a
configuration with the DG units installed at some places. The coded variables are the size
of the DG units for installation at candidate locations. The coding of the active power of
DGs is done in binary form using 12 binary bits for a single location to take care of real
power capacity available at the source node (substation), which can be repeated as many
Downloaded by [Dalhousie University] at 16:30 13 September 2012

times as the number of locations in the case of more than one location. Twelve binary
bits are sufficient to represent a DG of 4096 kW.

3.2. Implementation
Step 1: Determine candidate locations.
 Run the load flow solver.
 Determine the total real power loss and loss sensitivity factors of various
buses.
 Arrange the buses in descending order of their sensitivities, and select the
bus with highest priority.
Step 2: Input GA control data.
Step 3: Initialize population with random strings and copy into mating pool.
Step 4: Do while generation number is less than maximum number of generations
chosen.
 Do while population number is less than population size.
 Pick up the string corresponding to population number from mating pool
and decode it into test configuration.
 Apply load demand.
 Call distribution load flow solver.
 Check voltage and line flows constraints.
 Compute fitness function.
 Increment population number by one.
 Use mating pool to create new population for next generation.
 Carry out reproduction, cross-over, and mutation in mating pool.
 Increment generation number by one.
Step 5: Obtain desired solution (i.e., optimal DG size, minimum system loss, and
savings).
Step 6: Repeat Step 1 to Step 5 with DG at the bus obtained earlier.
Step 7: Stop.

3.3. GA Parameters
Convergence of the GA-based solution depends upon the proper choice of its parame-
ters. The parameters that govern the performance of the GA are the number of gen-
266 T. N. Shukla et al.

erations, population size, mutation and cross-over probabilities, types of cross-over,


and mutation. The type of cross-over and mutation are the user’s choice, however,
other parameters are problem dependent, and there is no rule by which they can be
fixed. As a result, they are obtained by trial and error. Following this procedure, the
following values of the parameters above were found suitable for the problem under-
taken:

 Number of generations: 50
 Population size: 100 individuals
 Cross-over probability: 86%
 Mutation probability: 0.6%
 Selection type: tournament (two individuals)
 Cross-over type: one point cross-over

Downloaded by [Dalhousie University] at 16:30 13 September 2012

Mutation type: constant

4. Results and Discussion

4.1. General Description


The effectiveness of the proposed methodology is tested on two widely used 33-bus–32-
branch and 69-bus–68-branch test systems at different loading conditions. The system
data of the 33-bus system from Kashem et al. [19] and the 69-bus system from Baran
and Wu [20] have been taken. The active power (sum of total connected load and base
system losses) at the source node in the beginning of the period is about 3936 kW for the
33-bus system and 4025 kW for the 69-bus system. The GA and load flow program are
written in Visual CCC language. This implementation makes use of load flow algorithm
proposed by Afsari et al. [18]. In order to demonstrate the long-term impact of DG, a
planning period of 15 years has been considered with an uniform load growth of 2.5% per
year while maintaining the same configuration of the network. DG sources are assumed
to always be available throughout the planning period.
The cost benefits of optimally allocated DG for a utility are explained by a reduction
in the energy bought by the amount produced by the DG units. This also implies a
reduction in losses and savings in energy, since the utility avoids buying from large
generators due to self-production. The cost of DG generated power of US$300.00 per kW
[21] has been taken for this study, which includes operation and maintenance cost. The
energy loss cost of US$0.05 per unit has been taken for the cost benefit analysis.
The savings (monetary benefits) are evaluated as the difference of energy loss cost without
DG, the sum of energy loss cost with DG, and DG generation cost including operation
and maintenance.
The base case loss sensitivity results of the six most sensitive buses arranged in
descending order are shown in Table 1 for both systems studied. Although this article aims
to illustrate the technical and economic benefits of DG allocation at multiple locations,
the demonstration also deals with the single-location case in order to compare the results
with those of [16].
To include the effect of load variation in the analysis, the approximated load duration
curve is divided into three load levels (in discrete form): L1 , L2 , and L3 as average
(62.5%), base (100%), and peak (125%) of system load demands with the duration of
1000 hr, 6760 hr, and 1000 hr, respectively, and these are tabulated in Table 2.
Optimal DG Placement in Distribution Systems 267

