Power System Stability Vol III Kimbark

POWER SYSTEM STABILITY
Volume III
Synchronous ~achines
IEEE Press
445 Hoes Lane, PO Box 1331
Piscataway, NJ 08855-1331
Editorial Board
John B. Anderson, Editor in Chief
R. S. Blicq J. D. Irwin J. M. F. Moura

M. Eden S. V. Kartalopoulos J.Peden
D. M. Etter P. Laplante E. Sanchez-Sinencio
G. F. Hoffnagle A. J. Laub L. Shaw
R. F. Hoyt M. Lightner D. J. Wells
Dudley R. Kay, Director ofBook Publishing

Carrie Briggs, Administrative Assistant
Lisa S. Mizrahi, Review and Publicity Coordinator
Valerie Zaborski, Production Editor
IEEE Press Power Systems Engineering Series

Dr. Paul M. Anderson, Series Editor
Power Math Associates, Inc.
Series Editorial Advisory Committee
Dr. Roy Billinton Dr. George Karady
University of Saskatchewan ArizonaState University
Dr. Atif S. Debs Dr. Donald W. Novotny

GeorgiaInstituteof Technology University of Wisconsin
Dr. M. El-Hawary Dr. A. G. Phadke

TechnicalUniversity of VirginiaPolytechnic and
Nova Scotia State University
Mr. Richard G. Farmer Dr. Chanan Singh

ArizonaPublic Texas A & M University
ServiceCompany
Dr. Charles A. Gross Dr. E. Keith Stanek

AuburnUniversity University of Missouri-Rolla
Dr. G. T. Heydt Dr. 1. E. Van Ness

Purdue University Northwestern University
POWER
SYSTEM
STABILITY
Volume III
Synchronous Machines
Edward Wilson Kimbark
A. IEEE
'V PRESS
roWiLEY-
~INTERSCIENCE
A JOHN WILEY & SONS, INC., PUBLICATION
IEEE Press Power Systems Engineering Series

Dr. Paul M. Anderson, Series Editor
The Institute of Electrical and Electronics Engineers, Inc., New York
© 1995 by the Institute of Electrical and Electronics Engineers, Inc.
345 East 47th Street, New York, NY 10017-2394
© 1956 by Edward Wilson Kimbark
This is the IEEE reprinting of a book previously published by John Wiley &
Sons, Inc. under the title Power System Stability, Volume III: Synchronous
Machines.
All rights reserved. No part of this book may be reproduced in any form,
nor may it be stored in a retrieval system or transmitted in any form,
without written permission from the publisher.
Printed in the United States of America
10 9 8 7 6 5 4 3 2
ISBN 0·7803·1135·3
Library of Congress Cataloging-In-Publication Data
Kimbark, Edward Wilson

Powersystemstability I Edward WilsonKimbark.
p. em. - (IEEEPresspowersystems engineering series)
Originally published: New York : Wiley, 1948-1956.
Includes bibliographical references and index.
Contents: v. L Elements of stability calculations - v. 2. Power
circuitbreakers and protective relays- v. 3. Synchronous
machines.
ISBN 0-7803-1135-3 (set)
1.Electric powersystemstability. I. Title. II. Series.
TKI010.K56 1995
621.319--dc20 94.-42999
CIP
To my wife
RUTH MERRICK KIMBARK
PREFACE
This is the third volume of a three-volume work on power-system

stability intended for use by power-system engineers and by graduate
students. It grew out of lectures given by the author at Northwestern
University in graduate evening courses.
Volume I, which appeared in 1948, covers the elements of the
stability problem, the principal factors affecting stability, the
ordinary simplified methods of making stability calculations, and
illustrations of the application of these methods in studies which
have been made on actual power systems.
Volume II, which appeared in 1950, covers power circuit breakers
and protective relays and the influence of these devices on power-
system stability.
Volume III, the present one, deals with the theory of synchronous
machines and their excitation systems, an understanding of which is
necessary for the justification of the simplifying assumptions ordi-
narily used in stability calculations and for carrying out calculations
for the extraordinary cases in which greater accuracy is desired than
that afforded by the simplified methods. This volume discusses such
effects as saliency, damping, saturation, and high-speed excitation.
It thus endeavors to give the reader a deeper understanding of power-
system stability than that afforded by Volume I alone. Steady-state
stability, which was discussed very sketchily in Volume I, is treated
more fully in the present volume.
Volume I should be considered to be a prerequisite to the present
volume, but Volume II should not.
I wish to acknowledge my indebtedness in connection with this
volume, as well as with the preceding ones, to the .following persons:
To my wife, Ruth Merrick Kimbark, for typing the entire manu-
script and for her advice and inspiration.
To J. E. Hobson, W. A. Lewis, and E. T. B. Gross for reviewing
the manuscript in its earlier form and for making many suggestions
for its improvement. (They did not, however, review recent changes
vii
viii PREFACE
and additions and thus 'cannot be blamed for any shortcomings which
this volume may now have.)
To manufacturers, authors, and publishers who supplied illustra-
tions or gave permission for the use of material previously published
elsewhere. Credit for such material is given at the place where
it appears.
EDWARD WILSON KIMBARK
Seattle, Waahington
November, 1955
CONTENTS
CHAPTER PAGE
XII Synchronous Machines 1
Reactances, Resistances, and Time Constants 2
Sudden Three-Phase Short Circuit 40
Mathematical Theory 52
Vector Diagrams 69
Applications to Transient Stability Studies 78
Saturation 118
XIII Excitation Systems 137

XIV Damper Windings and Damping 214
XV Steady-State Stability 247
INDEX 317
ix
CHAPTER XII
SYNCHRONOUS MACHINES
Since the problem of power-system stability is to determine whether

or not the various synchronous machines on the system will remain in
synchronism with one another, the characteristics of those synchro-
nous machines obviously play an important part in the problem.
FIG.
(a)
o (6)
1. Cross sections of synchronous machines, (a) round-rotor type, (b) salient-

pole type.
Synchronous machines are classified into two principal types-

round-rotor machines .and salient-pole machines (Fig. 1). Generators
driven by steam turbines (turbogenerators) have cylindrical (round)
rotors with slots in which distributed field windings are placed. Most
cylindrical rotors are made of solid steel forgings, though a few of
them are built up from thick steel disks. The number of poles is two,
four, or six.. Most new machines are two-pole. Generators driven by
water wheels (hydraulic turbines) have laminated salient-pole rotors
with concentrated field windings and, usually, a large number of poles.
Some water-wheel generators are provided with amortisseur windings
1
2 SYNCHRONOUS MACHINES
or damper windings; others are not. Synchronous motors and con-
densers have salient-pole rotors with amortisseur windings. Synchro-
nous converters have rotating armatures and salient-pole stationary
fields with amortisseurs.
In the foregoing chapters it was assumed that each synchronous
machine could be represented by a constant reactance in series with a
constant voltage, and, furthermore, that the mechanical angle of the
rotor of each machine coincided with the electrical phase of that con-
stant voltage of the machine. In view of the importance of synchro-
nous machines in the stability problem, these assumptions require
further investigation, both to justify the use of the assumptions in the
ordinary case and to provide more rigorous methods of calculation in
those extraordinary cases where the usual assumptions do not give
accurate enough results.
The usual assumptions may be in error because of three phenomena,
each of which requires consideration:
1. Saliency.
2. Field decrement and voltage-regulator action.
3. Saturation.
The word saliency is used as a short expression for the fact that the
rotor of a synchronous machine has different electric and magnetic
properties on two axes 90 elec. deg. apart - the direct axis, or axis of
symmetry of a field pole, and the quadrature axis, or axis of symmetry
midway between two field poles. This differencebetween the two axes
is present not only in salient-pole machines but also, to a lesser extent,
in round-rotor machines, because of the presence of the field winding
on the direct axis only. In this respect both kinds of synchronous
machines differ from induction machines. The effect of saliency will
be considered first while the effects of field decrement, voltage-regu-
lator action, and magnetic saturation are disregarded; later, these
other effects will also be taken into account. A knowledge of syn-
chronous-machine reactances, resistances, time constants, and vector
diagrams is necessary to the understanding of all these effects.
REACTANCES, RESISTANCES, AND TIME CONSTANTS
As has already been noted, a synchronous machine is often repre-
sented in a circuit diagram (the positive-sequence network) by a
constant reactance in series with a constant voltage. What reactance
should be used for that purpose? A great many synchronous-machine
reactances have been defined; for example, direct- and quadrature-
axis synchronous, transient, and subtransient reactances. Before
CONCEPTS OF INDUCTANCE 3
proceeding to define these reactances we may profitably review the

fundamental concepts of inductance and of inductive reactance.
Fundamental concepts of inductance. The concepts of electric cur-

rent and of lines of magnetic flux are familiar to all students of elec-
tricity. Since a line of flux is always a closed curve and an electric
circuit is likewise a closed ·path, * a given flux line either does not link
a given electric circuit at all or it does link it one or more times. Thus
Circuit Fluxline
FIG. 2. Illustrating flux linkages. FIG. 3. A line of magnetic flux con-

stitutes a positive flux linkage with an
electric circuit if it is related to the
circuit according to the right-hand rule.
A diagram similar to this one appears
on the emblem of the A.I.E.E.
in Fig. 2 flux line 1 links circuit C once, line 2 links it four times, and
line 3 does not link it at all. The flux linkage 1/J of a circuit is defined
as the algebraic sum of the numbers of linkages of all the lines of flux
linking the 'circuit, the positive direction of linkage' being taken ac-
cording to Fig. 3. Thus in Fig. 2, if there were no flux lines linking the
circuit except those shown, and if each line represented one weber of
flux, the flux linkage of the circuit would be -1 + +
4 0 = 3 weber-
turns. In general, the flux should be divided into infinitesimal parts
d(j), each represented by a line or tube linking the circuit n times; the
flux linkage of the circuit is then given by
1/1 f
= n dcP weber-turns [1]
*The flux linkage of a part of a circuit, such as a winding of a machine, which is

not a closed path, can be defined as the linkage of the closed path formed by the
winding and a short linefrom one terminal of the winding to the other.
According to Faraday's law of electromagnetic induction, any change

in the flux linkage of a circuit induces an electromotive force e, the
sign and magnitude of which are given by
d1/l
e= - - volts [2]
dt
where t is time in seconds. A positive e.m.f. tends to set up a positive
current (that is, a current in the direction of the circuit, Fig. 3). The
applied e.m.f. required to overcome the induced e.m.f. is of opposite
sign from the latter. Faraday's law is perfectly general: it holds
whether the change of flux is caused by a change of current in the cir-
cuit considered, by a change of current in another circuit, by a defor-
mation of the circuit, by relative motion of one circuit with respect
to another, or by relative motion of magnetic materials or permanent
magnets with respect to the circuit.
There are some circuits the flux linkages of which are substantially
proportional to the current:
1/1 = Li weber-turns [3]
where L is the self-inductance of the circuit in henrys and i is the cur-
rent of the circuit in amperes. The self-inductance of such a circuit
may be regarded as the flux linkage per ampere:
1/1
L = -; henrys [4]
~
It is always a positive quantity, because, according to the right-hand

rule and the sign convention of Fig. 3, a positive current sets up a
positive linkage, and a negative current, a negative linkage.
Substitution of eq. 3 into eq. 2 yields, for the e.m.f. of self-induction,
e= - ~ (Li) volts [5]

dt
If the self-inductance L does not vary with time, this equation may
be put in the customary form,
di
e = -L - volts [6]
dt
It would be decidedly unwise to assume constant self-inductance for
the windings of rotating machines, where iron parts of the magnetic
circuit are in motion relative to the windings. For such windings the
more general equations 2 or 5 should be used. In cases where eq. 6
MUTUAL INDUCTANCE 5
holds, however, the self-inductance can be defined by

e
L = - di/dt [7]
This relation is ordinarily used in defining the unit of inductance, the

henry, as the inductance of a coil in which an e.m.f. of 1 volt is in-
duced by a current changing at the rate of 1 amp. per second.
There are some circuits in which the flux linkage depends only upon
the current in the same circuit, but is not directly proportional to
the current because of the presence of
saturating iron (see Fig. 4). The self- t/I
inductance of such a circuit can be de-
fined by
L = d,p [8]
di
rrhen, since
d'" = d'" . di = L di [9]
dt di dt dt
eq. 6 is still valid. However, the concept
FIG. 4. Relation between flux
of inductance is useful chiefly where the linkage and current of a winding
linear relation of eq. 3 holds. on a closed iron core.
Mutual inductance. There are some groups of inductively coupled

circuits where the flux linkage of any circuit is a linear function of the
currents in all the circuits, thus:
,pI + L I 2i2 + Lisis · · .

= Lllil [lOa]
,p2 = L 21i1 + L 22 i 2 + L2S~S • • • [lOb]
1/Ia = Lalit + L a2i2 + Lasia · · · flOc]
Here L 11, L 22 , L 33 are the self-inductances of circuits 1, 2, 3; L 12 is

the mutual inductance between circuits 1 and 2; LIB is the mutual in-
ductance between circuits 1 and 3, and so on. Although the numerical
values of self-inductance are always positive, those of mutual induct-
ance may be either positive or negative, depending on the positive
directions chosen arbitrarily for each of two coupled circuits as well
as on their geometrical arrangement. The mutual inductance between
any two circuits is the same in both directions; that is, L 12 = L 2 1,

and so on.*
From eqs. 10 the self- and mutual inductances can be defined by
L 11
a1/11
= -. [lla]
atl
a1/1 1
L 12 = - . [lIb]
at2
and so on. The definition of self-inductance, eq. 11a, is more general

than that of eq. 4 for the single linear circuit and that of eq. 8 for the
single nonlinear circuit, and yet it is entirely consistent with them both.
Relation of inductance to the magnetic circuit. The magnetomotive

force of a current it amperes flowing in nl turns of a conductor sur-
rounding a magnetic circuit is n1i1 ampere-turns, and the flux cPll due
to this m.m.f. is
cPll = 9'11 nl i1 webers [12]
where 9'11 is the permeance of the magnetic circuit in m.k.s. units
(webers per ampere-turn, or henrys per turn-squared). The flux
linkage of the same electric circuit is
1/11 = nlcPll = 9'11 n 1 2i l weber-turns [13]
and the self-inductance of the circuit is
L 11 = -.- =
a1/11 r1)
'-F 11nl
2
henrys [14]
at1
The flux linking a second electric circuit of n2 turns due to current it

in the first circuit is
</>21 = g>21nl il webers [15]
where fP 21 is the permeance of the mutual flux paths. The flux link-
age of circuit 2 due to the current in circuit 1 is
"'2 = n2cP21 = 9'21 n2n1 i 1 weber-turns [16]
and the mutual inductance of circuits 1 and 2 is
at/l2
L 21 = -.- = 9'21n2nl henrys [17]
atl
*This relation can be proved by deriving and comparing expressions for the energy
put into two coupled circuits in order to build up the same magnetic field by each
of two different procedures: (1) first establishing current 11 in circuit 1 while the
current in circuit 2 is zero, and then maintaining 11 constant while current 12 is
established in circuit 2; (2) establishing currents 12 and 11 in the opposite order.
SUMMARY 7
Thus both self- and mutual inductance are directly proportional to the
permeance, or reciprocal of reluctance, of the associated magnetic cir-
cuits and the product of the numbers of turns of the two electric
circuits (or square of the number of turns in case of self-inductance).
The permeance fP and reluctance ffi. of a magnetic circuit depend
upon the permeability and dimensions of the circuit, and are given by
[18]
where JJ. is the permeability, A is the cross-sectional area, and l is the

length of the magnetic circuit. If p. and A do not vary along the path,
9' = ~A [19]
l
It is often useful to compare inductances of various circuits by com-
paring the permeabilities, lengths, and cross-sectional areas of the
associated magnetic circuits.
Inductive reactance in ohms of a circuit of inductance L is given by

X = wL '= 21rfL [20]
where f is the frequency in cycles per second. There are self- and
mutual inductive reactances corresponding to self- and mutual induct-
ances.' Mutual inductive reactance has the same algebraic sign as the
corresponding mutual inductance.
The reactance of a circuit can be determined by circulating in it
sinusoidal alternating current, finding the fundamental-frequency com-
ponent of terminal voltage in quadrature with the current, and taking
the ratio of this voltage to the current. The reactance is positive if
the voltage leads the current.
Vsin8 .
X = - - = ZSln8 [21]
I
where I. is the current in amperes, V is the fundamental-frequency
component of voltage drop in volts, and 8 is the angle by which V
leads I. V and I may be expressed in either peak or effective values.
Summary of concepts of inductance and inductive reactance. The

concept of inductance has been presented from the following view-
points:
1. Flux linkage per unit current.
2. Induced voltage per unit rate of change of current.
8 SYNCHRONOUS l\IACHINES
3. Derivative of flux linkage with respect to current.

4. Permeance of the associated magnetic circuit multiplied by the
square of the number of turns.
The concept of inductive reactance has been presented from the
following viewpoints:
1. Product of angular frequency and inductance.
2. Quadrrature voltage per unit current.
Inductances of two coupled circuits (the transformer). Consider

two inductively coupled circuits having constant resistances, self-
inductances, and mutual inductance, and variable currents and termi-
FIG. 5. Two inductively coupled circuits.
nal voltages, as marked in Fig. 5. The instantaneous terminal volt-

ages are given in terms of the instantaneous currents and circuit
constants by the following equations:
. d,p1 . di di
VI = Rl~1 + di = Rl~1 + L 11 dtl + L I 2 dt2 [22a]
. d,p2 . di 2 di l
V2 = R21,2 + di = R21,2 + L 22 dt + L 12 dt [22b]
In the steady state, with sinusoidal alternating currents and voltages,

the complex terminal voltages are:
VI = RIll + jW(L11I1 + L~212) [23a]
V2 = R 2I2 + jw(L2 21 2 + L 12I1 ) [23b]
or
VI = ZllII + Zl2I 2 [24a]
V2 = ZI2I l + Z22I2 [24b]
where
ZII = R I + jwLI I [25a]
Z22 = R 2 + jwL22 [25b]
Z12 = 0+ jwL12 [25c]
TWO COUPLED CIRCUITS 9
Short-circuit impedance and inductance. If the secondary circuit is

short-circuited, V2 = 0, and eq. 24b gives the following relation be-
tween primary and secondary currents:
12 =_ Z12 11 [26]
Z22
If this value of 12 is substituted into eq. 24a; the primary voltage is
VI = ( ZIt - -Z122) It [27]

Z22
and the driving-point impedance of the primary circuit with the sec-
ondary winding short-circuited is
VI Z12 2
Zsc = - = ZIt - - - [28]
II Z22
The first term of this expression, ZII, is the self-impedance of the

primary circuit, which is equal to the ratio Vt/I1 when the secondary
circuit is open or absent; and the second term is called the reflected
impedance of the secondary circuit. By use of eqs. 25b and 25c, the
reflected resistance and reactance may be found as the real and im-
aginary parts, respectively, of this term.
Z12 2 (jwL 12)2 w2L122R2 - jw3L122L22
---=- [29]
Z22 R2 + jwL22 R2
2
+
2
w L 22
2
Since w, R 2 , £22, and £12 2 are all positive numbers (whether £12 is
positive or negative), the reflected resistance is positive, but the re-
flected reactance is negative. Therefore the presence of a closed sec-
ondary circuit increases the apparent resistance of the primary circuit
and decreases its apparent reactance. The short-circuit reactance of an
iron-core transformer may be as low as 1/500th of its open-circuit
reactance.
If R 2 2 «oo 2L 222 , then the driving-point impedance becomes
2
z; ~ R 1 + ( LLI2) . ( L 11 -
2 R 2 + Jw LL12 ) [30]
22 22
The driving-point inductance is

L 122
t.; ~ L 11 - - L [31]
22
and the driving-point reactance is
X 12 2
X sc ~ XII - -
X- [32]
22
The last two equations are similar in form to eq. 28.

10 SYNCHRONOUS MACaINES
Coefficient of coupling and leakage coefficient. Consider two induc-

tively coupled coils 1 and 2 having nl and n2 turns, respectively.
Suppose that there is a current i 1 in coil 1 but no current in coil 2.
If, under these circumstances, all the flux cPt due to the current in
coil 1 should link all the turns of both coils, the self-inductance of
coil 1 would be
[33]
and the mutual inductance of coils 1 and 2 would be
[34]
The ratio would be
L 12 n2
-=- [35]
L 11 n1
Similarly, with current i 2 in coil 2 and no current in coill, if the flux

cP2 should link all the. turns of both coils, the self-inductance of coil 2
would be
[36]
and the mutual inductance would be
L 12 -- -
nlcP2
.- [37]
~2
The ratio would be
-
£12
--
nl
[38]
L 22 n2
By elimination of nl/n2 from eqs. 35 and 38, the value of mutual in-
ductance would be
[39]
which is the greatest value that £12 could possibly have.
Actually, all the flux will not link all the turns of both coils. With
current in coil 1 only, the linkages of both coils will be less than that
given by eqs. 33 and 34, but, because of leakage flux, the linkage of
coil 2 will be reduced more than that of coil l. Hence
L 12 = k n2 [40]
1
L 11 nl
where k 1 < 1. Similarly, with current in coil 2 only, the linkage of
TWO COUPLED CIRCUI1'S 11
coil 1 is reduced more than that of coil 2, and
L 12 = k2 n1 (41]
£22 17,2
where k2 < 1. Elimination of nt/n2 between eqs. 40 and 41 gives
L 12 = Vk 1k2L 11I.J22 = k vII..JI1L2; [42]

where
_ r;-:;- £12
k = V k 1k2 = _ rr-;:- <1 [43]
V L 11L22
is the coefficient of coupling.
The leakage coefficient is defined as
q =1- k~ [44]
'The short-circuit inductance given by eq. 31 may then be written

