AbstractAn effort to develop a unit commitment approach

capable of handling large power systems consisting of both thermal

and hydro generating units offers a large profitable return. In order to
be feasible, the method to be developed must be flexible, efficient
and reliable. In this paper, various proposed methods have been
described along with their strengths and weaknesses. As all of these
methods have some sort of weaknesses, a comprehensive algorithm
that combines the strengths of different methods and overcomes each
others weaknesses would be a suitable approach for solving
industry-grade unit commitment problem.

KeywordsUnit commitment, Solution methods, and
Comprehensive algorithm.
I. INTRODUCTION
HE unit commitment problem determines the combination
of available generating units and scheduling their
respective outputs to satisfy the forecasted demand with the
minimum total production cost under the operating constraints
enforced by the system for a specified period that usually
varies from 24 hours to one week. Attempts to develop rigid
unit operating schedules more than one week in advance are
extremely curtailed due to uncertainty in hourly load forecasts
at lead times greater than one week.
Besides achieving minimum total production cost, a
generation schedule needs to satisfy a number of operating
constraints. These constraints reduce freedom in the choice of
starting up and shutting down generating units. The
constraints to be satisfied are usually the status restriction of
individual generating units, minimum up time, minimum
down time, capacity limits, generation limit for the first and
last hour, limited ramp rate, group constraint, power balance
constraint, spinning reserve constraint, hydro constraint, etc.
The high dimensionality and combinatorial nature of the
unit commitment problem curtail attempts to develop any
rigorous mathematical optimization method capable of solving
the whole problem for any real-size system. Nevertheless, in
the literature, many methods using some sort of approximation
and simplification have been proposed. The available
approaches for solving unit commitment problem can usually
be classified into heuristic methods and mathematical
programming methods. The proposed mathematical
programming approaches are dynamic programming,
Lagrangian relaxation, Benders decomposition and mixed

Manuscript received August 12, 2007.
Sayeed Salamis with the Dept of Computer Science and Engineering,
BRAC University, 66 Mohakhali, Dhaka, Bangladesh (phone: 8802-8824051
Ext 4018; fax: 8802-8810383; e-mail: [email protected]).
integer programming [1]-[3].
In the literature, dynamic programming and Lagrangian
relaxation have been used extensively to develop industry-
grade unit commitment programs. Their major advantage
seems to be the requirement of reasonable computation time
when compared to other mathematical approaches.
In dynamic programming, it is relatively easy to add
constraints that affect operations at an hour (such as power
balance constraints) since these constraints mainly affect the
economic dispatch and solution method. However, the
dynamic programming suffers from the curse of
dimensionality. Hence, it is required to limit the commitments
considered at any hour through some simplification
techniques such as truncation and fixed priority ordering. This
simplification, particularly for large scale systems, can lead to
suboptimal schedules.
The utilization of Lagrangian relaxation in production unit
commitment problems is much more recent than the dynamic
programming methods. It has the advantage of being easily
modified to model characteristics of specific utilities.
Lagrangian relaxation method is more advantageous due to its
flexibility in dealing with different types of constraints. It is
relatively easy to add unit constraints. Lagrangian relaxation is
flexible to incorporating additional coupling constraints that
have not been considered so far. The only requirement is that
constraints must be additively separable in units. Such
constraints could be area reserve constraint, area interchange
constraint, etc. To incorporate such constraint into the
framework of Lagrangian relaxation, a Lagrangian multiplier
is defined for each constraint for each time period and the
constraints are adjoined into the objective function of the
relaxed problem. The Lagrangian relaxation method is also
more flexible than dynamic programming because no priority
ordering is imposed. The amount of computation varies
linearly with the number of units. Hence, it is computationally
much more attractive for large systems.
One weakness of the Lagrangian relaxation method is that
the dual optimal solution seldom satisfies the once relaxed
coupling constraints. Another weakness is the sensitivity
problem that may cause unnecessary commitments of some
units. Therefore only a near optimal feasible solution can be
expected. However, the degree of suboptimality decreases as
the number of units increases.
Aoki et al. [4] presented the most promising Lagrangian
relaxation approach developed so far. Tong and Shahidehpour
[2] improved it further through the inclusion of post-
processor. Two new efficient techniques namely non-
discretization of generation levels in the solution process of
Unit Commitment Solution Methods
Sayeed Salam
T
World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
301
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7


