IEEE Transactions on Energy Conversion, Vol. 3, No.

2, June 1988 323

NEW APPROACH TO THE STEADY STATE STABILITY ANALYSI". OF SYNCHRONOUS MACHINES

J. Tamura .
T Mura ta I. Takeda J.Hasegawa H. Fuj iwara
Member IEEE
Kitami Institute of Technology Hokkaido University
Kitami Japan Sapporo Japan

Abstract - This paper presents a new tems[l4], generators for hydroelectric power
method for steady state stability analysis of plants[l5] and wind power convertors[l61, etc.
synchronous machines. The method is based tn a Likewise this machine also has the same two
new swing equation, which is a second-order, types of instability problems as in an ordi-
non-linear differential equation and differs nary synchronous machine. However there are
from the swing equation conventio.ially used. only a few papers[l31 that discuss the steady
Two steady state stability criteria can be state stability problem of a doubly fed syn-
derived by evaluating the eigen values of a chronous machine. This new method can be
linearized version of the new swing equation, applied to this machine also.
one of which is for step-out instability
and the other for hunting. Although these two
instabilities have mostly been discussed THE FUNDAMENTAL THEORY OF THIS NEW APPROACH
independently so far, this new approach
discusses the two instabilities on a common It is assumed that the developed torque
basis. of synchronous machines is expressed a s
follows;
INTRODUCTION
This new method of analysis is base8 on a where
new swing equation, which is a second-arder
differential equation whose variables are the : electrically developed torque,
internal phase angle and the rotational slip.
The steady state stability can be analyzed by
8"s : internal phase angle,
: rotational slip.
using the linearized version of the swing
equation, and two stability criteria are The equation of motion is
obtained as a result, one of which is for
step-out instability and the other for
hunting. This paper applies a new method of
steady state stability analysis to two types where
of synchronous machires, the ordinary synchro-
nous machines and the doubly fed synchronous p : differential operator d/dt,
machines, and discusses its applicability. TD :
prime mover or load torque,
In the analyses of these two types of J : moment of inertia,
instabilities for an ordinary synchronous wm : angular velocity.
machine, the most conventional method for the
step-out type of instability analysis applies and generally there are the following rela-
dP/d6>O criterion [1],[2], which is based on tions among 6, s and wm as
the synchronizing torque and has Tery high
accuracy. On the other \and, hunting is pd=-sw (3)
analyzed by assuming a small mechanical oscil-
lation in the shaft[3]-[9]. Thi. technique 1 s Flr,=-OPS (4)
developed by considering the damping torque
and it also has sufficient accuracy. Until now where w is the synchronous angular velocity.
these two instability phenomena have been Therefore, we can rewrite eq.(2) as follows;
analyzed by using the familiar independent
method. However this new method deals with p6=-sw (5a)
both problems on a common basis.
Like the ordinary synchronous machine, a ~s={TG(~,s)-TD}/wJ (5b)
doubly fed synchronous machine also has been
studied for rather a long time [101-[161. This is the new swing equation which is the
Since this is a kind of "variable speed basis for this new approach.
synchronous machine", its applications have Now steady state stability can be
been proposed in many fields; for example, analyzed by linearizing the new swing equation
applications in flywheel energy storage sys- and evaluating its eigen values. The line-
arized version of eqs.(5a,b) at the steady
state operating point ( 6 = 6 0 , s = O ) , (i.e., the
linearized swing equation), can be obtained as
follows;

87 SM 619-0 A paper recommended and approved 6a)

tby the IEEE Rotating Machinery Committee of the
I E E E Power Engineering Socit ':y €or presentation at
6b)
the IEEE/PES 1987 Summer Meeting, San Francisco,
California, July 12 - 17, 1987. Manuscript
submitted February 28, 1986; made available €or
printing April 21, 1987.
(7)
0885-8969/88/0600-0323$01.00@1988 IEEE
3 24

