Electrical Power and Energy Systems 69 (2015) 213221

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

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

Transmission network expansion planning considering repowering

and reconguration
Diego Tejada a,1, Jess M. Lpez-Lezama b,, Marcos J. Rider c,2, Guillermo Vinasco d,3
a

XM S.A., Medellin, Colombia

Universidad de Antioquia (UdeA), Medelln, Colombia
c
Universidade Estadual Paulista (UNESP), Brazil
d
Interconexin Elctrica S.A. ISA, Medelln, Colombia
b

a r t i c l e

i n f o

Article history:
Received 7 November 2013
Received in revised form 6 December 2014
Accepted 11 January 2015
Available online 2 February 2015
Keywords:
Combinatorial optimization
Mixed binary linear programming
Network reconguration
Transmission network expansion planning
Repowering assets

a b s t r a c t
Transmission expansion planning (TEP) is a classic problem in electric power systems. In current optimization models used to approach the TEP problem, new transmission lines and two-winding transformers
are commonly used as the only candidate solutions. However, in practice, planners have resorted to nonconventional solutions such as network reconguration and/or repowering of existing network assets
(lines or transformers). These types of non-conventional solutions are currently not included in the classic mathematical models of the TEP problem. This paper presents the modeling of necessary equations,
using linear expressions, in order to include non-conventional candidate solutions in the disjunctive linear model of the TEP problem. The resulting model is a mixed integer linear programming problem,
which guarantees convergence to the optimal solution by means of available classical optimization tools.
The proposed model is implemented in the AMPL modeling language and is solved using CPLEX optimizer. The Garver test system, IEEE 24-busbar system, and a Colombian system are used to demonstrate
that the utilization of non-conventional candidate solutions can reduce investment costs of the TEP
problem.
2015 Elsevier Ltd. All rights reserved.

Introduction
Transmission expansion planning (TEP) is a classic problem in
electrical power systems; its goal is to nd the optimal plan for
expansion of lines and/or transformers to be installed in the network in order to allow a feasible operation in a pre-dened horizon
at a minimum cost. The optimal expansion plan should dene
where, how many, and when new network elements (lines or
transformers) must be installed. The necessary data for solving
the TEP problem include: current topology, candidate circuits, generation, demand for year-horizon, and investment constraints,
among others.

Corresponding author at: Calle 70 No. 52-21 Medelln, Colombia. Tel.: +57 (4)
2198557, +57 3008315893.
E-mail addresses: dtejada@xm.com.co (D. Tejada), jesusmarialopezl@yahoo.com
(J.M. Lpez-Lezama), mjrider@dee.feis.unesp.br (M.J. Rider), gevinasco@isa.com.co
(G. Vinasco).
1
Address: Calle 12 Sur No. 18-168 Bloque 2, Medelln, Colombia. Tel.: +57 (4)
3172929.
2
Address: Avenida Brasil, 56. Bairro: Centro 15385-000 Ilha Solteira, SP, Brazil.
Tel.: +55 (18) 3743 1000.
3
Address: Calle 12 Sur No. 18-168, Colombia. Tel.: +57 (4) 325 22 70.
http://dx.doi.org/10.1016/j.ijepes.2015.01.008
0142-0615/ 2015 Elsevier Ltd. All rights reserved.

An ideal mathematical model for TEP should use the load ow

equations from the Alternating Current (AC) Model. However, the
use of these equations would result on a mixed-integer non-linear
programming problem which optimal solution is not guaranteed
by available classical optimization tools. Consequently, to solve
the TEP in the long term, relaxed mathematical models are often
used. These models generally use only the active part of the operation of a power system (active power and phase angle). Main
models collected from the literature [13] are: (a) Transportation
model Garver [4]; (b) Linear hybrid model Villasana et al. [5]; (c)
Direct Current (DC) model; and (d) Linear Disjunctive Model Bahiense et al. [6].
The literature provides different approaches to the TEP problem
that have been developed using the models above. Such
approaches include, among others, multistage dynamic planning
[79], demand uncertainty [1012], and the inclusion of security
constraints (N  1 criterion) [9,1315]. Solutions for these mathematical models have been proposed in the literature using different optimization techniques such as metaheuristics [7,1620],
and mathematical programming, [4,6,15,21,22].
Despite of the great number of model adaptations and solution
approaches applied to the TEP problem, there is not reference in

