Electrical Power and Energy Systems 33 (2011) 939946

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

Hydro-thermal unit commitment problem using simulated annealing

embedded evolutionary programming approach
C. Christober Asir Rajan
Department of EEE, Pondicherry Engineering College, Pondicherry 605 014, India

a r t i c l e

i n f o

Article history:
Received 24 July 2006
Received in revised form 20 December 2010
Accepted 1 January 2011
Available online 12 February 2011
Keywords:
Hydro-thermal unit commitment
Evolutionary programming
Simulated annealing

a b s t r a c t
This paper presents a new approach to solve the hydro-thermal unit commitment problem using Simulated Annealing embedded Evolutionary Programming approach. The objective of this paper is to nd the
generation scheduling such that the total operating cost can be minimized, when subjected to a variety of
constraints. A utility power system with 11 generating units in India demonstrates the effectiveness of
the proposed approach; extensive studies have also been performed for different IEEE test systems consist of 25, 44 and 65 units. Numerical results are shown comparing the cost solutions and computation
time obtained by conventional methods.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Hydro-thermal scheduling is concerned with the commitment
and dispatch of generating units. The optimal operation planning
of hydro-thermal power system is to reduce the operational costs.
Because of its complexity, the optimization problem is divided into
several tasks with different planning periods. The optimal scheduling of generation in a hydro-thermal system involves the allocation
of generation among the hydro-electric and thermal plants so as to
reduce total operation costs of thermal plants while satisfying the
variety of constraints on the hydraulic and power system network.
In tackling this problem, the thermal system may be represented
by an equivalent thermal generating unit. In the process of minimizing the fuel cost of the thermal generators in a hydro-thermal
system, equality and inequality constraints must be satised. The
equality constraints are power balance constraint, the total water
discharge constraint and the reservoir volume constraints. The
inequality constraints are due to the reservoir storage limits and
the operation limits of the equivalent thermal generator and those
of the hydro plant.
Unit Commitment (UC) can save a lot of money and bring in a
large prot. This money can in turn be used to improve the quality
of the supply by installing different quality control equipments.
Unit commitment in power systems refers to the optimization
problem for determining the on/off states of generating units that
minimize the operating cost for a given time horizon. The objective
of the unit commitment in regulated or state monopoly power
markets is to schedule the operation of the generating units in order to serve the load demand at minimum operation costs while

E-mail address: asir_70@hotmail.com

0142-0615/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijepes.2011.01.002

observing all plant and system constraints over a given scheduling

period, ranging from several hours to days ahead.
Research endeavours, therefore, have been focused on; efcient,
near-optimal UC algorithms, which can be applied to large-scale,
power systems and have reasonable storage and computation time
requirements. A survey of existing literature [114] on the problem
reveals that various numerical optimization techniques have been
employed to approach the complicated unit commitment problem.
More specically, these are the Dynamic Programming method
(DP), the Mixed Integer Programming method (MIP), the Lagrangian
Relaxation method (LR), the Augmented Lagrangian Relaxation method (ALR), the Articial Neural Network (ANN), the Simulated Annealing method (SA), the Tabu Search (TS), the TS and Decomposition
method (TSD), the Evolutionary Programming (EP) and so on. The major limitations of the numerical techniques are the problem dimensions, large computational time and complexity in programming.
The DP method [1] does not required the discretization of the
state and control variables and total required storage and computation time are reduced as compared to those in the successive
approximations algorithm. The MIP method [2] can give a near
optimal solution for in an acceptable time. It is suitable for regulated environment but also can be easily applied to deregulation
applications for decision supports with minor changes in the problem formulation. The LR method [3] is shown that decreasing of the
optimal distance function means that the algorithms converge to a
near optimal and feasible solution of the primal problem. Both the
solution accuracy and convergence performance are better or at
least of the same quality as those obtained using the sub gradient
method. It solves accurately and it can be used to obtain approximate solutions to short term planning [4].
The ALR method [5] produces feasible schedules and requires
no iteration with economic dispatch algorithms. Because of the