The steady state stability criteria can be Application to a Doubly Fed

obtained by evaluating the eigen values of the Synchronous Machine
above equations as follows;
The following analysis concerns a doubly
fed synchronous machine having a three phase
symmetrical structure both in the stator and
the rotor, connected to the source bus
directly and operating as a generator. The
Eq.(8) is the stability criterion for step-out basic equations are presented in Appendix 3.
instability and is virtually equivalent to the The source bus voltage (the stator terminal
conventional dP/dG>O criterion. While eq.(9) voltage of the machine) and the rotor terminal
is the stability criterion for hunting. voltage are assumed to be symmetrical and are
expressed as follows;

ANALYTICAL EXAMPLES Vla'J2 Vlacos(wt+P1 ) (lla)

In this section, the new method will be V1 b=fi v1 aCOS (Ut+P1-2T/3) (1lb)
applied to two rotating machines as an
example, that is, an ordinary synchronous v1 c=JzVlacos(02t+P1 -4n/3) (Ifc)
machine and a doubly fed synchronous machine.
Application to an Ordinary Synchronous Machine
The synchronous generator connected to an
infinite bus through the transmission line as
shown in Fig.1 is analyzed. It is assumed to
be a salient-pole-type machine and to have no Therefore, the angular velocity in the steady
damper windings, and control equipment, such state condition must be (1-so)W from a
as a governer, are neglected. The basic equa- viewpoint of synchronous operation; i.e., the
tions of this system are shown in Appendix l. slip in steady state condition becomes l l s ~ " .
This new method of analysis is based on The developed torque in the form of
the new swing equation of eqs.(5a,b). The most eq.(l) is also derived by using the same
important and complicated problem is to derive method as for an ordinary synchronous machine.
an expression for the developed torque T G as a Namely, at first we consider the case that the
function of 6 and s . But it is very difficult doubly fed machine is rotating with a very
in the general case when the slip 'lsl' is small slip deviation As from the steady state
varing with time. In this paper, it is assumed slip so in spite of being excited by the
that the time variation of s is sufficiently original steady state terminal voltages of
slow compared with the electrical transients, .
eqs. ( 1 1 ) ,( 12 ) Considering this we derived
and thus the derivation is developed by using TG using the same assumption as in the case of
the steady state theory in which " s " is fixed an ordinary synchronous machine; i.e., the
to a certain constant value. As the electrical time variation of As is assumed to be very
transients are considerably faster than the slow. As a result, the following expression
mechanical transients in general except in the for T~ is derived (see Appendix 4),
case when the inertia constant of the system
is small, the above assumption may be
considered to be satisfactory if the inertia
constant is relatively large.
On the basis of the above assumption, the
developed torque T G of the synchronous genera-
tor in Fig.1 can be obtained as follows (see
Appendix 2);
(loa) Each steady state stability criteria can be
obtained by partially differentiating above
T~ with respect to 6 and As.

NUMERICAL EXAMPLES
In general, the steady state stability
can be exactly determined by calculating all
the eigen values of the linearized differen-
-Jqexp(-j26)}] (10d) tial equations of the system and inspecting
thei real parts. In this section, the accu-
The steady state stability criteria can be racy of the proposed new stability criteria,
obtained by partially differentiating above eqs. 8 ) and (9), are evaluated for ordinary
T G as stated in eqs.(6)-(9). and doubly fed synchronous machines respec-
tively by comparing the steady state stable
regions derived by eqs.(8) and (9) with those
given by the above eigen values method.
Ordinary Synchronous Machines
The machine parameters of the ordinary
synchronous machine used in this section are
Fig.1 One machine infinite bus system. shown in Table 1 , which are typical values for
 325

the most common salient-pole-type machines. Table 1 . The parameters of the model.
The derived linearized versions of the
basic differential equations are shown in
Appendix 1. The exact solutions for the steady
Ordinary SM
_____-__.
I Doubly Fed SM
state stable regions can be derived by ra 0.01 PU rla 0.02 PU
calculating all the eigen values of eq.(22) ria 0 . 0 3 pu
at the various steady state operating points 1, 0.1 pu
( & = g o , Eo and inspecting their real parts. Lmd 0.9 pu Ila 0.1 pu
Five eigen values are obtained and two of them
are always very close to Lmq 0.6 pu 1;, 0.1 pu
rf 0.0007pu L, 2.7 pu
-1 /Tatjw (14) lf 0.26 PU J 1.5 & 5 sec
where Ta denotes an armature time constant. T; 1.5 sec
Since these are stable eigen values, the sta-
bility can be determined by the three remain- J 2 & 5 secI
ing eigen values. Fig.2 shows the resulting
stable regions. In it, R I denotes the unstable
region for step-out instability, where one of
the above three eigen values becomes a
positive real number, and R2 denotes the
unstable region for hunting, where two of 1 PU
h

