This article was downloaded by:[National Institute of Technology Calicut]

On: 31 March 2008

Access Details: [subscription number 773426689]
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.informaworld.com/smpp/title~content=t713399721

Determination of Location and Capacity of Power

Facilities by Genetic Algorithm

Tomonobu Senjyu a; Kai Shimabukuro a; Hirohito Yamashiro a; Katsumi Uezato a;

Toshihisa Funabashi b
a
Faculty of Engineering, University of the Ryukyus, Okinawa, Japan.
b
Meidensha Corporation, Tokyo, Japan.
Online Publication Date: 01 April 2004
To cite this Article: Senjyu, Tomonobu, Shimabukuro, Kai, Yamashiro, Hirohito,
Uezato, Katsumi and Funabashi, Toshihisa (2004) 'Determination of Location and
Capacity of Power Facilities by Genetic Algorithm', Electric Power Components and
Systems, 32:4, 1
To link to this article: DOI: 10.1080/759369249
URL: http://dx.doi.org/10.1080/759369249

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,
re-distribution, re-selling, loan or 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.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

EMP 32(4) #9735

Electric Power Components and Systems, 32:375390, 2004

c Taylor & Francis Inc.
Copyright

ISSN: 1532-5008 print/1532-5016 online
DOI: 10.1080/15325000490217470

Determination of Location and Capacity of

Power Facilities by Genetic Algorithm
TOMONOBU SENJYU
KAI SHIMABUKURO
HIROHITO YAMASHIRO
KATSUMI UEZATO
Faculty of Engineering
University of the Ryukyus
Okinawa, Japan

TOSHIHISA FUNABASHI
Meidensha Corporation
Tokyo, Japan
This article presents the determination of optimal location and capacity of
power facilities by using genetic algorithm (GA) based on reliability, loss of
load and dump power. We determine optimal capacity after optimal location;
however, since the optimal location and capacity are closely related, the optimal
location varies with the capacity and vice versa. Hence, we propose the method
that determines the optimal location and capacity at the same time. Using the
proposed method, expected loss of load in faults can be reduced by 32.3% in
comparison with that of separate optimization techniques.
Keywords
algorithm

power system reliability, loss of load, dump power, genetic

1. Introduction
In recent years, the demand of power has been increasing with rapid industrialization. Since electric power systems play a major role in a modern society, professional
engineers are responsible for proper planning, design, and operation of power systems. Further, the modern power systems are required to have better reliability.
Under these circumstances, new generation facilities and expansion of the transmission lines must be planned and constructed accordingly to maintain reliability
of power systems and reduce loss of load during faults. Large reserve can maintain
reliability and decrease loss of load during faults. However, this reserve increases
Manuscript received in nal form on 23 December 2002.
Address correspondence to Tomonobu Senjyu, Faculty of Engineering, University
of the Ryukyus, 1 Senbaru Nishihara-cho Nakagami Okinawa 903-0213 Japan. E-mail:
b985542@tec.u-ryukyu.ac.jp

375

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

376

T. Senjyu et al.

cost and dump power that indicates the excess electricity. Hence, there is an importance of not only location but also capacity on power system planning.
To solve these optimization problems, algorithms for heuristic techniques have
been presented. Among the various techniques are Tabu search, Lagrangian relaxation (LR) methods, and neural network [1]. However, since these methods are local
search techniques, it is dicult to obtain the optimal solution.
To achieve global search, we propose a genetic algorithm (GA) solution to the
optimal location and capacity problem. GA is an optimization technique based
on a model of evolutionary adaptation in nature [2, 3, 6]. The objective of the
problem mentioned above is maintenance of reliability and reducing the loss of
load and dump power. Since power system reliability and loss of load have diverse
dimensions, they cannot be evaluated at the same time. This is the multi-objective
optimization problem for that reason.
In this article, objective functions are incorporated for the evaluation of these
variables at the same time. Using the proposed method, optimal location and
capacity can be optimized at the same time, and the expected value of loss of
load in faults can be reduced by 32.3% in comparison with the method optimizing
the location and capacity separately.