single unit dynamic programming and inclusion of ramp rate
constraints are proposed by Guan et al. [5]. Kuloor et al. [6]
show environmental constraint incorporation technique in unit
commitment problem formulation.
Due to the imperfections of the dynamic programming
algorithm, the application of a unit commitment expert system
[7], [8] has been proposed to supplement this method. The
constraints, which are difficult or impractical to be
implemented in unit commitment algorithm, can be handled
by this expert system. In this paper, various proposed methods
for unit commitment have been described. Finally a
comprehensive algorithm [9], [10] for solving industry-grade
unit commitment problem is described.
II. HEURISTIC METHODS
Heuristic methods are non-rigorous computer aided
empirical methods, which make the unit commitment
decisions according to a pre-calculated priority list and
incorporate all the operating constraints heuristically.
Baldwin et al. [11] have used a heuristic approach for unit
commitment. All units are shut down and started up in strict
priority order. The priority list is prepared on the basis of the
average full load cost of each unit, where the average full load
cost is the net heat rate at full load times the fuel cost.
Kerr et al. [12] have proposed heuristic approach that
begins with an initial feasible schedule and then follows the
sequence of steps for adjusting the starting and stopping times
accordingly to reduce the cost of operation.
The heuristic method proposed by Happ et al. [13] uses a
sub-optimizer to get a feasible and near optimal commitment.
Then that method employs an optimizer to further optimize
the schedule. The optimizer uses a sequence of steps
repeatedly until no further reduction in cost is observed.
Heuristic approach to the short-term unit commitment
problem has also been implemented using an expert system
[14], [15].
Heuristic methods have the following advantages [16]:
are flexible and allow for the consideration of
practical operating constraints
feasible solutions if there are any are usually
obtained
computational requirements in terms of memory
and running time are modest
The main shortcoming of heuristic methods is that they
cannot guarantee the optimal solutions or even furnish an
estimate of the magnitude of their sub-optimality. This aspect
becomes rather significant in large-scale power systems, as a
small percentage, e.g. 0.5%, in the costs of unit commitment
schedules represents a substantial financial annual saving.
Therefore, it is advantageous to employ more rigorous
methods compared with heuristics methods to generate more
economical solutions as the size of a system grows, despite the
requirement of comparatively large computational efforts.
More recently, metaheuristic approaches have been used
such as simulated annealing [17], tabu search [18], genetic
algorithms [19]-[21], and greedy randomized adaptive search
procedure [22].
III. MIXED INTEGER PROGRAMMING
Integer programming optimizes integer function of integer
variables. A modification of standard integer programming
that allows non-integer function is known as mixed-integer
programming (MIP). MIP treates the objective and constraint
functions as continuous and the variables as integers.
Branch and bound is one of the techniques used for the
solution of the integer problem. It is a technique to solve a
discrete variable problem by solving a sequence of simpler
problems derived from the original problem. In solving using
the branch and bound method, one needs to define 1) the
problems in the branch and bound three, 2) the method of
solving each problem on the tree, and 3) the method of
searching the tree. The branch and bound tree is fully
described if one defines the problem corresponding to the top
node of the tree and the method of obtaining the children of
any node [1].
Dillon et al. [23] have formulated the unit commitment
problem as a linear MIP problem. Then they have used
standard integer programming algorithm for solving the
commitment schedule.
One of the proposed MIP methods [24] transforms the
linear optimization problems that arise during the search
procedure in the branch and bound algorithm into capacitated
transhipment problems. These are then solved efficiently by a
network-based solution procedure.
Bond and Fox [25] presented an algorithm based on a
combination of mixed integer-linear and dynamic
programming. Mixed integer-linear programming is used to
determine feasible combinations of units at each scheduling
point, while a novel dynamic programming approach
identifies promising scheduling routes in the time domain.
MIP models proposed so far have employed linear cost
functions although more accurate cost models are available.
To date, branch and bound techniques have only been
employed on moderate sized systems using linear models [1].
On the contrary, many available economic dispatch algorithms
use quadratic cost curves. Moreover, the present trend is
toward improved modeling of unit input/output characteristics
with more detailed non-linear models. The MIP based
approach, using only linear models, was found to take long
computation time. The MIP formulation of unit commitment
would become a very large problem demanding extremely
long computation time if applied to a typical generation mix
with more detailed non-linear models.
IV. BENDERS DECOMPOSITION
In Benders Decomposition method [26]-[29], the unit
commitment problem is decomposed into a master problem
involving only the discrete commitment variables and a
subproblem involving the continuous generation variables.
The subproblem corresponds to the economic dispatch
problem with a given commitment. The marginal costs, for
World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
302
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7


