Studies in Informatics and Control, 29(1) 131-140, March 2020 ISSN: 1220-1766 eISSN: 1841-429X 131

Profit Maximization of GENCO’s Using an

Elephant Herding Optimization Algorithm
Sundar RAVICHANDRAN1*, Manoharan SUBRAMANIAN2
1
Department of Electrical & Electronics Engineering,
Karpagam College of Engineering, Coimbatore, 641032, India*
sundareee1988@gmail.com (*Corresponding author)
2
Department of Electrical & Electronics Engineering,
Karpagam College of Engineering, Coimbatore, 641032, India
Abstract: When electrical power systems are restructured managers look for satisfying several objectives that include: the
working cost minimization and the profit for the unit commitment problems maximization. Power generating companies that
comply with the two above mentioned objectives are able to provide good quality and reliable power at a cheaper cost. The
individual power producers, schedule their generating units in such a way that they gain maximum profit. This is known
as Profit Based Unit Commitment (PBUC). The target of this proposed work is to obtain an optimum generating schedule
of the Power Generating Companies (GENCO’s) and maximize the profit of power generating companies when the system
is under various constraints like forecasted demand, minimum start-up /shutdown time, spot price, forecasted reserve and
ramp rate limits. In order to address this problem, a new meta-heuristic approach called Elephant Herding Optimization
(EHO) algorithm is presented. The method may help to solve the complex PBUC problems in the deregulated open market.
The effectiveness of projected EHO is tested on various systems with various market conditions. The comparison of the test
results with other optimization methods are presented and discussed taking into account their convergence characteristics,
solution superiority and reliability.
Keywords: Deregulation, Elephant herding optimization, Profit based unit commitment.

1. Introduction
Restructuring of electric power system network preserve the system precautions and consistency,
is taking place all over the world. During 90’s, a focal facilitator named Independent System
worldwide power network companies and many Operator (ISO) is used. In this vertically integrated
electric utilities started using deregulation instead system, the Customers can choose their individual
of vertically integrated structures. The traditional power supplier which improves the efficiency of
and vertically regulated power industry is the power generation and distribution, delivers at a
replaced by horizontally regulated system therein reduced price with high quality. In this deregulated
generation, transmission and distribution are structure competition is created between
unbundled. The deregulation of the power system GENCO’s. Most of the conventional optimization
created market-based competition in an open methods are to be modified to address the open
electricity pool. Due to the developments in the market competition. Sequentially, the unit
power industry, the entire network comprising commitment approach with profit maximization
the generation process, scheduling, and running plays an important role in the competitive pool
methods are required to be customized in power market.
regulated power system network. But it is a very
complex process since electricity policy and In a deregulated power system, the unit
the applications differ from country to country. commitment problem with multi-objective
This market-based power industry is not yet function is exposed to various system constraints.
implemented in Tamil Nadu, India, where the The determination of generator scheduling in a
system is owned by the state and operated as power system is a complex optimization problem.
vertically integrated. Earlier the electric utilities had an appeal to satisfy
the customer demand and forecasted reserve. Be
The restructuring of the power system has that as it may, in the competitive power market,
unbundled the responsibilities into three categories. the GENCO’s are not mandatory to equalize the
They are i. Generation companies (GENCO’s), power demand. The prepared load schedule may
ii. Transmission companies (TRANSCO), iii. generate less than the forecasted load requirement
Distribution companies (DISCO). In order to and reserve with more profit under various
balance the supply and demand of the system to constraints. This problem is stated as Profit based

https://doi.org/10.24846/v29i1y202013 ICI Bucharest © Copyright 2012-2020. All rights reserved