them become a pair of complex numbers whose 3 3 PU

real parts are positive. R3 denotes the a
sec
stable region where all the eigen values have 0
W
negative real parts.
It is clear that the larger the armature
resistance or the internal induced voltage is,
the more the hunting region enlarges; and the
larger the inertia constant is, the more it
narrows. But in the case when J is larger
than about lO(seconds), the change of the
hunting region against the change of J becomes
very small. Furthermore this region is larger
for motors ( - 1 8 0 < 6 0 < 0 )than generators ( O <
60<180). On the other hand, the step-out 3 PU
region varies little when J changes and varies 3 PU
a little when rT changes.
Fig.3 shows the stable regions obtained sec
by using t;e new stability criteria of eqs.(8)
and(9)., R I denotes the unstable region where
Tg ( 0 , R2 denotes the uvstable region where Tg
> O and Ts>O, and R3 denotes the stable
region where Tg'0 and Ts<O. Comparing them
with those of Fig.2, it is clear that the
stability criterion for step-out instability,
eq.(8), has very high accuracy in all cases
and that the stability criterion for hunting, (3)
eq.(9), has also sufficient accuracy in the
case of a relatively large inertia constant J. ryO.1 pu
If the inertia constant J is relatively small, 1 ~ = 0 . 3 PU
comparatively large errors are experienced in
the unstable region for hunting, since the J=2 sec
assumption introduced in the previous section
may not be satisfactory. But for the machiqe
which has rather short time constant of Td,
this error is not so large even if the system
inertia is relatively small.
Doubly Fed Synchronous Machines
The machine parameters for the doubly fed
synchronous machine used in this paper are (4)
shown in Table 1. The derived linearized rT=O.l pu
versions of the basic differential equations
are presented in Appendix 3 . 1~=0.3PU
The exact solutions for the steady state J=IO sec
stable regions when slip s0=0.1 are shown in
Fig.4. In it, R I denotes the unstable region
for step-out instability, where at least one
of the six eigen values becomes a positive
real number; and R 2 denotes the hunting
region, where two of them become a pair of Fig.2 The stable region of an Ordinary
complex numbers whose real parts are positive. Synchronous Machine obtained by the
R3 denotes the stable region where all of the eigen values method.
six eigen values have negative real parts.
Fig.5 shows the steady state stable
326

region obtained by usinglthe proposed criteria

of eqs.(8) and (9). R1 denotes thelunstable
region for step-out where Tg'O, and R2 denote?
the hunting region where Tg'0 and Ts>O. R3
denotes the stable region where T g > 0 and Ts<O.
Comparing Fig.4 and Fig.5, it is clear that
the stability criterion for step-out instabil-
ity, eq.(8), has very high accuracy, and that
the criterion for hunting, eq.(9), also has
sufficient accuracy in the case of a rel-
atively large inertia constant J.
It should also be noted that the stabil-
ity criterion for an ordinary ipduction
machine can be obtained by setting V2a=0 in
eq. (9).

(2) CONCLUSIONS
rT=0.3 pu
lT=0.3 pu This paper proposes a new approach for
the steady state stability analysis of syn-
J=5 sec chronous machines. This method can deal with
both types of instabilities of synchronous
machines; that is, step-out and hunting. It is
based on a new swing equation, from which two
stability criteria can be obtained, one is for
Fig.3 The stable region of an Ordinary step-out and the other for hunting. This paper
Synchronous Machine obtained by the has applied this new method to a doubly fed
proposed method. synchronous machine as well as to an ordinary
synchronous machine and has discussed its
applicability. As a result, it has been shown
that the criterion for step-out has a very
high accuracy and the criterion for hunting
also has sufficient accuracy if the system
inertia is relatively large.
(1)
Therefore it is concluded that the method
so=o. 1 presented here is very practical and should be
J=l - 5 sec useful as a unified method for the steady
state stability analysis of synchronous
machines.

