Professional Documents
Culture Documents
Abstract
This paper discusses the effects of thyristor controlled series compensator (TCSC), a series FACTS controller, on the transient stability of a
power system. Trajectory sensitivity analysis (TSA) has been used to measure the transient stability condition of the system. The TCSC is modeled
by a variable capacitor, the value of which changes with the firing angle. It is shown that TSA can be used in the design of the controller. The
optimal locations of the TCSC-controller for different fault conditions can also be identified with the help of TSA. The paper depicts the advantage
of the use of TCSC with a suitable controller over fixed capacitor operation.
2006 Elsevier B.V. All rights reserved.
Keywords: TCSC; Trajectory sensitivity analysis; Dynamic simulation; Transient stability margin; Optimal location
1. Introduction
The demand of electrical power is ever increasing. However,
the process of development of new infrastructure for power generation and dispatch is restricted due to mainly economic and
partially environmental constraints. These result in the need for
better utilization of the existing system. Flexible AC transmission system (FACTS) controllers are potent tools to achieve this
goal. These devices help in pushing the system to their limits
and thus to attain higher operational efficiency. The use of these
controllers increases the flexibility of operation by providing
more options to the power system operators.
In the steady state, FACTS controllers like TCSC help in
controlling and increasing the power flow through a line. However, the other important aspect of these controllers is their use
during large disturbances like faults because of their capability
to improve the transient stability condition of a power system.
The effectiveness of TCSC-controllers in enhancing the transient stability limit has been studied in Ref. [1]. Evaluation of
transient stability condition of the system is essential for understanding the effects of application of FACTS devices. Structure
preserving energy margin sensitivity has been used in Ref. [2]
Corresponding author. Tel.: +91 512 2597179; fax: +91 512 2590063.
E-mail address: aghosh@iitk.ac.in (A. Ghosh).
0378-7796/$ see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.epsr.2006.04.011
471
tory sensitivity with respect to fault clearing time has been used
to assess transient stability margin of power systems. At first, a
controller has been designed for effective use of TCSC under various fault conditions. TS has been used for finding suitable values
of the controller parameters. The effects of placement of TCSCcontrollers at various locations of a power system on the transient
stability are studied. A comparison of the effects of TCSCcontroller with those of fixed capacitors is also carried out.
The TCSC is represented by a fundamental frequency lumped
reactance model that varies with the change in the firing angle.
The system under consideration is the three-machine, ninebus WSCC system [9]. System loads are taken as constant
impedance.
2. Modeling of the TCSC and the power system
i = 1, . . . , m
(3)
KDi
2Hi dri
(ri s ),
= Pmi Pei
s dt
s
dEq i
xd
= i Eq i +
dt
xdi
Tdo
i
XC = 1 (XFC + 2 ) 4 5 XFC
i = 1, . . . , m
(1)
i = 1, . . . , m
(4)
xdi
1 Vi cos(i i ) + Efdi ,
xd i
(5)
where
2( ) + sin 2( )
XFC XP
,
, 2 =
XFC XP
XFC
, 4 = 3 tan [3 ( )] tan( ),
3 =
XP
1 =
5 =
422 cos2 ( )
XP
TAi
dEfdi
= Efdi + (Vrefi Vi )KAi ,
dt
sin(i i )
+ 0.5
Pei = Eq i Vi
xd i
i = 1, . . . , m
i = 1, . . . , m
1
1
xqi
xdi
(6)
Vi2 sin 2(i i ),
(7)
Fig. 2. (a) Flux-decay model of generator and (b) static exciter model.
