(1)
subject to
( )
~
, , , , , , L P D P Q V
p p t
X
k
= 0
(2)
( )
~
, , , , , , , G P D Q V
p p
P X
t k
0
(3)
where I
G
and I
D
are the sets of pool generator and load buses, respec-
tively; P
and D
pj
are active powers of pool generator-i with bid priceC
i
and pool load-j with bid price B
j
, respectively; P
p
, D
p
, P
t
, Q, V, and
are vectors of pool power injections, pool power extractions, bilateral
contracts, reactive powers, voltage magnitudes, and voltage angles, re-
spectively. Equality constraints (2) is a set of contracted transaction re-
lationships and power balance equations. Expression (3) is a set of
inequality constraints indicating the magnitude (mathematically upper
limits) of pool demands in addition to the usual system operating con-
58 0272-1724/01/$10.002001 IEEE IEEE Power Engineering Review, September 2001
Figure 4. Injection model of TCPAR
Figure 3. Equivalent circuit of TCPAR
Table 1. Generator data
Generator Bid Price ($/h) Specified Voltage
1 0.005 P
2
+ 3.71 P 1.08
2 0.005 P
2
+ 3.52 P 1.08
3 0.005 P
2
+ 3.89 P 1.08
4 0.005 P
2
+ 2.45 P 1.08
Figure 2. Injection model of TCSC
Figure 1. Model of TCSC
straints such as bus voltage levels and line overloads. X
k
is the control
parameter of FACTS device placed in line-k.
In the above formulation, it is assumed that the generation pattern
and FACTS devices alleviate the congestion of the systemcompletely.
Objective Function: Due to high cost of FACTS devices, it is neces-
sary to use cost-benefit analysis to analyze whether a new FACTS de-
vice is cost effective amongst several candidate locations when actually
installed. The TCSC cost in line-k is given by
C k c x k P
c l tcsc
Base_ power ( ) . ( ). . =
2
(4)
where c is the unit investment cost of FACTS ($/MVAR), x k
c
( ) is the
series capacitive reactance, and P
l
is the power flow in line-k. It is as-
sumed that the cost of TCPAR is same for all the lines. Therefore, for
TCPARthe objective function will be the same as (1) subject to the con-
straints (2) and (3).
The objective function for placement of TCSC will be
( ) ( )
min
P
i pi
i I
j pj
j I
pi
G D
C P B D C
+
tcsc
t
(5)
where
C
C k
t
tcsc
tcsc
=
. ( )
8760
and
=
+
+
r r
r
n
n
( )
( )
1
1 1
(6)
= the capital recovery factor (CRF),
r = the interest rate,
n = the capital recovery plan.
It is assumed that the investment cost of the TCSC is $50/KVAr.
Considering the interest rate r =0.05, the capital recovery period n =10
years, the capital recovery factor can be computed, i.e., =0.1295.
Thus the cost of TCSC is approximated as $17.74/MVAr/day.
Static Modeling of FACTS Devices: The power injection approach
has been suggested for static modeling of FACTS devices.
Thyristor Controlled Series Compensator: The model of a transmis-
sion line with a TCSC connected between bus-i and bus-j is shown in
Figure 1. In steady state, the TCSC can be considered as a static
reactance jx
c
. The change in the line flow due to series capacitance
can be represented as a line without series capacitance with additional
power (complex) injections at the receiving( ) S
jc
and sending( ) S
ic
ends
as shown in Figure 2. The real power injections due to series capacitor
at bus-i ( ) P
ic
and bus-j ( ) P
jc
can be written as [4]
[ ]
P V G VV G B
ic i ij i j ij ij ij ij
= +
2
cos sin
(7)
[ ]
P V G VV G B
jc j ij i j ij ij ij ij
=
2
cos sin
(8)
where
G
x r x x
r x r x x
ij
c ij c ij
ij ij ij ij c
=
+ +
( )
( )( ( ) )
2
2 2 2 2
and
( )
( ) ( )
B
x r x x x
r x r x x
ij
c ij ij c ij
ij ij ij ij c
=
+
+ +
2 2
2 2 2
2
.