References

S.B.Crary, "Power System Stability Vol.1

- Steady State Stability - ' I I John Wiley &
Sons, 1945.
E.W.Kimbark, I' Power System Stability ' I ,
(2) Dover Publ., 1968.
so=o. 1 C.A.Nickle et al., "Stability of Synchro-
nous Machine - Effect of Armature Circuit
J=5 sec Resistance - ' I , Trans.AIEE, ~01.49, p.338,
Jan. 1930.
C.F.Wagner, "Effect of Armature resistance
upon Hunting of Synchronous Machine",
ibid., vo1.49, p.1011, July 1930.
Fiq.4 The stable reaion of a Doublv Fed G.Kron, "Equivalent Circuit for the Hunt-
Synchronous Machine obtained b; the ing of Electrical Machinery", ibid.,
eigen values method. vo1.61 , p.290, May 1942.
C.Concordia, "Synchronous Machine Damping
and Synchronizing Torques", ibid., ~01.70,
p.731 , 1951.
R.V.Shepherd, "Synchronizing and Damping
Torque Coefficients of Synchronous Ma-
chines", ibid., v01.80, p.180, June 1961.
D.P.S.Gupta et al., 'I Synchronizing and
Damping Torques in Synchronous Machines",
Proc.IEE, ~01.114, no.10, p.1451, 1967.
D.P.S.Gupta et al., "Hunting Characteris-
tics of a Synchronous Machine with TWO
Field Windings" , ibid. ,vol. 1 1 7 ,no. 1 ,p. 1 1 9 ,
1970.
C.Concordia et al., The Doubly Fed
Machine", ibid., vo1.61 , p.286, May 1942.
Fig.5 The stable region of a Doubly Fed R.E.Bedford, "The Synchronous Double-Fed
Synchronous Machine obtained by the Induction Machine", ibid., vo1.75,p.1486,
proposed method. Feb. 1956.
 327

[12] B.M.Bird et al., "Analysis of Doubly Fed armature a-phase magnetic axis and the d-axis
Slip-ring Machines", Proc.IEE, vo1.113, (field magnetic axis), and its initial value,
no.6, p.1016, 1966. 8 0 , may be expressed by using the internal
1131 R.D.Jackson et al., "Steady State Stabil- phase angle 6 as
ity of the Doubly Fed Synchronous Ma-
chine", Elec.Lett. ,vol. 4 ,no. l ,p. 18, 1968. 0 o=-n /2+6 (19)
1141 P.T.Finlayson et al., "Cycloconvertor-
Controlled Synchronous Machine for Load Hence vd and vq become
Compensation on AC Power Systems", IEEE
Trans., vol.IA-lO,no.6,p.806, 1974. V d = 6 Vgsin6 (20a)
[151 W.B.Gish et al., "An Adjustable Speed
Synchronous Machine for Hydroelectric V q = G VBCOSG (20b)
Power Applications", ibid., vol.PAS-100,
no.5, p.2171, 1981. Making use of eqs.(l8b) and (19), w, becomes
[161 P.G.Holmes et al.,"Cycloconvertor-Excited
Divided-Winding Doubly-Fed Machine as a wm=pB=w+p6 (211
Wind Power Convertor", Proc.IEE, ~01.131,
Part-B, no.2, p.61, 1984. By linearizing eqs. (15),(16),(20) and
(21), the following equation is derived;
Appendix 1. The basic equations of an Ordinary
Synchronous Machine
The basic differential equations of the The expressions of the elements aij of matrix
synchronous machine which has no damper [AI are omitted.
windings and is connected to an infinite bus
through the transmission line may be expressed Appendix 2. Derivation of the Developed Torque
by using the two axis theory as follows; of an Ordinary Synchronous Machine