940

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

Nomenclature
bmid
value appropriately between bmax and bmin
bi
scaling factor
Ai, Bi, Ci the cost function parameters of unit i (Rs/MW2 h, Rs/
MW h, Rs/h)
C1,i . . . C6,i
power coefcients of unit i
Fit (Pit) production cost of unit i at a time t (Rs/h)
Ih (i, t)
Inow to the i th reservoir in time t
PDt
system peak demand at hour t (MW).
Ph (i, t) power produced by i th hydro unit at time t
Ph (i, t)max, Ph (i, t)min power limits of i th hydro unit
pi
number of hours in the reduced period for the unit of
interest
Psit
output power from thermal unit i at time t (MW)
output power from hydro unit i at time t (MW)
Piht
Psmax i
unit i maximum generation limit (MW)
Psmin i
unit i minimum generation limit (MW)
Qh (i, t) discharge of i th reservoir in time t
Qh (i, t)max, Qh (i, t)min discharge limits of i th reservoir in time t
Rt
system reserve at hour t (MW).
Sh (i, t) spillage of i th reservoir in time t

unit-wise decomposition structure of the problem, a more complex

hydro-thermal model and more unit constraints can be easily
incorporated in our approach. This approach [6] which penalizes
all coupling constraints provides more cost savings compared with
LR and ALR with equality constraint penalized. The ANN method
[7] takes care of the concurrent interaction among all the control
variables of the problem. And it will provide extremely fast computation for the optimal scheduling of large hydro-thermal power
systems. The SA algorithm [8] is much faster than the sequential
algorithm and near linear reduction in speed can be achieved. SA
is a powerful technique to solve combinatorial optimization problems. It minimizes the probability of falling into a local minimum.
It has good rule of nding a diversied and intensied neighbourhood so that a vast solution space can be explored.
In the TSD method [9], the more signicant constraints can be
considered in modelling the power system operation. And it will
provide a fast and feasible algorithm with a exible characteristic
which can be used for power scheduling in an open access
environment.
High speed of convergence, minimum cost and mean cost was
achieved by the EP method [10]. It was found that, the EP method
have reasonable accuracy and computational time for scheduling
large complex hydro-thermal system. It can be used for large-scale
complex problem. EP is better in obtaining global optimum, which
relies on mutation rather than crossover. EP is a probabilistic
search technique, which generates the initial parent vectors distributed uniformly in intervals within the limits and obtains global
optimum solution over number of iterations. The main stages of
this technique are initialization, creation of offspring vectors by
mutation and competition and selection of best vectors to evaluate
best tness solution. Due to inherent exibilities in tness function
and ease in coding, EP method produces the best solution with less
number of generations [11]. The evolutionary strategy (ES) is found
as an efcient and advantageous optimization method to solve the
short term planning task [12].
From the literature review, it has been observed that there exists a need for evolving simple and effective methods, for obtaining
an optimal solution for the Unit Commitment Problem (UCP).
Hence, in this paper, an attempt has been made to couple EP with
SA (EPSA) for meeting these requirements of the UCP, which gives

STi
T
Toffi
Toni
Tupi
TC(Sk)
TCmin

start-up cost of unit i at hour t.

total time schedule (24 h)
duration for which unit i is continuously OFF
duration for which unit i is continuously ON
minimum up time of unit i
total cost associate with schedule Sk
cost associated with the best feasible solution in the
population
Tdowni minimum down time of unit i
TFC
total fuel cost
TSDC
total shut-down cost
TSUC
total start-up cost
Uit
unit i status at hour t (1 if unit i is ON; 0 if it is OFF)
Vit
unit i start up/shut down status at hour t (1 if there is a
state change in Uit else 0)
Vh (i, t) volume of i th reservoir in time t
Vh (i, t)max, Vh (i, t)min volume limits of i th reservoir in time t
s(i, m) delay of water ow from i th to m th reservoir

the better solution than the individual EP and SA methods with

reasonable time. In case of SA, the temperature and demand are taken as control parameter. Hence the quality of solution is improved. The algorithm is based on the annealing neural network.
Classical optimisation methods are a direct means for solving this
problem. EP seems to be promising and is still evolving. EP [13,14]
has the great advantage of good convergent property and, hence,
the computation time is considerably reduced. EP does not suffer
from the drawback of handling non-continuous or non-differentiable objective functions as in some plants like Combined Cycle Cogeneration plants. Encoding and decoding schemes essential in the
GA approach are not needed, considerable computation time is
thus saved. The EP combines good solution quality for SA with rapid convergence for EP. And the selection process is done using ES
[14]. The validity and effectiveness of the proposed integrated
algorithm has been tested with a hydro power system of 4 generating units combined with a thermal power system of 7 generating
units Indian utility system, IEEE test system of 25 units, 44 units,
and 65 units systems. And the test results are compared with the
results obtained from other methods.