each hour, from the subproblem are used to constrain the
allowed commitments in the master problem. The master
problem supplies commitments to the economic dispatch
problem. The master problem and the economic dispatch
subproblem are solved iteratively until the solution converges.
Any economic dispatch routine can be applied for solving the
subproblem.
The major difficulty in Benders decomposition approach is
the determination of the solution of the master problem, which
is still regarded as a large scale integer optimization problem.
Turgeon [26] has solved the master problem by a variational
approach and a branch and bound algorithm whereas
Baptistella and Geromel [27] have solved it by relaxation in
the master level of Benders decomposition approach. In order
to improve the efficiency, some of the constraints which are
difficult to handle, such as nonlinear minimum up and down
time constraints, are replaced by simpler constraints in the
actual formulation of the scheduling algorithm. For instance,
Habibollahzadeh and Bubenko [28] did not use the minimum
up and down time constraints in their mathematical model but,
instead, included a constraint that allowed only one
commitment per day for each unit. Ma and Shahidehpour [29]
dealt with transmission-constraint by introducing a proper
constraint called Benders cut.
V. DYNAMIC PROGRAMMING
Dynamic programming acts as an important optimization
technique with broad application in many areas [30]. Dynamic
programming decomposes a problem into a series of smaller
problems, solves the small problems, and develops an optimal
solution to the original problem step-by-step. The optimal
solution is developed from the subproblem recursively.
In its fundamental form, the dynamic programming
algorithm for unit commitment problem examines every
possible state in every interval. Some of these states are
rejected instantly because they are found infeasible. But even,
for an average size utility, a large number of feasible states
will exist and the requirement of execution time will stretch
the capability of even the largest computers. Hence many
proposed techniques used some sort of simplification and
approximation to the fundamental dynamic programming
algorithm [31].
The approach, first used by Lowery [32], and later refined
by Ayoub and Patton [33], selected unit generation output as a
state variable and on-line capacity as the stages. Ayoub and
Patton included probabilistic techniques for reserve
determination in the developed code.
A typical approach [34], [35] determines some nominal
commitment, which is determined to be good for each hour.
Choices that have to be considered are minimum number of
units, in the priority ordering, needed to meet the reserve
constraints and the result of the above priority list
optimization. A set of units in the priority list about the
nominal commitment are then chosen for optimization - the
units below that set are assumed to be committed while the
units above that set are assumed to be off. If this set contains 5
units, then the state space size is 31 which is a reasonable
number.
The approaches [31], [36] used selection techniques for
choosing the most promising states from all possible states
and implemented approximate economic dispatch subroutines
to reduce computer running time requirement. Variable
window truncated dynamic programming that adjusts the
window size according to the incremental load demands in
adjacent hours and controls the program execution time have
been proposed. Kumar and Palanisamy [37] proposed a two-
step process that uses a direct computation Hopfield neural
network to generate economic dispatch. Then using dynamic
programming the generator schedule is produced.
The approaches [7], [8] have an integrated expert system
into the truncated dynamic programming based unit
commitment program to check and modify commitment
results.
VI. LAGRANGIAN RELAXATION
In Lagrangian relaxation approaches, unit commitment
problem is formulated in terms of 1) a cost function, that is the
sum of terms each involving a single unit, 2) a set of
constraints involving a single unit, and 3) a set of coupling
constraints (the generation and reserve requirements), one for
each hour in the study period, involving all units. Cohen and
Sherkat [1] have reported that an approximate solution to this
problem can be obtained by adjoining the coupling constraints
onto the cost using Lagrangian multipliers. The cost function
(primal objective function) of the unit commitment problem is
relaxed to the power balance and the generating constraints
via two sets of Lagrangian multipliers to form a Lagrangian
dual function. The dual problem is then decoupled into small
subproblems which are solved separately with the remaining
constraints. Meanwhile, the dual function is maximized with
respect to the Lagrangian multipliers, usually by a series of
iterations.
Unfortunately, duality theory has shown that for nonconvex
problems there will be a duality gap between the cost obtained
by solving the relaxed problem and the optimal cost of the
original problem. Since the commitment decision variables are
discrete, the unit commitment problem is nonconvex. Due to
the nonconvexity of the unit commitment problem, the dual
solution seldom satisfies the power balance constraints and the
reserve constraints. Hence, in addition to solving the dual
problem, a suboptimal feasible solution is usually searched
near the dual optimal point. Even though the dual optimal
solutions may violate the feasibilities of the original problem,
they may usually provide sharp lower bounds.
Muckstadt and Koenig [38] employed Lagrangian
relaxation to replace the common linear programming
relaxation approach, which dropped the integrality
requirements of the variables, in the fathoming process of
branch and bound algorithms. This causes a significant
improvement of computational efficiency compared with
previous branch and bound algorithms.
Based on the sharp bound provided by the Lagrangian dual
optimum, it is expected that a suboptimal feasible solution
near the dual optimal point can be accepted as a proper
World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
303
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7


