Professional Documents
Culture Documents
Economic Load Dispatch Using Swarm Intelligence
Economic Load Dispatch Using Swarm Intelligence
Algorithm
By
Mohammad Adnan Alam
EE-003
Farhan Hafeez
EE-011
Atif Latif
EE-023
Faizan Saleem
EE-026
Internal Advisor
Dr. Abdul Ghani Abro
ABSTRACT
The core objective of electric power utilities is to provide high-quality, reliable power supply to
the consumers at the lowest possible cost while meeting the limits and constraints imposed on
the generating units. This formulates the economic load dispatch (ELD) problem for finding the
optimal combination of all committed units in order to minimizes the production cost, while
satisfying all the constraints.
The fuel cost curve of generators are non linear and non smooth because of operation limits
imposed by ramp rate limits, valve point loading, multi-fuel effects etc. Hence it is highly
complex optimization problem which is difficult to solve using the conventional methods. The
traditional optimization techniques fail to solve this problem due to local optima solution
convergence.
Over the past two decades Evolutionary computation algorithms have proved to be more
efficient than the conventional methods and have received great consideration. Artificial Bee
Colony algorithm belongs to Swarm Intelligence. ABC algorithm is a meta-heuristic algorithm
which uses stochastic optimization to solve constrained optimization problems.
In this project we will utilize Artificial Bee Colony (ABC) algorithm to solve Economic Load
Dispatch (ELD) on various systems and results will be compared with results obtained from
conventional methods and other evolutionary computation techniques.
ACKNOWLEDGEMENT
All praise is for the Al-mighty Allah who has bestowed us with the ability and potential to do
this project. We are thankful to the Almighty for enabling us to complete our assigned
task within due course of time.
We are very grateful to our internal Dr. Abdul Ghani Abro for his cooperation, support and
continuous guidance during this whole work. We express our gratitude to all the researchers
and scholars whose research has helped us with our project. In the end we would like to thank
our parents without whom this would have not been possible.
ii
TABLE OF CONTENTS
ABSTRACT ....................................................................................................................................i
ACKNOWLEDGEMENT ................................................................................................................ ii
TABLE OF CONTENTS .............................................................................................................. iii
LIST OF FIGURES ......................................................................................................................... v
LIST OF TABLES .......................................................................................................................... vi
CHAPTER 1: INTRODUCTION........................................................................................................1
CHAPTER 2: BACKGROUND AND LITERATURE REVIEW ................................................................2
2.1
2.2
Constraints ....................................................................................................................3
2.2.1
2.2.2
2.3
2.4
2.5
2.6
2.6.1
2.6.2
iii
2.6.3
2.6.4
3.2
3.2.1
3.2.2
3.2.3
3.2.4
CONCLUSION ............................................................................................................................24
BIBLIOGRAPHY .........................................................................................................................25
iv
LIST OF FIGURES
LIST OF TABLES
vi
CHAPTER 1: INTRODUCTION
The main objective of electrical utilities is to provide high-quality and reliable service to the
consumers at the lowest possible rates. This requires long term planning and sophisticated
operation of the assets.
In this era power system operators not only have to meet the energy demand at minimum
production cost but also have to take certain factors into account including continuous
variations in load, generators inability to respond to them instantaneously, maintaining
reserves and plan for contingencies. Moreover the scheduling faces constraints because of
environmental laws and fuel limitations.
Economic Load dispatch is short term scheduling of all committed generating in such a way that
the fuel cost is minimized. The ELD is a constrained optimization problem which is difficult to
solve through traditional methods due to local optima convergence.
Over the past 2 decades Evolutionary computing techniques have been widely used for solving
constrained optimization problems and have yield better results than the conventional
methods. Evolutionary computation is subfield of Artificial intelligence. Evolutionary
computation techniques can be further divided into Swarm Intelligence and Artificial Immune
System.