2. Problem formulation
The objective of UCP entails the selection of generating units in a
power system, to be operated for a given time horizon, while satisfying the forecasted load demand and observing the units operating
requirements, such that the total production cost is minimum. Hydro-thermal scheduling is a complex power systems operation task.
Since hydro resources are almost zero cost, the problem is to schedule the water release and thermal generation such that the system
operational cost over the scheduling period is minimized subject to
numerous electrical and hydrological constraints. Here the thermal
generator variable costs are piecewise linear and convex functions of
output, that hydro generator outputs are piecewise linear and convex functions of water ow, that the intertemporal link (water conservation constraint) is linear and that other constraints are linear.
Thus the optimal scheduling of hydro-thermal power system is usually more complex than that for all-thermal system. It is basically a
non-linear programming problem involving non-linear objective

941

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

function and a mixture of linear and non-linear constraints. As hydro

generating units do not incur any fuel cost the hydro-thermal scheduling problem is aimed to minimize the total thermal cost while
making use of the availability of hydro resource as much as possible.
The objective function and associated constraints of the problem are
formulated as follows.

The objective is to nd the generation scheduling such that the

production cost (1) can be minimized, when subjected to a variety
of constraints.

F it Pit Ai P2it Bi Pit C i

Rs=h

2.2. Overall objective function

The objective is to nd the generation scheduling such that the
total operating cost (2) can be minimized, when subjected to a
variety of constraints. In the UCP consideration, an interesting
solution would be minimizing the total operating cost of the generating units with several constraints being satised. The major
component of the operating cost, for thermal units, is the power
production cost of the committed units and it is the rst term in
the objective function.

MinCu Min

H
X

U ih FCP ih  ST i U ih 1  U i;h1 

SDi U i;h1 1  U ih 

The start-up cost depends upon the downtime of the unit,

which can vary from a maximum value, when the unit I is started
from cold state, to a much smaller value, if the unit I has been
turned off recently. The start-up cost calculation depends upon
the treatment method for the thermal unit during downtime periods. The start-up cost is a function of the downtime of unit i is the
second term in the objective function. In the similar way shutdown cost which is a function of uptime of unit is the third term
in the function. The main objective here is to minimize all the
costs.
2.3. Constraints involved
Depending on the nature of the power system under study, the
unit commitment problem is subject to many constraints [15]. The
constraints met by thermal as well as the hydro units are discussed
below.
2.3.1. Thermal constraints
The temperature and pressure of the thermal units vary very
gradually and the units must be synchronized before they are
brought online. A time period of even 1 h is considered as the minimum down time of the units. The various thermal constraints to
be considered are:
(a) Load balance constraints: The real power generated by all the
units in the thermal power station must be sufcient enough to
meet the load demand according to the following equation

Pit U it PDt

(c) Minimum down time: If the units have already been shut
down, then there is a minimum time before which they can be
restarted.

(d) Minimum up time: If the units are running already, then

there is a minimum time before which they can be shut down.

Toni P Tupi

(e) Capacity limits: Generating units have maximum and minimum generating limits within which they should be operated to
ensure reasonable level of efciency as given by the following
relation.

Pmin i 6 Pi 6 Pmax i

2.3.2. Hydro constraints

(a) Hydro plant generation limits: Limitations on the power generation capacities,

Ph i; tmin 6 P h i; tmax

t2T

(b) Hydraulic network constraints: Physical limitations on reservoir storage volumes and discharge rates,

V h i; tmin 6 V h i; t 6 V h i; tmax

i1 h1

N
X

P max i U it P PDt Rt ; 1 6 t 6 T

i1

Toffi P Tdowni

2.1. Objective function

N
X

N
X

i1

(b) Spinning reserve constraints: The spinning reserve is the total amount of the real power generation available from all synchronized units minus the present load plus the losses (4). It must be
sufcient enough to meet the loss of the most heavily loaded unit
in the system. It has to satisfy the given equation.