214

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

Nomenclature
Constants
di
load in bus i
maximum generation in bus i
gi
cij
investment for the transmission in branch ij
f
maximum real power ow in branch ij
ij
n0ij
number of existing transmission lines in the branch ij
 ij
maximum number of reinforcement that can be added
n
in the branch ij
xpu
reactance of transmission branch ij
ij
Sbase
base power

h
maximum phase angles in radians

fr
maximum real power ow in repowered branch ij
ij
xrij
reactance of transmission repowered branch ij
crepij
investment for the upgrade of transmission in branch ij
crij
investment for the transmission in the repowered
branch ij
Variables
wij,y
binary decision to add a new line in branch ij and circuit
y
gi
active power generation in bus i
hi
phase angle at bus i in radians

the specialized literature to the use of non-conventional solutions

such as reconguration or repowering of existing network assets.
In consequence, when using the classic modeling of the TEP problem there might be a gap between the proposed solution candidates and the real life solution that can be implemented. In this
case real life solutions might include non-conventional candidates
(i.e. network reconguration or repowering of existing network
assets). Practice has shown that there are real life solutions including non-conventional candidates with a lower cost than proposed
by classic models. Various examples of these solutions around
the world include:
 Circuit reconguration Guavio Tunal 230 kV in Guavio Suria
230 kV and Suria Tunal 230 kV in Colombia, [23].
 Circuit reconguration Guajira Santa Marta 230 kV in Guajira
Termocol 230 kV and Termocol Santa Marta 230 kV in
Colombia, [23].
 New Grid Supply Point at Twinstead in UK including 132 kV
network reconguration, [24].
 Repowering the existing circuit Trujillo Norte Cajamarca
Norte 220 kV, Tingo Maria Vizcarra 220 kV, Vizcarra
Conococha 220 kV in Peru, [25].
 Repowering the circuit Itaip-Margen Derecha 500 kV in Brazil,
[26].
 Repowering the circuit Fortuna Panama y 230 kV Guasquitas
Panama in Panama, [27].
 Midway-Temblor 115 kV Line Reconductor (repowering) by
Pacic Gas and Electric Company in USA, [28].
 Reconductor (repowering) the Mickleton Gloucester 230 kV
parallel circuits with double bundle conductor by PJM in USA,
[29].
Although in practice there might be other types of unconventional candidates (eg voltage level shift or three phase transformers), the reconguration and repowering of circuits are generally
the most common unconventional candidates used in the TEP. As
regards reconguration, network topology optimization has been
explored in [30,31]. However, the main difference with the reconguration modeled in this paper is the fact that topology changes

f ij
fij,y
wak ek ;1
wbk ek ;1
wak bk ;1
rij
wrij;y
0r

f ij
r
f ij;y

Sets
Xb
Xl
Xk

real power ow in existing branch ij

real power ow in added elements in branch ij and circuit y
binary decision for reconguration between busbar a
and e
binary decision for reconguration between busbar b
and e
binary decision for reconguration between busbar a
and b
binary decision for repowering the existing branch ij
binary decision of repowered added line in branch ij and
circuit y
real power ow in repowered existing branch ij
real power ow in repowered added elements in branch
ij and circuit y

busbars set
branches set
set of candidate lines to be recongured

are proposed using existing and possible new circuits of the

network.
The primary motivations for resorting to reconguration and
repowering of circuits include the ever increasing environmental
constraints and the difculty in acquiring the easements and
rights-of-way in major consumption centers (i.e. cities, industrial
centers). Therefore, sometimes the only feasible solution for the
TEP in major consumption centers are, in the mid-term, the use
of FACTS [28], and in the mid and long term the reconguration
or repowering of existing network.
In this paper the authors propose a model for TEP that includes
non-conventional candidates. The main contributions of this paper
are as follows:
(i) It provides a novel modeling for the TEP problem in which,
besides new lines and two-winding transformers, the reconguration and repowering of existing network assets has
also been considered.
(ii) The new constraints are added in such a way that the linear
disjunctive model of the TEP problem remains a mixed-integer linear programming problem, which optimal solution
can be obtained by commercially available software.
The rest of the document is organized as follows: Section A TEP
model with commonly used candidates describes the classic TEP
model that considers commonly used candidate solutions. Section Introducing repowering and reconguration in the TEP problem describes the modeling of non-conventional candidates of the
TEP problem. Section Tests and results provides the tests and
results in different power systems. Finally, conclusions are presented in Section Conclusions.