[45]
Thus the leakage coefficient is the ratio of the apparent inductance
of the primary circuit when the secondary winding is short-circuited
to that when the secondary circuit is open. This ratio is the same no
matter which of the two windings is the primary. The tighter the
coupling, the smaller is the leakage coefficient.
FIG. 6. Equivalent T circuit of the inductively coupled circuits of Fig. 5.
Equivalent T circuits. Equations 22, 23, and 24, which were written
for the inductively coupled circuits of Fig. 5, are equally valid for the
T circuit of Fig. 6. Therefore, the two circuits are equivalent. The
T circuit is not always physically realizable because one of the series
inductances, L 11 - L 12 or L 22 - L 12 , may be negative. Although a
negative self-inductance may be replaced by a capacitance in the
steady state at a single fixed frequency, this substitution is not valid
under transient conditions.
The T circuit of Fig, 7 differs from that of Fig. 6 in that the mutual
inductance or impedance is multiplied by an arbitrary constant a
(usually taken as a positive real number), and the secondary resist-
ance, self-inductance, and impedance are multiplied by a2 • This cir-
cuit gives the same primary current and voltage as Figs. 5 and 6 do,
il L 11 - aL 12
FIG. 7. T circuit that gives the same primary current and voltage as the circuits of
Figs. 5 and 6, but different secondary current and voltage.
provided that the secondary current be divided by a and the voltage

multiplied by a, as indicated in the figure. The power is correct on
both primary and secondary sides. This circuit shows that it would
be possible to substitute one closed secondary winding for another
without making any difference in the primary current and voltage.
(l-k)L n
FIG. 8. The circuit of Fig. 7 with a = VL 11/L22. The inductances of the two
series arms are equal.
For example, if the number of secondary turns were doubled and their
cross-sectional area halved, and if the winding occupied the same space
as before, the mutual inductance and secondary e.m.f. would be
doubled, the secondary resistance and self-inductance quadrupled,
and the secondary current halved. In this case a = 2.
By choice of a in the range
k2 nl = L 12 < a < L 11 = ! nl [46]

n2 L 22 Li2 k1 n2
THEOREM OF CONSTANT LINKAGE 13
both the series inductances of Fig. 7 are rendered positive. The

greater is the coefficient of coupling, the narrower is this range. If
a = VL 11 /L22 , these two inductances become equal, and the equiva-
lent T circuit is as shown in Fig. 8. If a is given one of the limiting
values, then one of the series inductances becomes zero, as illustrated
in Fig. 9 for a = L 12/L22 • (Compare this figure with eqs. 30 and 45
FIG. 9. The circuit of Fig. 7 with a = L 12/ £ 22. The inductance of the secondary
series arm 'is zero.
for short-circuit impedance.) For iron-core transformers, it is cus-

tomary to take a as the turns ratio nl/n2' The two series inductances
(known as the primary and secondary leakage inductances) are then
positive but not necessarily equal. For rotating machines having
windings on opposite sides of an air gap, a may be taken as the ratio
of currents which separately produce equal fluxes crossing the air gap
per pole. The leakage fluxes are then fluxes which do not cross the
air gap. Since a is arbitrary, however, the leakage inductances are
arbitrary.
The theorem of constant flux linkage. 2 ,12 One of the most useful
principles for analyzing transient phenomena in inductive electric cir-
cuits is the theorem of constant flux linkages, which will now be de-
rived. Consider any closed circuit, or mesh, of an electric network,
having finite resistance, flux linkage due to any cause whatever, and
e.m.f.'s not due to change of linkage, but no series capacitance. The
sum of the potential differences around such a path is
[47]
Here the letters R, i, 1/1, t, and e have their usual meanings. The sum-
mation sign L: is used because the mesh under consideration may
consist of several branches, each having different resistance, current,
and e.m.f. The flux linkage, on the other hand, is that of the entire
mesh, as defined earlier in this chapter. Now integrate eq. 47 with

respect to t from 0 to ~t.
t" (At
LRJo i dt + ~1/1 = ~ J, edt [48]
If Rand e are finite, then as ~t approaches zero the definite integrals

approach zero, and therefore ~,y also approaches zero. This result
may be stated as follows: The flux linkage of any closed circuit of finite
resistance and e.m.j. cannot change instantly.
In the special case in which the resistance and e.m.f. are zero,
eq. 47 becomes simply
dl/; = 0 [49]
dt
which integrates to
1/1 = constant [50]
The flux linkage of any closed circuit having no resistance or e.m.j. re-
mains constant.
Actually every circuit has some resistance, and therefore the flux
linkage of the circuit changes at a rate ~(e - Ri) (eq. 47). In a
linear network having no e.m.f.'s, the rate of change of flux linkage
may be expressed in terms of the time constants of the network (dis-
cussed in the next section of this chapter).
The theorem of constant flux linkage is most useful in situations
where the self-inductance of a circuit or the mutual inductance be-
tween it and another circuit change because of relative motion of parts
of the circuit, or of two circuits, or of magnetic materials. If these
changes occur in a time which is short compared with the time con-
stant of the circuit, the flux linkages remain substantially constant
during the change, and this fact may be used to calculate the currents
after the change in terms of the currents before the change.
A few examples of applications of the theorem will now be given.
1. Suppose that a copper ring carrying no current is situated at
first so that it surrounds a permanent bar magnet. Suppose further
that the ring is then suddenly removed from the magnet. The current
in the ring immediately after removal will be of such magnitude and
in such direction that the magnetic flux passing through the ring is
equal to its previous value; namely, the flux of the permanent magnet.
This current is therefore equal to the flux of the magnet divided by the
inductance of the ring. Ultimately, of course, the current decays to
zero.
THEOR~M O~" CONS1'ANT LINI(AGE 15
2. Next consider a relay having a hinged iron armature and a wind-

ing connected to a battery. Initially the armature is in contact with
the magnet pole, and the current in the winding has its steady-state
value, equal to the battery voltage divided by the resistance of the
circuit. Suppose now that the armature is suddenly drawn away from
the magnet. This action causes the current in the relay coil suddenly
to increase to a value which maintains the same flux through the coil
in spite of the decreased permeance of the magnetic circuit. (See
Fig. 10.) While the armature is held away from the magnet, the cur-
E " .Armature pulled away

II _
E
R "'-Armature released
FIG. 10. Changes of current in relay coil when armature is suddenly drawn away
from magnet and later is allowed to fall back,
rent slowly decreases to its steady-state value, and the flux also de-
creases. If now the armature is released, allowing it to be drawn
back to the magnet, the current suddenly decreases to a value which
holds the flux at the reduced value in spite of the increased permeance.
Thereafter the current builds up· slowly to its steady-state value and
the flux builds up with it.
3. For the next example, consider a transformer having primary
self-inductance L 11, secondary self-inductance L 22 , and mutual in-
ductance L 12 . Initially the secondary winding is short-circuited, the
primary winding is open, and the currents are zero. Suddenly a di-
rect voltage E is applied to the primary terminals and is maintained
constant. What are the currents if resistance is neglected? The ini-
tial flux linkage of the closed secondary circuit is zero and, in the
absence of resistance, this value is maintained.
[51]
This requires a constant ratio of secondary current to primary current:
~2 · =- L 12
-~1
L 22
• [52]
The primary flux linkage is

1/11 = L 11i 1 + L 12i 2 [53]
which, by use of eq. 52, becomes
1/11 = (L i::)
11 - i1 = L soi 1 [54]
The apparent inductance of the primary winding with the secondary

winding short-circuited is thus seen to be the same when a direct
voltage is suddenly applied to the primary terminals as when a steady
alternating voltage is applied (eq. 31). 'Vith no resistance in the
windings and voltage E applied to the primary terminals, the primary
current would build up at a constant rate
di 1 E
de = L
so
[55]
and the secondary current would build up proportionally to the pri-

mary current (eq. 52). With resistance present, the rate of increase
of current is initially the same as with no resistance but falls off there-
after.
Primary
coil
~ ..J!condary
~ coil
FIG. 11. Flux paths in a short-circuited iron-core transformer with coils on

opposite legs.
If the transformer has a closed secondary winding with no resist-

ance, no flux can be forced through it; therefore the flux caused by
the primary current is forced into leakage paths, as illustrated in
Fig. 11 for an iron-core transformer with primary and secondary coils
on opposite legs of the core. The leakage paths have a higher re-
luctance than the path through the core which the flux could follow
if the secondary winding were open. Therefore more primary current
is required to produce a given amount of flux with the secondary wind-
ing closed than with the secondary winding open, and consequently
the primary inductance (linkage per unit current) is lower on short
circuit than on open circuit. This is true not only for alternating
voltage but also, initially, for suddenly applied direct voltage.
After the steady state has been reached, however, following the
application of direct voltage, the currents and linkages of both wind-
ings become the same as if the secondary circuit were open, and the
driving-point inductance, which is the primary flux linkage per ampere
THEOREM OF CONSTANT LINKAGE 17
of primary current, becomes the same as the open-circuit inductance.

Thus, when direct voltage is suddenly applied to a linear transformer
with short-circuited secondary winding, the inductance gradually in-
creases from the short-circuit value to the open-circuit value.
4. Now consider the same transformer under different operating
conditions. Initially the secondary winding is open and the primary
winding is excited in the steady state with alternating current. Then
the secondary winding is suddenly short-circuited. What are the
currents?
Before application of the short circuit, the secondary flux linkage
varies sinusoidally, and, in general, this linkage will not be zero at
the instant of short circuit. If the linkage is not zero, some flux is
"trapped" in the secondary winding by the short circuit and must
remain constant thereafter, provided that the secondary resistance is
negligible. At the same time, the primary linkage must continue to
vary sinusoidally in order to give a countervoltage equal and 'opposite
to the applied voltage. The maintenance of constant secondary link-
age requires a direct component of secondary current. The sudden
appearance of this component of secondary current would alter the
primary linkage did not a direct component of primary current appear
simultaneously. The production of the sinusoidal alternating primary
linkage requires an alternating component of primary current. This
would produce an alternating secondary linkage if it were not opposed
by an alternating component of secondary current. The alternating
components of current are sustained by the applied voltage and are
the steady-state currents. The direct components of current are
transient. Their initial values depend on the trapped flux, which, in
turn, depends on the instant of the cycle at which the short circuit
occurs. Thus the offset of the current waves depends upon the in-
stant of switching.
The principle of constant flux linkage has been applied very effec-
tively by R. E. Doherty in determining the short-circuit currents of
rotating machines.2 ,6 , l l A qualitative discussion of three-phase short
circuits of synchronous machines is given later in this chapter.
Before leaving the subject of the theorem of constant flux linkage,
the reader may be interested to note that the theorem is analogous to
the principle of conservation of momentum in mechanics. In the usual
analogy between electricity and mechanics, inductance is analogous
to mass, and current to velocity; accordingly, the product of induct-
ance and current, which is flux linkage, is analogous to the product of
mass and velocity, which is momentum.
Fundamental concepts of time constant. Although the theorem of

constant flux linkage is useful for finding the initial values of transient
currents, it does not give their rates of decay. In linear static circuits
io
OJ.---- T ----+- T ----..J

FIG. 12. Exponential decay of current.
with no capacitance the transient currents consist of components each

of which decays exponentially. The rate of decay of each component
can be expressed in terms of a time constant, thus:
it = itOE-t/T [56]
Here it is the transient current at any time t, itO is the initial value of
it, and T is the time constant. At time t = T, the transient current
has decreased to E-1 = 0.368 of its initial value. The rate of decay at
any time is
die itOE-t/T
----= [57]
dt - T T
If this rate continued without change, the current would decay to
zero in a time interval T.
The time constant of an exponential curve may be defined in either
of two equivalent ways. It is either
1. The time in which the variable decreases to 0.368 of its initial
value, or
2. The time in which the variable would decrease to zero if it
continued to decrease at its initial rate.
Both these definitions are illustrated in Fig. 12. At time T the
CONCEPTS OF TIME CONSTANrr 19
curve itself has decreased to 0.368 of its initial value and at the same
time the tangent to the curve at the initial point intersects the hori-
zontal axis. In the interpretation of these statements, any point on
the curve may he taken as the initial point.
ilio
1 I'.~
"
""
0.368 ;;;-T~'
"-
-, --
0.1
-, " "-
--
"'" --
"'
"<,
-,
FIG. 13. Plot of exponential on semilog paper.
If log i is plotted against t, or - what amounts to the same thing

- if i is plotted against t on semilog paper, the exponential curve be-
comes a straight line (Fig. 13).
Simple R-L circuit. In a series circuit of resistance R ohms and
inductance L henrys, the transient current i is given by the solution
of the differential equation
R ~" + Ldi
-dt =0 [58]
which, as may be verified by substitution, is

i = iOE-R t / L [59]
Here the time constant is
L
T = R seconds [60]
The time constant may be regarded also as

Li 1/1
T = Ri = Ri [61]
20 SYNCHRONOUS lVIACHINES
If the circuit contains a consumi e.m.j. E volts, the steady-state

current is 18 = E/R, and the actual current i is the sum of the steady-
state and transient components.
i = III + (10 - 18)E-RtIL [62]
where lois the initial value of total current (steady-state plus
transient).
If the circuit contains a sinusoidal e.m.f., the steady-state current
is found from well-known methods, and again the actual current is
the sum of the steady-state and the transient components.
If the circuit contains a varying e.m.f. whioh is any known function
of time e(t), the current can be found either graphically or by a point-
by-point calculation. The differential equation is
Ri + L ~~ = e(t) [63]
whence
di e(t) - Ri [e(t)/R] - i
dt - L L/R
di i,(t) - i
= -- [64]
dt - T T
At every instant the rate of change of current is equal to the difference
it between the steady-state current i 8 (t) which would be reached if
the e.m.f. remained at its instantaneous value and the actual current i,
divided by the time constant T.
Wagner'! has given a simple graphical solution of eq. 64, which he
calls the follow-up method: The coordinates of the desired curve are
current i and time t. On the same coordinates a curve of is(t) =
e(t)/R is first plotted against i, which curve, however, is shifted to the
right by a distance T (see Fig. 14). From a point representing the
initial current 1 0 at t = 0, a straight line is drawn to the point on
the curve of e(t)/R representing e(O)/R, which point, as already men-
tioned, is at a distance T to the right of the origin. The slope of this
line is therefore that given by eq. 64 and represents the initial slope
of the desired curve of i against t. The slope of the curve is assumed
to be constant for a short time interval ilt. At the point on the straight
line at t = ilt, a new value of slope is determined by drawing a second
straight line from this point to a point on the curve of is at t= T + ilt.
The new value of slope is assumed to be constant during a second in-
terval ilt, and so on. The process is illustrated in Fig. 14, with a
modification which will now be described.
EXAMPLE 1 21
Some cumulative error enters this process if, as described above,

the slope di/dt at the beginning of a time interval I1t is assumed equal
to the average slope throughout the interval. A much more accurate
value of average slope during an interval is obtained if the straight
e(O)
R""
eClAt)
-R- 3t
5 4t 10
6 7 9t
8 9
5f 6t
FIG. 14. The "follow-up method" for graphical determination of current in an

R-L circuit with varying e.m.f.
line is drawn to a. point on the curve of e(t)/R at the middle, instead

of at the beginning, of the interval. See, for example, the line in
Fig. 14 labelled "Avg. slope in 1st interval." This construction is
equivalent to using a time constant T + !l1t instead of T.
In point-by-point calculation of the current, the time constant is
modified as just described, so that eq. 64 becomes:
l1i [e(t + !l1t)/R] - i

[65]
I1t T + !l1t
EXAMPLE 1
To illustrate the accuracy of point-by-point calculations of transient
current, compute the decaying current in a circuit having a resistance of
1 ohm, an inductance of 1 henry, and no e.m.f., if the initial current is
1 ampere.
Exact solution. The time constant is
L/R = 1/1 = 1 sec.
The exact solution for the current is

(a)
Values, read from a table of exponential functions, are tabulated in the
last column of Table 1.
TABLE 1
POINT-By-POINT COMPUTATION OF EXPONENTIAL CURVE
(EXAMPLE 1)
o 1.000 -0.182 1.000

0.2 0.818 -0.149 0.819
0.4 0.669 -0.122 0.670
0.6 0.547 -0.100 0.549
0.8 0.447 -0.081 0.449
1.0 0.366 -0.067 0.368
1.2 0.299 -0.054 0.301
1.4 0.245 -0.045 0.247
1.6 0.200 -0.036 0.202
1.8 0.164 -0.030 0.165
2.0 0.134 0.135
Point-by-point solution. Let ~t = 0.2 sec. Then T + ~t/2 = 1.1 sec.

Equation 65 becomes
~i i
-=-- (b)
~t 1.1
~i = _ i ~t = _ 0.2 i = i
(c)
1.1 1.1 5.5
and
in = in- 1 + (~i)n (d)
The computation is shown in Table 1.

The accuracy is very good with as few as five points per time constant.
The maximum error in i is -0.002. If, however, the uncorrected value
T = 1 had been used instead of 1.1 in eq. b, the maximum error in i with
at = 0.2 would have been -0.038, or 19 times as large.
The point-by-point calculation of transients in circuits with satu-
rated iron will be illustrated in Chapter XIII.
Two coupled circuits. Consider two linear, inductively coupled cir-
cuits having resistances R 1 and R 2 , self-inductances L 11 and L 22 , and
mutual inductance L 12 • The transient currents in and i t 2 satisfy the
TIME CONSTANTS OF COUPLED CIRCUITS 23
differential equations
.
Rt'ttl + L 11 din di
di + L 12 dit 2 = 0 [66a]
.
R2'tt2 + L 22 didit2 + L 12 di
din
= 0 [66b]
and are of the form

itt = IlaEPat + I 1bE
Pbt
[67a]
it2 = 12aEPat + 12 bEPbt [67b]
where Pa and Pb are the roots of the determinantal equation
«. + L 11P L 12 P I= 0 [68]
IL 12P R2 + L22P
or
(R1 + L 11P) (R2 + L 22P) - L 122p 2 = 0 [69]
(L 11L22 - L 12 2 ) p 2 + (R 1L22 + R 2L11 )P + R 1R2 = 0 [70]
The roots are readily found, for given numerical values of R 1, R 2 ,
L l 1 , L 22 , and L 12 , through use of the formula for the roots of a quad-
ratic equation. The coefficients Ila, lib, 12a, and I 2b can be determined
from a knowledge of the initial values of in, i t 2 , and their derivatives,
which can be found as if there were no resistance.
The flux linkages of the two circuits, being linear functions of the
two currents, consist of exponential components having the same
values of p that the currents have. The initial values of the linkages
can often be found from the theorem of constant flux linkage.
The time constants of the transient current components of eqs. 67
are
1 1
T« = - - and Tb = - - [71]
Pa Pb
These time constants of the coupled circuits can be found in terms of
the time constants of the two separate circuits,
Tt = L11 and [72]

RI
as roots of the quadratic equation'"
[73]
where (J' is the leakage coefficient defined by eqs. 43 and 44. Equa-
tion 73 follows directly from eq. 70 and the substitutions indicated by

eqs. 71 and 72.
The roots of eq. 73 can be found when the values of T 1, T 2 , and a
are given. If (1 is small, the roots are approximately
Ta ~ T 1 + T2 [74a]
and
[74b]
EXAMPLE 2
What are the time constants of t\VO coupled R-L circuits if the leakage
coefficient is 0.15 and the time constants of the two separate circuits are
5 sec. and 1 sec.?
Solution. Given: T 1 = 5 seo., T 2 = 1 sec., (J = 0.15. Equation 73
becomes
T2 - 6T + 0.75 = 0 (a)
By the quadratic formula, the roots are
Ta = 3 + V9 - 0.75 = 5.872 sec. (b)
and
Tb = 0.75 = 0.128 sec. (c)
3 + V9 - 0.75
By the approximate equations (74), the time constants are
Ta =5+1 = 6 sec. (d)
and
T b = 0.15 X 5 X 1 = 0.125 sec. (e)
5+1
The errors of the approximate formulas are seen to be small if (1 is as small
as 0.15.
Reactances of synchronous machines. The concepts of inductive

reactance and of time constant, which have been developed for static
circuits, will now be extended to synchronous machines, where some
circuits move with respect to others. The reactances discussed ini-
tially will be specifically those of a three-phase, salient-pole rotating-
field machine without saturation or iron loss. At the same time, the
differences between salient-pole and round-rotor machines will be
pointed out. Later, the effects of saturation will be discussed.
The impedances of three-phase machines are classified, according
to the method of symmetrical components, into positive-sequence,
negative-sequence, and zero-sequence impedances. In the determina-
SYNCHRON()US REACTANCE 25
tion of any of these, the rotor circuits are closed but not excited, the
rotor is turned forward at synchronous speed, current of the proper
sequence is applied to the armature windings, and the armature termi-
nal voltage of the. same phase sequence as the current is found. The
ratio of the voltage to the current is, of course, the impedance. The
reactance is by far the most important component of every synchro-
nous-machine impedance, and the resistance is negligible for most
purposes.
Anyone machine has only one zero-sequence reactance and one
negative-sequence reactance, * but it has several different positive-
sequence reactances, depending upon the angular position of the rotor
and upon whether the positive-sequence armature currents are steady
or are suddenly applied.
Direct-axis synchronous reactance Xd. It is well known that

positive-sequence armature currents in a polyphase armature winding
set up a rotating magnetic field in the air gap. More accurately de-
scribed, this field consists of waves of m.m.f. and of flux, including
both the space fundamental and various space harmonics, some rotat-
ing forward and some backward at various speeds; but the most im-
portant of these waves are the space fundamentals which rotate for-
ward at synchronous speed with respect to the armature and are
therefore stationary with respect to the field structure (rotor). Given
armature currents produce the same fundamental m.m.f. wave, re-
gardless of the angular position of the rotor; but the fundamental flux
wave varies greatly with the rotor position. If the rotor is so rotated
that the direct, or pole, axis stays in line with the crest of the rotating
m.m.f. wave, a path of high permeance is offered (the flux paths are
approximately as shown in Fig. 15a), and the fundamental flux wave
has its greatest possible magnitude for a given armature current. Ac-
cordingly, the total flux linkage of each phase winding of the armature
(including both the flux which crosses the air gap and that which does
not) has the greatest possible value for a given current in the winding,
and the armature inductance and inductive reactance are greater than
they would be for any other position of the rotor. The flux linkage of
an armature phase per ampere of armature current under these con-
ditions is the direct-axis synchronous inductance Ld. The direct-axis
synchronous reactance is Xd = WLd.
Note that, under the stipulated conditions, the flux is stationary
with respect to the field structure, and that, therefore, it makes no
*As brought out on pp. 67-69, the negative-sequence reactance varies somewhat
with the harmonic content of the current.
26 l:lYNCHRONOUS MACHINBS
Stator m.m.f. wave

(steady state)
, I
~~-+---~
(b) Xq
Stator::--_.~~~~~ill~~~F~
winding
o
o
8""---
I I Field winding
o ~short-circuited
~I-----+---~ -i---_L
(d) Xq '
Amortisseur
winding
-1-
. ------I
(f) x q"
FIG. 15. Flux paths for various reactances of a salient-pole synchronous machine.
(a) Direct-axis synchronous reactance. (b) Quadrature-axis synchronous reactance .
(c) Direct-axis transient reactance. (d) Quadrature-axis transient reactance .
(e) Direct-axis subtransient reactance. (f) Quadrature-axis subtransient reactance.
(From Ref. 34 by permission.)
SYNCHRONOUS REACTANCE
difference whether the field winding is open or closed - the field flux
linkage is constant in either case. With respect to the armature, how-
ever, the flux is moving at synchronous speed. Consequently the flux
linkage of each armature phase winding varies sinusoidally with time.
The time variation of flux linkage is in phase with the time variation
of armature current because the space-fundamental flux wave is in
phase with the space-fundamental m.m.f. wave. Therefore the ratio
of flux to current is the same regardless of whether instantaneous,
maximum, or effective values are used. The armature voltage induced
by change of flux linkage is in quadrature with the flux and current.
The ratio of this voltage to the current, under the stipulated condi-
tions, is the direct-axis synchronous reactance Xd.
One method of measuring Xd has already' been suggested: Apply sus-
tained positive-sequence currents to the armature, drive the rotor
forward at synchronous speed, in such position that its direct axis
coincides with the crest of the space-fundamental forward-rotating
m.m.f. wave, and measure the sustained positive-sequence armature
voltage in quadrature with the current.
There is another method of measuring Xd, which is easier to carry
out than the method just described. The rotor is driven at rated
speed with the field winding excited, and the open-circuit armature
voltage is measured. Then a three-phase short circuit is applied to
the armature terminals and the sustained armature current is meas-
ured, the field current being the same as before. The ratio of open-
circuit voltage* to short-circuit current is Xd. This method does not
require a separate source of three-phase power, and the armature
m.m.f. is automatically aligned with the direct axis of the rotor.
That the second method gives the same result as the first in an un-
saturated machine can be shown by the principle of superposition,
which is valid for linear circuits. In Fig. 16a the field winding of the
generator is excited by application of a direct voltage, producing a
positive-sequence alternating voltage E at the open-circuited arma-
ture terminals. If now an external alternating e.m.f. E is applied to
the armature, exactly balancing the open-circuit voltage, no armature
currents flow. This subterfuge is necessary because the principle of
superposition is valid only when the impedances of the circuit are con-
stant: we cannot have an open circuit at one time and a short circuit
at another time. Any current (or voltage) in Fig. 16a, resulting from
the application of field and armature e.m.f.'s together, is equal to the
"If the machine is saturated, the voltage read from the air-gap line should be
used instead of the open-eircuit voltage read from the no-load saturation curve.
The result is then the unsaturated value of Xd.
28 SYNCHRONOlTS MACHINES
sum of the currents (or voltages) at the same place resulting from the
application of the field and armature e.m.f.'s separately (Figs. l6b and
l6e). In Fig. l6a the armature currents are zero; therefore the arma-
ture currents in Fig. l6b must be equal and opposite to the correspond-
ing armature currents in Fig. l6e. The armature voltage E in Fig. l6a
is likewise the sum of the armature voltages.in Figs. 16b and c, which
Armature
(a) open-
circuited
Armature
(b) short-
circuited
+
Voltage
applied to
armature
FIG•. 16. Application of the principle of superposition to show the equivalence of

two different methods of measuring Xd: first, open-circuit and sustained short-circuit
tests, shown in (a) and (b), respectively; second, sustained application of armature
voltage to unexcited machine, shown in (c).
are zero and E, respectively. The field current If in Fig. 16a is the
sum of the field currents in Figs. 16b and C, which are If and zero,
respectively. Therefore, the reactance Xd determined from the scalar
ratio of the voltage E in Fig. 16a to the current Is in Fig. 16b is the
same as the value of Xd determined from the same voltage E and cur-
rent Is in Fig. l6c, the field currents being equal in Figs. 16a and b.
SYNCHRONOUS REACTANCE 29
Figure 16c embodies the conditions used in defining Xd; Figs. 16a and
b portray the alternative method of measuring Xd by means of open-
circuit and short-circuit tests, which is the usual test method.
Quadrature-axis synchronous reactance X q • It has already been

noted that the magnitude of the space-fundamental wave of air-gap
flux depends on the position of the rotor with respect to the space-
fundamental wave of m.m.f. and is greatest when the direct axis of
the rotor coincides with the crest of the m.m.f. wave. On the other
hand, the flux wave is smallest when the quadrature, or interpolar,
axis of the rotor coincides with the crest of the m.m.f. wave. The
flux paths for this condition are shown in Fig. 15b. Under .this con-
dition the armature flux linkage per armature ampere is the quadrature-
axis synchronous inductance L q • The quadrature-axis synchronous
reactance is X q = wLq • This reactance is also equal to the component
of armature voltage in quadrature with the armature current, divided
by the current.
In round-rotor machines Xd and X q are almost equal. However, X q
is slightly less than Xd because of the effect of the rotor slots.
Note that both Xd and X q are defined with the field pole cores in a
symmetrical position with respect to the rotating m.m.f. wave, There-
fore, in neither case is there any magnetic torque tending to pull the
rotor into a different position. No mechanical power is developed in
either case, and the voltage is purely reactive in both cases. With the
rotor in any intermediate position, the flux wave would not be in
space phase with the m.m.f. wave, and therefore the voltage would
not be in quadrature with the current. Motor or generator action
would then occur. A torque, called a reluctance torque, would be
developed, tending to move the rotor into the position in which its
direct axis coincides with the m.m.f. wave. (See "Power-angle curves
of a salient-pole machine," p. 80.)
For the measurement of X q , a procedure suggests itself similar to the
first one for measuring Xd. Positive-sequence currents would be ap-
plied to the armature, and the rotor would be driven forward at syn-
chronous speed with the quadrature axis in line with the crest of the
rotating m.m.f, wave. Under these conditions, the sustained positive-
sequence armature voltage would be measured, and the ratio of arma-
ture voltage to armature current would be X q •
The procedure just described is very difficult to carry out because,
on account of reluctance torque, the rotor is in unstable equilibrium
when the quadrature axis coincides with the crest of the m.m.f. wave.
A more feasible procedure is the slip test, in which the rotor is driven
at a speed differing slightly from synchronous speed as determined

from the frequency of the applied armature currents. These currents
are modulated at slip frequency by the machine; they have maximum
amplitude when the quadrature axis is in line with the m.m.f. wave
and minimum amplitude when the direct axis is in line with the m.m.f.
wave. Because of impedance in the supply line, the armature voltage
(a)
(b)
Direct axis Quadrature axis
FIG. 17. Oscillogram for determination of Xd and X q by the "slip test."