Q h i; tmin 6 Q h i; t 6 Q h i; tmax

t2T

(c) System active power balance: The total active power generation must balance the predicted power demand plus losses, at
each time interval over the scheduling horizon.

X
i2Rs

P s i; t

Ph i; t PDt PLtt 2 T

10

iRh

(d) Hydro plant power generation: The power generated from a

hydro plant is related to the reservoir characteristics as well as the
water discharge rate. A number of models have been used to represent this relationship. In general, the hydro generator power output is a function of reservoir volume, V, and the rate of water
discharge, Q. The model can also be written in terms of reservoir
volume instead of the reservoir net head and a frequently used
functional is

Ph i; t C 1;i V h i; t2 C 2;i Q h i; t2 C 3;i V h i; t  Q h i; t

C 4;i V h i; t C 5;i Q h i; t C 6;i i 2 Rh

11

3. Evolutionary programming
EP [14] originally conceived by Lawrence J. Fogel in 1960, is a
stochastic optimization strategy. Evolutionary Programming is a
mutation-based evolutionary algorithm applied to discrete search
spaces similar to GA, but instead places emphasis on the behavioural linkage between parent and their offspring, rather than
seeking to emulate specic genetic operators as observed in nature. EP is similar to ES, although the two approaches were developed independently.
The main stages of this technique are initialization, creation of
offspring vectors by mutation and competition and selection of
best solution. EP is capable of determining the global or near global
solution. The ability of the EP method to nd the global optimum
solution is independent of the size of the discrete load step
assigned to each parent vector in the solution process. Also,

942

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

investigations have shown that EP was better among other Evolutionary Computation methods such as GA and micro GA [13,14].

Wk

Wf

V h i; t V h i; t  1 Ih i; t  Q h i; t  Sh i; t

Ru
X

Q h m; t  si; m Sh m; t  1i; m

12

m1

2. Calculate the power produced by each hydro unit (Ph(i,t))

from the values of discharge rates and volumes.

Ph i; t C 1;i V h i; t2 C 2;i Q h i; t2 C 3;i V h i; t

 Q h i; t C 4;i V h i; t C 5;i Q h i; t C 6;i

Psi; t

i2Rs

Phi; t PDt PLt

14

5. Obtain the power (Pdt) to be produced by thermal unit,

15

The thermal system UC is performed as below.

6. An initial population of parent solutions Sk, k = 1, 2, 3, . . .,
M (where M is the number of parents), is generated at
random.
7. The objective function value associated with each solution
Sk is calculated by economically dispatching the hourly load
to the operating units and by computing the total fuel and
start-up/shut-down costs, i.e.,

TCSk TFCSk TSUCSk TSDCSk

16

S0k

8. An offspring
is created from each parent by adding a
Gaussian random variable N(0, r2k ) to the elements aijk of
parent Sk:

aijk aijk N0; r2k

rk bi 

TCSk  pi
TC min

17
18

Here, the value of bi is chosen in such a manner that product

bi  pi should guarantee a minimum variance. Normally
constant scaling factor is used in conventional EP. In this
non-linear scaling factor is used for better convergence.
For the rst 40% of the total number of generations (N1)
the decrement in scaling factor g1 is calculated as

g1

bmax  bmid
N1

bmid  bmin
N2

1; if TCSk hTCSr
0; otherwise;

22

where r = [2Mu + 1], r not equal to k, [x] denotes the greatest

integer less than or equal to x, c is the number of competitions, and u is a uniform random number ranging over
[0,1]. Here, c is set at 1/10 of the population.
11. The feasible competitors are ranked in descending order of
Wk. The rst M solutions survive and are transcribed along
with their elements to form the basis of the next generation.
12. The process (steps 811) is repeated until the given maximum generation count is reached.