A TEP model with commonly used candidates

As it was already stated, there are different models for the TEP
problem (AC, transportation, linear hybrid, DC and linear disjunctive). In particular, the disjunctive linear model [6] is a linear
equivalent to the DC model. Because of this, the disjunctive linear

215

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

model can be solved by Mixed Integer Linear Programming (MILP)

maintaining the Kirchhoffs rst and second law. However, in order
to nd the solution using the disjunctive linear model it is necessary to determine the value of a sufciently large constant M
that, in some cases, might lead to numerical problems. The strategy
to avoid the use of such constant is presented in [8], and consists in
using the electrical parameter 2
h (i.e. twice the maximum angle
in a busbar for an electrical system) instead of M. Eqs. (1)(12)
show the linear disjunctive model with the modication proposed
in [8].

Minimize :

cij

ij 2 Xl

wij;y

y2Y

Subject to:

0
f ji

ji 2 Xl

X
y2Y

!
f ji;y

0
f ij

f ij;y

g i di

8i 2 Xb

y2Y

ij 2 Xl


0 hi  hj
Sbase nij
8ij 2 Xl
xpu
ij



f xpu

 ij;y ij
 hi  hj  6 2h1  wij;y


 Sbase
0
f ij

Fig. 1. Example of the network reconguration.

jf ij j 6 n0ijf ij

8ij 2 Xl ;

8y 2 Y

8ij 2 Xl

jf ij;y j 6 wij;yf ij

8ij 2 Xl ;

4
5

8y 2 Y

8i 2 Xb
0 6 g i 6 gi

8i 2 Xb
jhi j 6 h
X
ij
wij;y 6 n
8ij 2 Xl

6
7
8
9

y2Y

wij;y 6 wij;y1 8ij 2 Xl ; 8y 2 Y=y > 1

10

wij;y binary 8ij 2 Xl ; 8y 2 Y

11

hi 0

8i 2 Xb =i ref

12

The objective function given by (1) consists on minimizing the

cost of adding new lines. This problem is subject to constraints
(2)(12). The constraint given by (2) is the power balance equation
of nodes, also known as Kirchhoff Current Law (KCL). Constraints
(3) and (4), respectively, denote the expression of Ohms law for
the existing and candidate transmission lines and so Kirchhoffs
Voltage Law (KVL) is implicitly taken into account. Constraints
(5) and (6) represent the maximum power ow for existing lines
and candidates, respectively. The constraint given by (7) is the
maximum generation of each generator. The constraint given by
(8) represents the maximum values for the angles at the busbar.
Constraint (9) provides the maximum number of new lines that
can be added to the branch ij. The constraint given by (10) avoids
exploring the same solutions twice. Constraint (11) represents the
binary nature of the decision variables. Finally, the constraint given
by (12) denes a reference busbar in the system.
Introducing repowering and reconguration in the TEP
problem
In this section we present the necessary constraints that must
be added to the disjunctive model of the TEP, in order to include
non-conventional candidates (network reconguration and repowering of existing network assets).
New constraints in the TEP model to include network reconguration
Network reconguration is a possibility to expand the transmission grid, taking advantage of the existing network. In general, an
existing line is divided in order to become two new lines.

Fig. 2. Variables and parameters for the network reconguration model.

In Fig. 1, there is an existing line that connects busbars 1 and 2,

which has a given impedance (x12) and maximum power ow (f 12 ).
It is desired to connect busbar 3 to the system using the existing
line, without T connections. A mathematical model is developed
to introduce two dummy busbars a and b into the problem at
the reconguration point, as shown in Fig. 2. In order to include the
network reconguration in the TEP, the constraints illustrated in
Table 1 must be added.
According to the mathematical formulation presented in
Table 1, it is possible to extend the linear disjunctive model to
include reconguration of lines as candidates for TEP. First of all,
it is necessary to dene the lines susceptible to be recongured
in the set of branches. It is also necessary to dene a new set
(Xk) with nodes a, b, and e (node e is the generality of node
3 illustrated in Fig. 2) that allows the reconguration of an existing
line. Eqs. (13) and (14) provided below, correspond to the generalized constraints of Table 1:

wak ek ;1 wbk ek ;1
wak bk ;1 1  wak ek ;1

8k 2 Xk
8k 2 Xk

13
14

Constraints in the TEP model to Include the repowering of existing

network assets
The repowering of an existing power line is the possibility to
expand the transmission grid, by taking advantage of the existing
towers and easements. For example, the cables from a number of
existing lines (n0) between busbars 1 and 2, are changed in order
to increase the maximum power ow through the lines