2. Power System Reliability

There are many variations on the denition and evaluation of reliability. We adopt
the following denition of reliability; reliability is the probability of a device performing its purpose adequately for the period of time intended under the operating
conditions encountered [4, 5]. Since the power system is playing a major role in
modern society, its reliability evaluation is essential. Figure 1 shows the operating
condition and failure condition of power facilities. Reliability is decided with the
availability given by equation (1), which includes durability and maintainability of
power facilities.
A=

operating time
operating time + failure time

(1)

Equation (1) can be further written as,

A=

MTBF
MTBF + MTTR

Figure 1. Power system conditions.

(2)

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

377

where MTBF and MTTR are the mean time between failure and mean time to
repair, respectively. Here, we dene the new variable,
=

MTTR
MTBF

(3)

where is the coecient of maintainability.

Using the relationship given in equation (3), we obtain the following expression
for A as,
A=

1
.
1+

(4)

From equation (4), the availability A increases with decrease the coecient of
maintainability.

3. Genetic Algorithm
The model of power system used in this article is shown in Figure 2, where G, T , N ,
L represent power plant, transmission line, transformer, and load point, respectively.
Suppose that two generators and two transmission lines are added in Figure 2. We
would then determine the optimum setup point and capacity using GA. Figure 3
shows the owchart of genetic algorithm and each step of the owchart is described
as follows:
STEP 1

Generating Initial Population

Initial population is randomly generated using binary strings (0 and 1). The population strings consists of 14 bits. Figure 4 shows an initial population. Generators

Figure 2. Power system conguration.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

378
T. Senjyu et al.

Figure 3. GA owchart.

Figure 4. Initial population.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

379

1 through 9 in Figure 2 are candidates for new generator setup point. New transmission lines are set up in the existing transmission lines. Setup point and capacity
of those new facilities are expressed by bit row in Figure 4.
STEP 2

Evaluation

Populations are evaluated by the tness function. Fitness function is the most
important factor in GA.
STEP 3

Selection

The selection creates a new population from the old one. We adopt the elite selection
and roulette selection.
STEP 4

Crossover

Crossover is the most important operator in GA and many new solutions are
generated by crossover. This operator simply combines the parent symbol strings,
forming a new chromosome strings that inherits solution characteristic from both
parents. The crossover scheme used in this paper is an one-point crossover. We use
variable crossover rate. The crossover rate is given by
Pcr = Pc (Fave /Fmax )

(5)

where Pc is coecient of crossover rate. Fave and Fmax are the average and
maximum tnesses respectively. An example of the one-point crossover operator
is shown in Figure 5.
STEP 5

Mutation

The mutation operator is applied for the maintenance of diversity of the solutions.
Generally, the mutation operator is applied with a small probability. The mutation
scheme used is an one-point mutation. This operator randomly chooses bits of the
ospring genotypes, change from 0 to 1. We use a variable mutation rate given by
Pmr = Pm /(Fmax Fave )

Figure 5. One-point crossover operator.

(6)

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

380

T. Senjyu et al.

Figure 6. One-point mutation operator.

where Pm is coecient of mutation rate. Fave and Fmax are the average and
maximum tness respectively. If Fmax = Fave , then Pmr = Pm . An example of
the one-point operator is shown in Figure 6.
STEP 6

Final Decision

The above procedures repeat itself until nal generation is achieved.

