ICUE 2018 on Green Energy for Sustainable Development

Thavorn Palm Beach Resort Karon, Phuket, Thailand. 24 – 26 October 2018

Economic Load Dispatch Using Genetic Algorithm

Optimization Technique
Kishan Bhushan Sahay, Anurag Sonkar and Abhishek Kumar
Department of Electrical Engineering
Madan Mohan Malaviya University of Technology
Gorakhpur, India
[email protected]; [email protected]; [email protected]

Abstract —In the power system, “Economic load dispatch” method, Lagrange Multiplier method, Newton Raphson
means the generation of electricity is low of cost and the power method, Base point participation factor method. The
system constraints such as Real power and Reactive power vary conventional techniques are suffering form a serious problem
within the specified limits while fulfill the load demand of the if the loads are changed. Due to this oscillatory problem
power system. Main objective of the ELD problem is to conventional techniques have complex algorithm in
minimize the total fuel cost. The total cost of operation includes comparison to the Heuristic optimization technique such as
the cost of fuel, transmission cost, labor cost and cost of Genetic Algorithm, Particle Swarm Optimization, and Linear
maintenance. The two major factors to be considered while Programming etc. For solving of ELD problem, the
dispatching power to generating units are the cost of generation
application of intelligence technique, they do not suffer to
and the quantity of power supplied. The relation between the
any limitation on the nature due to their ability to find and an
cost of generation and quantity of power supplied is expressed in
a polynomial equation and it is solved by applying of
optimal solution. The genetic algorithm provides fast
mathematical technique. The suggested optimization techniques convergence to comparison of other algorithms [6]-[9].
have a merit in finding an optimal solution. In this research This research paper is organized into five sections as
paper we apply basically genetic algorithm and simulated follows: section 1st introduces the introduction, section 2nd
algorithm. We have tested the applied algorithm on IEEE 30 bus introduces the problem formulation, section 3rd introduces
system and 20-unit generator system. methodology, section 4 introduces result and discussion, and
last 5th section introduces conclusions.
Index Terms—Economic Load Dispatch, Fuel cost, Power
system constraints, Genetic Algorithm. II. PROBLEM FORMULATION
I. INTRODUCTION In the ELD problem, the main objective is to minimize
The ELD problem is defined as the process of allocating the cost of generation of each generating units of the plant
the generation level of the generating unit so that the system therefore the cost of generation should be optimum by
load is supplied entirely and the most economically. The fulfilling the demand power with satisfying the power system
electrical energy is generated by the entire committed constraints.
generating unit in an electrical power system so that the total Traditionally in the ELD problem, the cost of generation
generation cost of electricity is minimized. In a typical power is expressed in quadratic polynomial equation. The total
system, multiple generators are committed to providing generation costs are the sum of the cost of generation of each
sufficient total output to satisfy the given consumer demand. individual unit. The total cost of generation is expressed as
In the power system, the cost of generation is different for
each committed generating unit because the generating units Mathematically,
are not located at the same distance from the load center
therefore the cost total cost of generation is differ from each 𝐹𝑖 (𝑃𝑔𝑖 ) = 𝑎𝑖 𝑃𝑔𝑖 2 + 𝑏𝑖 𝑃𝑔𝑖 + 𝑐𝑖 $/hr (2.1)
other generating units. The production cost of each
Where, 𝐹𝑖 (𝑃𝑔𝑖 ) is the generation cost of the unit.
generating unit is expressed as the quadratic function form of
the power output from these generation units [1]-[5]. These 𝑎𝑖 , 𝑏𝑖 , 𝑐𝑖 are the cost coefficients of the 𝑖 𝑡ℎ generation unit
generating units can have a unique cost per hour and 𝑃𝑔𝑖 is the generated power of 𝑖 𝑡ℎ unit.
characteristics for its output range. For solving of the ELD
problem, we apply many conventional optimization The total cost of operation includes the fuel cost, cost of
techniques and artificial optimization techniques. The labour, maintenance and transmission cost and it is expresses
conventional technique is given as a Lambda Iteration mathematically,

Authorized licensed use limited to: Universidad del Valle. Downloaded on December 15,2023 at 15:39:45 UTC from IEEE Xplore. Restrictions apply.
 ICUE 2018 on Green Energy for Sustainable Development
Thavorn Palm Beach Resort Karon, Phuket, Thailand. 24 – 26 October 2018