solution for the primal problem. Merlin and Sandrin [39]
presented a more direct and fairly efficient technique based on
this idea. In this algorithm, a modified subgradient method
was used which incorporated the search of the suboptimal
feasible solution along the direction for maximizing the dual
function.
Some methods have been suggested which can be
implemented in systems with fuel constraints. An additional
set of multipliers has been introduced to associate the fuel
constraints with the primal function. Cohen and Wan [40]
proposed a successive approximation approach for altering the
three sets of multipliers. Three iteration loops were
constructed to update these multipliers independently,
according to their corresponding constraints. The iteration of
these three sets of multipliers was done by enclosing the three
iteration loops with an external loop. Once the dual optimal
had been found, the commitment of the fuel-constrained units
would be fixed. Then, any violations of the constraints were
adjusted by changing the commitment schedule and the
outputs of the remaining generating units, which included
thermal units operating without fuel constraints. These
adjustments were within an iteration loop that successively
solved the primal subproblems and modified the set of
multipliers associated with reserve constraints.
Aoki et al. [4] presented an algorithm which also dealt with
the fuel-constrained unit commitment problem. In the
algorithm, the dual optimal solution was determined by an
iteration loop which updated the multipliers associated with
power balance constraints and fuel constraints simultaneously.
The multipliers associated with fuel constraints were updated
by introducing an additional iteration loop within the previous
one. Compared with Cohen and Wans work, this method
introduced a more appropriate coupling between the three sets
of multipliers. In addition, the variable metric method was
employed instead of the subgradient method for updating the
multipliers. These two modifications improved the efficiency
of the process for finding the dual optimal. The feasible
solution was determined by successive modification of the
three sets of multipliers near the dual optimal and the solution
of the primal subproblem.
Tong and Shahidehpour [2] have developed an algorithm to
deal with limitations of the available Lagrangian relaxation
methods for scheduling large scale systems which consist of
thermal, fuel constrained and hydro generating units. The
basic building block of the algorithm follows the Lagrangian
relaxation approach directly [4]. The Lagrangian dual optimal
is first solved and a feasible solution is searched by modifying
the multipliers. Once a feasible solution is obtained, a new
post-processor based on the application of linear programming
is applied to find the final dispatch of units and improve the
schedule by decomitting unnecessary units for further saving
in cost.
In many approaches [2], [4], [39], the thermal subproblem is
solved by using dynamic programming that discretizes
generation levels and the ramp rate constraint is not
considered. The discretization of generation levels causes a
trade-off between computational requirements and the
accuracy. On the other hand, if the generation levels are not
discretized, the ramp rate constraint that couples the
generation levels of two consecutive hours is difficult to deal
with. A straightforward application of dynamic programming
technique may lead to suboptimal results as illustrated by
Guan et al. [5].
Guan et al. [5] have presented an approach that does not
discretize generation levels and handles ramp rate constraints
systematically. The thermal subproblem without ramp rate
constraint is solved by first constructing a state transition
diagram where the optimal generation levels of all up states
are computed without discretizing generation levels. Dynamic
programming technique is then applied with only a few well-
structured states. This eliminates the difficult trade-off
between computational requirements and the accuracy as
needed by most approaches that discretize generation levels.
Ramp rate constraints are relaxed by introducing an additional
set of multipliers for a unit with the constraints. The
subproblem is then solved as if there were no ramp rate
constraint. An intermediate level is introduced to update this
set of multipliers.
The algorithm presented by Yan et al. [41] follows the
Lagrangian relaxation approach [5]. But, in addition to
thermal units, hydro units have been included in the problem
formulation. Given the set of Lagrange multipliers, a hydro
unit subproblem is solved by a merit order allocation method,
and a thermal unit subproblem is solved by using dynamic
programming without discretizing generation levels.
Kuloor et al. [6] describe a method of solving
environmentally constrained thermal unit commitment.
Environmental consideration is added as a second objective
function to the conventional unit commitment. The problem is
thus converted from this dual-objective minimization model
into a single-objective model.
VII. COMPREHENSIVE ALGORITHM
As the size of a power system increases, the savings
potential of unit commitment in absolute terms increases
although in percentage terms it remains the same. The effort to
develop a unit commitment approach that is able to handle
large systems consisting of both thermal and hydro generating
units therefore offers a large profitable return. In order to be
feasible, the method to be developed must be flexible,
efficient and reliable. This requires that nonlinear functions
such as thermal generation cost, water discharge
characteristics, transmission loss and emission constraints
must be incorporated into the method. Furthermore, the
proposed method must be capable of dealing with other
operational constraints.
The potential of unit commitment can be realized by
employing a more rigorous mathematical programming
technique. Of the four techniques surveyed, Benders
decomposition is the least promising and is reflected by the
lack of published work reporting its success. Even though
mixed integer programming is rigorous, its demand on
computing resources is prohibitive. It may be feasible if
parallel computing using a fast processor is employed.
Dynamic programming also suffers the same problem as MIP,
but the truncated version reduces true computing requirement
albeit at the possible loss of accuracy. Lagrangian relaxation is
World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
304
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7


