You are on page 1of 5

2018 National Power Engineering Conference (NPEC)

Transmission Expansion Planning Using


Differential Evolution Multi-Objective
Differential Evolution and Multi-Criteria
Decision Making Methods
M. Divya Sri M. Veera Kumari
M. Tech Scholar, Dept of EEE, Assoc. Professor, Dept of EEE,
SIR C R Reddy College Of Engineering, SIR C R Reddy College Of Engineering,
Eluru, India Eluru, India
divyamoka08@gmail.com veerakumari.m@gmail.com

Abstract—Nowadays, Transmission Expansion is about considering entire planning horizon as single


Planning has been increasing widely in the time period. In dynamic planning multi stages of
electricity market. Multi-objective optimization of planning horizon are considered [2]. TEP purpose is
Transmission Expansion planning by installing to implement the planning of network to undergo
new transmission lines in the existing system at power economically so that there is improvement in
different locations by using the DE and the network reliability. Minimum investment costs of
MODE. In this paper, the Multi-criteria decision lines of the transmission network define about
making methods like AHP and SAW for ranking meeting the economical power. In power systems the
ultimate target of TEP is to find the best scheme so
the best compromise solutions are implemented on
that available transmission network can be expanded
test system like IEEE 24-Bus system are
to encounter the need of possible load growth for the
considered. This paper shows the feasibility and desired performance of the power system. In this
effectiveness of the proposed DE, MODE for the paper the model of TEP that applied and modified is
solution of the TEP problem. presented. Investment cost and Total cost are
considered in this optimization [3]. In this paper
Keywords—Transmission Expansion Planning, using the framework of different optimization
evolutionary optimization, Locational marginal techniques a analytical model and dynamic planning
prices, multi-criteria decision making methods of TEP is applied.
I. INTRODUCTION II. PROBLEM FORMULATION
The power system is mainly classified into three New objective functions are introduced for
parts they are the generation, the transmission, and reconstitute networks. The design that implemented
the distribution. The idea of the transmission system and reformed to dynamic model in this paper is
is transfer of the electrical energy from generating presented. Over a planning prospect TEP finds the set
stations to the distributing systems at different of optimum decision vectors that reduces the
locations that ultimately supplies the load. Over the investment cost and total cost under relevant
past few years, the electrical energy that has been reliability constraints. The objective function of this
transferred from generating units to major areas of paper is minimizing the investment cost and total cost
load has been growing drastically. Due to this there is of all combinations of with and without addition of
need for restructuring the transmission lines. In this new lines. Generally, TEP problem is a nonlinear,
paper the study of electric power transmission constrained multi-objective optimization problem; we
systems are done with regard to optimizing the consider minimization of investment cost and
transmission expansion planning (TEP) problem [1]. minimization of total cost in TEP problem as the
Minimizing the investment costs and operating costs objective functions and make the TEP problem as a
of the transmission systems is the objective of the non smooth optimization problem [4].MCDM
TEP. The transmission system planning is to set up techniques like Analytic Hierarchy Process (AHP)
new transmission lines according to the requirements and Simple additive Weighting (SAW) is applied at
of the public and private enterprises.TEP is broadly different locations of Transmission lines for the
classified into two categories based on the planning existing systems. The proposed Differential
horizon, they are static and dynamic planning i.e., Evolution (DE), Multi-Objective Differential
single stage and multi-stage planning. Static planning Evolution (MODE) algorithms and also Multi-criteria

978-1-5386-3803-3/18/$31.00 ©2018 IEEE


2018 National Power Engineering Conference (NPEC)

decision making method (MCDM) techniques are The algorithm is illustrated in following steps:
implemented on IEEE 24 bus system transmission 1) Initialization: By using the equation initial
network and minimize the overall investment cost population is generated and empty external Pareto
and total cost. optimal set is created.
A. Objective Function
The objective function is implemented to , (0) = + rand (0, 1) ( − ) (7)
curtail the total discounted line investment and total
discounted cost over throughout planning view [5]. Where, rand (0,1) is a random number systematically
The investment cost can be composed as: assigned between 0 and 1 [7].
CC=∑(i,j)ϵ ICij xij (1)
2) Fitness assignment: For different individuals
Where CC represents investment cost, ICij represents calculate fitness values. In the objective space
construction cost to build a line in the right-of-way individuals in the current population are evaluated
(i-j) ($/year), xij represents number of new line added and then scalar value is assigned which is known as
to the right-of-way (i-j), represents set of all new fitness.
right-of-ways
The congestion cost can be determined as follows 3) External set updating: Pareto optimal set is
updated as follows:
CoC= ∑(i,j) ϵAll corridors fij (LMPj – LMPi) (2)
where CoC represents Congestion cost, fij is active a) For the non-dominated individuals search
power flow in the right-of-way i-j, LMP is local the population and give them to the external set
marginal prices at i,j called 'external Pareto set.’
b) Search the external Pareto set for the non-
LMP’s are obtained from the following equation:
dominate individuals and remove all dominated
Min ∑ PGi ( aiPGi+ bi ) (3) solutions from the set.
s.t: sijγij (δi – δj) – PGi + PDi = 0 (4) c) If the externally stored individuals in the
Pareto set exceed the pre specified size of the
-fl ≤ γij (δi – δj) ≤fl (5) number, decrease the set by means of clustering.
PGimin≤PGi≤PGimax (6)
4) Perform DE mutation: According to the
The constraint (4) indicates the DC power flow following equation to generate the donor vector w (t)
equations. The constraint (5) indicates the power
for each ith-member (t) DE mutation operation is
flow limits of the network. The constraint (6)
performed
indicates generation boundary limit.
, (t+1) = , (t) + F( , (t) - , ( )) (8)
( )
× ( )
Min ∑ ∑( , )∈ ( +∑ (7) 5) Perform DE crossover: Perform the DE
)( ) ( )( )
crossover according to equation given below and find
the trail vector
Where Di indicates the discount factor; the first
, ( ) (0,1) <
expression indicates the investment cost, the second , (t) = (9)
, ( ),
expression indicates the total cost of the system.