Artificial Bee Colony algorithm belongs to Swarm intelligence (SI). Swam intelligence is a class of
bio inspired algorithms which utilizes stochastic optimization to solve constrained optimization
problems. ABC has been applied to large number of problems. ABC can yield better solution in
short interval with better convergence than other techniques. We have utilized this algorithm
for solving problem of economic load dispatch.
x 10
6
5
4
3
2
1
0
100
200
300
400
500
600
Output Power (MW)
700
800
900
1000
( )=
and
The total cost of thermal power plant with n number of generators can be given by the
following expression
( )=
+ )
The objective of economic load dispatch is to minimize the fuel cost while satisfying the
constraints imposed. The mathematical expression for ELD can be written as
=
( )
2.2 Constraints
The problem of economic load dispatch is subjected to constraints. These constraints can be
divided into two categories namely equality and inequality constraints. The two main
constraints considered for ELD are
1. Generators Power Limits
2. Power Balance
<
Where
P
Where
N is the total number of generators
P - is Power Balance
P - are transmission losses
P - is total real power Demand
The transmission losses impose a limit to the power generated by a unit, even though the unit
may be operating at higher efficiency than other units. The power loss can be calculated by B
matrix loss formula as
=
Where
B -is element of loss coefficient matrix
A more accurate formula for calculating transmission losses known as Krons formula is gven as
=
Where
Evolutionary algorithms
Gene expression programming
Genetic algorithm
Genetic programming
Evolutionary programming
5
Evolution strategy
Differential evolution
Swarm intelligence
Ant colony optimization
Particle swarm optimization
Artificial Bee Colony Algorithm
Bees algorithm
Artificial life (also see digital organism)
Artificial immune systems
Cultural algorithms
Harmony search
Learning classifier systems
Learnable Evolution Model
Self-organization such as self-organizing maps, competitive learning
= +
Where
l are lower bounds of parameters of x
u are upper bound of parameters of x
Where
- is random number between 0 and 1
- is parameter i of old food source
- is parameter i of new food source
The fitness of food source can be evaluated with following formulae
1
(
Here
)=
1+
1+
)
(
>0
)
<0
)
( )
After the probability is determined a new solution is produced using same method as used for
employed bees and then greedy selection is applied.
(0,1) ( )
CHAPTER 3: METHODOLOGY
3.1 Problem Formulation
The problem of economic load dispatch can be written in form of equation as
Minimize
( )=
+ )
Subjected to
<
<
=
Due to power balance ELD becomes a constrained optimization problem which can be made
unconstrained problem by using penalizing. Penalizing unfeasible solutions is a very popular
and effective approach to convert the multi-objective tness functions to the single-objective
tness function [7]. Hence the single objective fitness function for ELD can be written as
+ )+ (
Where
K is penalty Factor
The generators with multi valve steam turbines have highly non linear fuel cost curve. The