132 Sundar Ravichandran, Manoharan Subramanian

unit commitment problem. To increase the profit mimicking mechanism of biological evolution
of the GENCO, it is necessary to compute the for optimization problem known as evolutionary
amount of power required to be introduced in algorithm. Chen &Wang (2002) presented
the pool market based on the forecasted power a cooperative algorithm for solving the UC
demand and spot pricing at a particular time ‘t’. problems. Contreras.et.al (2006) proposed a
This is a more flexible and more complex problem technique which determined best feasible solution
under a deregulated environment. Different with least computational time for a UC problem.
solutions were obtained for the unit commitment
problem in the vertically regulated power system. Chendur Pandian et.al (2014) & Daneshi et.al
In the recent days, the researchers concentrate on presented a price-based solution for PBUC
the best possible unit commitment algorithms for problem using fuzzy logic application. Mixed
PBUC problems which will be more suitable for integer programming approach also addresses
large size power system with low storage space the problem and it is very practical with the
and less computational time. consideration of uncertainties in the parameters.
Sasaki et.al (2002) Used a Hopfield neural
Based on the reviews carried out, large number network approach to explore the probability
of numerical optimization approaches were to the UC problem when more inequality
implemented to give solutions for complicated constraints were considered. Yamin et.al (2007)
profit-based unit commitment. Many classical presented a method for Genco’s PBUC in a
approaches were developed and implemented day ahead open power market. The forecasted
effectively. Some of the frequently used demand and generated power are also taken into
approaches are deterministic approach and meta- account in the formulation to simulate the reserve
heuristic approach. The deterministic approaches uncertainty. Annakkage et.al (1995) investigated
include enumeration method, priority list, Benders the application of parallel simulated annealing
decomposition, branch and bound, dynamic for unit commitment problem to reduce the
programming method, Lagrangian relaxation computational time. Tabu search optimization
technique and mixed integer technique which has been applied to a combinatorial optimization
give local optimum solutions. The meta-heuristic problem. Mantawy et.al (1998) presented a
approach includes Genetic algorithm (GA), Ant unit commitment solution using Tabu search
colony algorithm (ACA), Fuzzy logic(FL), and introduced a new perturbation scheme for
artificial neural network(ANN), Tabu search(TS), conventional UC problems.
particle swarm optimization (PSO), Muller
method, Simulated annealing (SA), Memory Jing-yu et.al (2004) explored an approach with
management algorithm(MMA), Artificial immune PBUC multi-agents’ system having command
system(AIS). Hybrid meta-heuristic methods agent, mobile agent and generator agent. They are
like gravitational search LR-ANN, Dynamic placed with a distributed generator and operate
programming with particle swarm optimization- together to get the satisfying operation of PBUC
(PSO-DP), Particle swarm optimization based solution. Mori &Okawa (2009) developed a
Lagrangian relaxation (LR-PSO),Multi-agent Tabu search evolutionary PSO technique to
system(MAS), Improved pre-prepared power PBUC. Here Genetic algorithm is derived from
demand(IPPD) optimization, Teaching –learning the biological model of evolution and it operates
optimization(TLO), Binary fish swarm algorithm on the Darwinian principle of natural selection.
(BFSA) are also presented for PBUC problems Richter & Sheble (1997) formed a bidding
under restructured market. strategy using GA which maximizes the profit
of the generating companies in the competitive
The major limitation of this numerical approach pool electricity market. Richter & sheble (2000)
is that they are unable to handle large size power proposed a PBU using GA for Competitive
system network and it fails to give an accurate environment which considers the demand
solution within a short duration of time (ie.) constraints and it schedules for more profit. GA
Computational time is also more under the open to PBUC provided optimal UC and also optimal
market environment. Researchers developed a MW values for demand, reserve.

https://www.sic.ici.ro
 Profit Maximization of GENCO’s Using an Elephant Herding Optimization Algorithm 133

Vargas & Chen (2010) combined LR and GA 2. Problem formulation

to solve PBUC problems. Attaviriyanupap et.al
(2003) illustrated a hybrid LR-EP approach The objective function varies between a cost-
which actuates Genco’s -generating schedule minimized conventional market and profit
with the quantity of generated power and maximized restructured power market. The
spinning reserve to be sold by the bidding objective of the restructured power marketplace in
process to get the maximum profit. Valenzuela PBUC is not only to reduce the participation cost
but also to maximize the generating companies
et.al (2001) examined a new method of
(GENCO’s) profit under customary constraints
solution for individual power producer to
like forecasted demand, reserve capacity, ramp
tackle UC problem in electric power markets. rate limits, spot price, and minimum up/downtime.
Dimitroulas et.al (2011) explored a solution Revenue received from energy sold in the power
for PBUC problems by a hybrid model of GA market minus the net participation cost gives the
and narrow search algorithm. To obtain the profit of the GENCO.
maximum profit with the best possible solution
in the power market, Muller Optimization 2.1 Objective Functions
method is explored by Chandram et.al (2009).
The objective function for profit maximization is
To maintain the high search capability, a new
given by
nodal ant colony optimization is introduced by
Columbus et.al (2011). Srikanth Reddy et.al Max( PF )  max(TR  TC ) (1)
(2016) & (2019) proposed a new approach Where TR is total revenue, TC is the total
called Binary firework’s algorithm and binary operating cost for the power demand as well as
whale optimization algorithm for PBUC the reserve demand.
problem to obtain a maximum profit for N T
GENCO’s. The dimension of the problem, TR    {( P it * SP t )U it} (2)
complex programming and computation i 1 t 1