The SA technique is an algorithm [8], which exploits the

resemblance between the annealing of a metal and a minimization process. A metal is a system of many atoms. The total internal energy of the metal depends on its state in the form of
relative position, orientation and motion of the atoms in the metal. Although the prediction of a specic state is almost impossible due to the extremely rapid microscopic movement of the
atoms, the statistical properties of many replica systems of atoms
in their (thermodynamic) equilibrium can be characterized. It has
been observed that when a metal is annealed, or cooled slowly,
the energy of the metal tends to assume the global minimal value. Motivated by this observation, simulated annealing generates
feasible solutions of a minimization problem which correspond to
the states of a metal, with the cost of a feasible solution corresponding to the energy of the metal in a state. By moving among
the feasible solutions the way the states of a metal under annealing would evolve, the global optimum of the problem can be approached with high probabilities. SA has been successfully
applied to many difcult combinatorial optimization problems.
The method assumes no special problem structure and is highly
exible with respect to various constraints.
4.1. SA algorithm to solve UCP problem
1. Commit all the M hydro units and considering discharge
rates (Qh(i,t)) between the limits, calculate the volumes
(Vh(i,t)) of the reservoirs from 1 to M.

V h i; t V h i; t  1 Ih i; t  Q h i; t  Sh i; t

Ru
X
Q h m; t  si; m Sh m; t  1i; m

20

9. Each feasible offspring S0k is evaluated according to 7.

10. For each feasible candidate, parent or offspring, a value Wk
is assigned.

23

m1

2. Calculate the power produced by each hydro unit (Ph(i,t))

from the values of discharge rates and volumes.

Ph i; t C 1;i V h i; t2 C 2;i Q h i; t2 C 3;i V h i; t

 Q h i; t C 4;i V h i; t C 5;i Q h i; t C 6;i

19

For the remaining 60% of the total number of generations

(N2) the decrement in b is calculated as g2 as

g2

4. Simulated annealing

iRh

PDt PD  PDh

21

13

3. Sum up all the hydro powers for each period and subtract
the total hydro power from the power demand for each
period.
4. Find the remaining load demand to be met with thermal
power such that

Wf

f1

3.1. EP algorithm to solve UCP problem

1. Commit all the M hydro units and considering discharge
rates (Qh(i,t)) between the limits, calculate the volumes
(Vh(i,t)) of the reservoirs from 1 to M.

c
X

24

3. Sum up all the hydro powers for each period and subtract
the total hydro power from the power demand for each
period.
4. Find the remaining load demand to be met with thermal
power such that

X
i2Rs

Psi; t

Phi; t PDt PLt

25

iRh

and obtain the power (Pdt) to be produced by thermal unit,

943

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

PDt PD  PDh

26

5. Set an initial feasible solution Ui as current solution for the

thermal system scheduling.
6. Initialize temperature Ck and number of iteration for each
temperature.
7. Perform Economic Load Dispatch and calculate cost of
generation.

Pdti

n
X

Ui; j  Pi; j

27

j1

8. Calculate the Total Operating Cost Ft as the Summation of

Running Cost and Start Up Shut-Down Cost.

Ft

T X
N
X
F i;t Pi;t U i;t Si;t V i;t Rs
t1

28

i1

9. Initialize Fmin(i) = Ft and consider this Ui as best solution.

10. Randomly perturb the solution to get the neighbour solution Uj and check whether all the constraints are satised
and if they are satised go to next step, else repeat this
step.
11. Check if the cost Fj 6 Fi, if less replace Uj and Fj as current
solutions for Ui and Fi, if greater check if exp[(Fi  Fj)/Ck]
P U(0,1) if satised set Ui = Uj.
12. Decrement the iteration and if iteration reaches zero decrement the temperature and nd the converged optimal
value when the temperature reaches zero. The nal schedule is thus obtained and its cost value is also calculated.
5. EPSA algorithm to solve UCP problem

Psi; t

i2Rs

PDt PD  PDh

TCSk TFCSk TSUCSk TSDCSk

Ru
X

Q h m; t  si; m Sh m; t  1i; m

29

m1

2. Calculate the power produced by each hydro unit (Ph(i,t))

from the values of discharge rates and volumes.

Ph i; t C 1;i V h i; t2 C 2;i Q h i; t2 C 3;i V h i; t

 Q h i; t C 4;i V h i; t C 5;i Q h i; t C 6;i

30

3. Sum up all the hydro powers for each period and subtract
the total hydro power from the power demand for each
period.
4. Find the remaining load demand to be met with thermal
power such that