effect of multiple valves is to add ripples to fuel cost curve which is shown is figure 2. The fuel
cost for units with multi valve steam turbine can be given as
=
+ )+ (
) +
(|
( (
)|)
10
x 10
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
100
200
300
400
500
600
Output Power (MW)
700
800
900
1000
Another problem is emission of toxic gases such as CO, CO , SO NO . Various methods can be
used to limit the emission of these gases. The simplest and cheapest method is by dispatching
with the objective of minimum emission dispatch. The emission of toxic gases is modeled on
the basis of output power of generating units, as the output of a unit increases the emission of
gases increases depending upon the type of fuel being used. Following equation represents the
emission of toxic gases in form of Power of respective units
=
Where
, and are emission coefficient of ith generator
The final objective function including transmission losses, valve point effect and emission
constraints can be written as
11
+ )+ (
+
+
(
+
) )+
(|
( (
)|)
S.No
a (Rs/M
h)
b
(Rs/MWh)
c (Rs/h)
Pmax
(MW)
Pmin (MW)
0.006085
10.0450
136.9125
150
0.005915
9.7605
59.155
15
100
0.00525
8.6625
328.125
50
250
12
Sno
P1 (MW)
P2 (MW)
P3 (MW)
Pl (MW)
Fuel Cost
85.1477
85.0897
50.0000
10.3693
2744.0910
70.5094
100.0000
50.0000
10.6308
2744.9074
81.1511
89.1836
50.0000
10.4310
2743.6726
84.9936
84.7566
50.7311
10.5202
2744.7231
85.8454
84.3215
50.0000
10.3564
2744.4819
13
1st Run
2nd Run
3rd Run
4th Run
5th Run
3.8
10
3.7
10
3.6
10
3.5
10
3.4
10
3.3
10
100
200
300
400
500
600
Iterations
700
800
900
1000
3.6
10
3.45
3.5
10
2nd Run
10
3.44
10
0
500
Iterations
1000
3.45
10
3.44
500
Iterations
1000
500
Iterations
1000
5th Run
Log( Fuel Cost)
3.45
10
10
4th Run
10
3rd Run
3.46
1st Run
3.45
10
3.44
3.44
10
10
500
Iterations
1000
500
Iterations
1000
14
S.No
b
h) (Rs/MWh) c (Rs/h)
Pmin
(MW)
Pmax
(MW)
0.0016
7.92
561
300
0.032
100
600
0.0048
7.92
78
150
0.063
50
200
0.0019
7.85
310
200
0.042
100
400
Sno
P1 (MW)
P2 (MW)
P3 (MW)
Fuel Cost
494.2666
155.6900
400.0000
10124.7262
495.3596
154.5511
400.0000
10125.0231
492.5800
158.1196
398.9493
10129.9489
498.9380
150.6697
400.0000
10127.9254
496.5416
153.5062
400.0000
10127.9261
Table 4: Optimal Scheduling of Generators Considering Valve Point Effect 3 unit system
15
1st Run
2nd Run
3rd Run
4th Run
5th Run
10
100
200
300
400
500
600
Iterations
700
800
900
1000
4.06
4.9
10
500
Iterations
4.02
10
4.01
1000
3rd Run
4.5
10
4.1
10
0
500
Iterations
1000
500
Iterations
1000
4.07
4.05
10
10
4th Run
Log( Fuel Cost)
10
10
4.01
10
10
2nd Run
4.03
1st Run
4.03
10
4.01
5th Run
4.04
10
4.01
10
10
0
500
Iterations
1000
500
Iterations
1000
16
S.No
a (Rs/MWh)
b (Rs/MWh)
0.1525
0.106
0.028
c (Rs/h)
38.54
Pmin
(MW)
Pmax
(MW)
756.8
10
125
46.16 451.325
10
125
1050
35
225
0.0355
38.31 1243.53
35
210
0.0211
36.328 1658.57
130
325
0.018
38.27 1356.66
125
315
40.4
S
No
P1 (MW)
P2 (MW)
P3 (MW)
P4 (MW)
P5 (MW)
P6 (MW)
Fuel Cost
(Rs/h)
10.000
10.000
99.329
72.753
214.975
192.374
31868.908
24.983
12.768
84.873
102.950
201.698
171.767
31829.721
14.098
10.000
88.262
101.648
190.337