472
ri s
s
and
2Hi
i = 1, . . . , m
d ri
= Pmi Pei KDi ri ,
dt
i = 1, . . . , m
di1
dt
vc1 vs vc2 = Ri + L
eTabc = [ ea
eb
ec ],
iTabc = [ ia
ib
ic ]
vTabc = [ va
vc ],
vb
and
d
i1abc
dt
(16)
(8)
(9)
(10)
di
dt
(15)
d ri
1 dri
=
dt
s dt
d
1
(i i2 )
vc =
dt 2
Cch
(12)
d
1
vc1 =
(i1 i)
dt
Cch
(13)
d
1
vs =
i
dt
Ctcsc
(14)
d
i2abc
dt
(18)
(19)
d
1
iabc
vsabc =
dt
Ctcsc
(20)
1
d
(iabc i2abc )
vc abc =
dt 2
Cch
(21)
=
+ L1
dt i1q
e1q
v c1 q
L1
0
i1q
(22)
vc1 d
v c1 q
=
v c2 d
e2d
e2q
L
R
vsd
vsq
R
L
v c2 q
v c2 d
v c2 q
id
iq
=
d
+L
dt
id
(23)
iq
L2
L2
i2d
i2q
d
+ L2
dt
i2d
i2q
(24)
d
dt
(17)
1
d
vc1 abc =
(i1abc iabc )
dt
Cch
(11)
di2
vc2 e2 = L2
dt
d
iabc
dt
v c1 d
v c1 q
0
i1d id
1 i1q iq
Cch
0
vc1 d
+
v c1 q
0
1
Cch
=
(25)
d
dt
d
dt
vsd
vsq
v c2 d
v c2 q
Ctcsc
=
1
Cch
=
0
+
0
i
0
d +
1 i
vsd
vsq
0
i
d
1 i
Cch
Ctcsc
v c2 d
v c2 q
i2d
i2q
(27)
473
x(t0 ) = x0
(28)
where x is the state vector and is a vector of system parameters. The sensitivities of state trajectories with respect to system
parameters can be found by perturbing from its nominal value
0 . The equations for trajectory sensitivity can be found as [3]:
f
f
x =
(29)
x +
,
x (t0 ) = 0
x
where x = x/. Solution of (28) and (29) gives the state trajectory and trajectory sensitivity, respectively [3]:
However, the sensitivities can also be found in a simpler way
by using numerical method. To describe this, let us choose only
one parameter, i.e., becomes a scalar and the sensitivities with
respect to it are studied. Two values of are chosen (say 1 and
2 ). The corresponding state vectors x1 and x2 , respectively are
then computed. Now the sensitivity at 1 is defined as
sens =
x2 x1
x
=
2 1
(30)
(31)
ij
r
i
(32)
+
SN =
i=1
474
5
6
8
TCSC in line
45
57
46
69
78
89
0.6827
1.0743
1.0464
1.1081
1.2423
1.0346
1.0411
0.6850
1.0331
1.1363
1.0727
1.0795
0.9953
0.9493
1.0626
1.0065
0.9603
0.9462
Table 1B
Variation of normalized with TCSC location: fault cleared by line isolation
Fault near
bus
Line
isolated
TCSC in line
45
57
46
69
1.0955 1.2559
1.0203 0.9660 0.9764
1.0892 1.3018 0.7152
0.9777 0.9993
78
89
45
57
46
69
78
89
loop (i.e. without any controller) in one of the lines at a time and
the corresponding values of are computed for different fault
locations. These values are then normalized with respect to 0 .
Therefore, if the value of normalized in a particular operating
condition is greater than 1.0, it indicates that the absolute value
of is more than 0 . Table 1A shows the values of normalized
for different combinations of TCSC and fault locations when
the faults are of self-clearing type. Table 1B shows the same for
faults cleared by line isolation. The faults are considered to be
located near one of the load buses (bus 5, 6 or 8) in all the cases.
The value of tcl is taken as 0.15 s and the firing angle () of the
TCSC as 160 . As described in the previous section, gives the
transient stability margin of the system. Therefore, a normalized
> 1.0 means that the transient stability margin is higher than
that for the system without TCSC. Further, those locations of
TCSC are preferable (from transient stability improvement point
of view) for which the value of are higher. It can be seen from
Tables 1A and 1B that the preferable location of TCSC is line
69 in the cases of fault near bus 5 or bus 8 whereas it is line
57 for fault near bus 6.