Table 1
Sensitivities of top six buses of two systems

Systems Bus no. Sensitivity values

33-bus system 6 0.06779


2 0.0670792
5 0.0664193
20 0.0573545
8 0.0537697
7 0.0476795
69-bus system 50 0.0113279
11 0.064845
54 0.044784
Downloaded by [Dalhousie University] at 16:30 13 September 2012

37 0.0037352
10 0.0035403
12 0.0022563

4.2. Case Studies


Four cases have been considered for the study of the two systems undertaken for analysis:
Case 1: Single location at base load.
Case 2: Multiple locations at base load.
Case 3: Multiple locations at variable loads.
Case 4: Economic analysis and long-term impact.

Case 1. Being the most sensitive node, bus 6 (Table 1) is selected as the first candidate
location for DG placement in the 33-bus system. The same node has also been selected
as the most suited location by the heuristic method in [16]. Results of DG capacity, loss
reduction, and energy savings for proposed and heuristic methods are reported in Table 3.
The base case system loss of 211.20 kW reported in [16] is based on the approximate load
flow method, “Distflow,” that was used by the authors. However, this value was found
to be 216.00 kW using the backward sweep power flow method reported in [13], and
the same is shown in Table 3 for a fair comparison. The optimal DG size of 2380 kW
was obtained by the proposed method, whereas 2490 kW was reported by heuristic
method [16]. The system loss dropped from 216 kW to 132.83 kW in the heuristic
method and 132.64 kW in proposed method. It is to be noted that the evaluated DG
capacity is much less than reported in the heuristic method with a higher loss reduction

Table 2
Duration of various load levels
for the considered systems

Load levels L1 L2 L3

Load level 0.625 1.00 1.25


Duration time (hr) 1000 6760 1000
268 T. N. Shukla et al.

Table 3
Summary for 33-bus system at L2 (base case) load level for single location

Power loss (kW)


Loss Energy
Optimal DG size Without With reduction savings
Methodology location (kW) DG DG (kW) ($)

Heuristic [16] 6 2490 216.00 132.83 83.17 36,428.46


Proposed GA 6 2380 216.00 132.64 83.36 36,511.68

though marginal. It can also be seen that the energy savings obtained for one year by the
proposed method is $36,511.68, while it is $36,428.46 by the heuristic method.
Downloaded by [Dalhousie University] at 16:30 13 September 2012

The results for the 69-bus system under this loading condition are shown in Ta-
ble 4. The system loss without DG is 225 kW, whereas it reduced to 83.2246 kW
and 83.3723 kW with DG placed as suggested by the proposed and heuristic methods,
respectively. The location of DG is bus 50 in both methods; however, it is taken as bus 61
in [16] due to the different nomenclature of buses. Therefore, for clarification, the bus
number according to the nomenclature is this study is given outside the bracket, and that
of the heuristic method is given inside the bracket. It can be seen that the system loss
reduction and energy cost savings are greater, though marginal, in the proposed method,
even though the DG value is a little higher in the proposed method. This may be because
of the true optimal solution obtained by the proposed method.

Case 2. In order to study the effect of DG placement at multiple locations, an attempt


was made to place DG at more than one location. The next location was decided on
the basis of successive sensitivity, explained in Section 2.1. Following this procedure,
bus 8 (in the 33-bus system) and bus 11(in the 69-bus system) are found to be the next
locations for allocating DG at two locations. The GA produced DG value of 1718 kW and
840 kW at buses 6 and 8 (33-bus system) and 1777 kW and 555 kW at buses 50 and 11
(69-bus system), respectively. Proceeding in this manner, analysis was carried out to
four locations in the 33-bus system and to three locations in the 69-bus system, and the
results are shown in Tables 5 and 6. It can be seen from these tables that the line loss
decreases as the number of locations increases. However, the amount of decrease in line
loss reduces in successive locations. The saving in energy cost follows the reverse trend.
It increases with the increase in number of locations, but at a higher number of locations,
this increase is marginal. Since a marginal reduction in loss and a marginal increase in

Table 4
Summary for 69-bus system at L2 (base case) load level for single location

Power loss (kW)


DG Loss Energy
Optimal size Without With reduction savings
Methodology location (kW) DG DG (kW) ($)