Ng Now, the fitness function of the ELD defines to minimize the

FT ( Pgi=
) ∑ (ai Pgi2 + bi Pgi + ci ) sum of the total generation cost function 𝑒𝑞 𝑛 (2.2) and a
i =1 (2.2) demand function (equality constraint) 𝑒𝑞 𝑛 (2.3) as follows:
NG NG

Where,
Pgi
is the output power of the generator and
=i 1 =i 1
=F2
i i gi i gi i gi∑ F (a P
D + b P + c ) + K * ( P + P − ∑ Pgi )
ai , bi , ci are the cost coefficients of i th generation unit.
(2.7)
A. Power balance constraint (Equality constraint) Subjected, to the generator constraints. Where, K is the
The cost function is not affected by the reactive power Lagrangian Multiplier.
generation, but it is affected by real power generation. III. METHODOLOGY
According to this constraint, summation of the real power of
all generating units must be equal to the total real power Genetic Algorithms are adaptive search techniques and
demand plus transmission losses. Thus the power balance optimization algorithms based on the principles of natural
equation is given by, genetics and natural selections. In GA we convert the design
space into genetic space that genetic algorithms work with a
NG coding of variables. They are operating on string structure
∑ P= gi PD + PL (2.3) and the string is a combination of binary digits representing a
i =1 possible solution for a given problem. The genetic algorithms
start from randomly generated initial population. Typically,
Where, Pgi = Generation of real power by the 𝑖𝑡ℎ unit. the GA has three phases, first one is initialization, second is
evaluation, and third is genetic operation.
PD = Power demand (MW). A. Working steps involved in Genetic Algorithm
PL = Transmission losses associated with the system (MW). • Initialization
• Formation of population
B. Inequality constraints • Evaluate the fitness function
Inequality constants for the generating units can be given as,
• Applying genetic operators
Pgimin < Pgi < Pgimax • Repeat the step 3 and 4 until the process has
(2.4)
converged if it satisfied then stop
Pgimin Pgimax
Where, and is the minimum limit and maximum
th B. Flowchart of the Genetic Algorithms
limit of the power generation of the i generator.
C. Transmission Loss
The transmission losses can be expressed as a function of
generator power through B-coefficients,

PL = PT BP + PT B0 + B00
(2.5)
Above equation written in this form,
NG NG NG
PL
= ∑∑ Pgi Bij Pgj + ∑ B0i Pgi + B00
=i 1 =i 1 =i 1
(2.6)

Where,
𝑃𝐿 = Transmission loss (MW).

Bij = Loss coefficient square matrix of the i th element.

th Figure 1. Flowchart of the Genetic Algorithm.
B0i = Loss coefficient square vector of the i element.
B00 = Loss coefficient constant. C. Algorithm for ELD Problem
The Genetic Algorithm for ELD problem is given step by
step below.

Authorized licensed use limited to: Universidad del Valle. Downloaded on December 15,2023 at 15:39:45 UTC from IEEE Xplore. Restrictions apply.
 ICUE 2018 on Green Energy for Sustainable Development
Thavorn Palm Beach Resort Karon, Phuket, Thailand. 24 – 26 October 2018

1) Read the specified data namely, Population size, cost but in large interconnected network the transmission losses
coefficient of the fuel 𝑎𝑖 ,𝑏𝑖 , 𝑐𝑖 No. of iterations, will be also considered where the power is transmitted to over
Probability of crossover, Length of the string, long distance, transmission losses are a major factor and
min max affect the optimum dispatch of the power generation.
Mutation, Power demand, P and P .
2) Create the initial population randomly in the binary A. IEEE 30bus Test System
form. The Fuel cost coefficients and loss coefficients are given
3) Decode string or obtains the decimal integer from the below in the Table I.
binary strings.
4) Calculate the power generated from the decoded TABLE I. FUEL COST COEFFICIENTS OF IEEE30 BUS SYSTEM
population.
𝐜𝐢 𝐛𝐢 𝐚𝐢 𝐏𝐦𝐚𝐱 𝐏𝐦𝐢𝐧
𝑃𝑖𝑚𝑎𝑥 −𝑃
𝑗 𝑖𝑚𝑖𝑛 100 2 0.005 85 10
𝑃𝑖 = 𝑃𝑖𝑚𝑖𝑛 + 𝑌𝑖 𝑗 (i=1, 2…NG) (j= 1, 2…..L) 200 2 0.010 80 10
2𝑙−1
300 2 0.020 70 10
Where, L is the no. of string length, 𝑌𝑖 𝑗 is the binary 80 1.95 0.003 250 50
coded value of the 𝑖 𝑡ℎ substring. 100 1.45 0.015 150 5
5) Check the 𝑃𝑖 𝑗 , 120 0.95 0.010 100 15