time are notified as major limitations of these N T

methods. In this connection, the upgrading of TC    {(C it ( P it )  S it ) *U it} (3)

i 1 t 1
the existing methods is required in order to
obtain the optimal solution for PBUC problems. The total input cost is the summation of the power
generation cost and cost calculated with startup/
In this research article, a swarm based shut down constraint of all generating units over
meta-heuristic approach, Elephant Herding whole optimum scheduling time. The fuel cost of
Optimization algorithm is presented to maximize generating unit ‘i’ at hour‘t’ is calculated by using
the profit of GENCO’s. EHO algorithm is the quadratic cost function.
stimulated by the herding activity of elephants.
The food and shelter searching method is the Cit ( Pit )  ai  bi Pit  ci Pit2 (4)
main idea in this algorithm. This EHO method Where Cit(Pit) is the Power generation cost of unit
is implemented to solve the above -mentioned ‘i’ at hour ‘t’. Pit is the output power from the
optimal scheduling and profit maximization generating unit “i” at hour t; ai, bi and ci are fuel
under the deregulated power market. The cost function coefficients of unit “i”. SPt is the
article sequence is as follows: formulation of forecasted power price at hour‘t’.
the multi-objective function for Profit Based
Unit Commitment problem is dealt with in Startup cost:
section 2. Section 3 presents the idea of the   Toff ,i 
S  S oi 1  D i exp     E i (5)
it T
proposed Elephant Herding Algorithm. Section   down ,i  
4 discusses the implementation of EHO algorithm
Where Sit is Startup Cost, Soi is Cold startup cost,
for the PBUC problems under deregulated pool
Di& Ei are the startup cost coefficients.
market. Section 5 deals the meticulous outcomes
and discussion followed by conclusion with The various constraints considered for PBUC
comparative results of the work in section 6. problems are as follows:

ICI Bucharest © Copyright 2012-2020. All rights reserved

134

2.2 System Constraints 2.3.3 Ramp up/Down limits

2.2.1 Load constraints or Demand constraints The ramp up/down limits are the permissible
timely modification in power generating stations,
The balancing load demand constraints of PBUC
 min{P i , P i (t 1)  RU i}
max max
are given as P it (11)
 max{P i , P i (t 1)  .RD i}
N min min
P (12)
 P *U
i 1
it it
 PD t;1  t  T (6) it

PBUC time input step function τ is assumed to be

Where Uit equals to 1 if power generating unit 60 mins. Where, RUi & RDi are the ramp up and
‘i’ at hour t is ON and Uit equals to 0 if power down limits for unit ‘i’.
generating unit ‘i’ at hour t is OFF. These two
variables are known as Decision Variables. PDt 2.3.4 Crew constraints
is a power demand at hour ‘t’. In a deregulated
power system, it is not obligatory to generate the When more units are in ‘ON’ state at an equal time
same power as demand. period, then crew constraints are restricted.

2.2.2 Forecasted reserve constraints 3. Elephant Herding Optimization

The forecasted whole system reserve capacity Wang et.al (2015) introduced a metaheuristic
and GENCO’s reserve capacity together form the algorithm called Elephant Herding optimization
inequality constraint as follows, for solving multi-objective optimization
N
problems. It is a nature-inspired algorithm that
 P *U  SR ;1  t  T
it it t
(7) imitates the crowding activities of elephants
i 1 in groups. It has a mixed behavior of swarm
intelligence and evolutionary algorithm. The
Where SRt is a forecasted reserve of hour ‘t’. This elephant behavior modeling has both abuse
is also a Decision variable. (Group updating operator) and examination
(separating operator). In nature, elephants live
2.3 Thermal Unit Constraints together as a clan. Even though they belong to
various groups, they will live together under
2.3.1 Generation limit/ Dispatching limit the captainship of the eldest and largest female
The generation boundaries linked with the elephant matron of the group. The male elephants
committed generating units, leave their nuclear family unit when they reach
adulthood as shown in Figure 1.