the least rigorous of the mathematical programming
techniques but offers the best performance in computing
requirements. Under the current computing technology, it
seems that Lagrangian relaxation is the viable mathematical
programming technique to solve large-scale unit commitment.
This is reflected by the fact that they are widely reported in
the literature.
To get an industry-grade efficient algorithm based on
Lagrangian relaxation, the techniques like non-discretization
of generation levels, handling of ramp rate and environmental
constraints [5], [6] can be incorporated in the algorithm [2],
[4]. Transmission loss can be incorporated using a general
transmission loss formula [42] whose expression has a similar
quadratic form to the B matrix loss formulation. The
generation cost and water discharge rate can be represented as
continuously increasing quadratic function of the generation.
Finally, an efficient unit commitment expert system [8] can be
developed as a supplement to the Lagrangian relaxation
method.
To fulfill the aim cited above, hydrothermal scheduling
based Lagrangian relaxation (HTSLR) approach to solve the
unit commitment problem for a large practical system
comprising both thermal and hydro generating units is
proposed [9]. Commitment states of thermal units are obtained
by solving thermal subproblems only. To achieve the output
levels of hydro units, the hydrothermal scheduling is
performed with a thermal unit commitment schedule obtained
by solving only thermal subproblems. Extensive constraints
are taken into account such as status restriction of individual
generating units i.e. must run, must out, base load, cycling and
peaking, power balance, spinning reserve, minimum up/down
time, capacity limits, ramp rate, limited generation for the first
and last hour, sulfur oxide emission and hydro constraints.
Non-linear functions are employed for thermal generation cost
and water discharge rate. A general transmission loss formula
[42] whose expression has a similar quadratic form to the B
matrix loss formulation has been utilized for incorporating
transmission loss.
In the HTSLR approach, the commitment schedule may be
so sensitive to the variations of the Lagrange multipliers that a
slight modification of the multipliers may change the status of
several units. This sensitivity problem is more serious for
systems having several groups of identical units. Even though
fuel costs of identical units can be slightly modified to make
small differences among cost characteristics, this sensitivity
problem still exists. In other words, unnecessary commitment
of some units may be possible in the solution given by this
method. In order to overcome this difficulty, a refinement
process similar to that proposed by Tong and Shahidehpour
[2] has been developed for HTSLR approach. This refinement
process inspects some candidate units whose shutdown may
result in additional reduction of the operating cost.
A unit commitment expert system [8] has also been
developed [10]. It was employed as a preprocessor as well as a
postprocessor to the HTSLR based unit commitment program
to check and alter commitment results by adjusting the input
data if necessary. It handles constraints which are difficult or
impractical to be implemented in commitment algorithm such
as cycling of gas turbine and steam turbine units, group
constraints, etc.
VIII. CONCLUSION
The savings potential of unit commitment in absolute terms
increases with the increase of power system size. The effort to
develop a unit commitment approach that is able to handle
