Académique Documents
Professionnel Documents
Culture Documents
org
Published in IET Generation, Transmission & Distribution
Received on 12th June 2009
Revised on 18th January 2010
doi: 10.1049/iet-gtd.2009.0611
ISSN 1751-8687
Abstract: Transmission network expansion planning (TNEP) is a very important problem in power systems. It is a
mixed integer, non-linear, non-convex optimisation problem, which is very complex and computationally
demanding. Various meta-heuristic optimisation techniques have been tried out for this problem. However,
scope for even better algorithms still remains. In view of this, a new technique known as harmony search is
presented here for TNEP with security constraints. This technique has been reported to be robust and
computationally efcient compared to other meta-heuristic algorithms. Results for three sample test systems
are obtained and compared with those obtained with genetic algorithm and bacteria-foraging differential
evolution algorithm to verify the potential of the proposed algorithm.
Introduction
www.ietdl.org
Sf k + g = d
iteration. The rst phase is construction phase that nds out the
feasible solution. The second phase is a local search procedure
that seeks for improvements on the construction phase
solution by local search. The application of a new discrete
method in particle swarm optimisation for TNEP has been
discussed in [9].
A new technique known as harmony search (HS) used for
solving engineering optimisation problems was rst presented
in [10]. The HS algorithm is based on the musical process of
searching for a perfect state of harmony. The harmony in
music is analogous to the optimisation solution vector, and the
musicians improvisations are analogous to local and global
search schemes in optimisation techniques. Instead of a
gradient search, the HS algorithm uses a stochastic random
search [11] based on the harmony memory considering rate
(HMCR) and the pitch adjustment rate (PAR), so that
derivative information is unnecessary. Compared to earlier
meta-heuristic optimisation algorithms, the HS algorithm
imposes fewer mathematical requirements and can be easily
adapted for various types of engineering optimisation problems.
The HS algorithm has been very successful in wide variety of
optimisation problems [1217]. An improved version of
harmony search (IHS) is presented in [18] which employ a
novel method for generating new solution vectors that
enhances the accuracy and convergence rate of the classical HS.
Despite the promise shown by meta-heuristic methods for
TNEP, better techniques are still required. Hence, this paper
presents an application of IHS to TNEP with security
constraints. The transmission expansion problem is a very
complex problem. Hence, it is not possible to consider all
aspects of TNEP in this paper. The objective in this paper
is to investigate the potential of IHS algorithm for TNEP.
Hence, a simple STNEP with security constraints based on
DC model, is considered. The N 2 1 contingency analysis
is used to ensure system security. However, the algorithm is
general enough to consider all other aspects of TNEP. The
planners can use this technique for TNEP incorporating
the issues relevant to the individual systems.
2 Transmission network
expansion planning
The TNEP can be formulated as a mixed integer non-linear
optimisation problem. System security is an important aspect
and must be considered in a TNEP. The N 2 1 contingency
analysis looks at the system state after a single line outage. A
comprehensive model for TNEP with security constraints is
presented in [19], which is used as a base for formulating
TNEP with security constraints in this paper. The TNEP
with security constraints can be stated as follows
cl nl
(1)
min v =
l[V
s.t.
664
& The Institution of Engineering and Technology 2010
(2)
fl k gl (n0l + nl )(Dukl ) = 0,
for l [ 1, 2 . . . , nl and l = k
fl k gl (n0l + nl 1)(Dukl ) = 0,
|fl k | (n0l + nl )f l ,
for l = k
for l [ 1, 2 . . . , nl and l = k
for l = k
(3)
(4)
(5)
(6)
(7)
HS algorithm
www.ietdl.org
random selection
xti
xi [ Xi ,
(8)
i = 1, 2, . . . , N
(9)
(12)
(13)
..
HM =
.
HMS1
x1
xHMS
1
x12
x22
x1N 1
x2N 1
x2HMS1
...
...
..
.
...
HMS1
xN
1
xHMS
2
...
xHMS
N 1
..
.
..
.
x1N
x2N
(14)
PAR;
..
.
HMS1
xN
xHMS
N
(10)
var(X )
(15)
www.ietdl.org
solution to STNEP. Each element in x represents the rightof-way in which a candidate line is constructed. The range of
each variable dened by Xi indicates the list of available rightof-ways. If two lines are added in a particular right-of-way,
IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 6, pp. 663 673
doi: 10.1049/iet-gtd.2009.0611
www.ietdl.org
then two elements with same number (indicating number of
the right-of-way) will come in the vector x.
l
c l nl + W 1
NC
k=0
(abs(fl k ) f l ) + W2 (nl n
l )
ol
(16)
Here ol represents the set of overloaded lines.
