Documente Academic
Documente Profesional
Documente Cultură
Transient Stability
VOL.
Regions
PAS-85,
FEBRUARY, 1966
NO. 2
of Multimachine
Power
System s
AHMED H. EL-ABIAD,
K.
NAGAPPAN,
MEMBER, IEEE
k1VA (base)
(D1) diagonal (nxn) matrix representing damping coPaper 31 TP 65-655, recommended and approved by the Power
of n rotating machines.
efficients
System Engineering Committee of the IEEE Power Group for
presentation at the IEEE Summer Power Meeting, Detroit, Mich.,
D = 27rfd1 pu powerradian per electrical radian
June 27-July 1, 1965. Manuscript submitted November 2, 1964;
made available for printing April 14, 1965.
damping coefficient in pu power second
A. H. El-Abiad is with Purdue University,,Lafayette, Ind.
di
K. Nagap an is with the American Electric Power Service
Pmi= mechanical input power at the it" machine
Corporation, Kew York, N. Y.
169
170
FElBRUARY
1966
TRANSIENT
STABILITY REGIONS
1.
2.
3.
4.
5..
6.
171
172
IEEE
FEBRUARY
tem. The solution thus obtained is then tested for its given by V = b passes through the unstable equilibrium
steady-state stability. At the equilibrium state, if all the state closest to the steady-state stable equilibrium state
self-partial derivatives bfil/bi k 0, for all i, then the sys- of the postfault systenm. Thus, the region Q includes only
tem is steady-state stable. (This is obviously a sufficient the equilibrium state under investigation. At this equilibrium state, dV/dr is zero. This state is the only invariant
condition but may not be a necessary condition.)
Step 4: This unstable equilibrium state is found in the set in Q and, hence, the largest,
same way as in Step 3, but the initial guess of bi, for all i,
Step 6: As stated before, the region Q defines all the
for the minimization is chosen in such a way that the initial conditions of the postfault system for which it is
desired solution is obtained as shown in Appendix I.
asymptotically stable. In power system transient stability
Step 5: The Liapunov function for the multimachine studies, the possible initial conditions for the postfault
power system under postfault conditions is given by (5). system are along the trajectory (solution) of the faulted
(See Appendix II.)
system. Therefore, during the forward integration of the
faulted system (swing curves), the state at every step of
time is tested to determine whether or not it is inside the
) = E -2+
V(1, '52.. 5n
I, C02 *
kc=1
region Q by checking the corresponding value of V againist
_ n-1 n
the value of b which serves as the index for the regioil Q.
(Ek2Gkk - Pmk) ('k5
k) + E E EkE' X
Under some circumstances, the value of V will not reach
k=1 j=k+l
the
value b within a reasonable transient period (usually
Bk,j [cos (ks - ej) - cos (6k - 'j)] +
one
second). For this reason, time limit of the study is
Gkj [sin (3k - bjf) - sin ('k - 'j)] }* (5) specified in the digital computer program.
This function can be simplified by neglecting the effect
IV. NUMERICAL EXAMPLE
of network transfer conductances.1 This simplified form
The 1-line diagram of the system chosen as an example
of the V function is given by (6).
is given in Fig. 2. The line constants, machine details, and
the loads are given in Tables I-III.
E F+
V('51, 62
'5,
Wn) =MkWk
n Wi2i
The following assumptioins are made in the example:
n
n-1
_
1) The input power to all the machines in the system
(Ek 2 Gkk - Pmk) ('k &Oks) + E E EkEk Bkj X
k=l j=k+l
remain constant during the entire transient period. (Time
[cos ('kV - 'jf) - cos ('5 - '5) ]. (6) constants of the goverinors are considered large with respect to the period under study.)
where Yjk = Gjk + jBjk and b'S (i = 1, 2, ... n) is the
2) Each machine is represented by a constant reacsteady-state (stable equilibrium state under investigation) tance (direct axis transient reactance) in series with a convalue of the rotor angle of the ith machine.
stant EMF (voltage behind transient reactance; constant
These two V funletions satisfy the conditions given in flux linkages). Stator resistances of the machines are
the section on system neglect of the effect of transfer neglected.
conductances (see Appendix II). For the function given by
3) The rotor angle of each machine is fixed with
(6), the time derivative of V is negative definite, and for respect to the electrical phase angle of the voltage behind
the function given by (5), dV/dr-< 0 for
> E > 0 the transient reactance. These angles are measured with
(i = 1, 2, ... n) where E is a very small positive number. respect to common axis rotating at synchronous speed.