The results are verified by PSCAD/EMTDC simulation. A
fault of duration 0.15 s is simulated in line 69. The responses
of relative machine angle delta31 (3 1 ) are shown in Fig. 6
for two cases: (a) the TCSC placed in line 46 and (b) the TCSC
Fig. 6. Response of delta31 for fault in line 69 with TCSC in: (a) line 64 and (b) line 57.
475
placed in line 57. The firing angle of TCSC is 160 in both the
cases. It is clear that the maximum peak of the response is much
lower in case of TCSC in line 57, which indicates a higher
transient stability margin. This is in accordance with the higher
value of normalized for the TCSC in line 57 than in line 46
in the corresponding cases in Table 1B.
4. The control scheme and use of TSA in the choice of
control parameters
4.1. The control scheme
In the next step, a controller is employed along with the
TCSC. The block diagram of the control scheme used is shown
in Fig. 7. The active power flow (P) through the line containing
TCSC is taken as the control variable. It is compared with the
reference value of active power flow (Pref ) and the error is fed
to a PI controller. The output of the PI is the firing angle of the
TCSC, . This is passed through a limiter to keep it within
the capacitive operation zone of the TCSC (between 145 and
180 ). The output of the limiter is supplied to the firing circuit of
TCSC. The capacitance value of the TCSC (Ctcsc ) is computed
as described in Section 2.1. This capacitance is then included in
the line dynamics.
This scheme is sufficient if the fault is only of self-clearing
type, because there is no change in system configuration and
hence the steady state power flow should remain the same before
and after the fault. But when the fault is cleared by isolating
the faulty line, the system configuration changes resulting in a
change of the steady state power flow through the lines. Therefore, a corresponding change in Pref is needed. An additional
control action is taken to achieve this, which is shown in Fig. 8.
The current flowing through the line (containing TCSC) in dq
frame, id and iq , are used for this purpose. First, the current values in the pre-fault steady state, denoted by id0 and iq0 , are stored
to calculate
i0 =
i2d0 + i2q0
(33)
476
(35)
KI > 6.0). Fig. 9(c) is for a fault near bus 8 with TCSC in line 57.
In this case, higher values can be obtained for combinations
(KP = 0.5 and 1.5 < KI < 2.0) or (KP = 0.8 and KI = 3.0). Hence,
the TCSC will be most effective in transient stability improvement of the system when the controller constants are in these
ranges.
Next, we consider the cases in which the faults are cleared
by line isolation. As before, values for different KP and KI are
found. The variation of normalized with KI for different values
of KP are shown in Fig. 10 for a fault near bus 5 (line 45 isolated
to clear the fault) with the TCSC in line 69. It can be seen that
the suitable combinations of constants in this case are (KP = 0.1,
KI > 3.0), (KP = 0.5, KI < 1.0 or KI > 3.0) and (KP = 0.8, KI < 1.0
or KI > 6.0). Table 2 shows the most suitable combination of KP
and KI values found by this study for different fault locations
with TCSC in line 57 or 69.
4.3. Effect of fault clearing time on the relation of with
controller constants
Up to this point, we have observed the effect of TCSCcontroller constants on the value of (and hence on the transient
stability condition of the system) for a particular value of fault
clearing time (tcl ). However, tcl may also be different in different situations. Therefore, the same study, i.e., variation of
with controller constants, is carried out at different tcl values.
Fig. 11(a) and (b) shows the plots of variation of with KI (at
a particular KP ) for different values of tcl . Fig. 11(a) is for a
fault near bus 6 (line 46 isolated) with TCSC in line 57 and
KP = 0.8. The three plots correspond to tcl = 0.15, 0.20 and 0.25 s.
Fig. 11(b) is for a self-clearing fault near bus 8 with TCSC in line
69 and KP = 0.8. Here, the three plots correspond to tcl = 0.10,
0.15 and 0.20 s.
It is quite clear from the figures that the nature of variation
of normalized remains the same for different fault durations
(tcl ). Higher values of are obtained for the same ranges of controller constants even when the fault duration varies. Therefore,
controller constants chosen on the basis of variation of at a
particular tcl will be helpful in effective operation of the TCSC
in improving stability margin even when the tcl is different.