194.964
31823.876
10.000
10.000
95.985
108.051
194.356
181.159
31836.387
14.357
10.000
82.302
90.056
202.821
199.491
31838.401
17
1st Run
2nd Run
3rd Run
4th Run
5th Run
4.56
10
4.55
10
4.54
10
4.53
10
4.52
10
4.51
10
4.5
10
4.49
10
100
200
300
400
500
600
Iterations
700
800
900
1000
4.7
4.7
4.6
10
500
Iterations
4.6
10
1000
4.9
2nd Run
10
1st Run
Log( Fuel Cost)
10
500
Iterations
3rd Run
10
4.6
10
1000
500
Iterations
1000
4.53
10
4.508
10
5th Run
4th Run
4.52
10
4.51
10
4.503
10
0
500
Iterations
1000
500
Iterations
1000
18
a
(Rs/M
b
h) (Rs/MWh) c (Rs/h)
Pmin
(MW)
Pmax
(MW)
P1
0.00690
6.73
94.705
100
0.0000
36
114
P2
0.00690
6.73
94.705
100
0.0000
36
114
P3
0.02028
7.07
309.54
100
0.0000
60
120
P4
0.00942
818
369.03
150
0.0000
80
190
P5
0.01140
5.35
148.89
120
0.0000
47
97
P6
0.01142
8.05
222.33
100
0.0000
68
140
P7
0.00357
8.03
278.71
200
0.0000
110
300
P8
0.00492
6.99
391.98
200
0.0000
135
300
P9
0.00573
6.6
455.76
200
0.0000
135
300
P10
0.00605
12.9
722.82
200
0.0000
130
300
P11
0.00515
12.9
635.2
200
0.0420
94
375
P12
0.00569
12.8
654.69
200
0.0420
94
375
P13
0.00421
12.5
913.4
300
0.0350
125
500
P14
0.00752
8.84
1760.4
300
0.0350
125
500
P15
0.00708
9.15
1728.3
300
0.0350
125
500
P16
0.00708
9.15
1728.3
300
0.0350
125
500
P17
0.00313
7.97
647.85
300
0.0350
220
500
19
P18
0.00313
7.95
649.69
300
0.0350
220
500
P19
0.00313
7.97
647.83
300
0.0350
242
550
P20
0.00313
7.97
647.81
300
0.0350
242
550
P21
0.00298
6.63
785.96
300
0.0350
254
550
P22
0.00298
6.63
785.96
300
0.0350
254
550
P23
0.00284
6.66
794.53
300
0.0350
254
550
P24
0.00284
6.66
794.53
300
0.0350
254
550
P25
0.00277
7.1
801.32
300
0.0350
254
550
P26
0.00277
7.1
801.32
300
0.0350
254
550
P27
0.52124
3.33
1055.1
120
0.0770
10
150
P28
0.52124
3.33
1055.1
120
0.0770
10
150
P29
0.52124
3.33
1055.1
120
0.0770
10
150
P30
0.01140
5.35
148.89
120
0.0770
47
97
P31
0.00160
6.43
222.92
150
0.0630
60
190
P32
0.00160
6.43
222.92
150
0.0630
60
190
P33
0.00160
6.43
222.92
150
0.0630
60
190
P34
0.00010
8.95
107.87
200
0.0420
90
200
P35
0.00010
8.62
116.58
200
0.0420
90
200
P36
0.00010
8.62
116.58
200
0.0420
90
200
P37
0.01610
5.88
307.45
80
0.0980
25
110
P38
0.01610
5.88
307.45
80
0.0980
25
110
P39
0.01610
5.88
307.45
80
0.0980
25
110
P40
0.00313
7.97
647.83
300
0.0350
242
550
20
S.No
1st Runtime
2nd Runtime
3rd Runtime
4th Runtime
5th Runtime
P1
94.482
108.725
111.167
68.349
114.000
P2
86.672
82.870
93.352
69.458
48.358
P3
119.934
120.000
78.790
93.628
81.195
P4
84.219
80.000
80.000
80.000
89.408
P5
84.137
75.793
56.504
68.156
94.521
P6
99.415
140.000
102.098
118.697
134.806
P7
267.786
293.569
294.711
300.000
300.000
P8
278.079
228.315
255.885
300.000
235.824
P9
206.734
262.990
256.342
300.000
280.865
P10
254.722
257.081
217.548
300.000
136.115
P11
231.128
375.000
347.552
332.007
296.107
P12
375.000
235.281
146.093
375.000
372.960
P13
467.912
500.000
500.000
485.144
495.954
P14
494.231
256.542
480.478
496.028
302.097
P15
500.000
427.209
308.328
145.398
269.828
P16
426.668
195.548
433.432
500.000
342.665
P17
361.302
483.721
347.065
443.375
479.885
P18
468.843
500.000
412.252
477.870
419.312
P19
550.000
504.673
388.628
382.792
426.798
P20