P
min
it
max
 P it  P it ; i  1, 2, 3.....N , t  1, 2, 3....T (8)
min max
Where P it & P it are the min and max bound on
the output power of unit ‘i’.

2.3.2 Minimum up/ Minimum downtime

≥T i
on up
T i (9)

≥T i
off down
T i
(10)
on off
Where T i & T i are the continuous ON and
OFF time period. T i & T i are the Min Up and
UP down
Figure 1. Elephant herding behavior nature
Down time of unit ‘i’.

https://www.sic.ici.ro
 Profit Maximization of GENCO’s Using an Elephant Herding Optimization Algorithm 135

In spite of the fact, though the male elephants 3.2 Separating operator
live away, they make contact with their clan
through low-frequency pulsations. From this, When the separating operator is applied in each
interaction, the elephant is moving to a new
it is observed that exploration is done by male
position & replacing it with the worst fitness in
and exploitation is done by female respectively. the ith group.
When any of the male elephants finds the
enhanced location, then the whole clan will move γ worst , ci
=γ
min
+ (γ
max
−γ
min
+ 1) * r (16)
towards that position. The female elephant does
Where γmax and γmin are max and min limit
a local search of that region. To form the global
of the individual elephant’s location, respectively.
optimization method, the crowding activities of r€[01] is a kind of stochastic distribution.
elephants are considered with (i) Clan updating Therefore, the elephant herding algorithm implies
operator-which updates the elephant’s & matron’s the iterative applying 13-16 for a predefined no of
current position in each clan, (ii) Separating iterations. The Population size and Maximum no.
Operator- which enhances the inhabitant range of iterations are indirectly controlled by the no. of
at every search period. clans and clan size, whereas α and β are fixed for
certain application.
3.1 Clan updating operator
4. Implementation of EHO Algorithm
Initially, the total elephant population is assumed
as ‘n’ clans. While organizing the elephants, clan The PBUC optimization problem is accomplished
updating operator is applied based on their fitness using the EHO procedure following the steps
function. Each member j of the ith clan moves mentioned below:
according to the elephant matriarch, Ci with best Step 1:
fitness value as,
Read the GENCO’s unit and system data like
γ new, ci , j
= γ + α * (γ
c best ,c i
−γ )*r
ci , j (13)
Generation limits, cost coefficients, min up/
downtime, etc.
Step 2:
Where γnew, Ci,j and γCi,j are recently restructured
and old location of elephant j in group Ci, Read the EHO parameters such as maximum no.
respectively. α €[0,1] is a level parameter which of elephants, no. of clans, α and β.
decides the impact of ith matron Ci on γCi,j,γ Step 3:
best,Ci
represents the matron Ci which is the best Compute the feasible units for forecasted demand
individual elephant in group Ci and r€[0,1] or Market price of all objectives Function.
explained by R. Vijay, et.al (2018). The best
Step 4:
elephant can’t be updated in the group by eqn. 13
which means γ best,Ci =γCi,j . For the best elephant, Calculate the objective function (power
it can be updated accordingly, generation, cost, revenue, etc.) for entire load
scheduling time periods and Compute the PBUC
γ new, ci , j
= β *γ
center , ci (14) schedule prevailing the system constraints. If it is
completed, then go to the next step or else back
to step 3.
β€ [0,1] is another tuning parameter which decides
the impact of γ center,Ci on γ best, Ci,j. dth dimension Step 5:
is determined by the below equation, Call the EHO algorithm and Set the iteration count
i=1 and assign the population size. Calculate the
1 nci
γ center , ci , d
= * ∑γ
n j =1 c i , j , d
(15) Fitness function (Profit of Units) for all of the
solutions in each clan.
Step 6:
Where the dimension limits are1≤d≤D. Here D is
the total dimension of the problem. nci is the total Update the clan operator with the best and worst
quantity of elephants in the clan Ci. position of the elephants using eqn. 13 -16 for the
aforesaid objective function of PBUC problems.