216

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

Table 1
Constraints for reconguration model.
Constraints

Description

wa3,1 = wb3,1
wab,1 = 1  wa3,1

For a reconguration to take place, sections 3a and 3b must exist

Complementarity constraint: when a reconguration takes place (wa3,1 = 1), section a, b should not exist (wab,1 = 0) and vice versa

0r

f ji f ji

X

f ji;y f ji;y

!


y2Y

ji 2 Xl

0r

f ij f ij

X

f ij;y f ij;y

y2Y

ij 2 Xl

8i 2 X b

g i di

19

To complete the mathematical model, constraints (20)(25)

must be added.


 r
f xr pu 



 ij;y ij
 hi  hj  6 2h 1  wrij;y


 Sbase
 
 r 

8ij 2 Xl ; 8y 2 Y
f ij;y  6 wrij;y fr
ij
X
r
 ij
wij;y 6 n
8ij 2 Xl

8ij 2 Xl ; 8y 2 Y

20
21
22

y2Y

wrij;y 6 wrij;y1
Fig. 3. Repowering of an existing line.

wrij;y 6 r ij

8ij 2 Xl ; 8y 2 Y=y > 1

wij;y 6 1  rij
(f 012 > f 12 ). Also, the new lines impedance is adjusted because of
the cables change (x012 x12 ). Fig. 3 illustrates this situation.
To represent the repowering of an existing branch by using a
mathematical model, it is necessary to redene and add
constraints to the TEP model. First, the constraint described by
(3), KVL for existing lines, is redened as two new constraints
(15) and (16):


 0 pu

f x

 ij ij


h

h

8ij 2 Xl ; n0ij > 0


i
j  6 2hr ij
0

Sbase nij

 0r pu

f xr

 ij ij

h

h

8ij 2 Xl ; n0ij > 0

 6 2h1  rij

i
j

Sbase n0ij

15
16

Note that the reactance of the existing lines xpu

ij is used in (15). If
no repowering is performed rij = 0 and the right side of the inequality becomes zero. For the inequality to hold, the absolute value on
the left side must be zero, and then (15) becomes constraint (3).
The same reasoning can be applied to (16). If repowering is performed rij = 1 and the right side of (16) becomes zero, forcing the
absolute value on the left to be zero. In this case (16) becomes
(3) but with the reactance of the new repowered lines.
The constraint described by (5), maximum power ow for existing lines, also needs to be redened as two new constraints (17)
and (18):

 
 0
f ij  6 n0ij f ij 1  r ij
0r

 r
jf ij j 6 n0ij fr
ij ij

8ij 2 Xl

8ij 2 Xl

17

23

8ij 2 Xl ; 8y 2 Y

24

8ij 2 Xl ; 8y 2 Y

25

The constraint given by (20) is Ohms law for the new lines with
the same characteristics as the repowered lines. The constraint
given by (21) is the maximum power ow for the new lines with
the same characteristics as the repowered lines. Constraint (22)
limits the number of new lines, with the same characteristics as
the repowered lines that can be added to the branch ij. The constraint given by (23) avoids exploring the same solutions twice.
Constraint (24) Guaranties that new lines with the characteristics
of the repowered lines are considered only if such corridor was
effectively repowered (rij = 1). If the decision is no to repower
(rij = 0) new lines will not be built with the characteristic of repowered lines. Constraint (25) is the complement of (24). If a corridor is
repowered (rij = 1) it is not possible to have new lines with the
characteristics of the original. Conversely, if repowering is not performed (rij = 0) the new lines must have the same characteristic of
the originals.
Constraints (24) and (25) guarantee that new lines with the
same characteristics as the repowered lines are included only if
the original line between busbars i and j is repowered.
Finally, it is necessary to redene the objective function given
by (1) as the following equation:

Minimize :

X
ij 2 Xl

cij

X
y2Y

wij;y

n0ij crepij r ij

ij 2 Xl

X
ij 2 Xl

crij

wrij;y

y2Y

26
In this case the cost of repowering lines has been added to the
objective function.

18

It can be observed in (17) that if no repowering is performed

(rij = 0) the power ow of the original lines is limited by the original
constraint (5); however, if repowering is performed (rij = 1) the
power ow of the original lines must be zero.
As regards (18) if lines are repowered (rij = 1) the power ow of
such lines is limited between their minimum and maximum limits;
however if no repowering is performed the power ow of repowered lines is zero.
In order to include the power ow of the recongured lines in
the constraint given by KCL, it is necessary to redene Eq. (2) as
a new constraint (19):