Heuristic 50 (61) 1810 225.00 83.3723 141.628 62,033.00


Proposed GA 50 1872 225.00 83.2246 141.775 62,097.45
Optimal DG Placement in Distribution Systems 269

Table 5
Effect of variation in location at L2 (base case) for 33-bus system

One Two Three Four


No. of location (6) (6, 8) (6, 8, 20) (6, 8, 20, 24)

DG value (kW) 2380 1718, 840 1460, 838, 967 1391, 896, 298, 956
Loss (kW) 132.640 96.580 85.7508 84.830
Energy savings ($) 36,511.680 52,305.960 57,049.150 57,452.460

energy saving cost was observed when the number of locations was increased (to four in
the 33-bus system and three in the 69-bus system), a further increase in the number of
locations were dropped.
Downloaded by [Dalhousie University] at 16:30 13 September 2012

Case 3. A fixed load throughout the year is a hypothetical case as the load profile has
the impact of seasonal and time variations. In order to consider this effect, the load
of the entire year has been considered as combination of three load levels of different
durations, as shown in Table 2. The optimal values of DG to be placed for various
loading conditions were obtained using the proposed technique. The results are tabulated
in Tables 7 and 8 for the 33- and 69-bus systems, respectively. The effect of variations in
the number of locations on the DG values at the three above-mentioned loading conditions
are also tabulated in these tables. It is observed that the losses reduce and the savings
in energy cost increase at all load levels as the number of locations increase. The value
of loss reduction and increase in savings become smaller and smaller as the number of
locations increase, which can be seen in Tables 7 and 8. This is valid for both the 33-
and 69-bus systems. The values of loss reduction reduced to 0.381, 0.921, and 1.550 kW
at load levels L1 , L2 , and L3 , respectively, when numbers of locations were increased
from three to four in the 33-bus system, whereas it was 5.2839, 36.06, and 23.880 kW
when the number of locations increased from one to three. This suggests that the loss
reduction reduces as the number of locations increases, and beyond a certain number,
the loss reduction may cease. However, the limiting number is decided on the basis of
economic benefits, as discussed in the next section.

Case 4. Installed DGs will remain in the system as long as their performance is satis-
factory. Thus, the money invested once on DG will help an investor to harness its benefit
throughout its life. Although the systems losses reduce as the number of DGs at various
locations increase, the investment on DGs also increases. Thus, the reduction in cost of
energy loss is achieved at additional investment in DG, and an investor can only benefit

Table 6
Effect of variation in location at L2 (base case) for 69-bus system

No. of location One (50) Two (50, 11) Three (50, 11, 17)

DG value (kW) 1872 1777, 555 1727, 343, 514


Loss (kW) 84.431 71.7912 69.9841
Energy savings ($) 61,569.222 67,105.454 67,896.964
270 T. N. Shukla et al.

Table 7
Effect of optimal DG on DG size and losses for different locations for 33-bus system

DG size (kW) at bus no.


No. of Load Losses Energy
locations levels 6 8 20 24 (kW) savings ($)

One L1 1581 — — — 41.9792 46,108.00


L2 2380 — — — 132.640
L3 3291 — — — 177.460
Two L1 1023 531 — — 36.6953 52,284.12
L2 1718 840 — — 96.580
L3 2047 1111 — — 153.580
Three L1 897 511 599 — 32.584 57,153.152
Downloaded by [Dalhousie University] at 16:30 13 September 2012

L2 1536 846 946 — 85.751


L3 1882 1023 1280 — 136.569
Four L1 889 516 189 593 32.2027 57,452.460
L2 1391 896 298 956 84.830
L3 1839 1057 373 1216 135.019

as long as savings is more than investment. Therefore, there must be some limit on the
number of DGs beyond which it may be uneconomical.
In order to investigate this limit, an economic analysis has been performed. The
benefits achieved were obtained by varying loads and varying number of DG locations.
Considering a load growth of 2.5% per year, the loads have risen to 1.413 times the base
case loading of the two considered systems at the end of the 15-year planning period.
These values are plotted in Figures 1 and 2 for the 33- and 69-bus systems, along with
the loss cost with DG. It can be seen from Figure 1 that there is no benefit, even if the
number of locations are increased from one to three for the 33-bus system. However,
the value of negative benefit (compensation) reduces from one location to two locations

Table 8
Effect of optimally located DG at different locations on DG size
and losses for 69-bus system

Size of DG (kW) at bus no.