ICI Bucharest © Copyright 2012-2020. All rights reserved

136 Sundar Ravichandran, Manoharan Subramanian

Step 7: 5. Numerical Results and Discussion

Separate the worst (local optimum) elephant using
the separating operator by eqn. 16. The elephant The elephant Herding Optimization algorithm
will communicate with others to update the current was developed in MATLAB 7.10 and the
worst position among the iteration. machine configuration is Intel I5 processing unit
Step 8: with the 3.55GHz speed with 8 GB RAM. In this
article, 2 GENCO’s (Three units, and ten units)
Check for the total no. of a clan. If it is reached test systems were taken for simulation. The
then go to next step, otherwise go to step 5
computational outcomes of Profit Based Unit
with new values of α and β which are normally
Commitment acquired by the EHO algorithm
assumed €[0,1].
for 2 GENCO’s and the simulation outcomes
Step 9: were compared with the various standing
After updating the best and worst elephant (Global optimization methods.
and local optimum), check for the optimum
solution for the PBUC problem. If it is reached, 5.1 GENCO I (3 Units 12-hour Schedule)
then save the best simulation results and then stop
the process otherwise change the PBUC variables Elephant Herding Optimization algorithm
and proceed to step 4. chooses only the best fit optimum allocation
if the number of units is less than two. Before
executing the PBUC-EHO algorithm, it is
necessary to find an accurate hourly power
demand of GENCO’s and a scheduled period
spot price. The generator cost function is always
derived in the quadratic equation.

Table 1 is the system operating data for

GENCO-I consisting of 3 units 10 bus system.
When the generators are in Continuous
operation, the abuse of min up/downtime
limitations may be avoided.
Table 1. Unit cost and performance data of
GENCO-I

Parameter G1 G2 G3
Pmn (MW) 600 400 200
Pmx (MW) 100 100 50
a(Constant) 500 300 100
b(Linear) 10 8 6
C(quadratic) 0.002 0.0025 0.005
Initial Status -3 3 3
Min up/downtime (hr) 3/3 3/3 3/3
Startup cost ($) 450 400 300

Table 2 explains the simulation input parameter

of Elephant Herding optimization algorithm. In
view of the demand data and related spot pricing
in the power market, the generating units are
committing at regular time period. Table 3 clarifies
Figure 2. Flow Diagram of PBUC with the optimum allocation of GENCO-I at the end of
proposed method

https://www.sic.ici.ro
 Profit Maximization of GENCO’s Using an Elephant Herding Optimization Algorithm 137

each generation which defines the PBUC optimum

schedule with revenue and profit. Figure 3 shows
the performance of total cost, GENCO’s revenue
and its profit over 12 hours’ time period. The
profit is found to be enhanced from the graphical
observation. Figure 4 and Figure 5 gives a
comparative analysis from the optimum allocation
of the PBUC schedule.
Table 2. Input Parameters of EHO Figure 4. Comparison between power demand and
power generation
Population size 30
No. of Generations 50
α 0.5
β 0.1
Clan number 5
Number of the elephants in each clan 10

Figure 5. Comparison between Reserve power

demand and reserve allocation

Table 4 provides a comparison between profits

obtained with other optimization methods like
LR-EP, MPPD-ABC, MMA.
Table 4. Profit comparison between proposed
methods and existing methods for GENCO-I

S. No Scheme Profit ($)

1 Swarup et.al 9136
2 K Asokan et.al 9457.50
Figure 3. Performances of Total cost, Revenue, and 3 A. Amudha et.al 9168
Profit in GENCO-I 4 EHO Method 9735.5

Table 3. Optimum Allocation of GENCO-I (3 units 10 Bus System)