(a) Armature voltage, (b) armature current.
max. voltage min. voltage

Xd = Xq =
min. current max. current
is usually modulated at slip frequency too, being greatest when the

current is least and vice versa. Figure 17 represents an oscillogram
showing these variations of current and voltage. If the slip is small,
the maximum and minimum values of voltage and current can be read
from an indicating voltmeter and ammeter. Quadrature-axis synchro-
nous reactance is the ratio of minimum voltage to maximum current.
From this same test the direct-axis synchronous reactance is found as
the ratio of maximum voltage to minimum current.
In the slip test the field winding should be open so that slip-fre-
quency current is not induced in it.
Armature leakage reactance Xl is the reactance resulting from the

difference between the total flux produced by the armature current
TRANSIENT REACTANCE 31
acting alone and the space fundamental of the flux in the air gap. It
consists of three parts, due to the following fluxes:
1. Slot-leakage flux.
2. End-winding flux.
3. Differential-leakage flux, which comprises space harmonics of
air-gap flux that induce fundamental-frequency armature voltage.
Armature leakage reactance is a calculated quantity and cannot be

measured. It forms a part of the direct-axis and quadrature-axis
synchronous reactances, as well as of all other positive- and negative-
sequence reactances.
Potier reactance Xp is discussed on p. 122.
Direct-axis transient reactance Xd'. The conditions used in defining

direct-axis transient reactance are the same as those used in defining
direct-axis synchronous reactance except that the positive-sequence
armature currents are suddenly applied and the positive-sequence
armature voltage is measured immediately after application of the
current instead of after the voltage has reached its steady-state value.
In both cases the rotor is rotated forward at synchronous speed, with
its direct axis in line with the crest of the armature-m.m.f. wave and
with the field winding closed but not excited. (For the time being,
assume that there are no other rotor circuits.) Abrupt application of
the armature currents causes the sudden appearance of a m.m.f. oppo-
site each field pole, tending to establish flux through the pole core.
Such flux would link the field winding and is opposed by induced field
current, tending to maintain the flux linkage of the field winding con-
stant at zero value. By the theorem of constant flux linkage, at the
instant immediately after application of the armature currents the
field linkage is still zero. Therefore the only flux that can be estab-
lished immediately is that which does not link the field winding but
rather passes through low-permeance leakage paths, largely in air, as
shown in Fig. 15c. Under these conditions the flux per ampere is
small and is defined as the direct-axis transient inductance Ld' . The
direct-axis transient reactance is Xd' = WLd'. It is always less than
the direct-axis synchronous reactance: Xd' < Xd.
The synchronous machine, under the conditions described, is analo-
gous to a transformer with a short-circuited secondary winding and
with direct current suddenly applied to the primary winding. The
armature winding is the primary; the field winding, the secondary.
The direct-axis synchronous reactance Xd is analogous to the primary

self-reactance, or open-circuit reactance, XII; and the direct-axis tran-
sient reactance Xd' corresponds to the primary short-circuit reactance
2/
X sc = <TXII = XII - (X12 X22). Hence Xd' = <TXd. There is a differ-
ence between the transformer and the synchronous machine, however:
in the transformer, the frequency is the same in both windings; whereas,
in the machine, positive-sequence fundamental-frequency alternating
current in the armature corresponds to direct current in the field
winding; that is, both produce forward-rotating magnetic fields. Sud-
denly applied armature currents induce unidirectional field current
which opposes change of the field flux linkage. As time passes, how-
ever, unidirectional flux linkage of the field winding slowly builds up,
ultimately reaching a steady value, and, at the same time, the field
current decays to zero. The steady condition that is finally reached
if the amplitude of armature currents or voltages are held constant is
the condition under which direct-axis synchronous reactance was
defined, and under this condition it does not matter whether the
field winding is open or closed because its linkage is constant in either
case.
If positive-sequence armature current I is suddenly applied, the
positive-sequence armature voltage is initially xd'I, but gradually in-
creases to xdI. If positive-sequence voltage V is suddenly applied,
the positive-sequence current is initially V /x a', but gradually decreases
to V /Xd. The change is exponential with a time constant of the order
of several seconds (discussed later in this chapter under the heading
"Time constants of synchronous machines").
Direct-axis transient reactance, like direct-axis synchronous react-
ance, can be measured by means of open-circuit and short-circuit tests
with the field winding excited. This topic will be discussed more fully
under the heading "Sudden Three-Phase Short Circuit."
Direct-axis subtransient reactance Xd". In defining Xd' we assumed

that there were no rotor circuits except the main field winding, There
may be additional rotor circuits on both axes, however. Many salient-
pole machines have amortisseur (damper) windings, shown in Figs. 15e
and 15/. A few salient-pole machines have field collars. In round-
rotor machines the solid steel rotor core furnishes significant paths for
eddy currents.
If positive-sequence armature currents are suddenly applied in such
time phase that the crest of the rotating m.m.f. wave is in line with
the direct axis of the rotor, transient currents are induced in the addi-
tional direct-axis rotor circuits as well as in the main field winding.
33
These transient currents oppose the armature m.m.f., and initially

they are strong enough to keep the flux linkage of every rotor circuit
constant at zero value. The additional rotor circuits are situated
nearer the air gap than the field winding is. Consequently, the flux
set up by the armature current is initially forced into leakage paths of
smaller cross-sectional area and lower permeance than would be the
case if the only rotor circuit were the field winding (see Fig. 15e).
Under these conditions the armature flux linkage per armature ampere
is the direct-axis subtransient inductance Ld". The direct-axis subtran-
sient reactance is Xd" = wLd" . The subtransient inductance and react-
ance are always smaller than the corresponding transient quantities.
Even in water-wheel generators without amortisseur windings Xd" is
about 15% smaller than Xd' because of the effects of saturation* and
of such closed circuits as those through the pole rivets.
Invariably the decrement of the currents in the additional rotor
circuits is very rapid compared with that of the field current, so that
after a few cycles all rotor currents except the field current have be-
come negligibly small. The effective armature reactance has then in-
creased from the subtransient to the transient value. The term
subtransient is intended to convey the idea of a transient of very short
duration.
The transient reactance of machines having amortisseur windings is
practically equal to that which would exist if the amortisseur windings
were removed. It can be found, however, without removing the
amortisseurs, by plotting a curve of reactance against time (or current
against time after sudden application of a short circuit) and extrapo-
lating the slowly changing part of this curve (which does not include
the first few cycles) back to the instant at which current or voltage
was applied. The procedure is explained in more detail under the
heading "Sudden Three-Phase Short Circuit." The same procedure
is applicable to round-rotor machines, the additional rotor circuits of
which obviously cannot be removed.
Quadrature-axis transient reactance x q ' and subtransient reactance

x q " . These quantities are defined in a manner similar to Xd' and Xd",
respectively, except that the suddenly applied positive-sequence arma-
ture current is in such time phase that the crest of the space-funda-
"Rudenberg''? has shown that, even if there is no direct-axis rotor circuit except
the main field winding, the effect of saturation is to produce a more rapid decay
of armature current in the early part of the transient than in the following part,
which is practically exponential. The whole transient can be approximated by
the sum of two exponentials with different time constants just as in the case of two
rotor circuits.
mental m.m.f. wave is in line with the quadrature axis of the rotor
instead of the direct axis.
In a machine having laminated salient poles and no amortisseurs,
the changing flux can pass sideways through the field coils without
linking them and inducing field current. The flux paths (Fig. 15d)
are the same as those for a steady flux (Fig. 15b). This is true also
of a machine with amortisseurs after the current in them has become
negligible. Consequently, for salient-pole machines, x q ' is equal to X q •
Because field current is induced by a changing direct-axis flux, though
not by a changing quadrature-axis flux, Xd' is less than x q ' . Amortis-
seur windings restrict the quadrature-axis flux initially to low-perme-
ance paths, as shown in Fig. 15f, thereby making the quadrature-axis
subtransient reactance xq " much less than the transient reactance
x q ' . However, x q " is usually slightly greater than Xd". In a salient-
pole machine without amortisseur windings, x q " is equal to x q ' , which,
as already stated, is equal to X q .
In a machine with a solid round rotor, the changing reactance due
to quadrature-axis eddy currents can be represented fairly well as the
sum of two exponentials, one of which is rapid, the other slow. The
slower transient, extrapolated back to zero time, gives x q ' ; the initial
total reactance is x q " . In such machines the value of x q ' is between
the values of Xd' and X q , and x q " is slightly smaller than x q ' .
Negative-sequence reactance X2. Let the unexcited field structure

be rotated forward at synchronous speed with all rotor circuits closed
while negative-sequence currents are applied to the armature winding.
The negative-sequence currents produce a magnetomotive force ro-
tating backward at synchronous speed with respect to the armature
and hence backward at twice synchronous speed with respect to the
rotor. Seen from the rotor, the m.m.f. alternates at twice the rated
frequency of the machine. Currents of twice rated frequency are .in-
duced in all rotor circuits, keeping the flux linkages of those circuits
almost constant at zero value. The flux due to the armature current
is forced into paths of low permeance which do not link any rotor cir-
cuits. These paths are the same as those for subtransient reactance
(shown in Figs. 15e and 15i). The armature flux linkage per armature
ampere under these conditions is the negative-sequence inductance L 2 -
The negative-sequence reactance is X2 = wL2 •
Since the m.m.f. wave moves at twice synchronous speed with re-
spect to the rotor, it alternately meets permeances of the two rotor
transient reactances Xd " an d x q " • H ence
axes, correspon dimg to BUbtransi
the value of X2 lies between Xd" and Xq". It is usually taken as the
RESISTANCES OF MACHINES 35
arithmetical mean,
[75]
The test for measuring X2 is obvious from the defining conditions.
Zero-sequence reactance Xo. If zero-sequence currents are applied

to the armature, there is no space-fundamental m.m.f. Hence the
reactance is small and is hardly affected by the motion of the rotor.
There is, in general, a third space harmonic of air-gap m.m.f. which is
stationary but pulsating. This m.m.f, is opposed by currents induced
in the closed rotor circuits; therefore not much air-gap flux is produced.
The zero-sequence currents produce slot-leakage, end-winding-leakage,
and differential-leakage fluxes, but these are not the same as those
produced by positive-sequence currents. Altogether very little flux
is set up, and the zero-sequence reactance is the lowest of the synchro-
nous-machine reactances - even lower than Xd".
The actual value of Xo varies through a wider range than that of the
other reactances and depends upon the pitch of the armature coils.
The reactance is least for a pitch of tvyo-thirds because then each slot
has two coil sides carrying equal and opposite currents.
Zero-sequence reactance is readily measured by connecting the three
armature windings in series and sending single-phase current through
them. The ratio of terminal voltage of one phase winding to the cur-
rent is the zero-sequence impedance, which has very nearly the same
magnitude as the zero-sequence reactance.
Resistances of synchronous machines. Positive-sequence resistance

fl. The positive-sequence resistance of a synchronous machine is its
a-c. armature resistance r, which is greater than the d-e, armature
resistance ra because of nonuniform current density and local iron
loss. The positive-sequence resistance is nearly always neglected in
power-system studies, including both studies of normal operation and
stability studies. The 'only exception is the case of a three-phase short
circuit on or near the generator terminals. Neglecting resistance in
this case leads to pessimistic conclusions regarding stability because
the armature copper loss is large enough to make the accelerat-
ing power considerably less than the value calculated neglecting the
loss.
Negative-sequence resistance r2. Under negative-sequence conditions,
currents are induced in all rotor circuits. The negative-sequence
power input to the armature supplies half the copper losses of the
rotor circuits, the other half being supplied mechanically. * As these

losses are much greater than the armature copper loss, the negative-
sequence resistance (which is negative-sequence armature power input
per, phase divided by the square of the negative-sequence armature
current) is greater than the positive-sequence resistance. Its value
depends largely upon the resistance of the amortisseur windings if
such windings are present. Negative-sequence resistance is discussed
more fully in Chapter XIV of this book.
Zero-sequence resistance ro is equal to or somewhat greater than the
positive-sequence resistance, depending upon the rotor copper loss due
to rotor currents induced by zero-sequence armature currents. It is
usually neglected. If the neutral of a Y-connected armature winding
is grounded through a resistor, three times the resistance of the re-
sistor should be added to the zero-sequence resistance of the machine
itself. Such external zero-sequence resistance is not negligible.
Time constants of synchronous machines. During transient condi-

tions in synchronous machines, the currents and voltages have com-
ponents the magnitudes of which change in accordance with one or
more of the following time constants:
Direct-axis transient open-circuit time constant T dO' . If the armature
is open-circuited and if there is no amortisseur winding, the field cir-
cuit is not affected by any other circuit. Under these conditions the
change of field current in response to sudden application, removal, or
change of e.m.f. in the field circuit is governed by the field open-circuit
time constant, or direct-axis transient open-circuit time constant,
which is given by an expression similar to that of any simple R-L cir-
cuit, to wit:
TdO' = LII seconds [76]
R,
where L" is the self-inductance of the field circuit in henrys and RJ
is the resistance of the field circuit in ohms. The field circuit includes
the field winding, the field rheostat (if one is used), the armature of
the exciter, and the connecting wiring.
*The equivalent circuit of an induction motor (see Chapter XIV) is applicable
to a synchronous machine under negative-sequence conditions, the slip being 2
per unit. The rotor branch of the equivalent circuit has resistance R r /8, where R;
is the rotor resistance and s the slip. Rr/s may be split into two parts in series,
one having resistance R-, the other R; (1 - 8)/8. The loss in the resistances has the
following significance: I r2R r /8 = electric power input of rotor, I r2R r = rotor copper
loss, and I r 2R r (1 - 8)/8 = mechanical power output of rotor. Putting 8 = 2 into
these expressions proves the statement.
TIME CONSTANTS OF l\lACHINES 37
The amplitude of the open-circuit a-c. armature terminal voltage
of an unsaturated machine is directly proportional to the field current
and, therefore, changes with the same time constant as the field
current.
This time constant is greater than any of the other time constants
discussed below; it is of the order of 2 to 11 sec., with 6 sec. as an av-
erage value.
Direct-axis transient short-circuit time constant T d'. If both field and
armature circuits are closed, current in either circuit induces current
in the other. Because of the rotation of one circuit with respect to
the other, however, the two currents are not of the same frequency:
direct current in the field winding is associated with positive-sequence
alternating currents in the armature winding, and direct current in
the armature is associated with alternating current in the field. Each
set of currents has a different time constant. The one which governs
the rate of change of the direct current in the field and of the amplitude
of the alternating current in the armature is the field, or direct-axis
transient, time constant; the one which governs the direct current in
the armature and the alternating current in the field is the armature
time constant.
The field time constant is the ratio of the apparent inductance of
the field circuit when coupled to the closed armature circuits to the
apparent resistance of the field circuit under the same conditions.
Hence its value depends upon the impedance of the armature circuits.
The field open-circuit time constant, discussed above, is a particular
value of field time constant which is valid when the armature is open-
circuited. At the other extreme is the field short-circuit time constant,
which is valid when the armature is short-circuited. On short circuit
the effective inductance of the field circuit is much less than on open
circuit, for the same reason that the short-circuit inductance of a
transformer is less than its open-circuit inductance. Furthermore,
the fact that the' ratio of short-circuit inductance to open-circuit in-
ductance of a transformer is the same from both sides of the trans-
former (being equal to the leakage coefficient a in both cases) enables
us to find the ratio of field-circuit inductance with short-circuited
armature to that" with open-circuited armature. It is equal to the
ratio of armature-circuit inductance with short-circuited field to that
with open-circuited field for suddenly applied positive-sequence alter-
nating armature current. From the discussion of positive-sequence
armature reactances earlier in this chapter, it is clear that the ratio is
L d ' / L d or Xd' / Xd.
The apparent resistance of the field winding, unlike its inductance,
38 SYNCHRONOUS rvlACHINES
is practically the same whether the armature is open- or short-circuited.

At first sight this statement appears to contradict the fact that, in the
case of two coupled circuits at rest with respect to each other, the
apparent resistance of the primary circuit depends on that of the sec-
ondary (see eqs. 29 and 30). Circumstances are different, however, in
the rotating machine, where one winding carries direct current and the
other, alternating current. Here the resistance of the d-c. winding is
much more effective than that of the a-c. winding. Indeed, when the
a-c. winding is short-circuited, its resistance has negligible effect. For
in the a-c. winding the self- and mutual reactances are so much greater
than the resistance that the latter has negligible effect upon the mag-
nitude and phase of the current. In the d-e, winding, however, the
current changes much more slowly than in the a-c. winding, with
the result that 'the inductive drop is comparable with the resistive
drop. *
Because the ratio of the short-circuit inductance of the field wind-
ing to its open-circuit inductance is Xd' /Xd, while its resistance is un-
changed, the ratio of the short-circuit time constant to the open-circuit
time constant is equal to the ratio of the corresponding inductances.
Hence the direct-axis transient short-circuit, time constant is
[77]
It is about one-fourth as large as the open-circuit time constant, or

about 1.5 sec.
If the armature circuit is closed through finite balanced external
impedances, then the field time constant has a value between T d' and
TdO'. If the external impedances are purely reactive, the time con-
stant is given by a modification of eq. 77 in which both Xd' and Xd are
increased by the amount of external reactance per phase. If the
external impedances have much resistance, or if the machine is con-
nected to a network having other large synchronous machines (as is
always true in a stability study), the situation is more complex, but
it will be shown later in this chapter that in every case the field tran-
sient can be computed in terms of the open-circuit time constant.
Direct-axis subtransient time constants T dO" and T d". In a machine
wi th amortisseurs, there are on the direct axis of the rotor two coupled
*In this connection the reader may recall that the apparent reactance of the rotor
circuit of an induction motor, referred to the stator, is independent of the slip, but
the apparent resistance varies inversely as the slip and hence becomes great at
small slips.
TIME CONSTANTS OF MACHINES 39
circuits (the field and amortisseur windings) at rest with respect to

one another but both in rotation with respect to the armature. The
two coupled circuits have two time constants, as discussed earlier in
this chapter. The longer one is the transient time constant; the shorter
one, the subtransient time constant. Both time constants are affected
by the impedance of the armature circuit. If the armature circuit is
open, the time constants have their open..circuit values TdO' and TdO'"
If the armature is short-circuited, the time constants have their short-
circuit values Td' and T d" .
Typical open-circuit values, T dO' = 6 sec. and T dO" = 0.125 sec.
(7.5 cycles), were calculated in Example 2. Typical short-circuit val-
ues are Td' = 1.5 sec., Td" = 0.035 sec. (2 cycles).
Quadrature-axis time constants T qo ' , T q ' , T qo " , and t;". In a ma-
chine with a solid round rotor, the changing amplitude of the quad-
rature-axis component of alternating armature current or voltage can
be represented fairly well by the sum of two exponentials, as stated
previously in the discussion of x q ' and Xq". The time constants of
these exponentials are T qO' and T qO" when the armature circuit is
open; T q ' and T q " when the armature is short-circuited. The value
of T q ' is about one-half that of Td', or about 0.8 sec. The value of
T q " is very nearly equal to Td", or about 0.035 sec. (2 cycles).
In a salient-pole machine T q' is meaningless; but, if the machine
has amortisseurs, T q" is significant and is approximately equal to T d".
Armature short-circuit time constant T a • This time constant applies
to direct current in the armature windings and to alternating currents
in the field and amortisseur windings, both of which are closed. It is
equal to the ratio of armature inductance to armature resistance under
the stated conditions. The armature inductance under these condi-
tions is equal to the negative-sequence inductance L 2 because, although
the rotor currents are of fundamental frequency in the present case
and twice fundamental frequency in the negative-sequence case, in
both cases the flux linkages of the rotor circuits are constant and the
flux created by the armature current is forced into low-permeance
paths, as illustrated in Figs. 15e and f. The armature resistance is
unaffected by the rotor circuits, for reasons already explained, and is
the d-c. value rae Consequently the armature time constant is
_L2_~
Ta - - [78]
ra ""ra
It has a value of about 0.15 sec. with a short circuit on the armature
terminals, but is greatly shortened by any external resistance in the
armature circuit.
SYNCHRONOUS MACHINES
Typical values of reactances, resistances. and time constants of

synchronous machines of various types are listed in Table 2.
TABLE; 2
TYPICAJI CONSTANTS OF 'fHREE-PHASE SYNCHRONOUS MACHINER
(Adapted from Refs. ] 5, 34, and 41)
Synchronous
Turbo- Vlater-'\Vheel
Synchronous Motors
generators Generators
Condensers (general
(solid rotor) (with dampersj]
purpose)
Low Avg. High Low AIJg. High Low Avg. High Low Avg. High
Reactances in per unit
Xd 0.95 1.10 1.45 0.60 1.15 1.45 1.50 1.80 2.20 0.80 1.20 1.50
xq 0.92 1.08 1.42 0.40 0.75 1.00 0.95 1.15 1.40 0.60 0.90 1.10
,
I
Xd 0.12 0.23. 0.28 0.20 6.37 0.50t 0.30 0.40 0.60 0.25 0.35 0.45
xq 0.12 0.23 0.28 0.40 0.75 1.00 0.95 1.15 1.40 0.60 0.90 1.10
Xd " 0.07 0.12 0.17 0.13 0.24 0.35 0.18 0.25 0.38 0.20 0.30 0.40
x q" 0.10 0.15 0.20 0.23 0.34 0.45 0.23 0.30 0.43 0.30 0.40 0.50
Xp 0.07 0.14 0.21 0.17 0.32 0.40 0.23 0.34 0.45
X2 0.07 0.12 0.17 0.13 0.24 0.35 0.17 0.24 0.37 0.25 0.35 0.45
xo· 0.01 0.10 0.02 0.21 0.03 0.15 0.04 0.27
Resistances in per unit
r·
ra(d-c.) 0015 0.005 0.003 0.020 0.002 0.015
rCa-c.) 0.003 O.OOS 0.003 0.015 0.004 0.010
"2 0.025 0.045 0.012 0.20 0.025 0.07
Time constants in seconds
Tdo I 2.8 5.6 9.2 1.5 5.6 9.5 6.0 9.0 11.5
Td' 0.4 1.1 1.8 0.5 1.8 3.3 1.2 2.0 2.8
Td" = T q" 0.02 0.035 0.05 0.01 0.035 0.05 0.02 0.035 0.05
Ta 0.04 0.16 0.35 0.03 0.15 0.25 0.1 0.17 0.3
*xo varies from about 0.15 to 0.60 of Xd", depending upon winding pitch.
[For water-wheel generators without damper windings,
Xd" = 0.S5xd', Xq" = x q' = X q, X2 = (Xd ' + x q ) / 2,
and Xo is as listed.
tFor curves showing the normal value of Xd' of water-wheel-driven generators
as a function of kilovolt-ampere rating and speed, see Ref. 54.
SUDDEN .THREE-PHASE SHORT CIRCUIT