No. of Load Losses Energy
locations levels 50 11 17 (kW) savings ($)

One L1 1157 — — 32.3668 62,121.54


L2 1872 — — 83.22
L3 2359 — — 133.632
Two L1 1095 345 — 27.6251 67,172.73
L2 1777 555 — 71.7912
L3 2300 770 — 114.615
Three L1 1072 233 272 26.9376 68,015.5686
L2 1727 343 514 69.9841
L3 2047 466 641 110.986
Optimal DG Placement in Distribution Systems 271
Downloaded by [Dalhousie University] at 16:30 13 September 2012

Figure 1. Loss cost and benefits with DG of 33-bus system.

but increases further. This means that two locations are optimal in the case of the 33-bus
system. The reason for the negative benefit can be explained by the high cost of DG
units, which is US$300 per kW. In the future, with technological developments and the
mass production of DG, it may be reduced, and DG placement may become economical.
The same quantities obtained for the 69-bus system are plotted in Figure 2 with respect
to variations in number of locations. In this case, DG placement is beneficial (positive
benefit). However, that benefit is reduced as the number of locations increases. Thus, it
is economical to install DG at a single location at bus 50.
Another advantage of the improvement in voltage profiles due to the installation of
DG (at the end of 15-year planning period) has been demonstrated in Figures 3 and 4.
These figures show a variation in voltage profile at various loading conditions with the
variation of the number of locations for the 33- and 69-bus systems, respectively. Only

Figure 2. Loss cost and benefits with DG of 69-bus system.


272 T. N. Shukla et al.
Downloaded by [Dalhousie University] at 16:30 13 September 2012

Figure 3. Voltage profile improvement of 33-bus system with a number of locations.

the lowest voltage profiles of these systems at various loads are taken for these plots.
Thus, it can be understood that other voltages are higher. It can be very well understood
that the voltage profiles improve as the number of locations increases. This increase has
been observed from 0.8236 to 0.9189 and 0.8245 to 0.96365 for the 33- and 69-bus
systems, respectively, at a peak load level of L3 .

4.3. Computational Time


Computation time for the proposed method has been recorded for both systems. Variations
in time with respect to the number of locations have also been recorded for 1- and 15-
year planning periods with 2.5% load growth per year with varying load. These values
are plotted in Figures 5 and 6 for the 33- and 69-bus systems, respectively. It can be seen
from these figures that the computational time increases as the number of locations and
planning period increase. The maximum computational time is 4.094 sec for the 69-bus
system at three locations for the 15-year planning period. These studies were performed

Figure 4. Voltage profile improvement of 69-bus system with a number of locations.


Optimal DG Placement in Distribution Systems 273
Downloaded by [Dalhousie University] at 16:30 13 September 2012

Figure 5. Time versus number of locations for 33-bus system.

on an Intel IV dual core PC with 2.66 GHz and 1 GB RAM (Santa Clara, California,
USA). It is also observed that the computational time is greater for 69-bus system than
the 33-bus system due to the larger size of the system.

5. Conclusion
In this article, a GA-based method has been presented to minimize system losses and
increase the overall monetary benefits regarding the effects of DG. The results obtained
by the proposed method show that the presence of DG at appropriate locations reduces
energy loss costs significantly and increases savings, while the constraints of voltage
profile and branch current limits are satisfied. It is observed that the monetary benefit
also increases with the number of locations. However, this trend reverses after certain
number of locations, that is, the optimal number of locations. Noticeable improvement
in voltage profile is an added advantage of DG placement. The computational time was
found to be dependent on the system size and number of locations.

Figure 6. Time versus number of locations for 33-bus system.