TEP including repowering and reconguration

The objective function given by (26) and constraints (4), (6)
(24) and (25) correspond to the MILP model proposed in this paper
for TEP in power systems including non-conventional candidates
(network reconguration and repowering of existing network).
This MILP model guarantees convergence to optimality while using
existing optimization software. As it will be illustrated in Section Tests and results, this type of optimization problems can be
solved with the help of standard optimization software, as it is usually carried out in other works on this area.

217

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

Tests and results

The proposed MILP model for TEP problem considering nonconventional candidates was implemented in AMPL [32], and
solved using CPLEX 12.2 [33] (with default option settings). The
Garver system, IEEE 24 busbar system, and a Colombian reduced
case were used as test cases.
Garver test system
The Garver System consists of a set of 6 busbars and 15
ranches [4]. In this system it is desired to nd the minimal
investment for expansion. This system allows adding up to 5
lines per branch. The best known solution for this system,
taking into account conventional candidates, is US$110, which

is achieved by adding a line between busbars 3 and 5, and three

lines between busbars 4 and 6. In order to test the proposed
model, using the Garver system, a reconguration candidate
for the existing line 24 and the repowering of line 35 are
included. Fig. 4 shows the proposed reconguration parameters,
including two new dummy nodes 7 and 8, representing the
reconguration points.
The solution of the rst test is presented in Table 2. For this test,
the total investment cost is US$80. Table 2 shows the results of the
optimal solution including the reconguration within the system
expansion plan. In addition to the reconguration, it can be
observed that the solution includes the repowering of line 35
and new lines between 46 and 23. The total investment is
US$30 less than the best known solution for the Garver system
using the original conventional candidates.

Fig. 4. Parameters of the reconguration and repowering candidates.

Table 2
Garver system with reconguration and repowering test 1.
Busbar i

Busbar j

Existing branch

New circuits

Repowered circuits

Power ow (MW)

Max power ow (MW)

1
1
1
2
3
4
2
8
6
6

2
4
5
3
5
6
7
4
7
8

1
1
1
1
1
0
1
1
0
0

0
0
0
1
0
1
0
0
1
1

0
0
0
0
1
0
0
0
0
0

13.7
2.61
53.7
133.7
186.3
78.0
92.6
79.4
92.6
79.4

100
80
100
200
200
100
100
100
100
100

Table 3
Garver system with reconguration and repowering test 2.
Busbar i

Busbar j

Existing branch

New circuits

Repowered circuits

Power ow (MW)

Max power ow (MW)

1
1
1
2
3
4
2
7
8

2
4
5
3
5
6
7
8
4

1
1
1
1
1
0
1
0
1

0
0
0
0
1
3
0
1
0

0
0
0
0
0
0
0
0
0

40.0
40.0
66.7
100.0
173.3
300.0
100.0
100.0
100.0

100
80
100
200
200
100
100
100
100

218

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

To complement the previous test and validate the model, a second test is performed assuming that the cost of the reconguration
and repowering are 10 times the cost presented in Fig. 4. The
results of the second test are presented in Table 3. The total cost
of the investment is US$110 in the second test. When the reconguration cost is much higher than the cost of traditional candidates,
the optimal solution does not take into account the reconguration. For this test, the best known solution for the system was
found using only traditional candidates.
Table 4
Parameters of the IEEE 24 bus system test.
Parameter

Value


fr
ij
xrij

1.25 f ij

crepij
crij

0.5 cij
1 cij

0.95 xpu
ij

Table 5
Results of the IEEE 24 bus system.
Result

Only conventional
candidates

Including
non-conventional
candidates

Total investment (MUS$)

164

150

Added elements

610
2  78
1012
1113

610
2  78
1113

Repowered elements

15
1012

Fig. 5. Results for the IEEE 24 bus system using only conventional candidates.