One important application of the reactances and time constants of
synchronous machines is the prediction of short-circuit currents. Con-
versely, oscillograms of short-circuit current can be used to determine
the values of some of the reactances and time constants.
DESCRIPTION OF OSCILLOGRAM 41
A typical short-circuit oscillogram is shown in Fig. 18. A three-
phase short circuit was suddenly applied to the armature terminals of
an unloaded three-phase generator, and the currents of the three arma-
ture phases and the field circuit were recorded.
ia
ic
FIG. 18. Short-circuit currents of synchronous generator. Armature currents

i a , ib, i,, ; field current if.
Description of oscillogram. Armature currents. Each armature

current consists of a symmetrical, or alternating, component and an
asymmetrical, or direct, component. The a-c. components of all three
armature currents have identical envelopes, though the sine waves
bounded by the envelopes are l~O° apart in phase, constituting a
positive-sequence set of currents. The envelope of the a-c. component
is high at the instant of short circuit, and decays ultimately to the
sustained value I.. If I. is subtracted from the a-c. wave, the re-
mainder is found to consist of two exponential components: the tran-
sient component I' with a long time constant T d ' , and the subtransient
component I" with a very short time constant T d" . Thus the ampli-
tude of the a-c. component is given as a function of time t by
I ac -- I 8 + I 0 E-t/7'd' + I 0
I II -tITd"
E [79]
The d-c. components of the three armature currents are generally all
of different magnitudes. They all decrease to zero exponentially with
the same time constant T a. The initial values of these components
depend upon the point in the cycle at which the short circuit occurs
and each one is equal and opposite to the initial instantaneous value
of the a-c. component of the same phase, so that there are no discon-
tinuities in the current waves
-Ie at the instant of short circuit.
I
I The maximum possibleinitial
I
I value of the d-c. component is
I
I -I a equal to the initial value of
I
-----"oIIE----......
--+-=~---_...---
I Real axis the envelope of the a-c. com-
fA ponent. This value can occur
in only one armature phase
at a time. Thus, in Fig. 18,
wave i a is almost completely
offset, while i b is almost sym-
metrical. The initial d-e.
components ofall three phases
FIG. 19. Relation of initial d-e, components can be found by taking the
of short-circuit armature currents (lA, 1B, 10)
to vectors representing initial crest values of projections on the real axis of
the a-c. components (la, Ib, Ie). three vectors 1200 apart and
of length equal to the initial
crest value of the a-c. component (see Fig. 19). The algebraic sum
of the d-e, components is zero, both initially and thereafter, since the
time constant is the same for all. The last two statements assume
that the short circuit is applied simultaneously to all terminals, which
is not always strictly true.
If Xd ' l ~ X q ' l there are also second-harmonic components of armature
current, which are proportional to the d-e, components. The second-
harmonic components do not show in Fig. 18 and are appreciable only
in salient-pole machines without damper windings.
Field current. The field current, like the armature current, consists
of d-e, and a-c. components. The d-c. component may be further sub-
divided into sustained, transient, and subtransient components. The
sustained component, if no voltage regulator is used, is equal to the
field current before occurrence of the short circuit, and is the quotient
SHORT-CIRCUIT CURRENTS 43
of the exciter voltage by the field-circuit resistance. The transient

and sub transient d-c. components of field current decrease with the
same time constants, Td' and Td" respectively, as the corresponding
a-c. components of armature current. The a-c. component of field
a,
current decays with time constant T as do the d-e, components of
the armature currents. The initial crest value of the a-c. component
is equal to the initial value of the transient d-c. component.
Explanation of short-circuit currents. The reasons for the existence

of the various components of armature and field currents just de-
scribed can be explained by the principle of constant flux linkage. 2
Assume that before the occurrence of the short circuit the field wind-
ing is excited and the armature terminals are open-circuited. Then
the flux linkage of the field circuit is constant, but the linkages of the
armature windings vary sinusoidally with time with 1200 phase dif-
ference between adjacent windings. The flux linking each armature
winding at the instant of short circuit is "trapped" and must there-
after remain constant (to a degree of approximation depending upon
the resistance of the windings), The trapped flux comprises a set of
armature poles equal and. opposite in strength to the field poles and
located opposite the places where the field poles were at the instant
of short circuit, just as if the field poles had stamped their images
upon the armature. The armature poles are fixed with respect to the
armature while the field poles are fixed with respect to the field and
rotate with respect to the armature. The strength of both sets of
poles remains constant during the rotation because of the requirement
of constant flux linkages of both armature and field windings,
The maintenance of stationary armature poles requires d-c. com-
ponents of armature current, the relative magnitudes of which in the
three windings depend upon the position of the poles, hence upon the
position of the field poles at the instant of short circuit. In addition
to the d-e, components, a-c. components of armature currents are also
required to maintain the armature poles, for the following reason:
The constant flux from the field poles would continue to produce
three-phase positive-sequence fundamental-frequency flux linkages
with the armature did not three-phase positive-sequence fundamental-
frequency currents flow in the armature to oppose the field m.m.f.
and to prevent any change of armature flux linkage. Therefore three-
phase armature currents are induced. These currents, as is well
known, produce a m.m.f. rotating forward at synchronous speed with
respect to the armature but stationary with respect to the field and
centered on the direct axis of the field. This armature m.m.f, opposes
44 SYNCHRONOUS l\IACHINES
the field m.m.f, and tends to reduce the flux linkages of the field and
damper windings, To prevent such changes of linkages, an increase
of field current is induced and damper currents are also induced.
Thus we have the transient and subtransient d-e, components in the
rotor windings and the corresponding; a-c. components in the arma-
ture windings,
The stationary armature poles would produce a fundamental-fre-
quency alternating linkage of the field and damper windings if funda-
mental-frequency currents did not flow in these windings to prevent
it. The alternating current in the field winding creates a pulsating
m.m.f. which is stationary with respect to the field poles. This m.m.f.
may be resolved into two waves of m.m.f., one rotating forward, the
other backward, with respect to the field poles. The backward-rotat-
ing m.m.f, is stationary with respect to the armature and opposes the
m.m.f, of the d-e, components of armature current. The forward-
rotating m.m.f. travels at twice normal speed with respect to the
armature and induces positive-sequence second-harmonic currents in
the armature in order to prevent any change of armature flux linkage.
However, if the machine has damper windings, currents are induced
in them which almost prevent this field m.m.f, from inducing second-
harmonic armature currents.
Decrements. If the flux linkages remained absolutely constant, the

short-circuit currents would not decay. Actually they have decre-
ments, which ,vere explained in the section on "Time constants of
synchronous machines." The direct-axis short-circuit time constants
Tel' and T d" control the decrement of alternating armature currents
and direct field currents; the armature short-circuit time constant T a
controls the decrement of the direct armature currents and alternat-
ing field current.
Amplitudes. The amplitudes of the various alternating components

of armature current can be found in terms of the direct-axis reactances
Xel, Xd', and Xd" by use of the principle of superposition. This principle
was invoked earlier in this chapter to justify the measurement of
synchronous reactance Xd as the ratio of open-circuit armature voltage
to sustained short-circuit armature current, as shown in Fig. 16. Here
Xd = Ells, or Is = Elxd' where E is the open-circuit voltage and Is
is the sustained short-circuit armature current, both in effective values.
If the armature is suddenly short-circuited, the initial alternating
armature current of a machine without amortisseurs is given by
Is + I' = Elxd'; or the transient reactance is xa,' = EI(I, + If), the
Al.\tIPLITUDES 45
ratio of the open-circuit voltage to the initial short-circuit alternating

armature current. This method of measuring Xd' is shown in Fig. 20
to be equivalent to the method used in defining Xd' . (namely, the sud-
den application of armature voltage to an unexcited machine). In
Armature
Field suddenly
(a) excited short-
circuited
Armature
Field )00---" open-
(b) excited
circuited
+ Is +1'
O
It'
Voltage
Field suddenly
(c) unexcited applied to
armature
FIG. 20. Application of the principle of superposition to show the equivalence

of two different methods of measuring xi: first, open-circuit and sudden-short-
circuit tests, shown in (b) and (a); second, sudden application of armature voltage
to unexcited machine, shown in (c).
Fig. 20b the open-circuit condition is represented, as before, by the

artifice of applied armature e.m.f.'s which balance the generated
e.m.f.'s of the machine. The sudden application of a short circuit to
the armature terminals is represented in Fig. 20a by the sudden con-
nection of a second set of external e.m.f.'s in the armature circuit
in series with the first set and equal and opposite to them, resulting in
46 SYNCH;RONOUS MACHINES
a net terminal voltage of zero. The combination of e.mJ.'s shown in

Fig. 20a results in the same currents as the separate application of the
two sets of e.m.f.'s in Figs. 20b and c. Since the armature currents
are zero in Fig. 20b, the armature currents produced by sudden short
circuit (Fig. 20a) are equal to those produced by sudden application
of armature voltage to an unexcited machine (Fig. 20c), provided this
applied voltage is equal and opposite to the open-circuit voltage of
the excited machine (Fig. 20b).
If the machine has amortisseur windings, both methods give sub-
transient as well as transient components of armature current.
The initial values of the alternating components of armature cur-
rent are given by
E
I, = - [80]
Xd
, E
I, + 10 =-,
Xd
[81]
and
I, + 10, +,10 " E

=---,; [82]
Xcl
Hence eq. '79 for the amplitude of the a-c. component of short-circuit
armature current may be written in terms of reactances as
where E is the open-circuit voltage of an unsaturated machine, or

the voltage read from the air-gap line of a saturated machine. Effec-
tive values are generally used for lac and E. The d-e, component of
armature current is given by
I do -- V2 E cos a E-tfTa
II
[84]
Xd
where a is the switching angle. The effective value of armature cur-

rent, including a-c. and d-e, components, is
I = vlao
2
+ Ido 2
[85]
The greatest possible initial effective value of armature current, ob-
tained with a fully offset wave, is composed of an a-c. component
E
[Oao = -" [86]
Xd
EXAMPLE 3 47
and a d-e, component
v2E
10 de max = --,,- [87]
Xd
and hence is
lomax
_/
= 'v 2 2 v3E
lose + lOde max = --,,- [88]
Xtl
This value, slightly decreased to allow for decrement in the first half-
cycle, is used in figuring the required short-time momentary ratings
of circuit breakers.
Analysis of oscillogram. The oscillogram of short-circuit armature

currents can be separated into d-e, and a-c. components, and the a-c.
components can be separated into sustained, transient, and subtran-
sient components. From the open-circuit prefault armature voltage
and the initial values of the several components of armature current,
the values of the direct-axis reactances Xd, Xd', and Xd" can be calcu-
lated. From logarithmic plots of the transient, subtransient. and d-e,
components, the direct-axis short-circuit time constants T d', T d",
and T a can be calculated.
EXAMPLE 3
The short-circuit oscillogram of Fig. 18 was taken on a water-wheel
generator rated as follows: 56 Mva., 13.8 kv., three-phase, 60 cycles, 112.5
r.p.m., 90% power factor. The field current was set at 128 amp., giving an
open-circuit voltage of 4,600, and a three-phase short circuit was then
applied to the armature terminals, giving the armature and field currents
shown in the figure. The sustained short-circuit armature, current was
800 amp. r.m.s, Find the values of reactances Xd, Xd', and Xd" and time
constants T d ' , T d " , and T a • Plot the r.m.s, values of the a-c. and d-e. com-
ponents and of the total armature current as functions of time, all in per
unit, assuming that the d-e, component has the greatest possible value and
that the internal voltage is equal to rated voltage.
Solution. Currents and time constants. The envelopes of the current
waves were drawn and their ordinates were measured at times of 0, 1, 2, 3, 4,
5, 7, 10, 15, 20, 25, 30, 40, 50, 60, 80, 100, and 120 cycles after the instant of
short circuit. The a-c. (crest value) and d-e. components were calculated as
half the difference and half the sum, respectively, of the upper and lower
envelopes, multiplied by the proper oscillograph scale factors. The a-c.
components of the three armature currents were averaged. The resulting
components of armature current are plotted in Fig. 21.
The crest value of sustained armature current (V2 X 800 = 1,130 amp.)
was then subtracted from the envelope of the a-c. component, and the
48 SYNCHRONOUS MACInNES
remainder was plotted on semilog paper in Fig. 22. The points lie nearly on
a straight line except during the first few cycles'. This straight line, extended
back to zero time, gives the initial value of the transient component I',
which is 2,190 amp. crest, or 1,550 ~mp. r.m.s, The time constant is found
as twice the time required for I' to drop to i/V~, or 0.606, times its initial
value, or 1,330 amp. crest. This time if' 85.5 cycles, making; the time con-
stant T d' = 171 cycles = 2.85 sec.
I·
4
1
Time (seconds)
-2
-4
FIG. 21. D-c., or unsymmetrical, components and amplitude (crest value) of a-c.,
or symmetrical, component of short-circuit armature currents.
The values of I', read from the straight line, are then subtracted from the
ordinates of the curve I' +I" in the first few cycles and are replotted on
semilog paper in Fig. 23. These values lie on a straight line. The initial
value of I" is 1,470 amp. crest, or 1,040 amp. r.m.s, The time constant Td"
is the time taken for I" to decrease to liE or 0.368 of its initial value, or to
537 amp. crest. This time is T d " = 2.4 cycles, or 0.04 sec.
An approximate check on T d ' is obtained by plotting the d-c. component
of field current less the sustained value in Fig. 24. This gives TIi = 160
cycles, or 2.67 sec. This value is believed to be less accurate than that ob-
EXAMPLE 3 49
4,000 r - - - - - . . , . - - - - , - - - - r - . . . . . . . ; - . , . _ _ - - - - , r - - -__
.!!
~ 3,000 1----+-----+---.f----f----4--~
en
.Y
E
.s::.
=c
~ 2,000
.2
~
--r- -+--~
!i 1,500 1--_--+-_ _ +--_--l-~ _ __I_o-.6106 = v!E

...e
::J
(J
foE--~-- 7d'---=---~ i
1,00°0
120
FIG. 22. Semilogplot of excessof a-c. component of short-circuit armature current

over sustained current. The transient component is represented by the straight
line, from which time constant T l is determined.
2,000 ,...----,----.-----,----,------,.
CI)
~ 1,000
u
'E
.s::.
'c
~
.2 500
-u} I 0.135
e Td"~
Q)
a.
=-\
E
E
...
.s
c::
...:se 200
(.)
2Td"
100
0 2 3 4 5
Time (cycles)
FIG, 23, Semilog plot of subtransient component I" of short-circuit armature

current for determination of time constant Ti'.
tained from the a-c. armature current, because the zero axis was not recorded
on the original oscillogram but was drawn in later, possibly introducing an
error in the d-e, component. Note that the subtransient component of field
current is opposite in sign from the transient component and is small com-
pared with the sub transient armature current. It must be opposed by a large
sub transient current in the amortisseur windings.
300.------r---......-------...,..-------.....
14----+-- :
T.'
=80~
40 80 120
Time (cycles)
FIG. 24. Semilog plot of excess of d-e. component of field current over sustained
value. The transient component is represented by the straight line, from which
T d' can be determined. The subtransient component is small and of opposite sign
from the transient component.
The d-e. components of the three armature currents and the a-c. component
of field current are plotted logarithmically in Fig. 25. Except for i b, which
is too small to be determined accurately, the plots of these currents are
parallel but not quite straight. The plot of alternating field current if is
the straightest, possibly for reasons stated above, and was therefore used in
the determination of T a • The value of T a so found is 13.6 cycles, or 0.23 sec.
Values found from the plots of i a and i, range from 10 to 15 cycles, depending
upon how the straight line is drawn,
Reactances. The open-circuit voltage used in the test is 4,600/13,800, or
only one-third of normal. At such a low voltage· the machine is unsaturated.
The line-to-neutral value of open-circuit voltage is 4,600/ Va-
= 2,650 volts.
By eq, 80,
E 2,650
Xd =- = - - == 3.32 ohms
1, 800
By eq. 81,
x,/ = E = 2,650 = 1.13 ohms
I, +1 0' 2,350
EXAMPLE 3 51
5,000 ~---~----~'----.-----r----....,
4>
~
o
'E
:5 2,000
';:
'"
~
-"
CIl
~
~
E
.e 1,000 1------l------I-~~~e_+--~..,......jI__--____1
k-----_l__ To. --+-----'>-i"..
500 1 - 4·- - - - 1 - - - - +- ---1--+- - --1-- """....,...-- -1
o 5 10 15 20 25
Time (cycles)
FIG. 25. Semilog plot of d-e, components of armature currents i a , ib, i c and a-c.
component (crest value) of field current if for determination of time constant To..
By eq. 82,
xi' = ~ = 2,650 = 0.783 ohm
+ 10 + 1 0
II
I. 3,390
To convert the values of reactance into per unit, they must be divided by
the base impedance, which, from the machine rating, is
3
(13.8 X 10 ) 2 = 3.40 ohms
56 X 106
TABLE 3
CONSTANTS OF A 56-MVA. WATER-WHEEL GENERATOR,
AS DETERMINED FROM SHORT-CIRCUIT OSCILLOGRAM
Reactances Time Constants

(per unit) (sec.)
Xd = 0.98 Ti = 2.85
xi = 0.332 Ti' = 0.04
xi' = 0.230 To. = 0.23
The per-unit values of reactance and the time constants are summarized in
Table 3. All these values lie within the limits given in Table 2 for water-
wheel generators.
Substitution of these values and of E = 1 into eq. 83 gives, for the a-c.
component,
lac = 1.02 + 2.00E- t / 2•85 + 1.33E- t / O• 04 per unit
6 ~
,
\,~ \ ~
~/
'
'-..S >-
..
~
2 " '.P.. .c
~' ....
Sustained
o
" '---- ......
......... ----- ---- ~ ,J
o 10 20 30 40 50 60 00
Time (cycles)
FIG. 26. R.m.s. values of d-e, and a-c. components and of total short-circuit
armature current, fully offset wave (Example 3).
Equation 84, with E = 1 and cos a = 1, gives the d-e, component of a fully
offset wave as
I de = 6.14E- t / O• 23 per unit
The total current is given by eq. 85. The a-c. and d-e. components and the
total current are plotted in Fig. 26.
MATHEMATICAL THEORY
In the foregoing part of this chapter physical concepts of synchro-
nous-machine reactances and time constants were set forth and applied
to the determination of the three-phase short-circuit currents. Before
applying these concepts to stability calculations, it is desirable to have
a more comprehensive mathematical theory of synchronous machines
in the transient state than has been given. The following theory is
developed from the fundamental starting point that the machine con-
sists of several inductively coupled circuits, the self- and mutual in-
ductances of which vary periodically with the angular position of the
rotor.
A three-phase salient-pole machine without amortisseur windings
will be considered. Iron loss and saturation will be neglected. Such
a machine has four windings: the field winding and three armature
phase windings.
INDUCTANCES VS. ROTOR POSITION 53
The instantaneous terminal voltage of anyone of these windings

may be written in the form
v=n .+ -d1/l [89]
. dt
where r is the resistance of the winding, i is the current, and 1/1 is the
flux linkage of the winding, which depends upon the self-inductance
of the winding, the mutual inductances between it and other wind-
ings, and the currents in all the coupled windings; thus:
1/1 = ~Li [90]
Denoting the armature windings by subscripts a, b, c, and the field
winding by j, eqs, 89 and 90 yield the following:
Va = ria + d~a (91a]
Vb = rib + d~b (91h]
• d1/lc
Vc = ri; +-ai [91c]
• d1/l1
VI = Tltl +-
dt
[91d]
where
1/Ia = Laaia + Labib + Lacic + Lalil [92a]
1/Ib = Lbaia + Lbbib + Lbcic + Lblij [92b]
1/Ic = Lcaia + Lcbib + Lccic + Lcjil [92c]
1/11 = Ljaia + Ljbib + Ljcic + Ljji j [92d]
In eqs. 91 r represents the resistance of each armature phase and T/
represents the field resistance.
Because many of the inductances vary with time, it is impossible
to solve the differential equations 91 to find the currents for any given
terminal voltages without knowing these inductances as functions of
time. Therefore, they will be expressed in terms of the angular posi-
tion of the rotor fJ, which changes with time, thus:
fJ = wt + fJ o [93]
where w is the angular speed, which for present purposes may be con-
sidered constant, and 80 is the initial value of fJ.
Inductances of the windings as functions of the rotor position. The

conventions for positive directions of currents and angle will be taken
in accordance with Fig. 27. The magnetic axes of the three armature
phases are 120 elec. deg. apart. The direct and quadrature axes of
the field are 90 elec. deg. apart,
Axis
phase a with the quadrature axis leading.
The position of the rotor is given
by the angle 8 by which the direct
axis of the field has turned beyond
the axis of armature phase a.
The self-inductance of each ar-
mature' phase is always positive but
varies with the position of the rotor,
being greatest when the direct axis
of the field coincides with the axis
of the armature phase and least
when the quadrature axis coincides
with it. The variation of the self-
FIG. 27. Conventions for positive
angle, currents, and m.m.f.'s. (From inductance between these two posi-
Ref. 34 by permission.) tions can be either calculated from
design data or measured; by both
methods it is found to be practically sinusoidal. See Fig. 28, for
example. The variation of the self-inductance of phase a may be
expressed. by the following equation:
L aa = L + L m cos 28
8 [94]
where L 8 > Lm . It has maximum values at 8 = 0 and 180°, and min-
imum values at 8 = 90° and 270°. For phase b the variation of self-
3.5
i' 3.0I--~Pr--+----~----+---+-----+-~o.---I
e
~
-ec 2.5 1----+-~--+----__+---+---_a___+_--___1

:::J
"0
C
"
~ 2.0
30 60 90 120 150 180

Angle of direct axis from axis of phase a (elec. deg.)
Fro. 28. Variation of self-inductance of armature phase a with position of the field.
The test points, taken on a 4-pole 15-kva. 220-volt 1,800-rpm. synchronous motor,
compare favorably with the curve showing the theoretical variation of an ideal
machine: L aa = L, + L m cos 28 = 2.36 + 0.796 cos 28 mho (From Ref. 34 by
permission.)
INDUCTANCES VB. ROTOR POSITION 55
inductance is similar, except that the maximum value occurs where

the direct axis coincides with the axis of phase b, that is, at 8 = 120°
or -60°.. Hence, the proper equation for phase b is obtained by re-
placing 8 in eq. 94 by 8 - 120°.
L bb = L8 + L m cos 2(8 - 120°)
=L 8 + i ; cos (28 + 120°) [95]
Similarly, for the self-inductance of phase c,
Lee = L 8 + Lmcos 2(8 + 120°)
=L 8 + L m cos (28 - 120°) [96]
The mutual inductance between two armature phases also varies with
the position of the field. It is always negative, and its greatest abso-
lute value occurs where the direct axis lies midway between the axis
-2.0 r----,.-.--,----..,------,----~--__,
...-:-
s:
~ -1.5Q-----f-----r-.----+-----,t'r---+----o
Q)
<J
e
g -1.0
«I
t------:ln---J----f--.---+-~--+----+-----I
"C
.5
~ -0.5 t-----+--~--+--~_+_--_+_--__+--~
~
:IE
30 60 90 120 150 180'