33

8. An offspring
is created from each parent by adding a
Gaussian random variable N(0, r2k ) to the elements aijk of
parent Sk:

aijk aijk N0; r2k

rk bi 

TCSk  pi
TC min

34
35

Here, the value of bi is chosen in such a manner that product

bi  pi should guarantee a minimum variance. Normally
constant scaling factor is used in conventional EP. In this
non-linear scaling factor is used for better convergence.
For the rst 40% of the total number of generations (N1)
the decrement in scaling factor g1 is calculated as

g2

32

S0k

5.1. EPSA algorithm

V h i; t V h i; t  1 Ih i; t  Q h i; t  Sh i; t

31

and for the thermal system Unit Commitment is performed as

below.
6. An initial population of parent solutions Sk, k = 1, 2, 3, . . .,
M (where M is the number of parents), is generated at
random.
7. The objective function value associated with each solution
Sk is calculated by economically dispatching the hourly load
to the operating units and by computing the total fuel and
start-up/shut-down costs, i.e.,

g1

1. Commit all the M hydro units and considering discharge

rates (Qh(i,t)) between the limits, calculate the volumes
(Vh(i,t)) of the reservoirs from 1 to M.

Phi; t PDt PLt

iRh

5. Obtain the power (Pdt) to be produced by thermal unit,

In this proposed work the EP is assisted by the SA technique to

rapidly converge towards the optimum point. The search is basically done with EP but the SA is used to escape the search path
from the local optimum region.

The proposed integrated algorithm combines EP and SA techniques to solve the UCP problem. The EP technique, hold the main
responsibility of nding the optimal point and SA assists EP to converge towards the optimum point quickly. The search is basically
done with EP, but additionally the SA is used to escape the search
path from local optimum point. The algorithm for the proposed
method is as follows:

bmax  bmid
N1

36

For the remaining 60% of the total number of generations

(N2) the decrement in b is calculated as g2 as

bmid  bmin
N2

37

9. Each feasible offspring S0k is evaluated according to 7.

10. For each feasible candidate, parent or offspring, a value Wk
is assigned.

Wk

c
X

Wf

38

f1

Wf

1; if TCSk hTCSr
0; otherwise;

39

where r = [2Mu + 1], r not equal to k, [x] denotes the greatest

integer less than or equal to x, c is the number of competitions, and u is a uniform random number ranging over
[0,1]. Here, c is set at 1/10 of the population.
11. The feasible competitors are ranked in descending order of
Wk. The rst M solutions survive and are transcribed along
with their elements to form the basis for Simulated Annealing Algorithm.
12. In Simulated Annealing Algorithm the temperature variable
(Cp) is initially assigned to be relatively higher value.
13. The number of iteration n to be performed for rening each
individual solution is obtained and the process is done to
every individual independently.
14. The initial solution is assigned as the current best solution
Ui, the function to be checked is assumed to be minimum,
in our case it is the cost Fi.

944

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

15. Random perturbation is done to the current solution and the

neighbouring solution Uj is obtained whose feasibility is
examined by checking to see if there is an uptime or downtime constraint.
16. Check if the cost Fj 6 Fi, if less replace Uj and Fj as current
solutions for Ui and Fi, if greater check if exp [(Fi  Fj)/Ck]
P U(0,1), if satised, set Ui = Uj.
17. The iteration count n is decremented and another neighbouring solution is generated. When the iteration count n reaches
zero, the temperature variable Cp is lowered to a new value.
18. The entire process terminates when sufcient iterations
have occurred at the specied lowest temperature and this
process is repeated to all the individual solution till all the
Np solutions are rened.
19. The rened Np number of population is passed onto the EP
part as the parents for next generation. And this process is

repeated till the convergence in production cost is reached

along with the optimum schedule having satised the
constraints.
The diagrammatic description of the proposed hybrid EPSA
algorithm is shown in Fig. 1.
6. Case study
A hydro power system of 4 generating units combined with a
thermal power system of 7 generating units Indian utility system,
IEEE test system of 25 units, 44 units, and 65 units systems are
considered as the case study. The Unit Commitment Problem is
solved by considering a time period of 24 h. Table 1 gives the coefcients for hydro power generation of utility system. And these
coefcients are used to calculate hydro power by substituting in

Fig. 1. Flowchart for EPSA.

945

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946
Table 1
Hydro power generation coefcients.

Table 4
Thermal system data.

Plant no.