24-Busbar test system

The case study described in this section is based on the IEEE 24
busbar Reliability Test System (RTS) [34]. The transmission network contains 24 buses and 34 existing corridors. We consider that
all lines in the same corfridor are identical and the maximum number of lines per corridor is three. Lines investments cost for existing
corridors are related to the line length [35]. The IEEE 24-bus system is to be expanded to a future condition with generation and
load levels three times their original values (i.e., up to a total of
10215 MW maximum generation capacity and 8550 MW peak
demand). The solution for this system, taking into account conventional candidates, is US$164M, which is achieved by adding a line
between the busbars 6 and 10, two lines between the busbars 7
and 8, one line between busbars 10 and 12, and one line between
the busbars 11 and 13.
In order to include non-conventional candidates in the test, the
repowering of the 34 existing corridors is considered. The network
repowering parameters for the test are shown in Table 4. Table 5
shows the test results with the conventional candidates, and
results including non-conventional candidates (repowering of
existing network). Including non-conventional candidates the total
investment is reduce in US$14M compared to the solution using
only conventional candidates.
Figs. 5 and 6 shows the results for the IEEE 24 bus system. Comparing these results it is possible to conclude that for large power
systems, the repowering of existing network might not the only
solution to the TEP problem; instead, the optimal solution might
include a combination of repowering and construction of new
lines. This is due to the fact that repowering alone might lead to
congestion on some existing lines.

Fig. 6. Results for the IEEE 24 bus system including repowering of existing network.

219

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

Colombian reduced case for a generators connection to the electrical

network
The Termocol generator must be connected in the area of Guajira-Cesar-Magdalena (GCM), located near the city of Santa Marta
on the Caribbean coast, with a capacity of 202 MW. In order to con-

SABAN

FUNDA

nect this generator to the Colombian system, different connection

alternatives were evaluated including reconguration and repowering of existing network [23], as shown in Fig. 7.
Busbars 9 through 12 are the dummy nodes that are included as
candidates for the reconguration of the circuits Santa Marta
Termoguajira 230 kV 1 and 2. The busbars with load are Cuestec-

CUEST

TGUAJ

SMART

G
G

Generator
COPEY

Existing
Network
Candidates

TERMO
VALLE

Dummy
node
Repowering
candidates
Fig. 7. Alternatives for connecting Termocol generator.

Table 6
Results of the generators connection.
Busbar i

Busbar j

Existing branch

New circuits

Repowered circuits

Power ow (MW)

Max power ow (MW)

CUEST
CUEST
VALLE
COPEY
FUNDA
SABAN
SMART
10
SMART
11
12
TERMO
TERMO

TGUAJ
VALLE
COPEY
FUNDA
SMART
FUNDA
9
TGUAJ
11
12
TGUAJ
9
10

2
1
2
1
2
3
1
1
1
0
1
0
0

0
0
0
0
0
0
0
0
0
1
0
1
1

0
0
0
0
0
0
0
1
0
0
0
0
0

109.5
184.5
104.5
115.5
287.5
152.0
228.7
26.7
163.8
163.8
163.8
228.7
26.7

452
214
422
214
470
912
235
235
235
235
235
344
344

SABAN

FUNDA

TGUAJ

SMART

G
G

Generator
Existing
Network
Solution

COPEY

Dummy
node
Repowered
network
Fig. 8. Generator connection.