Angle of direct axisfrom axisof phase a (elec. deg.)
FIG. 29. Variation of mutual inductance between armature phases a and b with
position of the field. The test points, taken on the same machine as in Fig. 28,
are close to the curve which shows theoretical variation for an ideal machine:
Lab = -M, + L m cos (28 - 120°) = -1.14 + 0.796 cos (28 - 120°) mho {From
Ref. 34 by permission.)
of one phase and the reversed axis of the other phase. Thus the maxi-
mum absolute value of the mutual inductance between phases a and
b occurs at 8 = - 30° or 150°; between phases a and c, at 8 = -150°
or 30°; and between phases band c, at 8 = 90° or 270°.
It is also found, both by calculation and by test, that the amplitude
of variation of the mutual inductance between two stator phases is
practically the same as the variation of the self-inductance of a stator
phase. For example, see Fig. 29. Consequently, it will be denoted
by the same symbol, L m .
Lab = Lba = -[M s + L m cos 2(8 + 30°)]
-M8 + L cos (28 - 120°)
m [97]
Lbe = + L m cos 2(0 - 90°)]

L eb = -[M8
= -M + L m cos 288 [98]
Lea = Lac = -[M, + Lmcos 2(8 + 150°)]

= -M + L m cos (28 + 120°)
8 [99]
The mutual inductance between the field winding and any armature
phase is greatest when the direct axis coincides with the axis of that
phase. After the field turns 180° from this position the mutual in-
ductance is equal and opposite to its original value. The variation is
sinusoidal: it may be shown that sinusoidal variation of this mutual
inductance is necessary to give a sinusoidal open-circuit voltage, which
is usually attained closely in commercial machines.
[100]
Lbf = L j b = M j cos (0 - 120°) [101]
Lef = Lj e = M j cos (0 + 120°) [102]
The self-inductance of thefield winding L f j is constant.
On substituting the values of the inductances (eqs. 94 through 102)
into eqs. 92, we obtain the following expressions for the flux linkages:
1/Ia = + L m cos 28)ia + [-M + L m cos (20 - 120 b
(L 8 8
0)]i
+ [-M + t.; cos (28 + 120 c + (Mf cos 8)ij

8 [103a]
0)]i
1/Ib = [-M + L m cos (28 - 120 )]i a + [L + L m cos (28 + 120

8
0
8
0)]ib
+ (-M, + L m cos 28)ie + [Mj cos (0 - 120 j [103b] 0)]i
1/Ic = [-M + L m cos (28 + 120 )]i a + (-M + L m cos 28)i b

8
0
8
+ [L + Lm cos (28 - 120 )]i e + [Mf cos (8 + 120 )]i f [103c]

8
0 0
1/Ij = (Mf cos O)ia + [M j cos (8 - 1200 )]i b + [M j cos (8 + 1200 )]i c
+ Lfjif [103d]
Now assume that the voltages impressed on the four windings (three
armature phases and the field) are known as functions of time and
that expressions are wanted for the currents as functions of time.
Four differential equations are needed which have no unknowns
except the four currents and their time derivatives. To obtain such
+
equations, put () = wt 80 in eqs. 103, and, assuming that w is con-
stant, take the time derivative of these equations and substitute the
result into eqs. 91. The resulting equations are very complicated; for
PARK'S TRANSFORMATION 57
example, the equation for Va is:

Va = [r- 2wLm sin 2 (wt+8 o)]i a -[2wLm sin (2wt+28 o-1200)]ib
-f2wLm sin (2wt+28o+1200)]ie-[wMj sin (wt+Oo)]if
dia ° dib
+[Ls+Lm cos 2(wt+8o)] di +[Ms+Lm cos (2wt+28 o-120 )] di
die di/
+[M.+Lm cos (2wt+20o+1200)] di +[M, cos (wt+8o)] -;Ii [104]
While the solution of four simultaneous differential equations of this

kind is theoretically possible, it would not be feasible as they stand.
Park's transformation. The equations of the synchronous machine

may be greatly simplified by a proper substitution (transformation)
of variables. That is, a set of fictitious currents, voltages, and flux
linkages will be defined as functions of the actual currents, voltages,
and flux linkages, and the equations will be put into terms of the new
variables. The equations may then be solved for the new variables as
functions of time. From the results, the actual electrical quantities
can be found as functions of time. The substitution could be regarded
as a purely mathematical one, and from this point of view it would
not be necessary to give any physical interpretation to the fictitious
or substituted quantities. They could merely be chosen in such a
way that the equations would be simplified by the substitution. Actu-
ally, the particular substitution used is based on physical reasoning
and the substituted variables can be given physical interpretations.
The reader has already encountered the use of substitute variables
in the method of symmetrical components. The particular substitu-
tion that we are about to use is from R. H. Park. 10 It is therefore
fitting to call it "Park's transformation" and to call the substitute
variables "Park's variables."
Let the actual armature phase currents i a, ib, and i e be replaced by
new fictitious currents id , i q , and i o in accordance with the following
equations:
ia = id cos 8 - i q sin 8 + i o [105a]
i b = ie cos (8 - 120°) - i q sin (8 - 120°) + i o [105bl
i c = i d cos (8 + 120°) - i q sin (8 + 120°) + i o [lOSe]
No substitution will be made for the field current if. Similar substi-
tutions are made for armature voltages and for armature flux linkages.
58 SYNCHRONOUS MACHINEt;
The inverse transformation is:

id = ilia cos 8 + i b cos (8 - 120°) + i c cos (8 + 120°)] [106al
iq = -ilia sin 8 + i b sin (8 - 120°) + i csin (8 + 120°)] [106b]
io = leia + ib + i c ) [l06e]
for armature currents and similarly for armature voltages and arma-
ture flux linkages. Equations 106 may be obtained from eqs. 105 as
follows: 'I'o obtain i d , multiply the first, second, and third equations
+
by cos 8, cos (8 - 120°), and cos (8 120°), respectively, and add.
To obtain i q , multiply them by sin 8, sin (8 - 120°), and sin (8 + 120°),
respectively. To obtain i o, simply add the three equations together
The following identities are needed:
C08
2
8 + cos2 (8 - 120°) + C08 2 (8 + 120°) =! [107]
sin 2 (J + sin 2 (6 - 120°) + sin 2 (6 + 120°) = t [108]
sin (I cos (I + sin (8 - 120°) cos (8 - 120°)
+ sin (8 + 120°) cos (8 + 120°) = 0 [109]
cos 8 + cos (8 - 120°) + cos (8 + 120°) = 0 [110]
sin 8 + sin (8 - 120°) + sin (8 + 120°) = 0 [111]
These identities may be proved by writing the sines and cosines in
terms of exponentials.
Flux linkages in terms of Park's variables. Now we will proceed

with the change of variables. Before obtaining differential equations
of voltage in the new system, analogous to eqs. 91, let us obtain the
equations for flux linkages as functions of the currents, analogous to
eqs. 103. To accomplish this, first write equations similar to 106 for
1/Id' 1/Iq, and 1/10 in terms of 1/Ia, 1/Ib, and v;c. Then substitute eqs. 103
into them, and eqs. 105 into the result. The amount of algebraic and
trigonometric work required is formidable, but if the reader persists
he will find that the multiple products of trigonometric functions can
be reduced to simple constants, some of which were given in eqs. 107
through 111, and the following simple equations are finally obtained:
1/Id = (L 8 + M, + !Lm)id + Mlif [112a]
1/Iq = (L 8 + M fLm)i
8 - q [112b]
v;o = (L 8 - 2M8 )i o [112e]
1/11 = !M,id + L'lil [112d]
INTERPRETATION OF VARIABLES 59
These equations are simpler than eqs. 103 in two ways: first, in that
all the coefficients are constants, independent of the rotor position 8;
second,. in that there is almost a complete separation of variables.
The quantities in parentheses are the ratios of flux linkage to current;
obviously, they are the inductances in the new system of variables.
Furthermore, it may be suspected that they are identical with some
of the synchronous-machine inductances which were previously de-
fined, and this will be proved a little later.
L a+llf,+"iLm =Ld =direct-axis synchronous inductance [113]
La+M 8 -jLm = L q = quadrature-axis synchronous inductance [114]
L 8 - 2M 8 = Lo=zero-sequence inductance [115]
Thus, eqs. 112 may be written
[116a]
1/Iq = Lqiq [116b]
1/10 = Loio [II6c]
1/If = j-Mfid + Lffif [116d]
Physical interpretation of Park's variables. A physical interpreta-

tion of the new variables is now in order. The m.m.f. of each arma-
ture phase, being sinusoidally distributed in space, may be represented
by a vector the direction of which is that of the phase axis and the
magnitude of which is proportional to the instantaneous phase cur-
rent. The combined m.m.f. of the three phases may likewise be rep-
resented by a vector which is the vector sum of the phase-m.m.f.
vectors. The projections of the combined-m.m.f. vector on the direct
and quadrature axes of the field are equal to the sums of the projec-
tions of the phase-m.m.f. vectors on the respective axes as given by
the expressions for i d and i q , eqs. 106. The constant i is arbitrary.
Thus id may be interpreted as the instantaneous current in a fictitious
armature winding which rotates at the same speed as the field winding
and remains in such position that its axis always coincides with the
direct axis of the field, the value of the current in this winding being
such that it gives the same m.m.f. on this axis as do the actual three
instantaneous armature phase currents flowing in the actual armature
windings. The interpretation of i q is similar to that of id except that
it acts in the quadrature axis instead of in the direct axis. The i o of
the new variables is identical with the usual zero-sequence current,
except that it is an instantaneous value and is defined in terms of the

instantaneous phase currents. This current gives no space-funda-
mental air-gap flux.
The flux linkages of the fictitious armature windings in which i d
and i q flow are Y;d and Y;q, respectively.
In view of the foregoing interpretation of id and i q , it is apparent
that their m.m.f.'s are stationary with respect to the rotor and there-
fore act on paths of constant permeance. Hence the corresponding
inductances L d and L q are independent of rotor position.
The fictitious direct-axis stator winding and the field winding are
inductively coupled. Each has a self-inductance (L d and L f f ) , and
there is a mutual inductance between them. It should be noted that
the mutual inductance has different values in eqs. 116a and d, being
Mj in one and -IM f in the other. The difference could have been
avoided by a different choice of the constant coefficients in eqs. 105
and 106; however, we will retain the form of the variables given by
Park.
Circuit equations in terms of Park's variables. Now that the equa-

tions of the form Y; = 'ELi have been transformed from their original
form (eqs. 103) to equations in the new variables (eqs. 116), it re-
mains to make a similar transformation on eqs. 91, which have the
form
. dy;
v=r~+
dt
The Park voltages are given by equations similar to eqs, 106 for
currents, to wit:*
[117a]
[117c]
Substitute into eqs. 117 the expressions for phase voltages, eqs. 91,
•A different convention on the application of subscripts d and q to voltages was

used by Doherty and Nickle 3,4,6,11 and others. A speed voltage generated by
direct-axis flux t/ld they called a direct-axis voltage (ed), and one generated by
quadrature-axis flux t/lq they called a quadrature-axis voltage Ceq). Both Park's
and Doherty and Nickle's voltage notations have been widely used in the literature,
but Park's is deemed preferable and is used in this book.
CIRCUIT EQUATIONS 61
obtaining from eq. 117a:
Vel = 3"2 [ (r~a

. + dt/la)
di cos 8 + (.r~b + di
d1/Jb) cos (8 - 1200)
+ (ric + d~c) cos (fJ + 120°)J

+ i b cos (9 - 120°) + i cos (9 + 120°)]
= ri[ia cos (J c
+"32[d~
dt cos 8 +dt dtc cos (8 + 120°)]
dtb cos (8 - 120°) + dt
= r1,d. + (d1/!)
-dt d [118]
where (dl/l/dt)d, which is introduced as an abbreviation for the second

bracketed expression, should not be confused with dl/ld/dt. Indeed,
the latter, found by taking the time derivative of
l/Id = j(t/la cos 8 + V;b cos (8 - 120°) + v;c cos (8 + 120°)] [119]
where (J = wt + 80, is
dv;tl = ~ [d1/ta cos 8 + d1/tb cos (8 - 120°) + d1/tc cos (8 + 120 )J
0
dt 3 dt dt dt
-wi[,pa sin 8 + 1/Ib sin (8 - 120°) + ,pc sin (8 + 120°)]
= (~~\ + W1/tq [120]
Substituting the value of (dl/l/dt)d from eq. 120 into eq. 118, we obtain
· d1J;d
Vd = r1,d + dt - wl/lq [121]
Proceeding in the same manner with Vq , we obtain

· d1/lq
Vq = rt q + dt + W,pd [122]
Also
·
Vo = r1,o +-
d1f;o
dt
[123]
Equation 123 has the usual form, like eq. 89, but eqs. 121 and 122
are unusual because of the extra terms -w1f;q and wVtd, respectively.
The presence of these terms demands a physical interpretation.
62 f;YNCHRONOUS l\tIACHINES
In the interpretation of eqs. 116, it was suggested that id and i q

were the currents in fictitious rotating stator windings - if that para-
doxical expression may be used - which gave the same m.m.f.'s as
did the actual armature currents in the actual armature windings,
But it is not necessary to have the fictitious windings rotate. The
same effect can be achieved by conceiving the armature winding to
be stationary (as it actually is) and to be a closed-circuit. winding
with a commutator on which rest brushes which rotate with the field.
The magnetic axis of the stator will always coincide with the brush
axis. Thus i d may be interpreted as the current entering and leaving
the armature through a pair of brushes which are aligned with the
direct axis of the field. Similarly, i q may be regarded as the current
entering and leaving the armature through a second pair of brushes,
aligned with the quadrature axis of the field. In other words, the
armature may be thought of as like that of a synchronous converter,
having both commutator and slip rings, .but having brushes in both
axes instead of in the quadrature axis only. The actual phase cur-
rents, .entering the slip rings, give the same m.m.f.'s as do the substi-
tute currents i d and i q entering the commutator brushes.
This physical picture is also correct with respect to the voltages.
The terms -wY;q and WY;d occurring in eqs. 121and 122 may be regarded
as components of applied voltage required to balance the speed voltages.
The speed voltage across each pair of brushes is proportional to the
flux on the axis 90° ahead of the brush axis.
We may now substitute eqs. 116 for the flux linkages into eqs. 121,
122, 123, and 91d in order to obtain differential equations' of the volt-
ages in terms of the currents. The results are as follows:
. did·. dif
Vd = r~d + La di - wLq~q + MJ di [124a]
di + W(L·d~d + M·I~I )
q
·
vq = r~q + L q di [124b]
· o
Vo = r~o + L 0 di
di [124c]
. di, 3 did
vJ = TJ1,J + LJj di + 2M'di [124d]
These equations correspond with those like eq. 104 but are much
simpler than the latter as there are only two, three, or four terms per
equation instead of eight and as the currents and their derivatives
have constant coefficients.
PARK INDUCTANCES 63
Proof of expressions for Park inductances. When the 1/1 = ELi

equations (eqs. 116) were obtained in terms of Park's variables, the
inductances were suspected to be identical with some of the synchro-
nous-machine inductances previously defined. This will now be
proved.
Direct-axis synchronou8 inductance Ld. The conditions previously
specified in the definition of this inductance were:
1. Steady-state positive-sequence currents applied to the arma-
ture:
i a = I cos wt [125a]
i b = I cos (wt - 120°) [125b]
ic =I cos (wt + 120°) [125c]
2. Field circuit open:*
if = 0 [126]
3. Field rotated at synchronous speed with the direct axis in line
with the axis of the armature m.m.f. wave:
8 = wt [127]
Under these conditions the flux linkage per armature ampere is equal
to the direct-axis synchronous inductance Ld. Consider phase a, the
linkage of which is given by eq. 103a. Substitute the specified condi-
tions (eqs. 125 through 127) into it, obtaining:
t/la= (L 8 + Lmcos 2wt)I cos wt+[ -M8 + L mcos (2wt-1200)]
XI cos (wt-1200)+[ -M8 + L m cos (2wt+1200)]1 cos (wt+1200)
= {La cos wt- M a [cos (wt-1200) + cos (wt+ 120°)]
+Lm [cos 2wtcos wt+ cos (2wt-1200) cos (wt-1200)
+ cos (2wt+1200) cos (wt+1200)]}I [128]
Through use of the identities

cos (wt - 120°) + cos (wt + 120°) = -cos wt [129]
and
cos 2wt cos wt + cos (2wt - 120°) cos (wt - 120°)
+ cos (2wt + 120°) cos (wt + 120°) = ! cos wt [130]
•As stated previously, the field current is zero under the specified conditions re-
gardless of whether the field circuit is open or not.
eq. 128 reduces to

1/Ia = (L 8 + M. + !Lm)I cos wt
= (L + M + JLm)ia
8 8 [131]
Ld == ~a == La + M a + tLm
~a
[132]
which agrees with eq. 113. The same result can be obtained for
phase b or for phase c.
Quadrature-axis synchronous inductance L q • The conditions speci-
fied in the definition of L q are the same as those for La except that
the field is rotated with the quadrature axis in line with the axis of
the armature m.m.f. wave:
() = wt - 90°* [133]
For these conditions, eq. l03a becomes
-Pa = [L,+L mcos 2(wt-900)]I cos wt
+{ -Ma+Lm cos [2(wt-900) -1200]}I cos (wt-1200)
+{-Ma+Lm cos [2(wt-900)+1200]}I cos (wt+1200)
= {La coswt-M8 [cos (wt-1200)+ cos (wt+1200)]
+Lm [cos (2wt-1800) cos wt+ cos (2wt+600) cos (wt-1200)
+ cos (2wt-600) cos (wt+1200)]}I
= (L 8 + M s-!Lm)I cos wt
= (Ls+Ms-!Lm)i a [134]
- VJa
L q- . -
- L a + M '"2
_ SL m [135]
ta
which agrees with eq. 114.

Zero-sequence inductance £0. The conditions specified in the defini-
tion of L o·are:
1. Steady-state zero-sequence currents applied to the armature:
i a = i b = i c = I cos wt [136]
2. Field circuit open:
if = 0 [137]
3. Motion of field not specified:
8=8 [138]
*The same result could be obtained by letting 8 = wt as before and advancing the
phase of the armature currents 90°.
P.L\.RK INDUCT.A~CES 65
Equation 103a becomes:

YJa = (L, - 2M 8 ) I cos wt + Lm [cos 28 + cos (28 - 120°)
+ cos (28 + 120°)][ cos wt
== (L 8 - 2M,)ia [139]
£0 = ~a = £8 - 2M 8 [140]
ta
which agrees with eq. 115.

Direct-axis transient inductance La,'. This inductance is the initial
armature flux linkage per unit armature current (YJa/ia) under the
following conditions:
1. Positive-sequence currents suddenly applied to the armature:
ia = I cos wt 1 [141a]
ib = I cos (wt - 120°) 1 [141b]
ic =I cos(wt + 120°) 1 [141c]
2. Field circuit closed but not excited:

VI =0 [142]
3. Field rotated as specified under direct-axis synchronous in-
ductance:
(J = wt [143]
Condition 1, substituted in eqs. 106, yields:
ia, = Ii [144a]
iq = 0 [144bl
io = 0 [144c]
Condition 2 and the law of constant flux linkage show that initially
[145]
and from eq. 116d,
or
. 3 M J.
tl = - - - ta, [146]
2 LJJ
Substitution of the values of the currents (eqs. 146 and 144b and c)
into eqs. 116 yields:
1/Id = (L d - 3 Mj ) •
- - - ~d
2
[147a]
2 Lff
1/Iq =0 [l47b]
1/10 = 0 [147c]
Since, according to eqs. 144, i q = io = 0, from eq. 105a we have
i a = id cos 8 [148]
and similarly, since, according to eqs. 147, 1/Iq = 1/10 = 0,
1/Ia = 1/Id c.os 6 [149]
Therefore, upon dividing eq. 149 by eq. 148 and then using eq. 147a,
2
1/Ia
- = 1/Ia,
- = La - -3 -M f =
,
La [150]
ia ia, 2 L1 j
This is the direct-axis transient inductance. Since Llj is a positive

quantity,
La' < La, [151]
Equation 150 should be compared with eq. 31 for the equivalent in-
ductance of a transformer with short-circuited secondary winding.
La, corresponds to the self-inductance of the primary winding,Ljl to
that of the secondary winding, and M I to the mutual inductance
between the two windings. The only difference between the two
equations is the presence of the factor t in eq. 150, which appears in
consequence of the coefficients arbitrarily chosen in Park's trans-
formation.
Quadrature-axis transientinductance L q' and subtransient inductances
La') and L q " . It was assumed in the development of the mathemati-
cal theory that the only circuit on the rotor was the field winding in
the direct axis. Under this assumption the direct-axis subtransient
inductance is equal to the direct-axis transient inductance, and the
quadrature-axis transient and subtransient inductances are equal to
the quadrature-axis synchronous inductance. The addition of a
damper winding in the direct axis would make the direct-axis sub-
transient inductance less than the transient inductance. A damper
winding in the quadrature axis would make the quadrature-axis sub-
transient inductance less than the transient and synchronous induct-
ances although the last two would still be equal unless there were a
field winding in the quadrature axis (a very unusual condition) or
PARK INDUCTANCES 67
some other circuit of similar time constant, for example, the solid
rotor of a turbogenerator.
Negative-sequence inductance £2. The negative-sequence inductance
of a three-phase synchronous machine is the flux linkage of an arma-
ture phase per ampere of armature current under the following con-
ditions:
1. Steady-state negative-sequence currents applied to the arma-
ture windings:
i a = I cos wt [152a]
i b = I cos (wt + 120°) [152b]
ic =I cos (wt - 120°) [152c]
2. Field circuit closed, not excited:
VI =0 [153]
3. Field rotated at synchronous speed, in any position:
8=wt+a [154]
The Park currents corresponding to eqs. 152 are:
i d = I cos (2wt + a) [155a]
i q = - I sin (2wt + a) [155b]
io = 0 [155c]
Note that i d and i q are of twice the applied frequency. Current ill in-
duces a field current of the same frequency. At this frequency the
resistance of the field circuit is usually negligible compared with its
reactance and may therefore be neglected.
. dt/ll dt/ll
VJ = rJ~J + di ~ di = 0 [156]
Hence
'"f = constant
and, in the steady state,
1/11 = 0
Therefore, as shown before (eqs. 145, 146, 147a, and 150),
iJ =- ~ MJ i d [157]
2 LIJ
..II
'I'd ~d + M·
= L· J~f = (L d - 3 M-
- -
2 LJJ
/2) ·
'td
= La'ia [158]
Also, by eqs. 116b and c,
ljIq = Lqiq [159]
¥to = 0 [160]
The flux linkage of phase a is (compare eq. 1000):
1/Ia =1/Id, cos 8-,yq sin 8+1/10
= Ld,'id, cos 8- Lqiqsin 8
= [Ld,' cos (2wt+a) cos (wt+a)+L q sin (2wt+a) sin (wt+a)]I
= {Li ~Lq [cos (2wt+a) cos (wt+a) + sin (2wt+a) sin (wt+a)]
q
+ Li;L [cos (2wt+a) cos (wt+a) - sin (2wt+a) sin (wt+a)]} I
l
La +Lq La' -Lq ]
= [ 2 cos wt+ 2 cos (3wt+2a) I [161]
It may be seen from this result that, although the applied currents
are exclusively of fundamental frequency, .the armature flux linkage,
and consequently also the armature voltage, consists of a fundamental
and a third harmonic. For this reason the instantaneous value of
"'alia is not constant. It would be discreet now to modify the defini-
tion of negative-sequence inductance to: fundamental-frequency arma-
ture flux linkage per fundamental-frequency armature ampere. Then
La = lfa(fundamental) = Li + L q [162]
ia 2
That is, negative-sequence inductance is the arithmetical average of
direct-axis transient inductance and quadrature-axis synchronous in-
ductance. But this result is valid only if there are no rotor circuits
save the field winding.
If there are low-resistance damper windings in both axes, the Park
flux linkages are
1/Ia = La,/Iid, [163]
.11 L
r« = q ~q II· [164]
and the negative-sequence inductance becomes
L
a,
II + L q II
L 2= [165]
2
Likewise, the negative-sequence reactance is
Xd,
II + Xq II
X2 =- [166]
2
VECTOR DIAGRAfvlS 69
If the resistances of the damper windings should be appreciable