large systems consisting of both thermal and hydro generating
units therefore offers a large profitable return. In order to be
feasible, the method to be developed must be flexible,
efficient and reliable. Many proposed methods have been
described along with their strengths and weaknesses. It seems
that the comprehensive algorithm that combines the strengths
of different methods and overcomes each others weaknesses
would be a suitable approach for solving industry-grade unit
commitment problem.
REFERENCES
[1] A. I. Cohen and V. R. Sherkat, Optimization based methods for
operations scheduling, Proceedings of the IEEE, vol. 75, no. 12, pp.
1574-1591, 1987.
[2] S. K. Tong and S. M. Shahidehpour, An innovative approach to
generation scheduling in large-scale hydro-thermal power systems with
fuel constrained units, IEEE Trans. on Power Systems, vol. 5, no. 2, pp.
665-673, 1990.
[3] N. P. Padhy, Unit commitment - a bibliographical survey, IEEE Trans.
on Power Systems, vol. 19, no. 2, pp. 1196-1205, 2004.
[4] A. Aoki, T. Satoh, M. Itoh, T. Ichimori, and K. Masegi, Unit
commitment in a large scale power systemincluding fuel constrained
thermal and pumped storage hydro, IEEE Trans. on Power Systems,
vol. 2, no. 4, pp. 1077-1084, 1987.
[5] X. Guan, P. B. Luh, H. Yan, and J . A. Amalfi, An optimization-based
method for unit commitment International Journal of Electrical Power
and Energy Systems, vol. 14, no. 1, pp. 9-17, 1992.
[6] S. Kuloor, G. S. Hope, and O. P. Malik, Environmentally constrained
unit commitment, IEE Proceedings- Generation, Transmission &
Distribution, vol. 139, no. 2, pp. 122-128, 1992.
[7] S. Mokhtari, J . Singh, and B. Wollenberg, A unit commitment expert
system, IEEE Trans. on Power Systems, vol. 3, no. 1, pp. 272-277,
1988.
[8] M. S. Salam, A. R. Hamdan, and K. M. Nor, Integrating an expert
systeminto a thermal unit commitment algorithm, IEE Proceedings-
Generation, Transmission & Distribution, vol. 138, no. 6, pp. 553-559,
1991.
[9] M. S. Salam, K. M. Nor, and A. R. Hamdan, Hydrothermal scheduling
based Lagrangian relaxation approach to hydrothermal coordination,
IEEE Trans. on Power Systems, vol. 13, no. 1, pp. 226-235, 1998.
[10] M. S. Salam, K. M. Nor, and A. R. Hamdan, Comprehensive algorithm
for hydrothermal co-ordination, IEE Proceedings- Generation,
Transmission & Distribution, vol. 144, no. 5, pp. 482-488, Sept. 1997.
[11] C. J . Baldwin, K. M. Dale, and R. F. Dittrich, A study of the economic
shutdown of generating units in daily dispatch, AIEE Trans. Part III,
Power Apparatus and Systems, vol. 78, pp. 1272-1284, 1959.
[12] R. H. Kerr, J . L. Scheidt, A. J . Fontana, and J . K. Wiley, Unit
commitment, IEEE Trans. on Power Apparatus and Systems, vol. 85,
no. 5, pp. 417-421, 1966.
[13] H. H. Happ, R. C. J ohnson, and W. J . Wright, Large scale hydro-
thermal unit commitment: method and results, IEEE Trans. on Power
Apparatus and Systems, vol. 90, no. 3, pp. 1373-1384, 1971.
[14] S. K. Tong, S. M. Shahidehpour, and Z. Ouyang, A heuristic short-term
unit commitment, IEEE Trans. on Power Systems, vol. 6, no. 3, pp.
1210-1216, 1991.
[15] S. Li, S. M. Shahidehpour, and C. Wang, Promoting the application of
expert systems in short-termunit commitment, IEEE Trans. on Power
Systems, vol. 8, no. 1, pp. 286-292, 1993.
[16] E. Khodaverdian, A. Brameller, and R. M. Dunnett, Semi-rigorous
thermal unit commitment for large scale electrical power systems, IEE
World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
305
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7