TERMO
VALLE

CUEST

220

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

Table 7
Summary of results.
Test case

Solution with conventional candidates

Solution with non-conventional candidates

Garver test system

Objective function: 110 USD

Binary variables: 75
Execution time: 0.03 s

Objective function: 80 USD

Binary variables: 90
Execution time: 0.2 s

Colombian reduced test case

Objective function: 1.05 MUSD

Binary variables: 3
Execution time: 0.01 s

Objective function: 0.93 MUSD

Binary variables: 18
Execution time: 0.13 s

IEEE 24 bus system

Objective function: 164 USD

Binary variables: 170
Execution time: 0.48 s

Objective function: 83.6 USD

Binary variables: 374
Execution time: 6.55 s

itas (CUEST), Valledupar (VALLE), Copey (COPEY), Fundacin

(FUNDA) and Santa Marta (SMART). Sabanalarga (SABAN) and
Copey (COPEY) busbars allow the interconnection between other
areas. Cuestecitas (CUEST) busbar is a node that permits the connection of the international interchange with Venezuela. Additionally, this busbar can import and export power from the
neighboring country. A high demand scenario with all generators
in service and importing energy from Venezuela was used in order
to test the proposed model. Results are shown in Table 6.
The optimal solution is the reconguration of one line Termoguajira Santa Marta 230 kV and repowering the line section
between Santa Marta (SMART) and the reconguration point. The
cost of this solution is US$0.93 M. Fig. 8 shows the solution that
was found.
If non-conventional candidates are included, the optimal solution would be to build a circuit between the substations TERMO
and SMART at a total cost of US$1.05 M. Therefore, the solution
including the reconguration and the repowering reduces the total
investment by almost US$0.12M, fullling the same purpose by
allowing the connection of the generator to the system.
The nal recommendation made by the planning bureau in
Colombia was to recongure the circuit Santa Marta Termoguajira 230 kV 1 [23]. The Termocol new substation and the reconguration proposed will be in operation in November 2013.
Summary of results
Table 7 summarizes the results obtained in this paper. All tests
were performed using an Intel Core i5-2410M CPU@2.30 GHz
and 4 GB of RAM on a 64-bit operating system. The models proposed in this paper were written in AMPL and solved with CPLEX
12.2.0.0 using the default settings of the optimizer.
With non-conventional candidates it is possible to obtain better
solutions than those obtained by conventional candidates only. The
constraints, additional parameters and variables that must be
added to account for non-conventional candidates, increase the
number of binary variables and execution time (see Table 7).
Conclusions
This paper presented a novel modeling of the TEP problem. The
main contribution of this work consists on the inclusion of nonconventional candidate solutions of the TEP problem, namely
repowering and network reconguration. Such non-conventional
solutions were included in the linear disjunctive model of the
TEP by means of linear expressions, so that it is possible to obtain
optimal solutions by using commercially available software.
The inclusion of repowering and network reconguration as
solution candidates allows to explore a greater search space of
the TEP problem, and to nd better solutions than those commonly
obtained using conventional candidates only.

Reconguring or repowering existing networks has shown to be

a feasible solution in power systems worldwide. Because of this,
having an optimization model for TEP that includes conventional
and non-conventional candidates allows to nd solutions that
can be implemented in real life power systems.
In a future work, the model proposed in this paper will be combined with security constraints and/or multistage TEP.
Acknowledgments
The authors would like to acknowledge the Sostenibilidad Programme 20142015 of the University of Antioquia for nancial
support.
References
[1] Romero R, Monticelli A, Garca A, Haffner S. Test systems and mathematical
models for transmission network expansion planning. IEE Proc Gener Transm
Distrib 2002;149(1).
[2] Latorre G, Cruz R, Areiza J, Villegas A. Classication of publications and models
on
transmission
expansion
planning.
IEEE
Trans
Power
Syst
2003;18(1):93846.
[3] Molina JD, Rudnick H. Transmission of electric energy: a bibliographic review.
IEEE Latin Am Trans 2010;8(3):24558.
[4] Garver LL. Transmission network estimation using linear programming. IEEE
Trans Power Appar Syst 1970;PAS-89(7):168897.
[5] Villasana R, Garver LL, Salon SJ. Transmission network planning using linear
programming. IEEE Trans Power Appar Syst 1985;PAS-104(2):34956.
[6] Bahiense L, Oliveira GC, Pereira M, Granville S. A mixed integer disjunctive
model for transmission network expansion. IEEE Trans Power Syst
2001;16(3):5605.
[7] Escobar AH, Gallego RA, Romero R. Multistage and coordinated planning of the
expansion of transmission systems. IEEE Trans Power Syst 2001;16(3):5605.
[8] Vinasco G, Rider MJ, Romero R. A strategy to solve the multistage transmission
expansion planning problem. IEEE Trans Power Syst 2011;26(4):25746.
[9] Zhang H, Vittal V, Heydt GT, Quintero J. A mixed-integer linear programming
approach for multi-stage security-constrained transmission expansion
planning. IEEE Trans Power Syst 2012;27(2):112533.
[10] Silvia IJ, Rider MJ, Romero R. Transmission network expansion planning
considering uncertainness in demand. IEEE Trans Power Syst
2006;21(4):156573.
[11] lvarez Lpez J, Ponnambalam K, Quintana VH. Generation and transmission
expansion under risk using stochastic programming. IEEE Trans Power Syst
2007;22(3):136978.
[12] Yu H, Chung CY, Wong KP, Zhang JH. A chance constrained transmission
network expansion planning method with consideration of load and wind
farm uncertainties. IEEE Trans Power Syst 2009;24(3):156878.
[13] Silvia IJ, Rider MJ, Romero R. Transmission network expansion planning with
security constraints. IEE Proc Gener Transm Distrib 2005;152(6).
[14] Choi J, Mount TD, Thomas RJ. Transmission expansion planning using
contingency criteria. IEEE Trans Power Syst 2007;22(4):224961.
[15] Arroyo JM, Alguacil N, Carrin M. A risk-based approach for transmission
network expansion planning under deliberate outages. IEEE Trans Power Syst
2010;25(3):175966.
[16] Rider M, Gallego L, Romero R, Garcia A. Heuristic algorithm to solve the short
term transmission network expansion planning. In: IEEE power engineering
society general meeting, Tampa, FL; 2007.
[17] Gallego R, Monticelli A, Romero R. Comparative studies on nonconvex
optimization methods for transmission network expansion planning. IEEE
Trans Power Syst 1998;13(3):8228.
[18] Gallego RA, Monticelli A, Romero R. Transmision system expansion planning
by an extended genetic algorithm. IEE Proc Gener Transm Distrib
1998;145(3):32935.