compared with their self-reactances at double frequency, then the
negative-sequence reactance would depend somewhat upon the re-
sistances as well as upon the reactances of the windings,
It is of academic interest to note that if fundamental-frequency
negative-sequence voltages (or flux linkages) are impressed upon the
armature, the armature currents are not sinusoidal but contain a third
harmonic in addition to the fundamental. In this case the negative-
sequence inductance, defined again as the ratio of fundamental flux
linkage per fundamental unit current, has a different value than
before; it is now:
2Ld"Lq"
L2 = Li' + Lq" [167]
This is the "harmonic mean" of Ld" and t.,'. The harmonic mean
gives the lower limit, and the arithmetic mean (eq.. 165), the upper
limit, of values of L 2• In practice there is little difference between
the two values and the arithmetic mean, being simpler, is generally
used. The proof of eq. 167 is left as an exercise.
VECTOR DIAGRAMS
Vector diagrams for positive-sequence steady state. Let the field be
excited with constant current:
if = If [168]
and let the armature currents be steady and of positive sequence;
i a = Re(I~) = I cos (wt + (3) [169a]
i b = Re(a2I~) = I cos (wt + (3 - 120°) [169b]
ic = Re(aI~) = I cos (wt + (3 + 120°) [169c]
where
I = I /f3 [170]
Then the Park currents (eqs. 106) are:
id = I cos {j = constant = I d [171a]
i q = I sin 13 = constant = I q [171b]
io = 0 [171c]
The Park fluxes are:
,yd = Ldld + M,l, [172a]
t/lq = Lqlq [172b]
1/10 = 0 [172c]
1/11 = fMfla, + LfJlf [172d]
all of which are constant. The Park voltages (eqs. 121 and 122) are:
tJd = rIa, - w1/lq = rId -_. wLqIq = rId - xql q
= V d = constant [173]
and
Vq = rl q + wl/ld = rl q + wLd1d + wMjl j
= rlq + Xdld + E = V = constant
q q [174]
where
[175]
is the excitation voltage, the voltage behind synchronous impedance, or
the steady-state internal voltage. I t is directly proportional to the field
current 1/.
~ Imaginary or
I quadrature axis The positive-sequence armature
-xqlq
I
current is (by eqs. 170 and 171):
I = I /13 = I (cos 13 + j sin (3)
= l s + jlq [176]
and, similarly, the positive-se-
quence armature terminal voltage
is:
v = Va, + jVq [177]
and the positive-sequence excita-
tion voltage is:
E =0 + jE q = jwMfl f [178]
-.-.-.1_-1;>- Real or-direct a_xis Substituting eqs. 173 and 174 into
eq. 177,
FIG. 30. Steady-state vector diagram V=E + rI + jXdld - xql q [179]
of salient-pole synchronous motor with
lagging (magnetizing) current. I j,
The voltages, currents, and link-
ages in eqs. 172 to 179 are constants
I d, I q, and fJ are positive. This diagram
agrees with eq. 179. equal to the crest values of the
corresponding positive-sequence
quantities, as indicated in eqs. 169 for current I. However, the corre-
sponding effective values can be used equally well in any except the
equations involving field current If.
The vector diagram corresponding to eq. 179 is shown in Fig. 30. It
is drawn for positive If, la" I q , and {j. These conditions correspond
VECTOR DIAGRA1\.IS 71
to motor operation with lagging (magnetizing) current, as may be

seen from the following considerations: If If and I d are both positive,
eq. 172a shows that their effects are cumulative; hence I d is magnetiz-
1~n.g. The field and armature poles are as shown in Fig. ~1(1.. Now if
(b) I q
:FIG. 31. Location of field and armature poles for positive I J and (a) positive
I« only, (b) positive I q only.
I q is assumed to be positive and I d to be zero, then, since the quadra-

ture axis was taken 900 ahead of the direct axis, the poles will be lo-
cated as shown in Fig. 31b, corresponding to motor operation (unlike
poles attract). Negative I d would be demagnetizing, and negative I q
would represent generator action.
Note that the original circuit equations (91) were set up with such
conventions that positive power means power consumed (that is,
motor operation). As a consequence, eq. 179, which was derived from
the original equations, bears the same sign conventions and is of the
form
V=E+ZI [180]
In dealing with generators the conventions on signs are usually taken

so that eq. 180 becomes
V=E-ZI [181]
or
E=V+ZI [182]
requiring that either the sign of I or the signs of V and E be changed.

If the sign of I (that is, of both I d and I q ) is changed without changing
the sign of If, then positive Id is demagnetizing and positive I q rep-
resents generator action. This changed sign of l s and I q (including
instantaneou8 values id and iq) is in effect in the rest of this book and is
in general use elsewhere.
The vector diagram for generator action with lagging current is
72 SY'NCHRONOUS MACHINES
shown in Fig. 32. In this diagram the quadrature axis is drawn hori-
zontally, as is customary, and resistance is neglected.
In both vector diagrams the excitation voltage E lies along the
positive quadrature axis. If the vector jxqI - that is, the total arma-
ture current multiplied by the quadrature-axis synchronous reactance
- is added to the terminal-voltage vector V, a vector E qd is obtained
which also lies on the quadrature axis. If the generator is operated
I
I
I
I
I ",,"
1 /
I.
d ------
I
90.
\<:-..~/,,,;,
\
;'
t
",
v
Direct
axis
~FIG. 32. Steady-state vector diagram of salient-pole synchronous generator with

lagging (demagnetizing) current; resistance neglected.
with constant field current I I but with slowly changing load or power
factor, then E will be constant; and E qcl, though not constant, will
nevertheless remain in phase with E. The phase of either E· or Eqa
is indicative of the mechanical position of the rotor, more specifically,
the position of the quadrature axis of the rotor. Hence, when a steady-
state stability study is being made on an a-c. calculating board, each
unsaturated salient-pole synchronous machine could be represented
by an adjustable e.m.f. E qcl in series with reactance X q • If this repre-
sentation is used, the value of Eqcl must be adjusted from time to
time to correspond with the desired constant value of E, in accordance
with the following relation:
[183]
Because of the pronounced effect of saturation on steady-state sta-
bility, machines are seldom represented in this manner. (See Chap-
ter XV.) A similar representation, suitable for transient stability
studies, is described later in this chapter.
In a round-rotor machine Xcl and X q are very nearly equal and may
be assumed equal without serious error. In this case E qd = E.
REVISED EQUATIONS 73
Revised equations of a synchronous machine. Before proceeding

with our consideration of vector diagrams of a synchronous machine,
a resume of the equations which have been derived will be given in
which the algebraic signs of all armature currents (i a and iq) are re-
versed to conform with the usual practice in discussing generators, as
explained in the last section. Consequently, positive i q and i a will
now represent generator action with lagging (demagnetizing) current.
The sign of the field current is not changed.
Moreover, to make the equations applicable to a machine with
solid, round rotor, a circuit (denoted by subscript g) will be assumed
present on the quadrature axis of the rotor. This new circuit is analo-
gous to the main field winding (denoted by f) on the direct axis of the
rotor.
The most important equations, modified in the two respects just
described, are listed below:
Equation 116a becomes:
1/Ia = Mjif - Laid, [184]
116b, 1/Iq = Mgi g.. - Lqiq [185]
116d, 1/11 = Lljij - !Mj"id, [186]
Similarly, 1/Ig = Lggig - -!Mgiq [187]
. d1/la
121, Va = -Md + di - w1/lq [188]
122, [189]
124a, Va = ·
-r~a + M1dilL did, -
dt- - a rz:
dt wM g1.·g + wL ·
1,
q q
[190]
124b, [191]
. di, did,
124d, VI = rltl + £11-
dt
- !MI-
dt
[192]
. dig -3. diq

Similarly, o = rgtg + Lggdj - ~Mgdj [193]
173, Vd = -rId + xqlq + Ed, [194]

174, Vq = -rIq - xd,Id + Eq [195J
175, unchanged E q = wMfII [196]

Similarly to 175, Ed= -wMgIg [197]
176, unchanged, I = Id + jIq [198]
177, unchanged, V = Vd +jVq [199]
178, E = Ed +jEq [200]
179, V = E - rI - jXdId + xqIq [200a]
Vector diagram for the transient state. If the armature current I

changes very slowly (compared with the transient decrement), then
the field current I I remains constant. The fictitious excitation volt-
age E q , which by definition (eq. 196) is proportional to the field cur-
rent, also remains constant. If there is a quadrature-axis rotor circuit,
its current I 9 remains at zero, and so does the fictitious armature
voltage Ed (eq. 197). The terminal voltage V changes, however, for
it differs from the constant voltage E = jEq by a changing synchro-
nous-impedance drop (see eq. 200a). The steady-state equations (194
to 200a) and vector diagram (Fig. 32) are valid for very slow changes,
and If and E q are constant during such changes.
Now consider a faster change of armature current (fast compared
with the transient decrement, yet not faster than the subtransient
decrement). Such a change may be due to the application or removal
of a fault or to the swinging of the generator with respect to other
machines of the power system. During such a change, the flux linkage
1/If of the field winding remains substantially constant. A new ficti-
tious internal armature voltage will be defined, proportional to the
field flux linkage:
I WMf
E q =-L 1/11 [201]
II
Then this voltage will likewise remain constant during the change of
armature current. It will be shown that the quadrature-axis terminal
voltage V q differsfrom E q ' by the quadrature-axis transient-impedance
drop. The difference between E q and E q ' is, by use of eqs. 196and 201,
b,q -
E qI = WM/~f• wM
- - Lf 1/11 [202]
ff
which, upon substitution of eq. 186, becomes

2
Eq - Eq' =
3wM1
2L
•
~d [203]
ff
TRANSIENT STATE 7fi
and, by use of eq. 150,

Eq - Eq' = WeLd - Ld')id = (Xd - xd')id [204]
This result may he arranged thus:
.
=
fj1 , ,.
E_'Jq - Xdtd I-I;q - J;d 1'd [20n]
The relation between the quadrature-axis components of excitation
voltage and terminal voltage in the steady state is given by eq. 195,
which was derived by setting the time derivatives in eq. 191 equal to
zero. During changes of the rapidity which we are now considering,
these derivatives are still negligible compared with the terms in w;*
therefore the steady-state equations (194, 195, and 200a) and the
corresponding vector diagram are still valid. We shall denote the
currents and voltages by capital letters, even though they are varying
quantities, because they vary slowly and may be considered as the
effective values of alternating currents and voltages. (Most writers,
however, use lower-case e's and i's.) Granting that eq. 195 is still
applicable, we can make the substitution of eq. 205 in it, obtaining:
V q = -rl q - xd'I d + E q ' [206]
Similarly, if there is a quadrature-axis rotor circuit, its flux linkage
may be assumed constant and a voltage Ed' defined proportional to
the linkage, thus:
Ei = wMal/ta [207]
L gg
It follows, by analogy with eq. 204, that
- (Ed - Ed') = (xq - xq')iq [208]
and, from eq. 194,
Vd = -rId + xq'I q + Ed' [209]
Combining eqs. 206 and 209 into an equation in complex numbers
(analogous to eq. 200a), we get:
V = E' - rI - jXd'Id + xq'I q [210]
where
·E' = Ed' +iEq' [211]
may be called the voltage behind the transient impedance (or reactance)
*The relative sizes of the terms are approximately in the ratio 0 f w to liT, where
T is the time constant which expresses the rate of change of amplitude of the
current. In a 60-cycle machine w = 377, while liT lies between the limits of about
30 for the subtransient short-circuit time constant and about 0.1 for the transient
open-circuit time constant.
or the transient internal voltage. Its direct- and quadrature-axis com-

ponents are Ei and E/, respectively. This voltage is constant on
the assumption that the flux linkages of the rotor circuits, Vtf and Vto,
are constant. It is greater than the armature terminal voltage by
'tf>1
I
I
I
I
Id - _
I,, 1\
\~//
\
90'
/
-:
t \/
Direct
axis
FIG. 33. Vector diagram of synchronous generator in the transient state. Vector
E' remains constant during a rapid change of armature current I, but V and E
change.
the amount of the transient-impedance drop, which consists of re-

sistance and transient-reactance components. It is less than the
excitation voltage by the difference of the synchronous-impedance
and transient-impedance drops.
A vector diagram corresponding to eqs. 204, 206, 208, 209, and 210
is given in Fig. 33.
In the steady state preceding a change of armature current, the
excitation-voltage vector E lies on the quadrature axis; that is, Ed = O.
The corresponding values of Ei and E/ may be computed, respec-
tively, by adding lq(xq - x/) to 0 and by subtracting ld(xd - xi)
from E q • Then during a change of armature current E' remains
x q = xq
, E varies and may depart from the quadrature axis unless
constant;
•
Suppose that components I d and I q of the armature current undergo
changes of III d and IIIq, respectively. The corresponding changes in
other quantities are as follows:
IlVtf = 0 [212]
IlVto = 0 [213]
IlVtd = - Li tJ.ld [214]
IlVtq = - L/ III q 1215]
TRANSIENT STATE 77
3 MI La - La'
611 = - - Ma = Ma [216]
2 LII M1
3M L, - L/
st, = 2 L a si, M si, [217)
aa a
f:1E q = (xa - xi) f:1ia [218]
f:1E a = - (xq - x/) f:1i q [219]
f:1Ei = 0 [220]
l:1E/ = 0 [221]
f:1V q = - r f:1I q - xi f:1I d [222]
f:1Va = - r M d + x/ M q [223]
Salient-pole machine. Since a salient-pole machine has no quadra-
ture-axis field circuit, I a = 0 and Ea = 0, and the excitation voltage
E = jE q always lies on the quadrature axis in the transient state as
f--- - - -- - - Eq = E
k-- -- - - - Eqd - - - j - - -- - - , J
E~'= E '
FIG. 34. Vector diagram of salient-pole synchronous machine (x q' = x q ) . Voltage

vectors E, Eqd, and E' all lie on the quadrature axis, both in the transient state
and in the steady state.
well as in the steady state. Furthermore, since x/ = X q , Ei = Ed = 0,

and the voltage behind transient reactance, E' = jE/, also lies on the
quadrature axis, as does E qa , the voltage behind X q • The vector dia-
gram is shown in Fig. 34.
Turbogenerator. In the turbogenerator, which has a solid cylindrical
rotor, xi and x/ are very nearly equal, and Xa and X q are almost
equal. Consequently, the xl drops need not be split into direct- and
quadrature-axis components. The vector diagram for the transient
state is shown in Fig. 35.
Vector diagram for the subtransient state. In the preceding treat-

ment of the transient state, amortisseur windings or their equivalent
were assumed absent, because the equations that were used as a
starting point were derived under that assumption. If, however,
amortisseurs are present, their effect following a sudden change in
circuit conditions dies away in a period of a few cycles, after which
the relations just derived are valid.
f+---------Eq - - -- - - -
FIG. 35. Vector diagram of synchronous machine with solid, round rotor in the
transient state. Xd = Xq and xi = Xq'. In the steady state, Ed = 0 and E = E q,
but the same is not true in the transient state.
At the instant of a sudden change in circuit conditions, however,

the flux linkages of the amortisseur windings, as well as of the field
windings, are substantially constant. Under these conditions, the
fictitious voltage, E" = E/' + jE/', known as the voltage behind the
subtransient impedance, or the subtransient internal voltage, likewise
remains constant. Equations similar to those derived for the tran-
sient state (205, 206, 208, 209, 210, and 212 through 223) can be
written for the subtransient state by replacing E/ by E/', E/ by
E/', E' by E", x/ by x/', and x/ by x/'. The vector diagram for
the subtransient state is similar to Fig. 33, with the changes just noted.
APPLICATIONS TO TRANSmNT STABILITY STUDms
The period of mechanical oscillation of a machine during a dis-.
turbance is of the order of 1 sec., and the behavior of the machine
during the first quarter-period often suffices to show whether the sys-
TRANSIENT STABILITY STUDIES 79
tem is stable or unstable. The subtransient time constant of a ma-

chine is only-about 0.03 to 0.04 sec., which is short compared with the
period of mechanical oscillation. Therefore, in stability studies the
subtransient phenomena are disregarded. The armature time con-
stant is about 0.1 to 0.3 sec. for a short circuit at the armature termi-
nals, but it is greatly diminished by external resistance. Therefore
the d-e, component of armature current is ordinarily disregarded also,
though its effect may be appreciable in case of a fault at or near the
terminals of the machine.
The only electrical transient which need be considered in a stability
study is the "transient component." Its time constant is from about
0.5 to 10 sec., usually being longer than the period of mechanical oscil-
lation. Therefore the transient component cannot be ignored although
its decrement is frequently ignored.
The transient component can be taken into account in a number of
ways which differ as to the accuracy of the assumptions and the
amount of calculation required. The following different assumptions
can be made regarding a
Salient-pole machine:
1. The voltage behind direct-axis transient reactance may be
assumed constant (constant Ei~, Fig. 34); or
2. The flux linkage of the field winding may be assumed constant
(constant E q' , Fig. 34); or
3. Field decrement and voltage-regulator action (if any) maybe
calculated (varying E q ' ) .
These assumptions are listed in their order of increasing accuracy
and decreasing simplicity of calculation. Method 1 was employed in
earlier chapters of this book and is commonly used in practice. Meth-
ods 2 and 3 will be explained, illustrated, and compared with Method 1
in later sections of this chapter.
A machine with a solid cylindrical rotor has rotor circuits on both
direct and quadrature axes concerning which assumptions must be
made. Consequently, the following methods can be used for a
Round-rotor machine:
1. The voltage E l ' (Fig. 33 or 35) behind direct-axis transient
reactance may be assumed constant; or
2. The flux linkages of the rotor circuits on both axes may be
assumed constant (that is, constant Eel and E q' , Fig. 33); or
3. The effect of decrement and voltage-regulator action on the
flux linkage of the direct-axis rotor circuit (that is, on E q ' ) may be
80 SYNCHRONOUS l\'IACHINES
calculated, while the quadrature-axis circuit IS assumed open

(Ed' = 0); or
4. The decrement of the rotor flux linkages of both axes and the
effect of voltage-regulator action, if any, on the direct-axis linkage
may be calculated.
We will now consider Method 2 for a salient-pole machine. To do
so we need the transient power-angle curve of the machine.
Power-angle curves of a salient-pole machine. Steady state. The

electric power output of a salient-pole generator may be expressed in
terms of the terminal voltage V, the excitation voltage E q , the angle ~
between these two voltages, and the components of the synchronous
impedance r, Xd, and x q • All these quantities are shown in the vector
diagram, Fig. 34.
If eqs, 194 and 195 are solved for I d and I q, with Ed set equal to
zero, the result is:
Id =
- VdT (Eq - Vq)x q+ [224]
T + XdXq
2
I = (E q - Vq)r + VdXd [225]

q r2 + XdXq
The power output is
[226]
Substitution of the values of I q and I d into this expression yields:
P
_ VIl(- VdT + Eqxq - VqXq) + Vq(EqT - Vqr + VdXd )
r + XdXq
- 2
Eq(Vqr + VdXq) - (Vd2 + V (2)r + VdVq(Xd - X q)

= [227]
r2 + XdXq
Now
V q = V cos ~ and Vd = Vsino [228]
Hence
[229]
and
V d V q = V 2 cos ~ sin 0 = l V 2 sin 20 [230]
Substitution of eqs. 228, 229, and 230 into eq. 227 gives:
p =
EqV(r cos 0 + X q sin 0) - V 2r + lV2 ( Xd - xq) sin 2~
[231]
T
2
+ XdXq
POWER-ANGLE CURVES 81
By introduction of the abbreviations

Z2 = r 2 + XdXq [232]
Z q 2 = r2 +xq 2 [233]
'Y
· 1 -r = cos-1 -X q
= tan-1 -r = sln- [234]
Xq z, z,
eq. 231 may be further simplified to
= E q V Z2
Zq •
SIn (0 + l' z2r + V 2 Xd 2z- 2X
) 2 q •
P - V SIn 20 [235]
If a synchronous machine is operated with constant field current

and is connected to an infinite bus, both E q and V are independent of
o. The graph of power P delivered to the infinite bus versus angle a
between the excitation voltage and the bus voltage then consists of a
displaced sine wave (such as was previously found in the case of a
machine represented by constant impedance and internal voltage) plus
a second harmonic. The second-harmonic term represents the "reluc-
tance power" due to saliency, and is present even when the excitation
is removed. It depends upon the difference Xd - X q and thus disap-
pears in a nonsalient-pole machine.
The internal electric .power of the synchronous machine, which de-
termines the torque, is
[236]
This also, by substituting eqs. 224 and 225 in it, could be put in the
form of the sum of a constant term and of first and second harmonics
in o. However, the resulting expression is very complicated.
If the synchronous machine is connected to the infinite bus through
a series impedance, r includes the external resistance; likewise Xd and
X q include the external reactance. If the machine is connected to the
infinite bus through a network having shunt branches, the network
and the voltage of the infinite bus can be replaced by an equivalent
e.m.f. and series impedance according to Thevenin's theorem.
If the' series resistance is negligible compared with the reactance,
the power-angle equation (235) simplifies to
EqV . 2 Xd - Xq •
P=-slno+V SIn 28 [237]
Xd 2XdXq
The internal power, the output from the armature terminals, and the
input to the infinite bus are all equal in the absence of resistance, all
being given by eq. 237. The power-angle curve is plotted in Fig. 36.
It will be observed that the effect of the reluctance power is to steepen
the curve in the stable region near the origin.
Transient. The electric power output of a salient-pole machine
may be expressed in terms of the quadrature-axis voltage E q ' behind
transient reactance, the terminal voltage V, and the angle 0 between
Generator
.....
~--- ...---~,-----~----...-o
FIG. 36. Steady-state power-angle curve of salient-pole synchronous machine.
these two voltages. The expression is similar to eqs. 235 or 237 except
that E q in these equations is replaced by E q ' , and Xd by Xa'. The angle
ois the same in both cases because both E q and E q ' lie on the quadra-
ture axis. Thus eq. 237 becomes
Eq'V. V2 Xq - Xd' •
P = - , - SIll 0 - ,sIn 20 [238]
Xd 2Xcl xq
If both E q ' and V are independent of 0, the power-angle curve has

fundamental and second-harmonic components, as shown in Fig. 37.
The second harmonic of the transient curve is reversed in sign from
that of the steady-state curve because X q > ·Xd'. In other words, the
transient reluctance of the direct axis is greater than that of the
quadrature axis because of the presence of the closed field winding
whose flux linkage is constant: hence the transient "reluctance-power"
term is negative.
The effect of saliency. The knowledge of the transient power-angle

curve of a salient-pole synchronous machine enables us to use the
equal-area criterion to find either the critical switching angle or the
83
power limit of a system consisting of a salient-pole machine swinging

with respect to an infinite bus. The results of such an analysis may
then be compared with the corresponding results based upon the sim-
pler round-rotor theory; that is, upon the assumption of constant
voltage E/ behind direct-axis transient reactance. The simplest case
to analyze is t.hat. in which the machine is connected directly to the
--.r-- -7fL ~-_-~~-----a.- 0
FIG. 37. Transient power-angle curve of salient-pole synchronous machine.
infinite bus and in which resistance is negligible. This case also con-
stitutes a severe test of the validity of applying the usual round-rotor
analysis to a salient-pole machine because the further the ratio xa.'/x q
departs from unity, the more the power-angle curve of 'a salient-pole
machine departs from that of a round-rotor machine having the same
value of Xd'. The addition of external reactance makes Xd' /xq ap-
proach unity.
Two examples will now be given in which the results of a stability
calculation where saliency is neglected are compared with the corre-
sponding results where saliency is taken into account.
EXAMPLE 4
A salient-pole synchronous generator having the following per-unit
reactances
Xd = 1.15
Xq = 0.75 X q' = 0.75

is delivering current of 1.00 per unit at 0.91 power factor (lag) through a
circuit breaker to an infinite bus having a voltage of 1.00 per unit. If the
circuit breaker is then opened, how long may it be kept open before being
84 SYNCHRONOUS rvIACHINES
reclosed without loss of synchronism 'f Find the answer by plotting the
power-angle curve and using the equal-area criterion. Use each of the
following power-angle curves and compare the results.
1. Power for constant E q ' versus o.
2. Power for constant E/ versus 8i •
Additional data:
H = 2.5 kw-sec. per kva.
f = 60 cycles per second.
Solution. Initial conditions. Refer to the vector diagram, Fig. 34. Take
the terminal voltage V as reference vector. Then
V = 1.00/0 (a)
1= 1.00/-cos- 10.91 = 1.00/-24.5° (b)
c/> = 24.5° (c)
E qd = V + jxqI
= 1.00/0 + /90.0° X 0.75 X 1.00/-24.5~
= 1.00/0 + 0.75/65.5°
= 1.00 + 0.31 + jO.68
= 1.31 + jO.68 = 1.48/27.5° (d)
00 = 27.5° (e)
I q = I cos (ep + 0) = I cos (24.5° + 27.5°)