4. Simulation Results
4.1. Setup of Fitness Function
The objective of this article is the maintenance of reliability, reducing loss of load
and dump power. Hence, these variables are used within the tness function.
4.1.1. Fitness Function for Setup Point. Equation (7) is used as a tness function
to determine optimal setup point.
Ft =

M

{ALm DP (m) (1 ALm )LOL(m)}

(7)

m=1

where ALm is the reliability at load point m, DP (m) is the demand power at load
point m, LOL(m) is the total loss of load at load point m, and M is the total
number of load points in the system. Reliability is the probability of possible power
supply and the term ALm DP (m) is expected power supply. Subtracting reliability
from equation (1), we get loss of load probability. The term, (1 ALm )LOL(m) is
the expected loss of load.
4.1.2. Fitness Function for Capacity. Equation (8) is tness function to determine
the capacity
Ft =

M

{ESP (m) ELOL(m)} D PT

(8)

m=1

where ESP (m) is the expected supply power, ELOL(m) is the expected loss of
load, D is the dump power, PT is the margin of transmission capacity, and ,
, are weight factors. D and PT represent the upper bound constraint of the
capacity.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

381

4.2. Individual Optimization

At rst, we determine the optimal locations by using equation (7). The optimal
locations are determined by varying both the transmission and generation capacity
from 200 MW to 400 MW by a step of 50 MW. The total numbers of runs are 25
for each of transmission capacity and generation capacity. For each run, two setup
points, which indicate the location of generator or transmission line, are determined
as the optimal setup points. Figures 7 and 8 show the number of times the optimal
setup points for each setup point are selected. In Figure 7, the numbers on horizontal
axis corresponds to the load points in Figure 2. For example, a number 3 in Figure 7
corresponds to a load point L7 in Figure 2. In Figure 8, the numbers on horizontal
axis corresponds to the transmission lines in Figure 2 as is. From Figures 7 and 2,
since the load points L3 and L7 are selected as the optimal setup points among all
the setup points, the new generators are set up in the load points L3 and L7. From
Figures 8 and 2, since the transmission lines T 5 and T 7 are selected as the optimal
setup points among all the setup points, the new transmission lines are set up in
transmission lines T 5 and T 7.
After performing the optimal point selection, we determine the optimal capacity
of these facilities using equation (8) subject to the constraints given in Table 1. We
set = 9, = 0.0006, = 0.0001. Figure 9 shows the optimal capacity at load
factor 1.4. In Figure 9a, G.b and T.a are aligned to the same direction, which start
from 250 MW at 0 generation and then go down to 200 MW at 50 generation
and nally converge to 200 MW at 400 generation. G.a starts from 200 MW at
0 generation, and then up to 250 MW at 50 generation, nally converges to 250 MW
at 400 generation. T.b always maintains 200 MW. That means the capacity of
generators G.a and G.b are 250 MW and 200 MW, respectively, and that of both
capacity of transmission lines T.a and T.b are 200 MW. Figure 10 shows optimal

Figure 7. Number of times selected as optimal set up point (generator).

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

382

T. Senjyu et al.

Figure 8. Number of times selected as optimal set up point (transmission line).

capacity at load factor 1.6. From Figure 10a, the capacity of generators G.a and
G.b are 400 MW and 250 MW, respectively, and the capacity of transmission lines
T.a and T.b are 450 MW and 200 MW, respectively.
4.3. Optimization Simultaneously
In the previous section, we determined the optimal capacity after determining the
optimal locations as shown in Figure 11. However, since the optimal location and
capacity are closely related, the optimal location varies with the capacity and vice
versa. But this optimization problem is more dicult than the optimal setup point
problem or optimal capacity problem.
In this article, we propose the determination technique for the optimal location
and capacity at the same time, as shown in Figure 12. Equation (8) is used as
the tness function to determine the optimal location and capacity. We set = 9,
= 0.0006, = 0.0001 in equation (8), and compare with the dump power and
expected loss of load obtained in Section 4.2.
We determine the optimal location and capacity by equation (8). From Figure 13, the generators G.a and G.b are set up in the load point L3 and L8 when
the load factor is 1.4 and the corresponding capacities are 150 MW and 300 MW,
Table 1
Parameter constraints
Constraint of set up point

Constraint of capacity

1 G.a, G.b 9
1 T.a, T.b 10

100 MW G.a, G.b 550 MW

100 MW T.a, T.b 550 MW

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

383

(a)

(b)
Figure 9. Optimal capacity at load factor 1.4. (a) Optimal capacity; (b) maximum and
average tness.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

384

T. Senjyu et al.

(a)

(b)
Figure 10. Optimal capacity at load factor 1.6. (a) Optimal capacity; (b) maximum and
average tness.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

385

Figure 11. Conventional method to determine the optimal capacity and optimal location
separately.