The latter implies that the solutions of the system are
4) Damping or asynchronous torque is represented as
bounded [5]. The boundedness of all solutions is a kind of directly proportional to the rate of change of rotor angle.
stability, called Lagrange stability. The value of V is
5) Loads other than syinchronous machines are repreevaluated at the unstable equilibrium state found in step sented as constant impedances.
4, which is the index of the region of stability.
6) Saliency of the rotating machines is neglected.
b = V(1U5, 52U, ... 'nU W CW2 ..
u)
(7) -The critical switching time and the stability region are
The region Q is defined by V < b. It can be shown that evaluated for a 3-phase fault using the method described
the surfaces given by (5) and (6) are closed for V < b and
Load flow solution of the prefault system is given in
open
forV> b These surfaces completely
the region
Theuaso
open span
lrio
.
V
>
Table
IV. Inserting the transient reactances of the maU2. Hence, V8 > 0 in 9, except at the equilibrium state 1
chines
at the respective busses of the system, the internal
..Ol
62s *
2 vanishes. The surface
nSIos 2
Ct)ns) wher'ev
voltages of the machines behind their transient reactances
are calculated and given in Table V.
After calculating the equivalent impedances of the
'Transfer conductances are mainly caused by the conductances
of the interconnecting-network, which are generally small in comparison to the transmission susceptances, while the self conductances loads and inserting these impedances in the network, drivmay be high as a result of the effect of conductances of equivalent 'ng point and transfer admittances between the internal
load admittances. Therefore, neglecting the transfer conductances bse o
busses of the machines are calculated for the faulted and the
will not affect the results
Icow,I
1-
1966
173
TABLE I
TRANSMISSION LINE CONSTANTS FOR SYSTEM
Line
pu Impedances
To Buis
2
3
From Bus
1
2
3
4
5
6
2
TABLE II1
LOADS ON THE SYSTEM
X
0.20
0.50
0.80
0.30
0.40
0.15
0.50
R
0.05
0.10
0.20
0.10
0.20
0.10
0.20
4
5
6
1
5
Loads
TABLE II
SYNCHRONOUS MACHINE CONSTANTS
Generator
Number
10
7
8
9
At
Bus
1
2
3
4
MVA
Capacity
100
15
40
30
211
pu
75 350
1 130
2260
1508
Xd
pu
0.004
1.000
0.500
0.400
LOAD FLOW
P in mW
20
40
30
At Bus
2
5
6
pu
Bus
1.0
12.0
2,5
6.0
1
2
3
4
56
E
1.000
1.002
1.084
1.025
0.956
0.953
FOR
Q in Mvar
10
15
10
TABLE IV
PREFAULT CONDITION (EXCLUDING MACHINE
Degrees
0.0
-0.12
4.62
1.41
-2.80
-2.30
REACTANCES)
PG,
mW
33.2
10.0
30.0
20.0
0.0
0.0;
QG,
Mvar
9.1
5.0
20.0
10.0
0.0
0.0
PL,
mW
0. 0
20.0
0.0
0.0
40.0
30.0
QL, t
Mvar
0.0
10.0
0.0
0.0
15.0
10.0
174
FEBRUARY
TABLE V
INTERNAL Bus VOLTAGES FOR PREFAULT SYSTEM
Generator Internal
Bus
Number
10
10
7
7
8
8
9
9
E, pu
Pm (Input),
pu
0.332
0.100
0.300
0.200
Radians
0.0013
0.1030
0.1970
0.0772
1.0004
1.0410
1.1900
1.0710
TABLE VI
MATRIX FOR FAULTED SYSTEM (Y,,)
G Matrix
7
-0.001
0.027
0.000
0.008
10
1.456
-0.001
0.000
-0.036
10
7
8
9
B Matrix
9
-0.036
0.008
0.000
0.222
8
0.000
0.000
0.000
0.000
10
-3.582
0.546
0.000
0.303
10
7
8
9
TABLE VII
MATRIX FOR POSTFAULT SYSTEM
7
0.546
-0.871
0.000
0.062
7
-0.015
0.029
0.017