= 1.00 cos 52.0° = 0.616 (I)
I d = I sin (cP + 0)
= 1.00 sin 52.0° = 0.788 (g)
E q' = E qa - (xq - xd')I d

= 1.48 - 0.38 X 0.788
= 1.48 - 0.30 = 1.18 (h)
E/ =V + jXd'I
= 1.00 /0 + /90.0° X 0.37 X 1.00 / -24.5°
= 1.00/0 + 0.37 /65.5°
= 1.00 + 0.15 + jO.34
= 1.15 + jO.34 = 1.20 /16.3° (i)
DiD = 16.3° (j)
P = VI cos cP = 1.00 X 1.00 X 0.91 = 0.91 (k)

Also
P = Eqdl q = 1.48 X 0.616 = 0.91 (Check) (l)
EXAMPLE 4 85
and
P = E/I cos (cP + a·i ) = 1.20 X 1.00 cos 40.8°
= 0.91 (Check) (m)
Power-angle curves. Assumption 1. By eq. 238

Eq'V. ~ V 2 x q - , Xd' sm
P 1=--,slnu- • 2~
zo (n)
Xtl 2Xtl X q
= 1.18 X 1.00 ffinu-

. ~ (1.00)2 X 0.38. · 2~
mn u
0.37 2 X 0.37 X 0.75
= 3.18 sin 0 - 0.685 sin 20 (0)
The computation of P1 is given in Table 4.
TABLE 4
COMPUTATION OF PI, ASSUMPTION 1 (EXAMPLE 4)
a 20 sin 0 sin 20 3.18 sin 0 -0.685 sin 20 Pi

0 0 0 0 0 0 0
150 30 0 0.259 0.500 0.82 -0.3! 0.48
30 0 60 0 0.500 0.866 1.59 -0.59 1.00
450 90 0 0.707 1.000 2.25 -0.68 1.57
60 0 1200 0.866 0.866 2.76 -0.59 2.17
750 1500 0.966 0.500 3.08 -0.34 2.74
900 1800 1.000 0 3.18 0 3.18
1050 210 0 0.966 -0.500 3.08 0.34 3.42
1200 240 0 0.866 -0.866 2.76 0.59 3.35
1350 270 0 0.707 -l.000 2.25 0.68 2.93
1500 300 0 0.500 -0.866 1.59 0.59 2.18
1650 330 0 0.259 -0.500 0.82 0.34 1.16
1800 360 0 0 0 0 0 0
27.5 0 55 0 0.462 0.819 1.47 -0.56 0.91
..488umption~. The power-angle curve is a sine.

ElV. ~ 1.20 X 1.00 · ~ · ~
P 2 = --, SIn ui = SIn Ui = 3.24 SIn Ui (p)
Xd 0.37
Values of P2 are computed in Table 5.
Both power-angle curves are plotted in Fig. 38. When the breaker is open
the power output of the generator is zero; therefore the power-angle curve
for this condition is the horizontal axis.
Ctiticalswitching angles. The critical angles at which the breaker must be
closed are found graphically in Fig. 38 by use of the equal-area criterion.
For assumption 1 the angle is 116°; for assumption 2 it is 104°. For assump-
tion 2 the angle can be checked analytically, as follows:
TABLE 5
COMPUTATION OFP2, ASSUMPTION 2 (EXAMPLE 4)
ai sin 8i P2
o and 180 0
0 0
150 " 165 0 0.259 0.84
30° "u 1500 0.500 1.62
45 0 1350 0.707 2.29
u
60 0 120 0 0.866 2.80
75° " 105 0 0.966 3.13
90° 1.000 3.24
16.3° 0.281 0.91
The net area is the area under the sine curve to the right of the critical
angle less the area of the rectangle under the input line. Hence the area
under the curve equals the area of the rectangle, or
1
180 0 - 150
3.24 sin Oi dO i = 0.91(1800 - 200) (q)
c5 c
where 00 is the initial angle, which is 16.3°.

163.7 0
-3.24 X 57.30 cos Oi
J6c
= 0.91(1800 - 32.6°) (r)
-185.4° (cos 163.7° - cos oc) = 0.91 X 147.4°
0.960 + cos d, = 0.91 X 147.4° = 0.723

185.4°
cos oc = 0.723 - 0.960 = -0.237
s, = 1040 (Check) (8)
As long as the breaker is open, P« = 0, Pi = 0.91 per unit, and Po =

Pi - P u = 0.91 per unit. The acceleration during this time is constant at
the value
d 20
-=-
r, (t)
dt2 M
The angular speed at the instant of opening the breaker is zero. After the
breaker has been open for t seconds the speed is
~~
dt
=1&oMPa dt == Pat
. ~f
(u)
and the angular displacement is
o == 00 +
Jo
r Pat M
== 00 + Pat
2M
2
(v)
EXAMPLE 4 87
4 , . . - - - - - . - - - - - - . - - - - - - , - - - -,---- - -.--
3f----+----1-:;.,L---:~+--~
:!:
c:
:J
~ 2 f - - - -+---/-- Lt-----+---tf
~
30· 60· 90· 180·

oor 0;
FIG. 38. Determination of critical switching angle by the equal-area criterion
under two different assumptions. Curve 1 - saliency taken into account; P
plotted against 0 for constant. E,/. Curve 2 - saliency neglected; P plotted against
Oi for constant E;'.
The critical angle oe corresponds to the critical time te •
" _"uo + -
Ue - te
P a-
2
(w)
2M
(x)
The inertia constant is

M = GH = 1 X 2.5 = 2.32 X 10-4 (y)
180! 180 X 60
Substitution of the values of M and P II into eq. x gives
t =
e
2 X 2.32 X 1O-
4
(oe- (0) = ~oe - 00 (e)
0.91 1,965
For assumption 1, oe = 116 %
0 = 27.5°, and oe- 00 = 88.5°. The critical
time is
t
e
= ~1,965
88.5 = 0.212 sec. (00)
For assumption 2, Oe = 104% 0 = 16.3°, and oe - 00 = 87.7°. The critical

time is
~
7.7
te = - - = 0.211 sec. (bb)
1,965
Conclusion. In the present example the critical switching time computed
on the basis of constant flux linkage of the field winding (taking saliency
into account) agrees within the probable error of the graphical method with
the value of critical switching time computed on the basis of constant voltage
behind direct-axis transient reactance (saliency neglected) .
EXAMPLE 5
If the salient-pole synchronous generator of the last example is initially
operating as described in that example, to what value could the input power
suddenly be increased without loss of synchronism? Work the problem on
each of the two assumptions made in the last example and compare the
results.
Solution. The power-angle curves are the same as in the last example
(Fig. 38). They are copied in Figs. 39 and 40, and there the equal-area
criterion is applied graphically to find the limiting values of power. The
answer is 2.60 per unit by assumption 1 and 2.61 per unit by assumption 2.
The discrepancy is less than the probable error of the graphical solution.
4r----,....-------r-------r----~----.,---~
FIG. 39. Determination of power limit by the equal-area criterion, with saliency
taken into account (Example 5).
Examples 4 and 5 plainly show that practically the same results are
obtained through the use of the usual simple assumption of constant
voltage E/ behind direct-axis transient reactance as through use of
the more rigorous salient-pole theory in which the flux linkage of the
field winding is assumed constant. * It is reasonable to suppose that
·Concordia, in Ref. 59, p. 11, confirms the almost negligible effect of saliency by
giving calculated curves of transient power limit of a 600-mile line with saliency
first taken into account and then neglected.
SALIENT-POLE MACHINES 89
an equally good agreement would be obtained in systems more com-
plicated than the one considered, for example, in a system having
several salient-pole machines. Therefore, of the two assumptions, it
is natural to prefer the one which is simpler to use, namely, the assump-
tion of constant E/.
4 , - - - - , - -- - - ,--- -- - , - - -- - , -- - -- , - - - - - ,
FIG. 40. Determination of power limit by the equal-area criterion, with saliency
neglected (Example 5). The result agrees closely with that obtained in Fig. 39.
If, however, the assumption of constant flux linkage is not accurate

enough and it is desired to include in the analysis the effect of change
of linkage due to decrement or to decrement and voltage-regulator
action, then saliency must be taken into account; that is, the two
rotor axes must be considered separately. This is true for nonsalient-
pole machines as well as for salient-pole machines, and is true even
1if Xd I = X q I •
In the next fewsections of this chapter the methods of taking change
of rotor flux linkages into consideration will be set forth.
Representation of salient-pole machines. On a calculating board

there is no reactor that presents different reactances to components of
current Id and I q • A salient-pole machine can be represented, how-
ever, by a constant reactance* equal to the quadrature-axis synchro-
nous reactance X q , in series with a power source whose voltage is
E qd (Fig. 41). This representation has the advantage that the phase
of voltage E qd represents correctly the angular position of the field
*Armature resistance can be represented, if desired, but it is usually negligible.
~o SYNCHRONOUS l\lACHINES
structure. For, if the quantity jxqI be added vectorially to the termi-

nal voltage V, the vector sum Eqd always lies on the quadrature axis
(see the vector diagram, Fig. 34).
The magnitude of E qd is not constant, however, even if Eq ' is con-
stant. The relation hetweenEqd and E q ' is:
E qd = E q' + (x q - xd')ld [239]
If E q ' is constant, E qd changes with any change in I d caused by swing-
ing of the machines or by circuit changes, such as the clearing of a
fault.
Xq
I
FIG. 41. Method of representing a salient-pole synchronous generator on a

calculating board.
The power output of the machine is given correctly by the reading

of a wattmeter taking voltage from either side of the reactor (that is,
either E qd or V). Usually the wattmeter is connected to the terminals
of the power source (voltage E qd ) . The power can be calculated as
Eqdl q•
The procedure during point-by-point computation of swing curves
is as follows: Suppose that the adjustments have been made and that
the power outputs of all machines have been read for one instant of
time. The change of angular position of every machine is then com-
puted as usual. The power sources are turned into their new angular
positions, their voltage magnitudes (representing E qd ) being left un-
changed for the moment. Then I d of every machine is read. * On
the assumption that Eq' is constant, a new value of Eqd of every
machine is calculated by eq. 239. The voltages of the power sources
are adjusted to the new values of E qa. This adjustment alters the
values of I d. Hence I d of every machine should be measured again,
and if any significant changes are found, the values of E qa should be
*1d can be found from the readings of a voltmeter (reading Eqd) and a varmeter
the current coil of which'carries I and the potential coil Eqd. The varmeter reading
is Eqd I sin (cP + 8), which, divided by Eqd, gives I sin (cP + 8) = Id. On the
Westinghouse network calculator I d can be read directly from the vector meters.
CHANGE OF FIELD FLUX 91
recalculated and the voltages of the power sources readjusted accord-

ingly. This procedure is repeated as many times as necessary, but
one or two adjustments usually suffice. Then the power outputs of
the machines may be read again for the next point.
The method just described for representing a salient-pole machine
on a calculating board can be applied also to representation of the
network on paper. However, the solution of the network on paper is
more laborious than the solution on a calculating board, particularly
since the network may need to be solved more than once for each
point of the swing curves. The power output of each machine can be
calculated as Eqdlq.
If E q ' varies because of decrement or voltage-regulator action, the
procedure is the same as that just described except that new values
of E q ' must be calculated for each instant. The method of doing this
is described in the next section.
Calculation of change of field flux linkage. Kirchhoff's voltage law,

applied to the field circuit, gives the equation:
E ex ' = Rf~1
·
+ at
d1/;1 [240]
where E ex ' is the exciter armature e.m.I, in volts, RJ is the field-

circuit resistance in ohms, if is the field current in amperes, and 1/1/ is
the field flux linkage in weber-turns. This equation may be put into
terms of the armature circuit by multiplying it by CJJMJ/Rj and making
certain substitutions. Multiplication by wM,/R, gives
wMjEer&' _ ~{. + wM j d1/;,
[241]
Rj - uu J~I R, dt
Here, by eq. 196,

w},{JiJ = Eq [242]
Also, by use of eqs. 76 and 201,
wMI'Yt = LjJ wMp/lj = TclO'E,/ [2431

RI RJ Lj /
A new quantity, the exciter voltage referred to the armature circuit,
is defined by
wMjE.r&' _ E
Rj - ex [244]
It is the open-circuit armature voltage which would be produced by

exciter voltage E ex ' in the steady state. Substitution of eqs. 242, 243,
and 244 into eq. 241 gives
e; = e, + T do, dE
dJ: q'
[245]
or, solving for the derivative,

dE q ' _ E ex - Eq
[246]
dt - T dO'
Thus the rate of change of E q ' (which is proportional to the field flux
linkage) is expressed in terms of the open-circuit transient time con-
stant T dO' even though the armature may not be open-circuited. In
this equation E ex either may be constant or may vary due to voltage-
regulator action independently of the other variables in the equation.
In either case it is assumed that E ex is a known function of t. E q
depends upon E q ' and I d, thus (eq. 204):
E q = E q' + (Xd - xd/)I d [247]
and III depends upon the phase and magnitude of E qd of every machine
in the system. (The relation between E qa and E q ' is given by eq. 239.)
Three-phase short circuit at terminals of machine. It is known from
considerations stated in the section of this chapter entitled "Time
constants of synchronous machines" that when a three-phase short
circuit occurs on the armature terminals of a synchronous generator,
the exciter voltage being constant, the transient a-c. component of
armature current and the transient d-c. component of field current
both decay exponentially with the transient short-circuit time constant
[248]
That this is consistent with eq. 246 will now be shown. During the
short circuit the terminal voltage of the machine is zero and hence the
short-circuit current is
[249]
where E q ' has the same value just after the fault occurs as it had just
before the fault occurred. From the last equation,
,
E q / -- E q Xd [250]
Xd
CHANGE OF "FIELD FLUX 93
Substitute this into eq. 24o, obtaining:

Xd' dEq Ee:r; - Eq
--= , [251]
»« dt T dO
Dividing by Xd' jXd,
dEq E ex - E q
-= , [252]
dt (::) T dO'
The only dependent variable in the last equation is E q , and its varia-
tion is controlled by the short-circuit time constant Td'. The field
current If is always proportional to E q , by definition of E q , and, as
long as the short circuit is on, by eq. 249, E q ' and I d are proportional
to E q • Hence E q , E q ' , Is, and If all vary exponentially with time
constant T d'., which was to be shown.
When the armature is open-circuited, eq. 249 does not hold, but,
from eq. 247, since I d = 0, E q = E q ' . Then eq. 246 can be put in
terms of either E q or E q ' instead of both, and the equation shows that
the variation of both E q ' and E q (hence also of 1/) occurs with time
constant T dO'.
Point-by-point calculation. Earlier in this chapter it was stated that
the accuracy of point-by-point calculation of transient 'current in an
R-L circuit is improved if, in calculating the average rate of change of
current during a time interval ilt, half a time interval (ilt/2) is added
to the time constant and if the value of applied voltage at the middle
of the interval is used. Obviously the result of adding ilt/2 to T dO'
in eq. 246 would not be the same as the result of adding t1tj2 to T d'
in eq. 252. The procedure is correct for the latter equation, whose
form is similar to that of eq. 64, the solution of which was previously
considered. What, then, should be the correction to T dO' in order to
produce a similar improvement of accuracy in solving eq. 246? Let
numerator and denominator of the right-hand member of this equa-
tion be multiplied by E q ' /E q • Then
E q' )
e; -
I
dEq' ( E; Ell
- = [253]
dt (~:) TdO'
For a fixed ratio Eq'jEq, this equation shows that E q ' varies with a
time constant (E q ' /E q ) T dO' . (For the condition of short circuit,
E q' jEq = Xd' /Xd and hence the time constant is T d ' . But on open
circuit Eq'/Eq = 1, and the time constant is T dO' . ) For improved
accuracy, the denominator of eq. 253 should be increased by ~t/2,

and the value of E ex should be taken at time t ~t/2; thus: +
,
(Ii;
~ql)
MJ '
q
Eez(t + Iit/2) - Eq
--=
6.t (E-E
q' ) ~t
Tdo +--
,
2
[254]
Here the derivative has been replaced by the average slope. Now,
after numerator and denominator have been divided by E q ' /E q ,
eq. 254 becomes
aE q' Ee:e(t + at/2) - Eq
--=
~t
Tao
,
+
(E q ) ~t
Eq' "2
[255]
which shows that the time constant T dO'· in eq. 246 should be in-
creased by (Eq/Eq') (Ilt/2). Since the correction term is a minor one,
its value need not be changed at every step of a point-by-point cal-
culation even though the ratio Eq/Eq' varies somewhat.
EXAMPLE 6
A three-phase short circuit occurs at the terminals of a salient-pole gen-
erator which is initially operating at no load and at rated terminal voltage.
The constants of the generator are:
Xd = 1.05 per unit Xq = Xq' = 0.69 per unit
Xd' = 0.29 per unit T dO' = 5.0 sec.
Calculate and plot excitation voltage E q and quadrature-axis transient
voltage E q' as functions of time for each of the two following conditions:
(a) Constant exciter voltage Ee~.
(b) Linear exciter-voltage build-up of 2.5 units per second.
Solution. Before occurrence of the fault, at no load,
E f/ = E q = V = E6~ = 1.00 unit (a)
(b)
Immediately after occurrence of the fault, by the law of constant flux

linkage,
E q ' = 1.00 unit (c)
Also
Eu : = 1.00 unit (d)
v=o (e)
EXAMPLE 6 95
Id = -E q ' = -1.00 = 3.45 units

.
(I)
Xd' 0.29
I, = 0 (g)
Eq = Eq' + (Xd - xd')Id = xd'I d + (Xd - xd,')Id = xdl d

= 1.05 X 3.45 = 3.62 units (h)
E q
, = 3.62 = 3.62 (i)
s, 1.00
This ratio remains constant while the fault is on. For the point-by-point
calculation, take an interval ~t = 0.2 sec. The equation governing the
change of E q ' is eq. 255:
ss; + ~t/2) - E
E,z(t q E,z(t + dt/2) - E q
~ = T d O' + (E q/Eq')(At/2) 5.0 + 3.62 X 0.1

E,z(t + At/2) - E q
(j)
5.36
In condition a the exciter voltage has the constant value E ez = 1.00 unit.
During the first interval,
~Eq' 1.00 - 3.62 -2.62 · d
-- = = - - = - 0.489 unit per secon (k)
~t 5.36 5.36
hence
~E ' = ~Eq'
q~t
At = -0.489 X 0.2 = -0.0978 unit (l)
At the end of the first interval

E q ' = 1.00 - 0.098 = 0.902 unit (m)
The computation is continued in Table 6.

In condition b the exciter voltage increases linearly according to the
equation
E,z = 1.00 + 2.50t units ·(n)
During the first interval, the average value of E ez is

E,z(t + at/2) = 1.00 + 2.50 X 0.1 = 1.25 units (0)
and the average rate of change of E,/ is, by eq. i.
t:.Eq' = 1.25 - 3.62 = ~2.37 = -0.441 unit pel' second (p)
At 5.36 5.36
The computation is continued in Table 7.
Curves of E q, E q ' , and E ex for the two cases are plotted in Fig. 42.
Conclusions. The assumption of constant flux linkage of the field winding
appears at first sight to be a poor one because the flux linkage decr.eases in
TABLE 6
POINT-By-POINT COMPUTATION OF E(/ AND E q
CONDITION a: C.oNSTANT EXCITER VOLTAGE E ez = 1.00 (EXAMPLE 6)
t E'
q Id Eq s.; E ez - Eq 6Eq'
(sec.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.)
0+ 1.000 3.45 3.62 1.00 -2.62 -0.098

0.2 0.902 3.11 3.26 1.00 -2.26 -0.084
0.4 0.818 2.82 2.96 1.00 -1.96 -0.073
0.6 0.745 2.57 2.70 1.00 -1.70 -0.063
0.8 0.682 2.35 2.47 1.00 -1.47 -0.055
1.0 0.627 2.16 2.27 1.00 -1.27 -0.047
1.2 0.580 2.00 2.10 1.00 -1.10 -0.041
1.4 0.539 1.86 1.95 1.00 -0.95 -0.035
1.6 0.504 1.74 1.82 1.00 -0.82 -0.031
1.8 0.473 1.63 1.71 1.00 -0.71 -0.027
2.0 0.446 1.54 1.62 1.00 -0.62 -0.023
2.2 0.423 1.46 1.53 1.00 -0.53 -0.020
2.4 0.403 1.39 1.46 1.00 --0.46 -0.017
2.6 0.386 1.33 1.40 1.00 -0.40 -0.015
2.8 0.371 1.28 1.34 1.00 -0.34 -0.013
3.0 0.358 1.23 1.30 1.00
TABLE 7
POINT-By-POINT CoMPUTATION OF E g ' AND E g
CONDITION b: LINEAR BUILD-UP OF EXCITER VOLTAGE (EXAMPLE 6)
t E'
q Id Eq s.; s; -Eg 6E q'
(sec.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.)
0+ 1.000 3.45 3.62 1.25 -2.37 -0.088