Vd=-(Ra+pLD)id+PLmdif-evd (15a) When a synchronous machine is rotating

with a small slip "s", a displacement angle 8
Vq'- (Ra+PLQ)iq-evq ( 1 5b) and an angular velocity of revolution wm may
be written in the following form
vf=(rf+pLf)if-pL,did (15~)
e = ( 1 -S )wt+e 0 (23)
T D=TG+JPW, (15d)
wm=(l-S)w (24)
evd=-WmLQiq, evq=-Wm(-LDid+Lmdif) (16a)
Substituting the above equations into eq.(l8a)
yields
V d d T VBCOS ( SW t-8 0 ) ( 25a 1

vq=Js VBsin(swt-eo) (25b)

where These are AC voltages of a slip frequency of
sf(Hz) (f=w/2a). Since an applied field volt-
rT, 1~ :transmission line resistance and age vf is DC, there are two current components
inductance, in the d,q-axis circuits; i.e., the AC of the
ra, rf :armature and field resistances, slip frequency and the DC.
la, lf :armature and field leakage The network equations for the DC compo-
inductances, nents (to be described by adding a subscript
Lmd,Lmq :d-axis and q-axis magnetizing D ) can be derived by setting p=O, Vd=Vq=o and
inductances, wm=(l-s)w in eqs.(l5a-c); i.e.,
evd,evq :d-axis and q-axis speed electro-
motive forces.
We assume that the infinite bus voltages
have a constant r.m.s. value (1 pu) and are
expressed as follows;

Va=J2 VBCOSwt ( 7a 1 Eo denotes a no-load generated voltage.

Solving these equations with respect to idD
Vb=J2 VBCOs(wt-2a/3) ( 7b) and ' and substituting them into eq.
(1 6a )l'D;ield evdD and evqD.
vc=d2 VBCOs(wt-4n/3) ( 7c) On the other hand, the network equations
for sf(Hz) AC components (to be expressed by
Then the d,q-axis voltage components can be phasors) can be obtained by setting p=jsw, vf=
obtained by the following relations; 0, wm=(l-s)w in eqs.(l5a-c); i.e.,

Vd=-(Ra+JSWLD)Id+JSWLmdIf+(1 -S)WLQIq (27a)

0 = ( r f+ j swLf if- jswLmdid (27~)

From eq. (25), cd and cq may be expressed in
where e is a displacement angle between the the form of phasors as
328

ird=J3/2 vBexp ( - je 0 ) (28a) differential equations for a doubly fed syn-

chronous machine by using the two axis (d,q-
Cq=-j v 5 E VBexp ( - j80 ) (28b) axis) theory. The d-axis is fixed on the syn-
Solving eqs.(27a-c) with respect to
. .
IdrIq and
chronous reference frame which is rotating at
the synchronous speed w, and the q-axis is
If, and substituting them into fixed 90 degrees forward from the d-axis in
the phase sequence direction. The basic equa-
i=Jz7 ?exp(jsot)+i*exp( -jswt /2 (29) tions in this coordinate system for the doubly
fed machine may be obtained as follows;
where * signifies a complex conjugate, yields
the instantaneous value of them; i.e., idAriqA
and ifA. And substituting these currents into
eq. ( 16a) yields evdA and e, A
As a result of the above jeiivations, the
developed torque T G can be obtained from eq.
(16b) as follows;

T,=-( (evdD+evdA)(idD+idA)
+(evqD+evqA)(iqD+iqA)}/(l-s)w
where
=-I
(evdDidD+evqDiqD)+ (evdDidA+evdAidD
ria, ria :stator and rotor winding resist-
+evqDiqA+evqAiqD)+("vdAidA+evqAiqA)} ances (prime refers to the stator
D DA A side),
/(I-S)U=TG+TG + T G (30) liar lia :stator and rotor leakage induct-
ances ,
where Lm :magnetizing inductance,
and 02 denotes the displacement angle between
the rotor a-phase magnetic axis (rotor axis)
and the d-axis, and is expressed in terms of
81, the displacement angle between the stator
a-phase magnetic axis (stator axis) and the
rotor axis, as follows;
e2=wt-e1 (36)