6) Selection: Selection operation is made


III. MULTI-OBJECTIVE DIFFERENTIAL between trail vector (child) and target vector (parent)
EVOLUTION based on the dominance criteria as follows:
a) For solution ui if all the objectives are better
MODE algorithm solves the multi objective or equal to that of the corresponding objectives of Yi
functions with different individual giving optimal then Yi is dominated by ui and replaces it in the new
solutions. A contradiction is occurred between the population and vice-versa [8].
objective functions when there is sufficient difference b) If all objective are equal or some are better
in the optimal solutions corresponding to different and some are worse of solution Yi then vi and Yi are
objectives [6]. To initialize the optimization process, dominated by each other and vi continues in the new
control parameters and other parameters are defined.
population.

978-1-5386-3803-3/18/$31.00 ©2018 IEEE


2018 National Power Engineering Conference (NPEC)

7) Stopping criteria: Pareto-optimal set of C. Analysis of AHP and SAW


solutions are upgraded until the optimal value of cost In the AHP method the priority is given to total
is achieved. cost among all the attributes i.e., Investment cost,
Total cost, EENS (Energy not served Factor) and
LOLP (Loss of Load Probability) . In SAW method,
the priority is given to investment cost among all the
IV. MULTI-CRITERIA DECISION MAKING attributes Investment cost, total cost, EENS and
METHODS LOLP. Best rank is given to the optimal investment
cost, which is the second solution.
A. Analytic Hierarchy Process
A complex MCDM can be divided into system
hierarchies by using Analytic Hierarchy Process
V. RESULTS AND DISCUSSIONS
(AHP). Evaluations are converted into numerical
values by using AHP and this conversion can be
processed and compared throughout the problem . By The TEP is executed using DE, MODE
utilizing the respective significance of the algorithms along with MCDM techniques i.e., AHP
alternatives in terms of each criterion the matrix is and SAW which has been implemented in MATLAB
constructed. On priority theory Analytic Hierarchy 14b, 64 bit, 2GB RAM with windows 7 operating
Process (AHP) is based which is an MCDM method system. For the TEP problem IEEE 24-Bus standard
[9]. By consideration of multiple criteria/alternatives electrical test system is considered. Investment cost
simultaneously it deals with complex problems. To and total cost are the two objective functions that are
provide a scale for calculating intangibles and the examined in case studies.
procedure of creating priorities it has the ability to
IEEE 24-Bus system consists of 24 buses. 34
incorporate data and judgement of experts into the
existing branches, maximum demand is 8590MW
model in a logical way, analyses with correlation of
and maximum 5 lines are added at different locations.
elements in a system which allows revision of
The minimum generation Pgmin of each generator is
judgements in a short time to monitor the consistency
considered as zero and Pgmax of each generator as
in the decision-makers judgements to accommodate
specified for each candidate line. The complete data
group judgements if the groups cannot reach a natural
of the system is available in [5] and also shown in
consensus. By converting the evaluations into
Fig.1. To obtain optimal values of all objective
numerical values which are processed and compared.
functions the parameters for DE algorithm are taken
as scaling factor F=0.5; Crossover constant
CR=0.6;size of the population NP=20; discount
B. Simple Additive Weighting factor Di=0.015; and parameters for MODE
Among the MCDM techniques Simple Additive algorithm are taken as crossover constant CR=0.6;
Weighting (SAW) is one of the technique is scaling factor F=0.5; discount factor Di=0.085;
considered in this paper which consists of sum of population=200.
values which are assigning to each alternative, for
corresponding evaluation criterion each one is
associated and according to relative importance of
corresponding criterion weighted is done. SAW can
also be entitled as a weighted linear combination or
scoring method which is simple and most often
method used multi-attribute decision technique. SAW
depends on the weighted average using arithmetic
mean [10]. The virtue of this method itself it is a
proportional linear transformation of raw data which
means that the relative order of magnitude of the
standardized scores remains equal. This method is
more efficient because it requires less time, simple to
calculate and easier to examine in situations having
greater sensitivity constituents and optimizes decision
problems effectively. The ultimate target of SAW
method is to yield weighted sum of performance
ratings of each substitute over all attributes.
Fig.1. IEEE 24-bus system