Thyristor Controlled Phase Angle Regulator: The static model of a
TCPAR and transmission line between bus-i and bus-j is shown in
Figure 3. From basic circuit theory, the injection equivalent circuit of
Figure 4 can be obtained. The injected active power at bus-i ( ) P
is
and
bus-j ( ) P
js
of a line having a phase shifter are
[ ]
P V K G VV K G B
is i ij i j ij ij ij ij
=
2 2
sin cos
(9)
[ ]
P VV K G B
js i j ij ij ij ij
= + sin cos
(10)
IEEE Power Engineering Review, September 2001 59
Table 2. Sensitivity factors
Line -k Line
Rating
MW
Power
Flow
(MW)
TCSC
(a
k
c
)
TCPAR
(a
k
s
)
Line -k
Line Rating
MW
Power Flow
(MW)
TCSC
(a
k
c
)
TCPAR
(a
k
s
)
No. i - j No. i-j
1 8-3 150.0 23.5 -0.585 -2.683 11 2-9 150.0 95.0 1.419 1.485
2 9-6 150.0 49.9 2.721 5.535 12 6-7 50.0 49.9 2.722 5.645
3 9-7 150.0 27.9 -0.950 -3.372 13 7-10 50.0 14.4 0.442 2.658
4 4-8 150.0 100.3 -3.978 -4.066 14 3-11 150.0 59.8 -1.280 -2.139
5 2-8 150.0 96.8 1.435 1.502 15 3-12 50.0 19.7 -0.302 -1.511
6 1-9 150.0 90.3 0.441 0.482 16 3-13 50.0 45.2 0.877 1.865
7 8-9 150.0 0.6 0.015 -0.265 17 7-14 50.0 5.6 0.015 -0.077
8 4-2 90.0 96.8 2.502 3.460 18 11-10 50.0 4.7 -1.129 -2.407
9 2-1 90.0 6.9 0.053 0.943 19 12-13 50.0 3.3 -0.037 -1.520
10 6-5 50.0 0.0 0.000 0.000 20 13-14 50.0 20.2 0.066 0.277
Table 3. Optimal generation schedule
Generators Base Case TCSC in Line-4
TCPAR in
Line-4 Line-8
G1 89.9 + j 27.8 87.0 + j 28.8 86.1 + j 23.6 86.0 + j 27.3
G2 107.5 + j 88.6 110.2 + j 88.5 108.1 + j 72.7 107.8 + j 126.1
G3 110.7 + j 80.2 104.8 + j 80.9 102.9 + j 74.7 103.6 + j 79.6
G4 190.2 - j 1.4 196.6 - j 1.1 201.2 + j 2.6 201.1 -j 24.3
Total Cost ($/h) 1948.68 1948.18 1943.1 1945.1
where K tan .
Optimal Locations of Facts Devices: The severity of the system
loading under normal and contingency cases can be described by a real
power line flow performance index [7], as given below.
PI
w
n
P
P
m
m
N
lm
lm
n
l
j
(
,
\
,
(
2
1
2
max
(11)
where P
lm
is the real power flow, P
lm
max
is the rated capacity of line-m, n is
the exponent, and w
m
a real non-negative weighting coefficient that
may be used to reflect the importance of lines. PI will be small when all
the lines are within their limits and reach a high value when there are
overloads. Thus, it provides a good measure of severity of the line over-
loads for a given state of the power system. In this study, the value of the
exponent has been taken as 2 and w
m
= 1.0.
The real power flowPI sensitivity factors with respect to the param-
eters of TCSCand TCPARplaced in line-k, one at a time, are defined as
a
PI
x
k
c
ck
x
ck
0
(12)
a
PI
k
s
k
k
0
.
(13)
Using (11), the sensitivity of PI with respect to FACTS device pa-
rameter X
k
(x
ck
for TCSC and
k
for TCPAR) connected between bus-i
and bus-j for the case n 2, can be written as
j
(
,
\
,
(
PI
X
w P
P
P
X
k
m lm
m
N
lm
lm
k
l
3
1
4
1
max
.
(14)
The real power flow in a line-m (P
lm
) can be represented in terms of
real power injections using dc power flow equations [7] where s is the
slack bus, as
P
S P m k
S P P m k
lm
mn n
n
n s
N
mn n
n
n s
N
j
1
1
for
for
(15)
where S
mn
is the mnth element of matrix [S] which relates line flow
with power injections at the buses without FACTS devices and N is the
number of buses in the system.
Using (15), the following relationship can be derived,
j
(
,
\
,
(
+
P
X
S
P
X
S
P
X
m k
S
P
X
S
lm
k
mi
i
k
mj
j
k
mi
i
k
for
mj
j
k
j
k
P
X
P
X
m k
j
(
,
\
,
( +
for .
(16)
The terms
P
x
i
ck
x
ck
0
,
P
x
j
ck
x
ck
0
,
P
i
k
k
0
, and
P
j
k
k
0
can be obtained using (7)-(10), and the sensitivity factors a
k
c
and a
k
s
can
now be found.