D. Tejada et al. / Electrical Power and Energy Systems 69 (2015) 213221

[19] Areiza JM, Luiz da Silva E, Couto de Oliveira G, Binato S. Transmission network
expansion planning under a tabu search approach. IEEE Trans Power Syst
2001;16(1):628.
[20] Moeini-Aghtaie M, Abbaspour A, Fotuhi-Firuzabad M. Incorporating largescale distant wind farms in probabilistic transmission expansion planning
Part I: Theory and algorithm. IEEE Trans Power Syst 2012;27(3):158593.
[21] Rider MJ, Garca AV, Romero R. Power system transmission network expansion
planning using AC model. Inst Eng Technol 2007;1(5):73142.
[22] Khodaei A, Shahidehpour M, Kamalinia S. Transmission switching in expansion
planning. IEEE Trans Power Syst 2010;25(3):172233.
[23] Unidad de Planeacin Minero Energtica UPME, Plan de Expansin de
Referencia Generacin Transmisin 20122025, Bogot: Ministerio de Minas
y Energa, 2012.
[24] UK Power Networks. <http://www.ukpowernetworks.co.uk/internet/en/helpand-advice/documents/132kV-network-reconguration-to-accomodatewider-systems-works-July-2012.pdf>; 2012 [accessed 14.08.13].
[25] Comit de Operacin Econmica del Sistema Interconectado Nacional (COES),
Informe DP012012: Propuesta de la Actualizacin del Plan de Transmisin
20132022, Direccin de Planicacin de Transmisin, Lima, 2012.
[26] Sertich A, Ortiz A, Torres E. Estudio de repotenciacin de las lneas de
transmisin Itaip Margen Derecha en 500kV con conductores
Termoresistentes. In: Dcimo Tercer Encuentro Regional Iberoamericano de
CIGR, Puerto Iguaz; 2009.

221

[27] Empresa de Transmisin Elctrica S.A. (ETESA). Informe de proyectos al 31 de

diciembre de 2011. Gerencia de Proyectos, Panam; 2011.
[28] Pacic Gas and Electric Company. <http://www.caiso.com/Documents/
PGE2012-2013ProposedSolutions-SanJoaquinValley.pdf>; 2012 [accessed
14.08.13].
[29] PJM. PJM 2012 State RTEP Summaries. <http://www.pjm.com/documents/
reports/rtep-documents/2012-rtep.aspx>; 2013 [accessed 14.08.13].
[30] Villumsen J, Brnmo G, Philpott A. Line capacity expansion and transmission
switching in power systems with large-scale wind power. IEEE Trans Power
Syst 2013;28(2):7319.
[31] Hedman KW, Oren SS, ONeill RP. A review of transmission switching and
network topology optimization. In: IEEE power and energy society general
meeting, July 2011.
[32] Fourer R, Gay D, Kernighan B. AMPL: a modeling language for mathematical
programming. 2nd ed. Pacic Grove, CA: Brooks/Cole-Thomson Learning;
2003.
[33] CPLEX optimization subroutine library guide and reference. Incline Village,
NV: CPLEX Division, ILOG, Inc.; 2009.
[34] Reliability Test System Task Force. The IEEE reliability test system 1996. IEEE
Trans Power Syst 1999;14(3):101020.
[35] Fang R, Hill DJ. A new strategy for transmission expansion in competitive
electricity markets. IEEE Trans Power Syst 2003;18(1):37489.