Furthermore the developed torque TG may be

written as
1
T G = L m ( i 1 qi ;d-i 1 dizq) 37)

And the equation of motion is expressed as

follows ;
TD=JPW,+TG 38)

where the rotation speed w, may be described

by using 81 as

From eqs. (21 ) and (24), we can get In a steady state condition in which slip
takes the constant value s o ,
wm=w+p6=('l-s)w
e = ( 1 - s o ) wt+eO (40)
and thus
and thus
PG=-Sw
e2=wt-e1=so~t-e0 (41 1
Pwm=-wPs
The d,q-axis voltage components can be
Substituting eq.(33b) into eq.(l5d) yields obtained by substituting eqs. ( 1 1 a )- ( 12c), wt
eq.(5b). By using eqs.(18b),(19) and (23), one and 82 into eq.(l8a) as follows;
can obtain
V l d = n VlacosP1 ( 42a 1
swt-eo= IT/.^- 6 (34)
Substituting eq.(34) into eqs.(3lb) and (31c)
yields eqs.(lOc,d).

-Appendix
-- 3. Basic equations for a Doubly Fed
Synchronous Machine
In the same manner as an ordinary syn-
chronous machine, we can derive the basic 6 denotes the spatial displacement angle
 329

between the stator and the rotor voltages can be obtained. On the other hand, the net-
vectors as shown in Fig.6. work equations for Asf(Hz) AC components can
In a transient state condition, since we be derived by setting p=jAsw, v1d=vlq=O, PO2
assume that 80 is time dependent, such as =(so+As)w and by rewriting all the variables
80(t), and that pi and P 2 are constant, to phasors, in eqs.(35a-d). yhile thelAC cur-
therefore rent components fldA!ilqAri2dA and i2 A are
obtained by substituting the phasor so?utions
e 1=( 1 -so)wt+eo ( t) (44a) of the above network equations into eq.(29).
The developed torque T G is obtained by
eo(t=o) =eo (44b) substituting the above current solutions into
eq.(37) as follows;
1 = ( 1 -s 0 +Po 0 ( t ) = ( 1 - s 0 w+p6 (45)
By linearizing eqs.(35),(37),(38),(42) and
(451, the linearized version of the system
equations can be obtained in the form as
I I
P[ild ilq i2d i2q 6 Wml
1 1
where
=[Bl[ild iiq i2d izq 6 mm1 (46)
n
z s = zs1 +zs2 (53a1
The concrete expression for the matrix [B] is
omitted.
stator-axis
rotor-axis 4

+j { ( Sg+AS wL2r1 a+wL1r2a) (53h)

Fig.6 Diagrammatic explanation of 6. 2 2
Lo=L~L~-L~ (53i)
1 2
T G and T G are shown in eqs.(13brc). Further-
more, by making use of eqs.(45) and (50), the
Appendix 4. Derivation of the Developed Torque following relations can be derived,
of Doubly Fed Synchronous Machines
P6=-wAS (54)
When we assume that the rotation slip 2
changes infinitesimally, A s , but the frequen- pwm=P &=-UPAS (55)
cies both of the stator and the rotor terminal
voltages remain unchanged, then the new Consequently eq.(5b) is obtained. And from
rotation slip s is eqs.(44a) and (48),

s=so+As (47) Oo(t) =-ASwt+Oo (56)

and thus Substituting eq.(56) into eq.(52b) yields
eq. (1 3d).
e 1 = ( 1 -so-As)wt+Oo (48)
e2=(S ~ + A 1S ~ t - e ~ (49)
wm=(l - s o - A s ) ~ (50)
Substituting eqs.(l2a-c) and ( 4 9 ) into eq.
(18a) yields

v;a=fi V;aco~(A~~t-80-p2) (51a)

vld and vlq are the same as eqs.(42a,b).

Therefore two frequency components are exist
in d and q axis circuits; i.e., DC and sf(Hz).
The network equations for the DC, co?po-
nents can be derived by setting p=O, V2d=V2q=O
and p82=(so+As)w in eqs.(35a-d), land then fhe
DC current components ildD,ilqDri2dD and i 2 q ~