C1

C3

C3

C4

C5

C6

1
2
3
4

0.0042
0.004
0.0016
0.003

0.42
0.03
0.3
0.31

0.03
0.0015
0.014
0.027

0.9
1.14
0.55
1.44

10
9.5
5.5
14

50
70
40
90

the equation. Table 2 gives the reservoir inows of utility system.

Fig. 2 shows the layout of hydro reservoirs. Table 3 gives the minimum and maximum limits of volume of water in the reservoir, the
discharge from the reservoir, and the power generated by each unit
of utility system. Table 4 gives the upper and lower limits of the
power generated by each unit, the fuel cost coefcients and the
start-up cost of each unit which are used for calculating the production cost of each unit of utility system. Table 5 gives the load
demand at each hour per day of utility system.
The comparison done between the combined EP and SA technique with the intelligent techniques EP and SA separately as well
as the conventional method (DP) is illustrated in the Table 6. Here
the generation cost obtained from the DP technique is taken as the

Table 2
Reservoir inows (104 m3).
Period

Reservoir

1
2
3
4
5
6
7
8

10
9
8
7
6
7
6
5

8
8
9
9
8
7
8
9

8.1
8.2
4
2
3
4
4
5

2.8
2.4
1.6
0
0
0
0
0

Units

1
2
3
4
5
6
7

Upper limit
(Mw)

Lower limit
(Mw)

Running cost
A
(Rs/
Mw h2)

B
(Rs/
Mw h)

C
(Rs)

60
80
100
120
150
150
200

15
20
30
25
50
50
75

750
1250
2000
1600
1450
4950
4100

70
75
70
70
75
65
60

0.255
0.198
0.198
0.191
0.106
0.068
0.074

Start up
cost (Rs)

34590
41100
57800
32200
39000
26270
27200

Table 5
Total power demand.
Period (J)

Load (Mw)

Period (J)

Load (Mw)

1
2
3
4
5
6
7
8
9
10
11
12

1249
1166
1184
1168
1165
1173
1152
1173
1165
1135
969
966

13
14
15
16
17
18
19
20
21
22
23
24

908
901
896
860
843
833
855
852
849
804
900
1210

Table 6
Comparisons of costs and CPU time.
System

Methods

Total cost
(P.u.)

CPU time (s)

7 Thermal + 4 hydro systems (Utility)

DP
LR
SA
EP
EPSA
DP
LR
SA
EP
EPSA
DP
LR
SA
EP
EPSA
DP
LR
SA
EP
EPSA

1.00000
0.96481
0.95000
0.94902
0.92690
1.00000
0.94123
0.93210
0.93118
0.90905
1.00000
0.95968
0.94570
0.94462
0.91273
1.00000
0.99910
0.98015
0.97992
0.95790

325
279
270
224
218
509
495
489
448
440
605
578
570
526
519
1452
1368
1370
1332
1320

22 Thermal + 3 hydro systems (EEE)

32 Thermal + 12 hydro systems (EEE)

45 Thermal + 20 hydro systems (EEE)

Fig. 2. Layout of hydro reservoirs.

Table 3
Reservoir storage capacity limits, discharge limits, generation limits and reservoir end
conditions (104 m3).
Plant no.

Pmin (mw)

Pmax (mw)

1
2
3
4

80
60
100
70

130
120
240
160

100
80
170
120

120
70
170
140

5
6
10
13

15
15
30
25

0
0
0
0

500
500
500
500

base for the calculation of per unit cost. It is clearly observed from
the Table 6 that the costs of generation and CPU time obtained
from both EP as well as SA techniques individually are better than
that of the conventional method. And the efciency of the combined EP and SA technique is also understood from the table. Here
since the combination of two equally efcient intelligent techniques have been used the optimal schedule obtained and the cost
of generation is better than that of the individual techniques. The
convergence graph for individual EP and the combined EPSA techniques are depicted in the Fig. 3. Here SA is not considered in the
convergence graph since we give a random schedule as the input

946

C.C. Asir Rajan / Electrical Power and Energy Systems 33 (2011) 939946

Fig. 3. Convergence graph.