Forecasted Power Forecasted Reserve Allocated
Total Cost Revenue Profit Profit
Hour Demand Generated spot Price Demand Reserve
($) ($) ($) (Rs)
(MW) (MW) ($/MW-hr) (MW) (MW)
1 170 170 10.55 20 20 1253.50 1793.50 540.00 37530.00
2 250 200 10.35 25 0 1500.00 2070.00 570.00 39615.00
3 400 200 9.00 40 0 1500.00 1800.00 300.00 20850.00
4 520 200 9.45 55 55 1500.00 1890.00 390.00 27105.00
5 700 600 10.00 70 20 5400.00 6000.00 600.00 41700.00
6 1050 600 11.25 95 0 5400.00 6750.00 1350.00 93825.00
7 1100 600 11.30 100 0 5400.00 6780.00 1380.00 95910.00
8 800 600 10.65 80 0 5400.00 6390.00 990.00 68805.00
9 650 600 10.35 65 0 5400.00 6210.00 810.00 56295.00
10 330 330 11.20 35 35 2882.25 3964.00 1081.75 75181.63
11 400 400 10.75 40 40 3700.00 4500.00 800.00 55600.00
12 550 550 10.60 55 55 4906.25 5830.00 923.75 64200.63
Total 53977.50 44242.00 9735.5 676617.25

ICI Bucharest © Copyright 2012-2020. All rights reserved

138 Sundar Ravichandran, Manoharan Subramanian

5.2 GENCO-II (10 units 24-hr schedule) Table 6 gives the optimum load dispatch schedule
of the GENCO-II with total operating costs,
Table 5 gives the system working information and revenue, and profit over a period of 24 hours.
the power demand data for 10 units’ system. The GENCO-II receives high profit even though only
ramp rate limits are calculated by using eqn.11 a few units are operating at a particular period.
and 12. With this ramp limit, the continuous load Figure 6 and Figure 7 show comparisons of
scheduling for PBUC problem can be obtained various test results provided in Table 6. From
under the deregulated power market. Table 7, it can be observed that the total cost

Table 5. Unit cost and performance data of GENCO-II

Parameter G1 G2 G3 G4 G5 G6 G7 G8 G9 G10
Pmn (MW) 455 455 130 130 162 80 85 55 55 55
Pmx (MW) 150 150 20 25 25 20 25 10 10 10
a (constant) 0.00048 0.00031 0.00200 0.00211 0.00398 0.00712 0.00079 0.00413 0.00222 0.00173
b (linear) 16.19 17.26 16.60 16.50 19.70 22.26 27.74 25.92 27.27 27.79
c (quadratic) 1000 970 700 680 450 370 480 660 665 670
Up /
8/8 8/8 5/5 /55 6/6 3/3 3/3 1/1 1/1 1/1
downtime
Rampup/
40/60 62/73 75/91 51/109 133/142 119/257 270/276 51/83 158/145 152/90
down
Startup cost
4500 5000 550 560 900 170 260 30 30 30
($)
Initial Status 8 8 -5 -5 -6 -3 -3 -1 -1 -1

Table 6. Operating costs, Revenue and Profit for GENCO-II

Power Generated Reserve Allocated Forecasted Total

Time Startup Revenue Profit
Demand Power Demand Reserve Spot Price operating
(Hour) Cost($) ($) ($)
(MW) (MW) (MW) (MW) ($/MW) Cost ($)
1 700 700 70 70 22.15 0 15246 17056 1810
2 750 750 75 75 22.00 0 15864 18150 2286
3 850 850 85 60 23.10 0 17353 21021 3668
4 950 910 95 0 22.65 0 17353 20364 3011
5 1000 910 100 0 23.25 0 17353 21158 3805
6 1100 1040 110 0 22.95 1120 20214 23868 3654
7 1150 1150 115 0 22.50 1100 22709 24756 2047
8 1200 1170 120 0 22.15 0 23106 27344 4238
9 1300 1300 130 0 22.80 1800 26184 29640 3456
10 1400 1400 140 120 29.35 340 29048 41442 12394
11 1450 1412 145 0 30.15 0 29048 42572 13524
12 1500 1412 150 0 31.65 0 29048 44690 15642
13 1400 1400 140 120 24.60 0 29048 36953 7905
14 1300 1300 130 0 24.50 0 26184 31850 5666
15 1200 1170 120 0 22.50 0 23106 26325 3219
16 1050 1050 105 105 22.30 0 22809 26091 3282
17 1000 1000 100 100 22.25 0 20214 23366 3152
18 1100 1040 110 0 22.05 0 20214 22392 2178
19 1200 1040 120 0 22.20 0 20214 23088 2874
20 1400 1040 140 0 22.65 0 20214 23556 3342
21 1300 1040 130 0 23.10 0 20214 24024 3810
22 1100 1040 110 0 22.95 0 20214 23868 3654
23 900 900 90 10 22.75 0 17353 20703 3350
24 800 800 80 80 22.55 0 16827 19844 3017
Total 4360 519137 634121 114984