330.950
390.375
543.622
550.000
550.000
P21
550.000
550.000
550.000
517.606
421.091
P22
384.363
509.390
372.414
550.000
531.869
P23
506.791
483.264
434.965
403.546
550.000
21
P24
549.559
368.640
550.000
390.138
413.180
P25
345.352
463.935
550.000
381.300
549.199
P26
530.440
547.939
550.000
375.585
550.000
P27
110.993
149.796
112.600
83.681
39.337
P28
34.275
63.035
125.756
97.302
128.659
P29
71.410
59.765
123.330
67.903
14.989
P30
97.000
96.701
61.582
83.391
63.094
P31
151.562
141.561
185.700
88.789
102.649
P32
190.000
180.924
190.000
171.859
183.610
P33
107.497
190.000
154.872
177.433
128.043
P34
145.069
117.393
158.666
179.498
177.015
P35
101.912
130.926
150.302
102.727
193.944
P36
110.650
148.299
166.335
177.856
147.931
P37
97.960
84.952
110.000
62.733
103.201
P38
93.061
110.000
78.728
43.071
92.918
P39
48.926
36.696
74.274
109.722
110.000
494.361
550.000
345.399
550.000
526.816
207862.792
207926.570
214042.132
205814.370
209199.210
P40
Cost
(Rs/h)
Table 8: Optimal Scheduling of Generators Considering Valve Point Effect 40 unit system
22
10
1st Run
2nd Run
3rd Run
4th Run
5th Run
7
10
10
10
100
200
300
400
500
600
Iterations
700
800
900
1000
10
10
10
10
10
10
10
10
500
Iterations
10
1000
10
10
500
Iterations
1000
10
500
Iterations
1000
10
5th Run
Log( Fuel Cost)
4th Run
Log( Fuel Cost)
10
10
10
10
3rd Run
Log( Fuel Cost)
2nd Run
Log( Fuel Cost)
1st Run
7
10
10
500
Iterations
1000
10
500
Iterations
1000
23
CONCLUSION
The above case studies of 3,6 and 40 unit systems shows the performance of ABC with varying
conditions and different constraints. The most encouraging point is ABC never failed and is able
to reach near optimal solution each time no matter how many generators or which constraints
are considered. Also it is seen from the convergence graphs that ABC is able to converge to near
best solution within first 200 iterations. The highly non-linear problems like valve point effect
and emission constraints can be easily addressed by ABC algorithm. Hence ABC is a stable and
reliable algorithm with high quality solution and good convergence.
24
BIBLIOGRAPHY
1.
Kothari, D.P. and J.S. Dhillon, Power System Optimization. 2004: Prentice-Hall of India.
2.
3.
Beni, G. and J. Wang, Swarm intelligence in cellular robotic systems, in Robots and
Biological Systems: Towards a New Bionics? 1993, Springer. p. 703-712.
4.
5.
Zhu, Y.-f. and X.-m. Tang. Overview of swarm intelligence. in Computer Application and
System Modeling (ICCASM), 2010 International Conference on. 2010. IEEE.
6.
Karaboga, D., Artificial bee colony algorithm. scholarpedia, 2010. 5(3): p. 6915.
7.
8.
Mallikarjuna, B., K.H. Reddy, and O. Hemakesavulu, Economic Load Dispatch Problem
with Valve-Point Effect Using a Binary Bat Algorithm. International Journal on Electrical
& Power Engineering, 2013. 4(3).
9.
Demirel, Y. and A. Demiroren, Economic and Minimum Emission Dispatch. energy, 2004.
2(1): p. 12.
10.
dos Santos Coelho, L. and V.C. Mariani, Combining of chaotic differential evolution and
quadratic programming for economic dispatch optimization with valve-point effect.
Power Systems, IEEE Transactions on, 2006. 21(2): p. 989-996.
25