If Pi j > Pi max , then set the Pi j = Pi max

Loss coefficient matrix of IEEE30 Bus System
j
If Pi j < Pi min , then set the Pi = Pi min  0.14
0.17
0.17 0.15 0.19 0.26 0.22 
 0.6 0.13 0.16 0.15 0.2 
6) Find the fitness or cost function from equation 2.2
 0.15 0.13 0.65 0.17 0.24 0.9 
7) Find population with maximum fitness and average 10 ^ −4  
fitness of the population.  0.19 0.16 0.17 0.71 0.3 0.25
8) Perform the reproduction processes which include the  0.26 0.15 0.24 0.3 0.69 0.32 
following steps.  
a. Select the selection rate and number of matting  0.22 0.2 0.19 0.25 0.32 0.85
in a pool
Power demand = 550 MW, Optimal result for IEEE 30
b. Define the total fitness as sum of values obtain
bus Test System is given in Table II.
by using the above steps for chromosome which
are selected.
TABLE II. OPTIMUM RESULT OF IEEE30 BUS SYSTEM
c. Select the percentage of each chromosome which
is equal to the ratio of its fitness value to the total 𝑷𝒈𝟏 83.5577
fitness value. 𝑷𝒈𝟐 58.8564
𝐹𝑖𝑡𝑛𝑒𝑠𝑠
d. Probability = ∑ 𝑷𝒈𝟑 39.5052
𝐹𝑖𝑡𝑛𝑒𝑠𝑠
𝑷𝒈𝟒 205.2727
e. Calculate cumulative sum two numbers between 𝑷𝒈𝟓 72.7878
0.0 and 1.0. 𝑷𝑮𝟔 100.00
9) Perform crossover process Power demand 550 MW
a. Choose the pair of random no between o and 1 to Total loss 9.9798 MW
select one mother and one father chromosome to Total cost 2271.3
create an offspring.
b. Pairing the chromosome from different location
so that different location of chromosome will be
selected randomly. Generate offspring by
applying crossover.
10) Perform mutation by randomly selecting the mutation
point from the total number of bits in the population
matrix.
11) Updating the population.
12) If the number of iterations attempt maximum, then
go to step 13. Otherwise, go to step 6.
13) The fitness that generates the minimum total
generation cost is a solution of the problem.
IV. RESULT AND DISCUSSION
In the Economic Load Dispatch problem, the transmission
losses are neglected when transmission losses are very small

Authorized licensed use limited to: Universidad del Valle. Downloaded on December 15,2023 at 15:39:45 UTC from IEEE Xplore. Restrictions apply.
 ICUE 2018 on Green Energy for Sustainable Development
Thavorn Palm Beach Resort Karon, Phuket, Thailand. 24 – 26 October 2018

MATLAB Figure-

Optimal result for 20 Generator Test System is given in

Table IV.