Proceedings- Generation, Transmission & Distribution, vol. 133, no. 4,
pp. 157-164, 1986.
[17] F. Zhuang and F. D. Galiana, Unit Commitment by Simulated
Annealing, IEEE Trans. on Power Systems, vol. 5, no. 1, pp. 311-317,
1990.
[18] B. Xiaomin, S. M. Shahidehpour, and Y. Erkeng, Constrained unit
commitment by using tabu search algorithm, in Proc. Int Conf. on
Electrical Engineering, vol. 2, 1996, pp.1088-1092.
[19] S. A. Kazarlis, A. G. Bakirtzis, and V. Petridis, A genetic algorithm
solution to the unit commitment problem, IEEE Trans. on Power
Systems, vol. 11, no. 1, pp. 83-90, 1996.
[20] S. O. Orero and M. R. Irving, A genetic algorithmmodeling framework
and solution technique for short termoptimal hydrothermal scheduling,
IEEE Trans. on Power Systems, vol. 13, no. 2, pp. 501-516, 1998.
[21] S. Liyong, Z. Yan, and J . Chuanwen, A matrix real-coded genetic
algorithmto the unit commitment problem, Electric Power Systems
Research, vol. 76, no. 9-10, pp. 716-728, 2006.
[22] A. Viana, J . P. de Sousa, and M. Matos, A new metaheuristic approach
to the unit commitment problem, 14th Power Systems Computation
Conference, Sevilla, Spain, 24-28 J une 2002, Session 05, Paper 5.
[23] T. S. Dillon, K. W. Edwin, H. D. Kochs, and R. J . Taud, Integer
programming approach to the problemof optimal unit commitment with
probabilistic reserve determination, IEEE Trans. on Power Apparatus
and Systems, vol. 97, no. 6, pp. 2154-2166, 1978.
[24] D. Lidgate and K. M. Nor, Unit commitment in a thermal generation
system with multiple pumped storage power stations, International
Journal of Electrical Power and Energy Systems, vol. 6, no. 2, pp. 101-
111, 1984.
[25] S. D. Bond and B. Fox, Optimal thermal unit scheduling using
improved dynamic programming algorithm, IEE Proceedings-
Generation, Transmission & Distribution, vol. 133, no. 1, pp. 1-5, 1986.
[26] A. Turgeon, Optimal scheduling of thermal generating units, IEEE
Trans. on Automatic Control, vol. 23, no. 6, pp. 1000-1005, 1978.
[27] L. F. B. Baptistella and J . C. Geromel, Decomposition approach to
problemof unit commitment schedule for hydrothermal systems, IEE
Proceedings-D, vol. 127, no. 6, pp. 250-258, 1980.
[28] H. Habibollahzadeh and J . A. Bubenko, Application of decomposition
techniques to short-term operation planning of hydrothermal power
system, IEEE Trans. on Power Systems, vol. 1, no. 1, pp. 41-47, 1986.
[29] H. Ma and S. M. Shahidehpour, Transmission-constrained unit
commitment based on Benders decomposition, International Journal of