Figure 12. Proposed method to determine the optimal capacity and optimal location
simultaneously.

respectively. New transmission lines T.a and T.b are set up in transmission line T 5
and T 9. The transmission capacities of these new transmission lines are 500 MW
and 450 MW, respectively.
From Figure 14, the generators G.a and G.b are set up in the load point L3
and power plant G2 when the load factor is 1.6, and these generation capacities
are 400 MW and 350 MW, respectively. New transmission lines T.a and T.b are
set up in the transmission line T 3 and T 5. Transmission capacities of these new
transmission lines are 300 MW and 500 MW, respectively.
From Figure 15, we see that expected value of loss of load are equal in both
methods but the dump power can be reduced by 4.5% in comparison with that of
individual optimization technique in Section 4.2. In Figure 15b, the change in the
shape of stairs in case of simultaneous occurrence indicates that the search of GA
proceeds successfully.
Figure 16 shows the dump power and expected loss of load when load factor
equal to 1.6. There are slight dierences in the dump power between these two
methods; however, the loss of load can be reduced to 32.3% by simultaneous
optimization.

5. Conclusions
This article presents a determination of optimal location and capacity of power facilities by using genetic algorithm. Since power system plays a major role in modern
society, power supply reliability and loss of load in fault are important factors in
power system planning. Large reserves can maintain reliability and decreasing loss
of load during faults; however, this reserve increases cost and dump power. Moreover, since reliability and loss of load have dierent dimension, they cannot be evaluated at the same time. To achieve an optimal conguration of a power system, the
new evaluation method and optimization technique are introduced. Using the proposed method, the maintenance of power supply reliability, reducing loss of load and
dump power are done favorably in comparison with the other optimization method.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

386

T. Senjyu et al.

(a)

(b)

(c)
Figure 13. Load factor at 1.4. (a) Optimal set up points; (b) optimal capacity; and
(c) maximum and average tness.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

387

(a)

(b)

(c)
Figure 14. Load factor at 1.6. (a) Optimal set up points; (b) optimal capacity; and
(c) maximum and average tness.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

388

T. Senjyu et al.

(a)

(b)
Figure 15. Load factor at 1.4. (a) Dump power; (b) expected value of loss of load.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

Determination of Power Facilities

389

(a)

(b)
Figure 16. Load factor at 1.6. (a) Dump power; (b) expected value of loss of load.

Downloaded By: [National Institute of Technology Calicut] At: 09:55 31 March 2008

390

T. Senjyu et al.

References
[1] T. Yalcinoz and M. J. Short, Neural networks approach for solving economic dispatch problem with transmission capacity constraints, IEEE Transactions on Power
Systems, vol. 13, no. 2, pp. 307313, 1998.
[2] D. C. Walters and G. B. Sheble, Genetic algorithms solution of economic dispatch
with valve point boding, IEEE Transactions on Power Systems, vol. 8, no. 3, pp. 1325
1332, 1993.
[3] I. J. Ramirez-Rosado and J. L. Bernal-Agustin, Genetic algorithms applied to the
design of large power distribution systems, IEEE Transactions on Power Systems,
vol. 13, no. 2, pp. 696703, 1998.
[4] R. Billinton and R. N. Allan, Reliability Evaluation of Engineering Systems, Second
Edition, New York: Plenum Publishing Co., 1992.
[5] R. Billinton and R. N. Allan, Reliability Evaluation of Power Systems, Second Edition,
New York: Plenum Publishing Co., 1996.
[6] S. A. Kazarlis, A. G. Bakirtzis, and V. Petridis, A genetic algorithm solution to
the unit commitment problem, IEEE Transactions on Power Systems, vol. 11, no. 1,
pp. 8392, 1996.