TABLE IV. OPTIMAL RESULT OF 20 GENERATOR UNIT TEST SYSTEM

P1(MW) 494.4971
P2(MW) 199.7003
Figure 2. Fuel cost for IEEE 30 bus System. P3(MW) 118.7977
P4(MW) 84.5726
P5(MW) 114.4614
B. 20 Generator Unit System P6(MW) 63.4021
The fuel cost coefficient and loss coefficient of the 20- P7(MW) 115.4217
generator unit system will be given in the Table III and the P8(MW) 126.2267
power demand will be also given 2500MW. P9(MW) 102.8472
P10(MW) 104.2190
P11(MW) 174.9621
TABLE III. 20 GENERATOR UNIT SYSTEM FUEL COST COEFFICIENTS
P12(MW) 300.0408
P13(MW) 133.2762
UNIT 𝒄𝒊 𝒃𝒊 𝒂𝒊 𝑷𝒎𝒂𝒙 𝑷𝒎𝒊𝒏
P14(MW) 37.6271
1. 1000 18.19 0.00068 600 150
P15(MW) 108.0312
2. 970 19.26 0.00071 200 50
P16(MW) 40.7737
3. 600 19.80 0.00650 200 50
P17(MW) 49.1411
4. 700 19.10 0.00500 200 50
P18(MW) 86.9830
5. 420 18.10 0.00738 160 50
P19(MW) 96.7349
6. 360 19.26 0.00612 100 20
P20(MW) 44.7629
7. 490 17.14 0.00790 125 25
Power Demand 2500 MW
8. 660 18.92 0.00813 150 50
Total Power 2403.53
9. 765 18.27 0.00522 200 50
Total Loss 96.4789
10. 770 18.92 0.00573 150 30
Total Cost 62476$/hr.
11. 800 16.69 0.00480 300 100
12. 970 16.76 0.00310 500 150 V. CONCLUSSION AND FUTURE SCOPE
13. 900 17.36 0.00850 160 40
14. 700 18.70 0.00511 130 20
A. Conclussion
15. 450 18.70 0.00398 185 25 In this paper, an approach based on the genetic algorithm
16. 370 14.26 0.07120 80 20 optimization technique has been successfully applied and also
17. 480 19.14 0.00890 85 30 tested. The Genetic Algorithm optimization technique
18. 680 18.92 0.00713 120 30 provides fast convergence of result and gives better result.
19. 700 18.47 0.00622 120 40
20. 850 19.79 0.00773 100 30 B. Future Scope
• Extend the ELD problem for the large size of IEEE
20 generator unit system loss coefficients data should be buses such as 25 Bus, 30 Bus, and 57 Bus etc.
given below: - • Using new optimization techniques such as Ant
Colony Optimization techniques, Artificial Bee
Colony Optimization methods etc, to solve the ELD
problem.

Authorized licensed use limited to: Universidad del Valle. Downloaded on December 15,2023 at 15:39:45 UTC from IEEE Xplore. Restrictions apply.
 ICUE 2018 on Green Energy for Sustainable Development
Thavorn Palm Beach Resort Karon, Phuket, Thailand. 24 – 26 October 2018

• By using of facts devices in the power system to

minimize the transmission losses and cost of
operation.
REFERENCES
[1] K. Nimish, U. Nangia and K. Bhushan Sahay “Economic Load
Dispatch Using Improved Particle Swarm Optimization Algorithms”,
2014 6th IEEE Power India International Conference (PIICON), 2014.
[2] S. Kumar Singh, L. Phunchok, K. Chaobasingh and Y R Sood “Genetic
Algorithm A Noble Approach For Economic Load Dispatch”
International Conference on Emerging Trends in Engineering &
Technology, May 2012.
[3] V. Kumar, J. Singh, Y. Singh and S. Sood “Optimal Economic Load
Dispatch Using Genetic Algorithms”, International Journal of
Electrical, Computer, Energetic, Electronic and Communication
Engineering, Vol. 9, 2015.
[4] K. Bhuahan Sahay, N. Kumar and M. M Tripathi “Implementation of
Different Optimization Techniques to Solve ELD Problem”, 6th IEEE
Power India International Conference (PIICON), 2014.
[5] A.J. Wood and B.F. Wollenberg, “Power generation, operation and
control” Second edition John Wiley & Sons.
[6] D.P. Kothari and J.S. Dhillon, “Power System Optimization”, Second
Edition PHI learning Private limited, 2011.
[7] D.P. Kothari and I. J. Nagrath, “Modern Power System Analysis”, Tata
McGraw-Hill, Third Edition, 2003.
[8] S. Pratap Singh, R. Tyagi and A. Goel, “Genetic Algorithm for Solving
Economic Load Dispatch”, International Journal of Electronic and
Electrical Engineering, Volume 7, Nov.7(2014).
[9] G. Sahu and K. Swarnkar, “Economic load Dispatch by Genetic
Algorithm in Power System”, International Journal of Science,
Engineering and Technology Research (IJSETR), Vol. 3, (8), August
2014.

Authorized licensed use limited to: Universidad del Valle. Downloaded on December 15,2023 at 15:39:45 UTC from IEEE Xplore. Restrictions apply.