rather than that of the input given to both EP and EPSA techniques.
Thus the above results, gures and discussions show that the combined technique of EPSA is an efcient algorithm for solving the
highly combinatorial and non-linear problem of UC. Here since
the process of SA is done to each and every Np number of solutions for each iteration the time taken for convergence is greater
than that of the individual algorithm. In spite of this problem the
search for the optimal global minimum solution is much more
effective and extensive than that of the individual algorithms.
7. Conclusion
This paper is concerned with obtaining a better efcient, fast
and robust solution for unit commitment problem through EPSA
Technique. In the EP approach to the UCP, the essential processes
simulated are mutation, competition, and selection. The mutation
rate is computed as a function of the ratio of the total cost by the
schedule of interest to the cost of the best schedule in the current
population. Competition and selection are applied to select from
among the parents and the offspring, the best solutions, to form
the basis of the subsequent generation. The SA algorithm is incorporated for improving the existing operating schedule into a feasible operating schedule and introduction of new rules for
generating feasible solutions which result in considerable saving
of CPU time. In EPSA, the solution obtained through EP is fed as
the initial solution to SA. SA is also used to verify certain constraints, which is time consuming when done by EP. The use of
Gaussian distribution with non-linear scaling factor in the process
of mutation incorporated in the EP helps in reducing the computing time. On comparing the results attained by the different techniques the EPSA technique obviously displays a satisfactory
performance. In EP, there is no obvious limitation on the size of
the problem that must be addressed, for its data structure is such
that the search space is reduced to a minimum. The population of
feasible solutions is produced at each generation and throughout
the evolution process. No relaxation of constraints is required SA

is known to converge to the global minimum with a probability

of unity and thus improves any given solution. Thus, the solution
obtained through EPSA Technique is having better quality and in
terms of economy and computation time.
References
[1] Ferrero RW, Rivera JF, Shahidehpour SM. A dynamic programming two-stage
algorithm for long-term hydrothermal scheduling of multi reservoir systems.
IEEE Trans Power Syst 1998;13(4):153440.
[2] Chan Gary W, Aganagic Mohamed, Waight James G, Medina Jose, Burton Tony, Reeves
Steve, et al., Experiences with mixed integer linear programming based approaches
on short-term hydro scheduling. IEEE Trans Power Syst 2001;16(4):7439.
[3] Ruzic Slobodan, Rajakovic Nikola. Optimal distance method for lagrangian
multipliers updating in short-term hydrothermal coordination. IEEE Trans
Power Syst 1998;13(4):143944.
[4] Ngundam JM, Kenfack F, Tatietse TT. Optimal scheduling of large-scale
hydrothermal power systems using the lagrangian relaxation technique. Int J
Elect Power Energy Syst 2000;22:23745.
[5] Salem Al-Agtash. Hydrothermal scheduling by augmented lagrangian
consideration of transmission constraints and pumped-storage units. IEEE
Trans Power Syst 2001;16(4):7506.
[6] Salem Al-Agtash, Renjeng Su. Augmented lagrangian approach to hydrothermal scheduling. IEEE Trans Power Syst 1998;13(4):1392400.
[7] Naresh R, Sharma J. Two-phase neural network based solution technique for
short-term hydrothermal scheduling. Proc Inst Elect Eng Gen Trans Dist
1999;146(6):65763.
[8] Wong KP, Wong YW. Short-term hydrothermal scheduling part II: parallel
simulated annealing approach. Proc Inst Elect Eng Gen Trans Dist
1994;141(5):5026.
[9] Xaiomin Bai, Shahidehpour SM. Hydrothermal scheduling by tabu search and
decomposition method. IEEE Trans Power Syst 1996;11(2):96874.
[10] Nidul Sinha, Chakrabarti, Chattopadhyay PK. Fast evolutionary programming
techniques for short-term hydrothermal scheduling. IEEE Trans Power Syst
2003;18(1):21420.
[11] Cau TDH, Kaya RJ. Evolutionary optimisation method for multi storage
hydrothermal scheduling. Proc Inst Elect Eng Gen Trans Dist 2002;149(2):1526.
[12] Werner TG, Verstege JF. An evolution strategy for short-term operation planning
of hydrothermal power systems. IEEE Trans Power Syst 1999;14(4):13627.
[13] Fogel DB. Evolutionary computation. Toward a new philosophy of machine
intelligence. New York: IEEE Press; 1995.
[14] Back T. Evolutionary algorithms in theory and practice. New York: Oxford
University Press; 1996.
[15] Wood AJ, Woolenberg BF. Power generation and control. 2nd ed. New
York: John Wiley and Sons; 1996.