Electrical Power and Energy Systems, vol. 20, no. 4, pp. 287-294, 1998.
[30] R. E. Bellman and S. E. Dreyfus, Applied dynamic programming.
Princeton University Press, New J ersey, 1962, pp. 4-16.
[31] W. L. Snyder, H. D. Powell, and J . C. Rayburn, Dynamic programming
approach to unit commitment, IEEE Trans. on Power Systems, vol. 2,
no. 2, pp. 339-351, 1987.
[32] P. G. Lowery, Generating unit commitment by dynamic programming,
IEEE Trans. on Power Apparatus and Systems, vol. 85, no. 5, pp. 422-
426, 1966.
[33] A. K. Ayoub and A. D. Patton, Optimal thermal generating unit
commitment, IEEE Trans. on Power Apparatus and Systems, vol. 90,
no. 4, pp. 1752-1756, 1971.
[34] C. K. Pang and H. C. Chen, Optimal short term thermal unit
commitment, IEEE Trans. on Power Apparatus and Systems, vol. 95,
no. 4, pp. 1336-1346, 1976.
[35] C. K. Pang, G. B. Sheble, and F. Albuyeh, Evaluation of dynamic
programming based methods and multiple area representation for
thermal unit commitment, IEEE Trans. on Power Apparatus and
Systems, vol. 100, no. 3, pp. 1212-1218, 1981.
[36] W. J . Hobbs, G. Hermon, S. Warner, and G. B. Sheble, An enhanced
dynamic programming approach for unit commitment, IEEE Trans. on
Power Systems, vol. 3, no. 3, pp. 1201-1205, 1988.
[37] S. S. Kumar and V. Palanisamy, A dynamic programming based fast
computation Hopfield neural network for unit commitment and
economic dispatch, Electric Power Systems Research, vol. 77, no. 8,
pp. 917-925, 2007.
[38] .A. Muckstadt and S. A. Koenig, An application of Lagrangian
relaxation to scheduling in power-generation systems, Operation
Research, vol. 25, no. 3, pp. 387-403, 1977.
[39] A. Merlin and P. Sandrin, A new method for unit commitment at
Electricite De France, IEEE Trans. on Power Apparatus and Systems,
vol. 102, no. 5, pp. 1218-1225, 1983.
[40] A. I. Cohen and S. H. Wan, A method for solving the fuel constrained
unit commitment problem IEEE Trans. on Power Systems, vol. 2, no. 3,
pp. 608-614, 1987.
[41] H. Yan, P. B. Luh, X. Guan, and P. M. Rogan, Scheduling of
hydrothermal power systems, IEEE Trans. on Power Systems, vol. 8,
no. 3, pp. 1358-1365, 1993.
[42] O.I. Elgerd, Electric energy systems theory: an introduction. McGraw-
Hill Book Company, New York, 1971, pp. 294-296.


World Academy of Science, Engineering and Technology
Vol:1 2007-11-22
306
I
n
t
e
r
n
a
t
i
o
n
a
l

S
c
i
e
n
c
e

I
n
d
e
x

V
o
l
:
1
,

N
o
:
1
1
,

2
0
0
7

w
a
s
e
t
.
o
r
g
/
P
u
b
l
i
c
a
t
i
o
n
/
1
1
2
0
7