https://www.sic.ici.ro
 Profit Maximization of GENCO’s Using an Elephant Herding Optimization Algorithm 139

and profit of the EHO based on PBUC solution 6. Conclusion

for GENCO-II are higher than those of the
conventional existing methods. In this article a new meta-heuristic approach,
the Elephant Herding optimization algorithm is
utilized to get solution for the PBUC problem
under deregulated power marketplace. This
works on grouping behavior of the elephants
in the clan. Based on the exploration and
exploitation operator of the Elephant clan,
PBUC solution procedure was developed
using the EHO algorithm. This makes way for
extensive simulation experiment for various
Figure 6. Comparison of Total cost, revenue and
profit of 3 units’ 24-hours economic conditions. The numerical outcomes
are presented with reference to the solution
excellence and its features of various EHO
algorithms. EHO algorithms optimally allocate
the generators to evaluate the fitness value of
the objective function (Profit Maximization) in
a balanced and oscillated power market. The
numerical results are tested on a proposed system
which includes optimum UC schedule, power
generation total cost, startup cost, revenue, and
Figure 7. Comparison of demand & generation for profit. The comparative study is also done with
24 hours
other benchmark existing approaches. From the
Table 7. Comparison of total cost and profit with
existing search methods for PBUC
solution, it is evident that the proposed elephant
herding algorithm has more ability, accuracy,
S. No Authors Total Cost ($) Profit ($)
robustness with less computational time for the
1 Tim & Sheble 623441 27889
2 Kazarlis et.al 610500 40830
solution of power system optimization problem
3 Swarup et. al 609023 42306 in a deregulated open pool market. Future
4 Ganguly et. al 591715 59615 enrichment of this work will concentrate on
5 Logavani et.al 581541 69788 the performance improvement in a large power
PBUC-EHO system with reserve uncertainty and sensitivity
6 519137 114984
(Proposed) for reserve changes.

REFERENCES
Annakkage, U. D., Numnonda, T. & Attaviriyanupap, P., Kita, H., Tanka, E. &
Pahalawaththa, N.C. (1995). Unit commitment by Hasegawa, J. (2003). A hybrid LR-EP for solving
parallel simulated annealing, IEEE Proceedings- new profit-based UC problem under competitive
Generation, Transmission & Distribution, environment, IEEE Transactions of Power
142(6), 595-600. System, 18(1), 229-237.
Amudha, A. & Christober Asir Rajan, A. Chandram, K., Subrahmanyam, N. & Sydulu, M.
(2013), Generator scheduling under competitive (2009). Improved Pre-Prepared Power Demand
environment using Memory Management Table and Muller’s Method to Solve the Profit
Algorithm, Alexandria Engineering Journal, 52, Based Unit Commitment Problem, Journal of
337-346. Electrical Engineering & Technology, 4(2),
159-167.
Asokan, K. & Ashokkumar, R. (2014). Emission
controlled profit based unit commitment for Chen, H. & Wang, X. (2002), Cooperative
Gencos using MPPD table with ABC algorithm evolutionary Algorithm for Unit Commitment,
under competitive environment, WSEAS IEEE Transactions on Power Systems, 16,
Transactions on Systems, 13, 523-543. 128-133.