Criteria for Optimal Location: The FACTS device should be placed
on the most sensitive lines. With the sensitivity indices computed for
each type of FACTS device, TCPAR should be placed in a line (k) hav-
ing largest absolute value of the sensitivity factor. TCSC should be
placed in a line (k) having largest negative value of the sensitivity fac-
tor, however. The final placement is decided after solving the
optimization problem.
Simulation Results: The approach has been examined on a modi-
fied IEEE 14-bus system [5] consisting of four real power generator
buses and one voltage-regulating bus. Generators at buses 1, 2, 3, and 4
bid into the pool. Generator at bus-1 is taken as the reference bus. Volt-
age magnitudes at load buses are kept within the range of 0.95-1.10.
The prices bid by generators are given in Table 1 where P is in MWand
$ is a monetary unit that may be scaled by any arbitrary constant with-
out affecting the results.
The generation schedule obtained from optimal dispatch without
considering the line flow limit and FACTS devices was 86.0, 108.0,
103.7, and 200.4 MWfor generators 1, 2, 3, and 4, respectively. With this
generation schedule, it was found that the real power flows in lines were
within the rating limit except line-2, which was 96.8 MW. The rating of
each line is given in Table 2. Sensitivities were calculated for FACTS de-
vices (TCSCand TCPAR) placed in every line one at a time for this oper-
ating condition. The sensitivities of the real power performance index
with respect to TCSC and TCPAR are presented in Table 2. The highest
negative sensitivities in the case of TCSC and the highest absolute value
of sensitivities in the case of TCPAR are presented in bold type.
The sensitivity factor of TCSC for line-4 is the most negative com-
pared to the other lines and, hence, the most suitable for the TCSC
placement. Branches 2 and 12 are the most sensitive for TCPAR place-
ment but have not been considered due to connecting the tertiary wind-
ing of the transformer. The next choices for placement of TCPAR
would be line-4 and line-8. Table 2 also shows that the placement of
TCPAR in line-8 is the other choice as the magnitude of sensitivity fac-
tors is the next highest. This sensitivity is positive, which indicates that
the phase angle shift of the TCPAR should be negative.
The optimal dispatch, described in the second section of this letter
[(1)-(3)], with TCSCin line-4, TCPARin line-4, and in line-8, taken one
at a time, was obtained and presented in Table 3. Table 3 also shows the
optimal dispatch without FACTS devices (base case). The optimal value
of capacitive reactance of TCSCplaced in line-4 was 0.0132 pu. This in-
dicates that 5.93 % of compensation is optimal. However, the optimal
values of phase angle of TCPARplacement in line-4 and line-8 were 5.77
and -3.95 , respectively. The placement of TCPAR in line-4 gives less
pool generation price compared to other cases. The variation of optimal
pool real power generation with each FACTS device is very small.
Conclusions: In this letter, a sensitivity-based approach has been
developed for determining the optimal placement of FACTS devices in
an electricity market having pool and contractual dispatches. In a sys-
tem, first, the few locations of FACTS devices can be decided based on
the sensitivity factors a
k
c
and a
k
s
and then the optimal dispatch problem
is solved to select the optimal location and parameter settings. Test re-
sults obtained on a test system show that new sensitivity factors along
with FACTS device cost could be effectively used for optimal location
of FACTS devices.
References:
[1] N.G. Hingorani, Flexible AC transmission, IEEE Spectrum,
pp. 40-45, Apr. 1993.
[2] R. Rajaraman, F. Alvarado, A. Maniaci, R. Camfield, and S.
Jalali, Determination of location and amount of series compensation
to increase power transfer capability, IEEETrans. Power Syst., vol. 13,
pp. 294-299, May 1998.
[3] E.J. de Oliveri, J.W.M. Lima, and J.L.R. Pereira, Flexible AC
Transmission system devices: Allocation and transmission pricing,
Electric Power and Energy Syst., vol. 21, pp. 111-118, 1999.
[4] S.N. Singh and A.K. David, Congestion management by
otimising FACTS device allocation, in Proc. Int. Conf. Electric Utility
Deregulation and Restructuring and Power Technologies 2000, City
University, London, 4-7 April 2000, pp. 23-28.
[5] R.S. Fang and A.K. David, Optimal dispatch under transmis-
sion contracts, IEEE Trans. Power Syst., vol. 14, pp. 732-737, May
1999.
Copyright Statement: ISSN 0282-1724/01/$10.00 2001 IEEE.
Manuscript received 18 January 2001. This paper is published herein in
its entirety.
60 IEEE Power Engineering Review, September 2001