274 T. N. Shukla et al.

References
1. Ackermann, T., Anderson, G., and Soder, L., “Distributed generation: Definition,” Elect. Power
Syst. Res., Vol. 57, No. 3, pp. 195–204, 2001.
2. Ackerman, T., and Knyazkin, V., “Interaction between distributed generation and the distri-
bution network: Operation aspects,” Proceedings of the Second International Symposium on
Distributed Generation Power System Market Aspects, Stockholm, Sweden, 2–4 October 2002.
3. Jóos, G., McGillis, B. T., Ga, D., Liana, F. D., and Marceau, R., “The potential of distributed
generation to provide ancillary services,” Proceedings of the IEEE Power Engineering Society
Summer Meeting, Seattle, WA, 16–20 July 2000.
4. Hoff, T., Wenger, H. J., and Farmer, B. K., “The value of grid-support photovoltaic in providing
distribution system voltage support,” Clean Power Research Conference, Napa, CA, 1995.
5. Liew, S., and Strbac, G., “Maximizing penetration of wind generation on existing distribution
networks,” Proc. IEE Generat. Transm. Distrib., Vol. 149, pp. 256–262, 2002.
6. Rau, N. S., and Wan, Y. H., “Optimum location of resources in distributed planning,” IEEE
Downloaded by [Dalhousie University] at 16:30 13 September 2012

Trans. Power Syst., Vol. 9, No. 4, pp. 2014–2020, 1994.


7. El-Khattam, W., Hegazy, Y. G., and Salama, M. M. A., “Investigating distributed generation
systems performance using Monte Carlo simulation,” IEEE Trans. Power Syst., Vol. 21, No. 2,
pp. 524–532, 2006.
8. Wang, C., and Nehrir, M. H., “Analytical approaches for optimal placement of distributed
generation sources in power systems,” IEEE Trans. Power Syst., Vol. 19, pp. 2068–2076,
2004.
9. Carpinelli, G., Celli, G., Pilo, F., and Russo, A., “Distributed generation siting and sizing under
uncertainty,” IEEE Porto Power Tech Conference, Porto, Portugal, 10–13 September 2001.
10. Celli, G., Ghiani, E., Mocci, U., and Pilo, F., “A multi-objective evolutionary algorithm for the
sizing and siting of distributed generation,” IEEE Trans. Power Syst., Vol. 20, pp. 750–757,
2005.
11. Keane, A., and O’Malley, M., “Optimal allocation of embedded generation on distribution
networks,” IEEE Trans. Power Syst., Vol. 20, pp. 1640–1646, 2005.
12. Zhu, D., Broadwater, R. P., Tam, K.-S., Seguin, R., and Asgeirsson, H., “Impact of DG
placement on reliability and efficiency with time-varying loads,” IEEE Trans. Power Syst.,
Vol. 21, pp. 419–427, 2006.
13. Goldenberg, D. E., Genetic Algorithm in Search, Optimization and Machine Learning, Reading,
MA: Addison-Wesley Publishing Co. Inc., 1989.
14. Deb, K., Optimization for Engineering Design: Algorithm and Examples, New Delhi: Prentice
Hall of India Ltd., 1995.
15. Kennedy, J., and Eberhart, R., “Particle swarm optimization,” Proceedings of IEEE Interna-
tional Conference on Neural Networks, Vol. IV, pp. 1942–1948, Perth, Australia, 29 November–
1 December 1995.
16. Acharya, N., Mahat, P., and Mithulanathan, N., “An analytical approach for DG allocation in
primary distribution network,” Elect. Power Energy Syst., Vol. 28, pp. 669–678, 2006.
17. Elgerd, I. O., Electric Energy System Theory: An Introduction, New Delhi: McGraw-Hill,
1971.
18. Afsari, M., Singh, S. P., Raju, G. S., and Rao, G. K., “A fast power flow solution of radial
distribution networks,” Elect. Power Compon. Syst., Vol. 30, pp. 1065–1074, 2002.
19. Kashem, M. A., Ganpathy, V., Jamson, G. B., and Buhari, M. I., “A novel method for loss
minimization in distribution networks,” Proceedings of the International Conference on Electric
Utility Deregulation and Restructuring and Power Technologies, pp. 251–255, London, 4–7
April 2000.
20. Baran, M. E., and Wu, F. F., “Optimal capacitor placement on radial distribution systems,”
IEEE Trans. Power Delivery, Vol. 4, pp. 725–734, 1989.
21. Pipttanasomporn, M., Willingham, M., and Rahman, S., “Implication of on-site distributed
generation for commercial/industrial facilities,” IEEE Trans. Power Syst., Vol. 20, No. 1,
pp. 206–212, 2005.

View publication stats

You might also like