ICI Bucharest © Copyright 2012-2020. All rights reserved

140 Sundar Ravichandran, Manoharan Subramanian

Chendur Pandian, S. & Duraiswamy, K. (2004). In IEEE Conference Power Tech, Bucharest (pp.
Fuzzy Logic Implementation for Solving the 1-6).
Unit Commitment Problem. In International
Conference on Power System Technology - Reddy, S. K., Panwar, L. K., Kumar, R. &
POWERCON, Singapore (pp. 413-418). Panigrahi, B. K. (2016). Binary fireworks
algorithm for profit based unit commitment
Columbus, C. C. & Simon, S. P. (2011). Profit problem, International Journal of Electrical
based unit commitment for GENCOs using Power & Energy Systems, 83, 270-282.
parallel PSO in a distributed cluster, ACEEE
International Journal of Electrical and Power Reddy, S., Panwar, L., Panigrahi, B. K. & Kumar,
Engineering, 2(3), 24-30. R. (2019). Binary whale optimization algorithm:
a new metaheuristic approach for profit based
Contreras-Hernandez, E. J. & Cedenfo- unit commitment problems in competitive
Maldonado J. R. (2006). A Sequential electricity markets, Engineering optimization,
Evolutionary Programming Approach to 51(3), 369-389.
Profit-Based Unit Commitment. In IEEE PES
Transmission and Distribution Conference and Richter, C W. & Sheble, G. B. (1997). Genetic
Exposition Latin America, Venezuela (pp. 1-8). Algorithm Evolution of Utility Bidding Strategies
for the Competitive Market Place. In 1997 IEEE/
Dimitroulas Dionisios, K. & Georgilakis Pavlos, PES Summer Meeting, Berlin, Germany, PE-752-
S. (2011). A new memetic algorithm approach PWRS-1-05-1997, IEEE, NewYork.
for the price based unit commitment problem,
Application of Energy, 88(12), 4687-2699. Richter, C. W. & Sheble, G. B. (2000). A Profit
based Unit Commitment GA for Competitive
Ganguly, D., Sarkar, V. & Pal, J. (2004). A Environment, IEEE Transactions on Power
New Genetic Approach for Solving the Unit Systems, 15(2), 715-721.
Commitment Problem. In Proceedings of the
International Conference on Power System Sasaki, H., Watanabe, M., Kubokawa, J. & Yorino,
Technology – POWERCON 2004, Singapor, 1, (pp. N. (2002). A solution method of unit Commitment
542-547). DOI: 10.1109/ICPST.2004.1460054 by Artificial Neural Networks, IEEE Transactions
on Power Systems, 7, 974-981.
Jing, Y., Zhou, J., Wu, W., Yang, J. & Yu, W. (2004).
Solution of the profit-based unit commitment Swarup, K. S. & Yamashiro, S. (2002). Unit
problem by using multi-agent system. In Commitment Solution Methodology Using
Proceedings of the Fifth World Congress on Genetic Algorithm, IEEE Transactions on Power
Intelligent Control and Automation (IEEE Cat. Systems, 17(1), 87-91.
No.04EX788), Hangzhou, China, China, 079-
5083. DOI: 10.1109/WCICA.2004.1343686 Valenzuela, J. & Mazumdar, M. (2001). Making
Unit Commitment Decisions When Electricity is
Kazarlis, S. A., Bakirtzis, A. G. & Petridis, V. Traded at Spat Market Prices. In Proceedings of
(1996). A Genetic Algorithm Solution to the Unit IEEE Winter Meeting.
Commitment Problem, IEEE Transactions on
Power Systems, 11(1), 83-92. Vijay, R. & Abhilash, M. (2018). Elephant
Herding Optimization for optimum allocation of
Logavani, K. & Senthil Kumar, S. (2014), Solving electrical distributed generation on distributed
Short Term Unit Commitment Problem Using power networks, Asian Journal of electrical
Lambda Optimization and Modified Dynamic sciences, 7(2), 70-76.
Programming, International Journal of Applied
Engineering Research, 9(24), 27569-27579. Wang, G.-G., Deb, S. & Coelho, L. (2015),
Elephant Herding Optimization. In 2015 3rd
Mantawy, A. H., Abdel-Magid, Y. L. & Selim, International Symposium on Computational
S. Z. (1998). Unit commitment by Tabu search. and Business Intelligence (ISCBI 2015), Bali,
In Proceedings of Institutions of Electrical Indonesia (pp. 1-5). DOI: 10.1109/ISCBI.2015.8
Engineering, Generation, Transmission and
Distribution, 145(1), (pp. 56-64). Yamin, H. Y., El-Dwairi, Q. & Shahidehpour, S.
M. (2007). A new approach for GENCOs profit
Marifield, T. T. & Sheble, G. B. (1996). Genetic based unit commitment in day-ahead competitive
Based Unit Commitment Algorithm, IEEE electricity markets considering reserve
Transactions on Power Systems, 11(3), 1359-1370. uncertainty, International Journal of Electrical
Power Energy System, 29(8), 609-616.
Mori, H. & Okawa, K. (2009). New Meta-
Heuristic Method for Profit- Based Unit
Commitment Under Competitive Environment.

https://www.sic.ici.ro