4.4. Comparison with the effects of TCSC in open loop
Fig. 10. Variation of normalized with KI for different KP ; fault in line 45 and
TCSC in 69.
477
Table 2
Suitable controller constants and corresponding normalized for different fault and TCSC locations
TCSC in line
57
69
Self-clearing fault
Fault bus
(KP , KI )
Faulted line
(KP , KI )
0.8, 9.00
1.6357
45
0.8, 9.00
1.4641
0.8, 0.12
1.7357
46
69
0.8, 9.00
0.8, 9.00
1.8279
1.9094
0.5, 1.80
1.0906
78
89
0.8, 1.80
0.5, 1.20
0.9758
1.0014
0.1, 6.00
1.2944
45
57
0.1, 6.00
0.1, 6.00
1.2972
1.4221
0.8, 9.00
1.3001
46
0.8, 9.00
1.2487
0.8, 1.80
1.1583
78
89
0.8, 1.80
0.8, 0.60
1.2560
1.2636
Fig. 11. Variation of normalized with controller constants at different fault clearing time: (a) fault near bus 6, line 46 isolated, TCSC in line 57 and (b) self-clearing
fault near bus 8, TCSC in line 69.
Therefore, it can be concluded that the improvement in the transient stability condition of the system is much more with the
TCSC-controller combination. This can also be verified from
Fig. 9(b) and (c) and Fig. 10, where for fixed capacitor (with-
Table 3
Comparison of for system compensated by fixed capacitor and TCSC-controller
Type of fault
Fault location
TCSC in line
57
69
Fixed capacitor
With controller
Fixed capacitor
With controller
Selfclearing
Bus 5
Bus 6
Bus 8
1.1081
1.2423
1.0346
1.6357
1.7357
1.0906
1.1363
1.0727
1.0795
1.2944
1.3001
1.1583
Fault
with
line
isolation
Line 45
Line 57
Line 46
Line 69
Line 78
Line 89
1.0139
1.2559
1.3018
0.9474
0.9668
1.4641
1.8279
1.9094
0.9758
1.0014
1.1567
1.2399
1.0203
1.1550
1.1534
1.2972
1.4221
1.2487
1.2560
1.2636
478
scf + f
2
Fig. 12. Variation of normalized with controller constants for different types
of faults.
Fig. 13. (a) Variation of normalized and OI with KI and (b) variation of OI with controller constants.
479
Table 4
Optimized controller constants and corresponding for different fault and TCSC locations
TCSC in line
57
69
Self-clearing fault
Fault bus
(KP , KI )
Faulted line
(KP , KI )
0.8, 9.00
1.6357
45
0.8, 9.00
1.4641
0.8, 9.00
1.7225
46
69
0.8, 9.00
0.8, 9.00
1.8279
1.9094
0.5, 1.80
1.0906
78
89
0.5, 1.80
0.5, 1.80
0.9559
0.9908
0.1, 6.00
1.2944
45
57
0.1, 6.00
0.1, 6.00
1.2972
1.4221
0.8, 9.00
1.3001
46
0.8, 9.00
1.2487
0.8, 1.80
1.1583
78
89
0.8, 1.80
0.8, 1.80
1.2560
1.2452
480
Fig. 14. (a) Variation of relative rotor angle delta2-1 for a self-clearing fault near bus 5 and (b) variation of relative rotor angle delta2-1 for a fault in line 46.
controller at line 69 (with optimized constant values), are plotted. The relative rotor angle are found to diverge and become
unbounded in case of system without TCSC whereas it remains
bounded and stable when a TCSC is placed in line 69. The
plots clearly show the effectiveness of the TCSC with suitably
designed controller to improve transient stability condition of the
system.
5.3. Comparison with a self-tuning controller
A PI controller is applied here for effective use of a TCSC
in a multi-machine power system to improve transient stability
margin. Trajectory sensitivity has been used for the tuning of the
PI controller. To assess the performance of this method, a selftuning controller is employed in place of the PI controller in the
same system under the same operating conditions and the results
are compared. A pole-shift controller is used for this purpose.
The detailed equations of the pole-shift self-tuning controller
are given in Appendix A.