The objective of STNEP is to nd the set of transmission
lines to be constructed such that the cost of expansion plan is
minimum and no overloads are produced during the planning
horizon. Hence, the rst term in (16) indicates the total
investment cost of a transmission expansion plan. The
second term is added to the objective function for the real
power ow constraint violations in the base case, and
N 2 1 contingency cases. The third term is added to the
objective function if maximum number of circuits that can
be added in lth right-of-way exceeds the maximum limit.
W1 and W2 are constants. The second and third terms are
added to the tness function only in case of violations.
A stepwise owchart of IHS algorithm for TNEP (Fig. 1)
is given in Section 3.2 as follows.
through
an
x 2NGU+1 , . . ., x genNGU+1
] indicates the set of decision
NGU+1
variables that dene all other candidate lines connected to
the network for a given topology (this includes indirect
connections to the generators also).
It must be noted that this division is carried out only
before evaluating the tness function using DC load ow
to ensure the connectivity of unconnected generators. For
all the other meta-heuristic operations like improvisation,
and so on the harmony vector is a single vector only, like
in the case where unconnected generators/loads are not
present.
Note: Unconnected loads are also treated as unconnected
generators.
www.ietdl.org
Table 1 Results for TNEP with security constraints for generation plan G1 G4
Plan
TNEP with GA
n7 8 3, n1 5 2, n1 2 1,
n3 24 2, n4 9 1, n6 10 2,
n9 12 1, n10 11 1,
n11 14 1, n12 13 1,
n14 16 2, n15 24 1,
n15 21 2, n15 16 1,
n19 22 1
n7 8 3, n6 10 2, n1 5 2,
n15 24 2, n15 21 2,
n15 16 1, n4 9 1, n2 6 1,
n3 24 2, n14 16 2,
n10 11 1, n11 13 1,
n19 22 1
n7 8 3, n6 10 2, n1 5 2,
n1 2 1, n15 24 2, n15 21 2,
n15 16 1, n4 9 1, n2 6 1,
n3 24 2, n14 16 2,
n10 11 1, n11 13 1,
n19 22 1
total number
of lines
22
21
22
investment
cost [106 US$]
964
975
978
n7 8 2, n1 5 1, n2 4 1,
n3 24 2, n3 9 1, n6 10 2,
n10 12 2, n12 13 1,
n14 16 2, n15 24 2,
n15 21 1, n16 17 2,
n17 18 2, n21 22 1
n7 8 2, n6 10 2, n1 5 1,
n15 24 2, n15 21 2,
n15 16 1, n3 24 2,
n14 16 2, n10 12 2,
n19 22 1, n3 9 1, n2 4 1,
n12 13 1
n7 8 2, n6 10 2, n1 5 1,
n1 2 1, n15 24 2, n15 21 2,
n15 16 1, n3 24 2,
n14 16 2, n10 11 1,
n10 12 1, n19 22 1, n3 9 1,
n2 4 1, n11 13 1
total number
of lines
22
20
21
investment
cost [106 US$]
942
974
977
G1
G2
G3
n1 5 2, n3 9 2, n3 24 1,
n4 9 1, n6 10 2, n7 8 3,
n9 12 1, n10 12 1,
n12 23 1, n13 14 1,
n14 23 1, n15 21 1,
n20 23 1, n21 22 1
n7 8 3, n6 10 3, n1 5 2,
n7 8 3, n6 10 2, n1 5 1,
n1 2 1, n15 21 1, n20 23 1,
n15 21 1, n20 23 1,
n1 3 1, n3 9 1, n14 23 1,
n3 24 1, n16 19 1,
n13 14 1, n9 12 1,
n14 23 2, n15 24 1,
n12 23 1, n3 24 1,
n9 12 1, n10 12 1,
n10 12 1, n21 22 1, n2 4 1, n12 13 1, n19 22 1, n4 9 1
n10 11 1
total number
of lines
19
20
20
investment
cost [106 US$]
837
898
903
n7 8 3, n6 10 2, n15 24 2,
n3 24 2, n12 13 1, n2 4 1,
n1 5 1, n1 2 1, n14 23 2,
n10 12 1, n15 21 1,
n21 22 1, n10 11 1
n7 8 3, n6 10 2, n15 24 2,
n3 24 2, n12 13 1, n2 4 1,
n1 5 1, n1 2 1, n14 23 2,
n10 12 1, n15 21 1,
n21 22 1, n10 11 1
n7 8 3, n6 10 2, n1 2 2,
n1 5 1, n15 21 1, n10 11 1,
n3 9 2, n19 22 1, n14 16 2,
n12 23 1, n4 9 1, n20 23 1,
n10 12 1, n9 12 1,
n11 13 1
total number
of lines
19
19
21
investment
cost [106 US$]
882
882
899
G4
www.ietdl.org
4
Results
The proposed algorithm has been tested for three standard test
systems, IEEE 24 bus system, south Brazilian 46 bus system
and 93 bus Colombian system. The comparison of results is
presented for IEEE 24 bus system and 46 bus south
Brazilian system, with the one obtained with basic binary
GA and bacteria foraging-differential evaluation algorithm
(BF-DEA) to conrm the potential of the proposed
approach. The GA and BF-DEA were implemented to
compare the results. The detailed results are shown only for
the case of TNEP with security constraints; however, the
results for TNEP without security constraints are also available.