0.009
10
7
8
9
B Matrix
8
-0.081
0.017
0.104
0.003
9
-0.121
0.009
0.003
0.255
10
7
8
radians
10
7
8
9
0.05630
0.15013
0.21430
0.02497
co, radians
per radians
0.0
0.0
0.0
0.0
Ofi/d5s
0.30014
8.01844
4.41392
6.27845
X
X
X
X
10-10-4
10-4
10-4
Remarks
Stable
Stable
Stable
Stable
Internal
Bus
10
7
8
9
TABLE IX
UNSTABLE EQUILIBRIUM STATE CLOSEST TO THE STABLE
EQUILIBRIUM STATE OF TABLE VIII, CASE
Internal
Bus
10
7
8
9
5,
radians
0.06610
0.20136
3.08290
-0.02425
w,
per
radians
radians
0.0
0.0
0.0
0.0
Ofs/a5i
0.9444 X 10,0.5378 X 10-.1
-0.4419 X 10-3'
0.5237 X 10. 3
Remarks
Stable
Stable
Unstable
Stable
10
-2.310
0.664
0.656
0.751
7
0.664
-0.880
0.121
0.062
8
0.656
0.121
-0.868
0.062
9
0.751
0.062
0.062
-0.984
TABLE X
STABLE EQUILIBRIUM STATE OF THE POSTFAULT SYSTEM,
CASE 2: TRANSFER CONDUCTANCES INCLUDED
TABLE VIII
STABLE EQUILIBRIUM STATE OF THE POSTFAULT SYSTEM,
CASE 1: TRANSFER CONDUCTANCES NEGLECTED
Internal
Bus
9
0.303
0.062
0.000
-1.216
(Y,,)
G Matrix
10
0.864
-0.015
-0.081
-0.121
8
0.000
0.000
-2.000
0.000
5,
radians
0.00030
0.10243
0.24106
0.05197
co, radians
per radians
0.0
0.0
0.0
0.0
afs/a5s
0.294X
0.805 X
0.444 X
0.635 X
10-4
10-3
10-3
10-3
Remarks
Stable
Stable
Stable
Stable
TABLE XI
UNSTABLE EQUILIBRIUM STATE CLOSEST TO THE STABLE
EQUILIBRIUM STATE OF TABLE X, CASE 2
Internal
Bus
10
7
8
9
5,
w, radians
radians per radiains
afil/oa
0.914 X
-0.00136
0.0
0.539 X
0.0
0.17606
0.0
-0.445 X
3.09250
0.0
0.532 X
0.05892
10-5
10-3
10-3
10-'
Remarks
Stable
Stable
Unstable
Stable
1966
2.00
175
S 25
a:~~~~~~~~~~:
MAHN 8~
w~~~~
AHIE1
0 .0
-I.0
2.0
Fig. 3. Swing curves, Case 2: stable fault cleared at 160 radians (0.4244 seconds).
2.0
3.0-
2.0MACHINE 8
En
1.0
<Hz0
o0
100
TIME
200
300
400
500
600
a:
1.0
POSTFAULT SYSTEM
Fig. 4. Swing curves, Case 2: unstable fault cleared at 165 radians (0.4377 seconds).
RADIANS
176
V. CONCLUSION
Liapunov's direct (second) method for the study of
stability of nonlinear systems has been successfully applied
to multimachine power systems. The advantages over
previous methods are
SYSTEMSE
FEBRUA.RY
POWER OUTPUT
M
A/
a.
\B
POWER INPUT
- -
MR 311
i = 1, 2, ... n and i # R
where Pai anid Mi are the accelerating power and angular
momentum constant, respectively, of the ith machiiie.
Then, considering the ith machine with respect to the reference machine and neglecting the effect caused by the
difference of rotor angles of the other machines, for i 1,
2, ... n andi7z R, (8) yields
dT2
i)
[Pmj
(6i- R) =
(MRPmi
sin-'
-
MiPmR
or
(6 i
R) =
Sill-,
(MRPmi
APPENDIX II
CONSTRUCTION OF THE V FUNCTION (LIAPUNOV FUNCTION)
The system given by (1) can be written as a set of first
order differential equations (state vector form) as follows:
d 5i
dcoi~
~ ~ d
1966
fi(K1, a2,
aPn mi = 0
i = 1, 2, 3, 4, ..., n.
63,
(15)
We are interested in two equilibrium states, one corresponding to the steady-state value and the other closest to
the former as determined in Appendix I.
Denote the steady-state stable point by (3,S; i = 1, 2, 3,
... , n and coi8 = O for i = 1, 2, 3 . . ., n) and the unstable
state closest to the stable state by (bi, i = 1, 2, 3, .. ., n
and c.i== 0, i = 1, 2, 3, . ., n).