978-1-5386-3803-3/18/$31.00 ©2018 IEEE


2018 National Power Engineering Conference (NPEC)

B. MODE Results
A. DE Results
TEP using MODE with various parameters, system
Table 1 shows the Investment cost (CC) and total parameters and constraints EENS and LOLP are
cost (TC) with adding 3 lines in the existing lines of considered. Table 2 shows the evaluation of
IEEE 24-Bus system. The following Fig2 and Fig3 Investment cost (CC), Total cost (TC) and best
represent the convergence characteristics of two compromise solution using MODE. The results are
objective functions along with their fitness function summarized in Table 2 for optimal minimum and
maximum values of investment cost, total cost and
TABLE1. INVESTMENT COST AND TOTAL COST WITH
also the best compromise solution at different
ADDITION OF LINES locations of the system.

New added lines CC ($) TC ($)


10-12 2.41e+5 8.57e+8 TABLE 2.EVALUATION OF OBJECTIVES USING MODE
6-10 2.57e+5 8.93e+8
16-17 2.75e+5 9.42e+8 Objectives MODE Best
Min Max compromise
solution
CC ($) 1.51e+6 1.98e+6 1.93e+5

TC ($) 6.99e+8 1.08e+9 7.29e+8

Fig.2. Convergence characteristics of objective 1 using DE

Fig.4.Convergence characteristics of Investment cost versus Total


cost using MODE

C. AHP and SAW Results

The optimal attributes considered are Investment


cost, Total cost, EENS and LOLP at different
locations are considered for AHP and SAW given in
Table 3. Priority is given to Investment cost in AHP
and SAW techniques.
Fig.3. Convergence characteristics of objective 2 using DE

978-1-5386-3803-3/18/$31.00 ©2018 IEEE


2018 National Power Engineering Conference (NPEC)

TABLE 3. RANKING OF INVESTMENT COST AND TOTAL [5] Abdollah Rastghou, Jamal Moshtagh
COST USING AHP AND SAW
“Application of firefly algorithm for multi-stage
transmission expansion planning with adequacy
system constraints in deregulated environments”
CC ($) TC ($) EENS LOLP Rank Rank
AHP SAW
applied soft computing 41(2016)© Elsevier

2.37e+5 8.52e+8 0.1048 0.1007 1 2 [6] M.A.Abido, N.A. Al-Ali “Multi-objective


2.39e+5 8.93e+8 0.0873 0.1092 3 1 Differential Evolution for optimal Power flow”
March 18-20©2009 IEEE
2.44e+5 8.84e+8 0.1041 0.1032 2 3
2.54e+5 8.75e+8 0.0958 0.1136 5 4 [7] Wenyin Gong, Zhihua Cai “A multi objective
2.64e+5 8.91e+8 0.1048 0.1007 4 5 differential evolution algorithm of constrained
optimization”© 2008 IEEE

VI. CONCLUSION [8] Feng Xue, Arthur C. Sanderson, Robert J. Graves


“Pareto-based Differential Evolution” ©2003 IEEE
The proposed MODE algorithm solved the optimal
transmission expansion planning with the two [9] S.A. Torabi and M. Madadi Da-Yong Chang
objective functions along with reliability constraints. “Applications of the extent analysis method on AHP”
The objective functions considered is minimization European Journal of Operational Research 95 (1996)
of investment cost and total cost for planning horizon
of two years corresponding to EENS and LOLP as [10] Hepu Deng “Multi-criteria analysis with fuzzy
constraints. The standard test system considered is pair wise comparison” International Journal of
IEEE 24-Bus system. In MODE algorithm one Approximate Reasoning 21(1999)
objective is minimized and other is maximized. The
best compromise solution of best combination of
transmission lines at different locations is evaluated
by AHP and SAW, multi-criteria decision making
methods. Also ranking analysis by using AHP and
SAW were presented.

REFERENCES

[1] Hong Fan, Hao-zhong cheng, Liang Gao, Jie-tan


Zhang “Study on transmission expansion planning
considering uncertainties” © 2011 IEEE

[2] G.A. Orfanos, P.S. Georgilakis, N.D.


Hatziargyriou “Transmission expansion planning by
enhanced Differential Evolution” 2007-2-TREN-
218903 (IRENE 40-Project)

[3] Zongyu Liu, Georgios Papaefthymiou, Lou van


der Sluis “Collaborative Transmission network
planning for congestion alleviation using Lagrange
multipliers”© 2012 IEEE

[4] Russell Bent, G. Loren Toole, Alan Berscheid


“Transmission network expansion planning with
complex power flow models” IEEE
TRANSACTIONS ON POWER SYSTEMS,
VOL.27. NO.2, MAY 2012

978-1-5386-3803-3/18/$31.00 ©2018 IEEE

You might also like