Fig. 15. Relative rotor angle delta2-1 for a fault in bus 5, fault-clearing time
0.42 s.
Fig. 16. Relative rotor angle delta2-1 for a self-clearing fault in bus 6, TCSC in
line 57 with: (a) pole-shift controller and (b) PI controller.
481
(36)
where y(k) is the sampled value of the system output, u(k) is the
control input, e(k) is a zero-mean, uncorrelated random noise
input. The polynomials A and B are given in backward shift
operator notation z1 as
A(z1 ) = 1 + a1 z1 + + an zn ,
B(z1 ) = b1 z1 + + bm zm
(37)
(38)
an
b1
bm ] ,
(39)
482
Fig. 18. Identified system parameters when a fault is applied in line 78 with TCSC in line 57.
the recursive least squares (RLS) method, where the parameters are computed recursively at each sampling instant using the
following equations [13]:
T
1) + K(k)(k),
(k)
= (k
K(k) =
P(k 1)(k)
,
+ T (k)P(k 1)(k)
(40)
In the above equation, is a scalar less than 1 used for discounting old data.
Once the parameters are estimated, they are then used in the
pole-shift control in which the control law is given by
u(k) =
S(z1 )
{yref (k) y(k)}
R(z1 )
(44)
where 0 < < 1 is called the pole-shift factor that determines the
penalty on control.
In the simulation studies performed in Section 5.3, the system
order is chosen with n = m = 3. The parameters and are chosen
as 0.8 and 0.985, respectively for the self-clearing fault, the result
of which is shown in Fig. 16(a). These parameters are chosen as
0.8 and 0.999 for fault and subsequent line removal, the result of
which is shown in Fig. 17(a). Some of the identified parameters
for this case are shown in Fig. 18. In this the fault is applied at 4 s
and is removed by line isolation at 4.35 s. Most of the parameters
can be observed to vary in three distinct steps associated with
pre-fault, during-fault and post-fault conditions.
References
) = 1 + a 1 z1 + + n a n zn
T (z1 ) = A(z
(41)
(43)
and B
are the estimates of the polynomials A and B and
where A
are obtained from (40).
In (43), the polynomial T is the closed-loop system characteristic equation. We can define it by shifting the open-loop poles
[1] K.R. Padiyar, K. Uma Rao, Discrete control of TCSC for stability improvement in power systems, in: Proceedings of the Fourth IEEE Conference on
Control Applications, September 2829, 1995.
[2] K.N. Shubhanga, A.M. Kulkarni, Application of structure preserving
energy margin sensitivity to determine the effectiveness of shunt and series
FACTS devices, IEEE Trans. Power Syst. 17 (3) (2002) 730738.
[3] M.J. Laufenberg, M.A. Pai, A new approach to dynamic security assessment using trajectory sensitivities, IEEE Trans. Power Syst. 13 (3) (1998)
953958.
[4] M.A. Pai, T.B. Nguyen, Trajectory sensitivity theory in nonlinear dynamical systems: some power system application, in: N. Michel, D. Liu, P.J.
Antsaklis (Eds.), Stability and Control of Dynamical Systems with Applications: A Tribute to Anthony, Birkhauser, Boston, 2003.
[5] I.A. Hiskens, M.A. Pai, Trajectory sensitivity analysis of hybrid systems,
IEEE Trans. Circuits Syst. Part 1: Fundam. Theory Appl. 47 (2) (2000)
204220.
483
[10] N. Christl, R. Hedin, P.E. Krause, S.M. McKenna, Advanced series compensation (ASC) with thyristor controlled impedance, in: Proceedings of
the CIGRE General Session, Paris, 1992.
[11] P. Kundur, Power System Stability and Control, McGraw-Hill, New York,
1994.
[12] K.R. Padiyar, Power System Dynamics, Stability and Control, BS Publications, Hyderabad, 2002.
[13] A. Ghosh, G. Ledwich, O.P. Malik, G.S. Hope, Power system stabilizers
based on adaptive control techniques, IEEE Trans. Power Appl. Syst. PAS103 (1984) 19831989.