The second method of Liapunov to determine stability
is illustrated as follows. Let b(t) be the general solution of a
set of differential equations a = f(a); f(be) = 0, where be
is the equilibrium state. Suppose there exists a function
V(a) satisfying the conditions (1, 2, and 3 of Section II)
in the neighborhood of be. Then V(a) = c, where c > 0 is a
constant parameter, defines, for sufficiently small values of
c, a family of closed surfaces [13] in the neighborhood of be
since V(a) is positive definite. The equilibrium state is itself located in the interior of each of the closed surfaces
and corresponds to the value c = 0. Then by condition
[2 ] of Section II, all the trajectories originating from points
in the neighborhood of the equilibrium state cross the
surface V(a) = c from exterior towards the interior for
increasing values of time when dV/dr < 0 and stay in
the same neighborhood around the equilibrium state
when dV/dr = 0.
From the preceding, it is intuitively clear that, near an
equilibrium state of a physical system, if the transient
energy of the system is always decreasing, then the equilibrium state is stable. Thus, Liapunov's theory of stability
is a generalized extension of the energy concept. The main
step of the second method of Liapunov is the construction
of the Liapunov function to satisfy the conditions given
in Section II. The construction of a Liapunov function
has to. be- adjusted to the peculiarities of every specific system. Liapunov theorems do not containi any hint on how
to find a Liapunov function. As such, there is no systematic
method of constructing this function. In fact, the general
question on the existence of such functions remains completely open. On the other hand, suitable functions can be
determined by trials for numerous specific cases and also
for certain general types of equations. In general, a particular Liapunov function for a physical system yields a
strong sufficient condition for stability, which is desired
in practice, but- it may not be a necessary condition.
With this point in mind, the following Liapunov function is found to be the suitable one for the system given by
(14).
as)
k +
aj) ]
k=li=k+l
G,j [sin
(6k1 - bs) -
sin
(ak
i) -
COS
177
STABILITY REGIONS
The first summation term is the kinetic energy of the system. The last two summation terms on the right-hand side
of (16) are equal to
rk
IBkj
(Ik-8j)
n-1 n
Ei j=k+1
E EkEj
k=l
kas_ s)
(xi
cos
xj) + Gk j
Xk)d(xj -x)
V(a3, 0) = 0
/Vdav
V , + aV dcW,
-= 1 ) d-r bco dr /
dV
dr
(17)
(18)
EjEjGijwi cos
(bi -X). (19)
For large values of ci, dV/dr is negative definite. Inspection of (19) shows that V is negative definite for || (6, c)| I| >
r > 0, r being relatively small constant within the region
Q where (a, co)|| is the norm of the vector (6, c) defined
by
n
v(61s, a2', .. .
V(6 C) = k=i
E -2k=
Mk@W2 + E (E, 2Gk - Pmk) X
(ak
as, 0, 0,
...
0) =
also
(sk -
ai) ] } (16)
dV
d
dT-
(20)
178
REFERENCES
[1] N. Minorsky, Introduction to Non-Linear Mechanics. Ann Arbor,
Discussion
A. J. Pennington (The Detroit Edison Company, Detroit, Mich.):
This is a valuable paper because it contributes to the development
of "direct" methods for power system stability analysis. As our
systems grow larger, pointwise numerical integration of the equations
of motion becomes increasingly costly, and the results become less
useful because of the sheer bulk of information which must be
evaluated. Hence, the interest in "direct" methods, the mathematical
foundation of which is the 1892 contribution of Liapunov [1].
Kalman and Bertram [2] and other Western mathematicians and
control engineers began to discuss and extend Liapunov's ideas
about five years ago, and an impressive body of theory has grown
up since then. Unfortunately, there has not been a corresponding
development of successful applications to actual engineering systems.
To some extent, this reflects the current preoccupation with theory,
but it is mainly a consequence of the great difficulty of large non-
FEBRUARY
[1] A. A. Liapunov, "Probl6m6 general de la stabilite du mouvement," in Annals of Mathematics Study No. 17, Princeton,
N. J.: Princeton University Press, 1949 (1907 French translation
of the original 1892 paper in Russian by Liapunov).
1966
[5]
[6]
[7]
[8]
179
exciter.
These limitations of source representation are not especially
important for many cases of stability testing. For long-distance
transmission and EHV transmission, i.e., for the so-called highimpedance systems frequently encountered with hydro plants or
mine-mouth plants, the simplified source representation is often
quite inadequate for both transient and steady-state stability tests.
The question then arises: If the electrical output power fi is to
include the flux-linkage equations (including saturation) for circuits
in the 2-machine axes and the transfer functions for the exciter-
IR?TGi
i
1,2, .. .
TGi
n
E RkTGkPvTP.
When the system equations include the flux linkage equations,
saliency, saturation and exciter-regulator-stabilizer feed back loop,
the differential equations become time variable in their parameters,
which would present difficulty in constructing Liapunov function.
Investigations by the authors are in progress.