Table 2 Comparison of number of tness function required by three algorithms, for plans G1 G4
Cost [106 US$]
Plan
IHS
BF-DEA
GA
IHS
BF-DEA
GA
G1
964
975
978
118 280
1 157 900
1 945 090
G2
942
974
977
20 450
737 300
313 167
G3
837
898
903
58 400
553 500
2 753 166
G4
882
882
899
220 500
1 293 100
2 690 833
Compared to BF-DEA
Compared to GA
Compared to BF-DEA
Compared to GA
G1
1.13
1.43
89.78
93.92
G2
3.29
3.58
97.23
93.47
G3
6.79
7.31
89.45
97.88
G4
no change
1.89
82.95
91.81
IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 6, pp. 663 673
doi: 10.1049/iet-gtd.2009.0611
669
www.ietdl.org
Table 4 Effect of HMCR variation on performance of IHS algorithm for TNEP (with HMS 50)
HMCR
6
0.99
0.98
0.95
0.9
837
837
837
1026
38.31
61 250
25.82
58 400
76.77
79.69
113 950
114 300
Table 5 Effect of HMS variation on performance of IHS algorithm for TNEP (with HMCR 0.98)
HMS
cost of expansion plan [US$ 106]
standard deviation
number of tness function evaluations
25
50
75
100
837
837
837
925
82.88
40 250
25.82
58 400
19.50
117 975
85.0815
149 300
www.ietdl.org
Table 6 Effect of HMCR variation on performance of IHS algorithm for TNEP (with HMS 50)
HMCR
0.99
0.98
0.95
0.9
337.809
337.809
337.809
340.679
21.39
239 550
18
96 800
15.55
48
172 600
230 700
Table 7 Effect of HMS variation on performance of IHS algorithm for TNEP (with HMCR 0.98)
HMS
25
6
340.679
standard deviation
number of tness function evaluations
50
337.809
42
18
79 900
96 800
75
100
337.809
337.809
19.50
17.809
117 975
155 300
Number of tness
function
evaluations
Cost of expansion
plan [US$ 106]
HS
2.40 105
337.809
BF-DEA
2.98 10
361.863
basic binary GA
2.67 106
432.350
671
www.ietdl.org
case studies (TNEP with/without security constraints).
Hence, for this system the results are obtained with
IHS only.
IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 6, pp. 663 673
doi: 10.1049/iet-gtd.2009.0611
www.ietdl.org
5
Conclusions
References
[19] SILVA I.D.J., RIDER M.J., ROMERO R., GARCIA A.V., MURARI C.A.:
Transmission network expansion planning with security
constraints, IEE Proc. Gener. Transm. Distrib., 2005, 152,
(6), pp. 828 836
[20] PANDI V.R., PANIGRAHI B.K., MALLICK M.K., ABRAHAM A., DAS S.:
Improved harmony search for economic power dispatch.
Int. Conf. Hybrid Intelligent Systems, 2009, DOI 10.1109/
HIS.2009.294
[9] JIN Y.X., CHENG H.Z., YAN Y.J., ZHANG L.: New discrete particle
swarm optimization and its application in transmission
network expansion planning, Electr. Power Syst. Res.,
2007, 77, pp. 227 233
[23] HAFFNER S., MONTICELLI A., GARCIA A., MANTOVANI J., ROMERO R.:
Branch and bound algorithm for transmission system
expansion planning using a transportation model, IEE
Proc. Gener. Transm. Distrib., 2000, 147, (3), pp. 149 156
IET Gener. Transm. Distrib., 2010, Vol. 4, Iss. 6, pp. 663 673
doi: 10.1049/iet-gtd.2009.0611
673