0.2 0.912 3.14 3.30 1.75 -1.55 -0.058
0.4 0.854 2.94 3.09 2.25 -0.84 -0.031
0.6 0.823 2.84 2.98 2.75 -0.23 -0.009
0.8 0.814 2.81 2.95 3.25 +0.30 +0.011
1.0 0.825 2.84 2.98 3.75 +0.77 +0.029
1.2 0.854 2.94 3.09 4.25 +1.16 +0.043
1.4 0.897 3.09 3.25 4.75 +1.50 +0.056
1.6 0.953 3.28 3.45 5.25 +1.80 +0.067
1.8 1.020 3.52 3.70 5.75 +2.05 +0.076
2.0 1.096 3.78 3.97 6.25 +2.28 +0.085
2.2 1.181 4.07 4.28 6.75 +2.47 +0.092
2.4 1.273 4.39 4.61 7.25 +2.64 +0.098
2.6 1.371 4.73 4.96 7.75 +2.79 +0.104
2.8 1.475 5.08 5.33 8.25 +2.92 +0.109
3.0 1.584 5.46 5.74
EXAMPLE 7 97
/
5
,
4 ,/
, /
/
..
Eex ."".,.,.-
""".".",. "".",,--
--...,.......
====------L
(a)
~......-.---._-------- ~---------
Eq=Eq'=Eex
11III::::-.::::- -7iJ -CEq'
(a) --- _
o
Time t (seconds)
FIG. 42. Variation of E q and E q' during three-phase short circuit (a) with constant
exciter voltage, E ez = 1, (b) with linear build-up of exciter voltage, E ez = 1 2.5t +
(Example 6).
0.5 sec. to 77% of its initial value when the exciter voltage is constant and
to 83% when the exciter voltage rises at the assumed rate. It should be
remembered, however, that in stability studies the fault is often assumed to
be one of less severe type than a three-phase fault and at a less severe location
than at the terminals of the generator; furthermore, the fault is not sustained,
but is cleared in a few tenths of a second.
EXAMPLE 7
The generator described in the last example undergoes a three-phase short
circuit at its terminals which is cleared 0.2 sec. later. The exciter voltage is
constant. Calculate and plot E q ' as a function of time.
Solution. While the short circuit is on the generator, E q and E q ' vary as
calculated in the last example, part a. At the end of 0.2 see., E'l = 3.26 and
E q' = 0.902. Just after the clearing of the fault, E q = E q ' = 0.902. Since
Eq/Eq' = 1, eq. j of the last example becomes:
dE'l' E ez - E'l' 1 - E q'
(a)
at = TdO' + at/2 = 5.0 + 0.1
I1E,/ = 1 - E,/ I1t = (1 _ E,/) 0.2 =1 - E,/ (b)
5.1 5.1 25.5
The computation is carried out in Table 8. A curve of E q ' versus t is plotted

in Fig. 43.
TABLE 8
POINT-By-POINT COMPUTATION OF E q' AND Eq (EXAMPLE 7)
l
(sec.)
I. --- E,/
0+ 1.000 -0.098
0.2 0.902 +0.098 +0.004
0.4 0.906 +0.094 +0.004
0.6 0.910 +0.090 +0.004
0.8 0.914 +0.086 +0.003
1.0 0.917 +0.083 +0.003
1.2 0.920 +0.080 +0.003
1.4 0.923 +0.077 +0.003
1.6 0.926 +0.074 +0.003
1.8 0.929 +0.071 +0.003
2.0 0.932
The flux linkage drops to 90% of its initial value while the fault is on;
then, after the fault is cleared, it rises very slowly.
1.0 -----------~-------
0.8 \Fault cleared
~ 0.6
:::s
l
"-'
~O.4
0.2
0'----.....
0 1 2 ------....100----
Time i (seconds)
FIG. 43. Variation of E q' during and after three-phase short circuit cleared in
0.2 sec. (Example 7).
If the fault were cleared in 0.1 sec., the flux linkage would decrease only
to about 95% of its initial value.
EXAMPLE 8
The salient-pole synchronous generator described in Example 4 is deliver-
ing power to an infinite bus under the initial conditions specified in that
EXAMPLE 8 99
example. A three-phase short circuit, occurring at the terminals of the

generator, is cleared in 0.20 sec. without disconnecting the generator from
the bus. Calculate and plot swing curves of the generator for each of the
following two conditions: (a) field flux linkage of generator assumed con-
stant; (b) field decrement taken into account, Tdlj' = 5.0 sec.
Solution. During fault. During the 0.2-sec. period in which there is a
three-phase fault on the terminals of the generator, the output of the gen-
erator is zero and the accelerating power Pais equal to the mechanical power
input Pi, which is assumed to be constant. From Example 4, the input is
0.91 per unit, the initial angular displacement is 00 = 27.5°, and the inertia
constant is M = 2.32 X 10- 4 • The swing curve can be calculated by eq. w
of Example 4:
o= 0 + P a t2 = 27.50 + 0.91 t2
o 2M 2 X 2.32 X 10- 4
= 27.5° + 1,964t2 (a)
This calculation is performed in Table 9.
TABLE 9
COMPUTATION OF SWING CURVE
DURING PERIOD WHEN FAULT IS ON
(EXAMPLE 8)
t
t2
1,964t 2 a
(sec.) (deg.) (deg.)
0 0 0 27.5
0.05 0.0025 4.9 32.4
0.10 0.0100 19.6 47.1
0.15 0.0225 44.1 71.6
0.20 0.0400 78.5 106.0
The swing curve during this period is the same whether or not field decre-
ment is taken into account. The decrement during this period can be com-
puted as a simple exponential,
E q' = Eq'(O)E-t / Tl (b)
The initial value of E q' , from Example 4, is 1.18 per unit. The short-circuit
time constant is
TdI = -Xd' T' 0.37
dO = - X 5.0 = 1.61 sec. (c)
Xd 1.15
Hence
(d)
Values of E q' while the fault is on are calculated in Table 10. Eq'decreases
about 12% during this period.
TABLE 10
COMPUTATION OF E t/ DURING PERIOD WHEN FAULT IS ON
(EXAMPLE 8)
t E- t / 1•61 E'q
t/1.61
(sec.) (p.u.)
0 0 1.000 1.18
0.05 0.031 0.969 1.14
0.10 0.062 0.940 1.11
0.15 0.096 0.909 1.07
0.20 0.124 0.883 1.04
After clearing of fault - field flux linkage con8tant. The swing curve will be
computed point by point, using a time interval At = 0.05 sec. The second
difference of angular position ~ is
2
.!l2a = (.!It) P = (0.05)2 P = lO.8P (e)
M a 2.32 X 10- 4 a a
where Pais the accelerating power in per unit and equals Pi - P u = 0.91 -
P u. The power output P u will be calculated from power-angle equation 0
of Example 4:
P u = 3.18 sin 0 - 0.685 sin 20 (f)
which was plotted in Fig. 38, Curve 1. The point-by-point calculation,
given in Table 11, starts at t = 0.20 sec. with ~ = 106.0° (from the last line
TABLE 11
COMPUTATION OF SWING CURVE AFTER CLEARING OF FAULT
FIELD FLUX LINKAGE ASSUMED CONSTANT (EXAMPLE 8)
t -0.685 P u Pa A 28 A6 8
sin 8 sin 28 3.18 sin 8
(sec.) sin 28 (p.u.) (p.u.) (deg.) (deg.) (deg.)
--
0.20- .. .. .. .. 0 +0.91
0.20+ 0.961 -0.530 3.06 +0.36 3.42 -2.51 +34.4
0.20 avg. .. . . .. .. 1.71 -0.80 ~8.6 106.0
+25.8
0.25 0.746 -0.994 2.37 +0.68 3.05 -2.14 -23.1 131.8
+2.7
0.30 0.713 -1.000 2.27 +0.68 2.95 -2.04 -22.0 134.5
-19.3
0.35 0.905 -0.770 2.88 +0.53 3.41 -2.50 -27.0 115.2
-46.3
0.40 0.933 +0.672 2.96 +0.46 2.50 -1.59 -17.2 68.9
-63.5
0.45 .. .. .. .. .. .. .. .. 5.4
EXAMPLE 8 101
of Table 9). The preceding difference ~5 is taken as the difference of the

last two values of 0 in Table 9 : 106.0° - 71.6° = 34.4°. Because there is a
discontinuity in P a at t = 0.20 sec., the average of the values before and
after clearing the fault must be used.
After clearing of fault - field decrement takeninto account. The swing curve
will be computed point by point as before except that the power output is
now given by
P« = 2.70Eq' sin 0 - 0.685 sin 20 (g)
where E q ' is no longer constant but is computed point by point as follows:
By eq. 255 the change of E q ' in time Llt is
ss, = E. z - Eq t!.t =
q)
(1.80 - E q )0.05 (h)
T '+ (Eq)~t
dO E' 2 q
5.0 + (E 0.025
EqI
The exciter voltage E ez in this equation is assumed constant and is equal to
the value of E q before occurrence of the fault. From information in Example
4, this value is
E q = E qd + Id(xd - xq) = 1.48 + 0.788(1.15 - 0.75)
= 1.48 + 0.32 = 1.80 per unit
The excitation voltage after occurrence of the fault is computed as follows:
E q = E q' + Id(xd - Xa') = E q' + 0.78I a (i)
and, by eqs. 224 and 228, with r = 0 and E q and Xd replaced by E q ' and
Xd', respectively,
I = E q ' - Vcoso E q' - 1.00 cos 0 (j)

d
Xd' 0.37
For example, just after the fault has been cleared, E q ' = 1.04 per unit and
o= 106.0°, the same as just before clearing. Equation j gives:
1.04 + 0.28 3 57 it
I d= 1.04 - 1.00 cos 106.0° =
1.32
=-=. perunl
0.37 0.37 0.37
By eq. i,
E q = 1.04 + 0.78 X 3.57 = 1.04 + 2.78 = 3.82 per unit
and by eq. b,
t!.Eq' = (1.80 - 3.82)0.05 -2.02 X 0.05
-0.020 per unit
5.0 + 3.82 X 0.025 5.092
1.04
Therefore, at t = 0.25 sec.,
E q ' = 1.04 - 0.02 = 1.02 per unit
The remainder of the point-by-point calculation is carried out in Table 12.
c'
t\j
TABLE 12
COMPUTATION OF SWING CURVE AFTER CLEARING OF FAULT
(FIELD DECREMENT TAKEN INTO ACCOUNT) (EXAMPLE 8)
t E'
q E q' - coso Id 0.78 I d Eq 1.80 - E q !lEq' 2.70E q ' -0.685 P u Pa ~20 ~o 0
cos 0 sino sin 20 rn
(sec.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) sin 0 sin 20 (p.u.) (p.u.) (deg.) (deg.) (deg.) ~
-- -- - - -- az
0.20- 1.04 0 +0.91 tI:
0.20+ 1.04 -0.28 1.32 3.57 2.78 3.82 -2.02 0.961 -0.53 2.70 +0.36 3.06 -2.15 34.4 =:c
0.20avg. 1.04 1.53 -0.62 -7.3 106.0 o
z
-0.02 27.1 o
0.25 1.02 -0.68 1.70 4.60 3.58 4.60 -2.80 0.730 -0.998 2.01 +0.68 2.69 -1.78 -19.2 133.1 ~
-0.03 7.9
~
0.30 0.99 -0.78 1.77 4.78 3.73 4.72 -2.92 0.629 -0.978 1.68 +0.67 2.35 -1.44 -15.5 141.0 ;>
-0.03 -7.6 o
::Q
-2.64 -1.66 -17.9 133.4 ~
0.35 0.96 -0.69 1.65 4.46 3.48 4.44 0.727 -0.998 1.89 +0.68 2.57
Z
-0.03 -25.5 t:tj
00
0.40 0.93 -0.31 1.24 3.35 2.62 3.55 -1.75 0.951 -0.585 2.39 +0.40 2.79 -1.88 -21.4 107.9
-0.02 -46.9
0.45 0.91 .. .. .. .. .. .. .. 0.875 +0.848 2.15 -0.58 1.57 -0.66 -7.1 61.0
-54.0
0.50 .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. 7.0
EXAMPLE 8 103
Conclusions. The swing curves for the two conditions are plotted in Fig.
44. The difference between them appears considerable. However, the
difference in the maximum angular displacements is only 4° or 5°.
In Example 4, with the same initial value of power as in the present
example, the critical time during which the generator could be disconnected
from the bus WIl.B found to be 0.212 sec. In t.he present example I\. three-
150 r - - - , - - - r - - - - ,-- --,-
125
"'",
~100
'"u
~
'"
.e
,
C>
~
'0
c: 75
III
";ii'
'"
'"
liD
'"
~
'0 50
251---+---+----+---+--+-+-1
0.1 0.2 0.3 0.4 0.5

Time (seconds) after occurrence of fault
FIG. 44. Effect of field decrement on swing curves (Example 8).
phase short circuit is assumed instead of the disconnection of the generator

from the bus. In both cases the power output of the generator is zero during
the disturbance. Therefore, for equal switching times the swing curves
would be identical for both types of disturbance, if the field flux linkage
were assumed constant in both cases. Furthermore, with the assumption
of constant flux linkage, the critical switching time is the same for both dis-
turbances; namely, 0.212 sec. Since the three-phase fault actually decreases
the flux linkage, the system would be unstable with 0.212-sec. clearing if
decrement were taken into account. Nevertheless, decreasing the clearing
time to 0.200 (a decrease of only 6%) was found to make the system stable
with some margin. The error in determining the critical switching time due
to neglecting field decrement in this example is, then, less than 6% - perhaps
4% - in spite of the fact that the flux linkage has decreased about 16% at
the instant of maximum swing.
The last example has shown that the error introduced hy the as-
sumption of constant flux linkage is hardly great enough, in the case
considered, to warrant the additional complication of calculating the
field decrement. But are cases likely to arise in which the error
would be much greater?
A less severe type or location of short circuit would permit either
slower clearing or the transmission of more power without loss of syn-
chronism. If the first alternative is chosen, the increased fault dura-
tion is offset by the decreased m.m.f. of the fault current, and the
decrease of flux linkage during the fault period is changed but little.
If the second alternative is chosen, there is less decrease of flux linkage
during the fault.
An increase in the inertia of the generator would also make possible
either slower clearing or the transmission of more power without loss
of synchronism. The first alternative would give greater reduction of
flux linkage during the fault than that encountered with 0.2-sec. clear-
ing. ,If the inertia were multiplied by four, giving H = 4 X 2.5 =
10 kw-sec, per kva., the critical clearing time for constant flux linkage
would be doubled, making it 2 X 0.21 = 0.42 sec., and the decrease in
flux linkage during the fault would also be very nearly doubled, mak-
ing it 24%. The time to reach maximum angular displacement would
likewise be doubled, and the decrease in flux linkage up to this time
would be nearly doubled, making the decrease about 30%. Without
doubt, the critical clearing time would be decreased by a significant
amount, perhaps by 7 or 8%.
The modern trend, however, is to clear faults as rapidly as possible.
Clearing times greater than 0.2 sec. are seldom used on important
circuits where system stability is a consideration.
Generator voltage regulators, if used, tend to counteract the de-
crease of flux linkage resulting from the fault and from the subsequent
swinging apart of the machines. Regulator and exciter action is
treated in the next chapter. Damping, which has been neglected"
somewhat improves stability.
It may be concluded that the assumption of constant flux linkage
is a reasonably accurate one unless the time of clearing severe faults
is greater than 0.2 sec.
One more example of decrement calculation will be given in order
EXAMPLE 9 105
to illustrate the procedure applicable to a multimachine system and

to illustrate the effect of a less severe fault than that in the last example.
EXAMPLE 9
A hydroelectric station is supplying power over a double-circuit trans-
mission line to a metropolitan receiving system which may be regarded
as an infinite bus. Initially the hydroelectric station has an output of 1.00
per unit and a high-tension bus voltage of 1.05 per unit. The voltage of
Fault
)(
(a)
0.40
(b)
0.75 0.30
(c)
FIG. 45. The power system of Example 9. (a) One-line diagram, (b) positive-
sequence network with reactances in per unit, and (c) the same after reduction.
EA represents Eqd of the hydroelectric generators; EB' the voltage of the infinite
receiving bus; and Ee, the voltage of the high-tension sending-end bus.
the receiving system is 1.00 per unit. A two-line-to-ground fault occurs

at the middle of one transmission circuit and is cleared in 0.15 sec. by the
simultaneous opening of the circuit breakers at both ends. Compute the
quadrature-axis transient voltage E q ' and the displacement angle 0 as
functions of time for each of the following assumptions:
1. Constant exciter voltage E~I:'
2. Constant direct-axis transient voltage Eq'.
Each transmission circuit has negligible resistance and capacitance and
the following reactances: Xl = X 2 = 0.40 per unit j X 0 = 0.65 per unit.
jO.751
jO.301
EB = 1.00
ECEB. (a)
P =--sma
X
1.00 = 1.05 X 1.00 sin

·
a (b)
0.30
SIn a = 0.30 = 0.286 (e)

1.05
(d)
The line current I, expressed with the voltage EB as reference, is
Eo - E B 1.05/16.6° - 1.00 /0
I = = --------
jX jO.30
= 1.006 + jO.30 - 1.00 = jO.30 = 1.00/0 (e)

jO.30 jO.30 -
EXAMPLE 9 101
and is in phase with EB. Then

EA = Eqd = EB + j1.051 = 1.00 + j1.05 = 1.45!46.5° (/)
o = 46.5° (g)
Id = I sin 46.!l° == .1.00 X 0.725 = 0.725 (h)
Iq = I cos 46.5° = 1.00 X 0.688 = 0.688 (i)
E,/ = Eqcl - Icl(xfJ. - xq') = 1.45 - 0.725(0.69 - 0.29)
= 1.45 - 0.725 X 0.40 = 1.45 - 0.29 = 1.16 (i)
Eq = Eqd + Id(Xd - x q ) = 1.45 + 0.725(1.05 - 0.69)
= 1.45 + 0.725 X 0.36 = 1.45 + 0.26 = 1.71 (k)
Fault condition. The three sequence networks, connected in parallel
to represent a two-line-to-ground fault at the middle of one circuit, are
shown in Fig. 47. The reduction of the network is carried out in Figs. 48
through 54.
FIG. 47. The sequence networks of the power system of Fig. 45a, interconnected
to represent a two-line-to-ground fault (Example 9).
The current of the hydroelectric generators lA, in terms of their internal

voltage EA = Eqd and the voltage of the infinite bus EB' is
I A = YAAEA + YABE B (l)
where the admittances are
YA A = _.1_ + _.1_ = -j(0.622 + 0.448) = -jl.070

;1.61 ;2.23
FIG. 48. Reduction of the network of Fig. 47 - first step.
0.85 0.20
FIG. 49. Reduction of the network of FIG. 50. Reduction of the network of
Fig. 47 - second step. Fig. 47 - third step.
and
YAB = -._1_ = jO.448
)2.23
Hence
I A = -j1.070E A + jO.448E B (m)
Fault-cleared condition. The positive-sequence network and its reduction

are shown in Fig. 55 with the faulted line disconnected. Under this condition
EXAMPLE 9 109
FIG. 51. Reduction of the network of FIG. 52. Reduction of the network of
Fig. 47 - fourth step. Fig. 47 - fifth step.
FIG. 53. Reduction of the network of FIG..54. Reduction of the network of

Fig. 47 - sixth step. Fig. 47 - seventh and last step.
FIG. 55. Positive-sequence network of the power system of Fig. 45a after the fault
has been cleared.
the admittances are

1
YAA = - Y A B = - - = -jO.80
j1.25
and the current of the generators is
(n)
Constants for swing calculations. The inertia constant is

M = GH = 1 X 3.00 = _1_
180! 180 X 60 3,600
(dt)2 = (~)2 3 600 = 9.00

M 20 J
Assumption 1: Constant exciter voltage. The prefault values may be

summarized as follows:
Eq ' = 1.16, Ed' = Ed = 0, E q = E e,; = 1.71
la = 0.725, lq = 0.688, 0 = 46.5 0
Immediately after the instant when the fault occurs, E q ' , E e,;, and ~ retain
their prefault values, but E q , Ls, and lq undergo sudden changes. To find
the new values of these quantities, a value of E qd will be assumed. By use
of this, the network will be solved to obtain I d and 1q; and the value of E q'
will then be calculated. This process will be repeated until E q ' comes out
with the correct value, 1.16.
First assume Eqd = 1.80 /0. By eq. m, since E A = E(ld,
I = -j1.070 X 1.80 /0 + jO.448 X 1.00 / -46.5 0
= -j1.925 + 0.448(0.725 + jO.688)
= -j1.925 + jO.308 + 0.325
= 0.325 - j1.617
I q = 0.325, I d = 1.617
E q' = E qd - Xd,')
Id(x q -
= 1.800 - 1.617 X 0.40

= 1.800 - 0.647 = 1.153 (Should be 1.16)
Next try Eqd = 1.810.
I = -j1.070 X 1.810/0 +jO.448 X 1.00/-46.50
= -j1.933 + 0.325 jO.308 +
= 0.325 - jl.625
I q = 0.325, Id, = 1.625
E q ' = 1.810 - 1.625 X 0.40
= 1.810 - 0.650 = 1.160 (Correct)
E q = E qd + Id(xd, - x q)
= 1.810 + 1.625 X 0.36
= 1.810 + 0.585 = 2.395
P« = Eqd,Iq = 1.810 X 0.325 = 0.588
Pi = 1.000
P a = Pi, - P = 0.412 during fault, 0 before fault, 0.206 average
11,
ROUND-ROTOR MACHINE 111
Take ~t = 0.05 sec. During the first interval,

dE ' = (E 6 :J; - Eq)Jit (1.71 - 2.40)0.05
q TdO' + (E q/Eq')(dt/2) 5.0 + 2.4 0.025
1.2
-0.69 X 0.05
- - - - - = -.
0 007 per unlit
5.05
At the end of the first interval,
Eq' = 1.160 - 0.007 = 1.153
The angular swing during the first interval is
~al = ~ao + (dt)2 Pal = 0 + 9.00 X 0.206 = 1.~o

M
Hence, at the end of the first interval,
01 = 00 + da 1 = 46.5° + 1.9 = 48.4°
0
E q ' = 1.153/0
EB = 1.00/ -48.4°
Now assume Eqd still to be 1.81. Then

I = -j1.070 X 1.810/0 + jO.448 X 1.00 / -48.4°
= -jl.933 + jO.448(0.664 -jO.748)
= -jl.933 + jO.298 + 0.335
= 0.335 -j1.635
10 = 0.335, III = 1.635

E,/ = 1.810 - 1.635 X 0.40 = 1.810 - 0.654 = 1.156
This is close enough to the correct value, 1.153.

P; = Eqdlq = 1.81 X 0.335 = 0.606
The process is repeated again and again, with results as given in Table 13.
The system is stable. The maximum value of a is about 89°. The field
flux linkage decreases only from 1.160 to 1.116 per unit, a change of 3.8%
of the initial value.
Assumption 2: Constant flux linkage. The flux linkage is so nearly
constant under assumption 1 that to assume it exactly constant would not
change the swing curve significantly. Therefore the swing-curve calcula-
tions will not be repeated.
Round-rotor machine with constant flux linkages on both axes of

the rotor. Attention will now be turned from the salient-pole machine
to the nonsalient-pole, or round-rotor, machine. The alternative as-
l12 SYNCHRONOUS MACHINES
sumptions which can be made regarding the round-rotor machine were

listed on page 79. Briefly summarized, they are as follows:
1. Constant E/.
2. Constant Ed' and E q ' .
3. Varying E q ' , but Ed' = o.
4. Varying Ed' and E q ' .
TABLE 13
POINT-By-POINT CALCULATION OF SWING CURVE (EXAMPLE 9)
Assumption 1: Constant Exciter Voltage
t Eq' Eqd Eq l:iEq' i, i, Pu Pa t12a t1a a

(sec.)
- - -- - - - - - - - - - - - - - -- - - -
(p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (p.u.) (deg.) (deg.) (deg.)
0- 1.160 1.45 1.71 0.725 0.688 1.000 0 46.5
0+
o avg.
1.160
1.160
1.81 2.40 -0.007 1.625 0.325 0.588 +0.412
+0..206 +1.9
0
..
II
+1.9
0.05 1.153 1.81 2.40 -0.007 1.636 0.335 0.606 +0.394 +3.5 48.4
+5.4
0.10 1.146 1.81 2.41 -0.007 1.668 0.361 0.654 +0.346 +3.1 53.8
+8.5
0.15-
0.15+
1.139
1.139
1.85
1.45
2.49
1.73
1.769 0.396 0.733 +0.267
0.000 0.789 0.709 1.029 -0.029 ..
62.3
0.15 avg. 1.139 +0.119 +1.1 It
+9.6
0.20 1.139 1.53 1.88 -0.002 0.975 0.760 1.161 -0.161 -1.4 71.9
+8.2
0.25 1.137 1.59 2.00 -0.003 1.135 0.787 1.250 -0.250 -2.2 80.1
+6.0
0.30 1.134 1.64 2.09 -0.004 1.260 0.799 1.310 -0.310 -2.8 86.1
+2.2
0.35 1.130 1.65 2.12 -0.004 1.296 0.800 1.320 -0.320 -2.9 88.3
-0.7
0.40 1.126 1.635 2.09 -0.004 1.275 0.799 1.308 -0.308 -2.8 87.6
-3.5
0.45 1.122 1.60 2.03 -0.003 1.196 0.796 1.273 -0.273 -2.5 84.1
-6.0
0.50 1.119 1.55 1.94 -0.002 1.176 0.782 1.211 -0.211 -1.9 78.1
-7.9
0.55 1.117 1.48 1.81 -0.001 0.914 0.753 1.115 -0.115 -1.0 70.2
-8.9
0.60 1.116 1.41 1.68 0 0.736 0.702 0.990 +0.010 +0.1 61.3
-8.8
0.65 1.116 52.5
Assumption 1 is the simplest and .commonest. It was used in the

earlier chapters of this book and is ordinarily used in practice.
Assumption 2 implies constant flux linkages of the main circuits on
both axes of the rotor. Observation of the vector diagram, Fig. 33,
shows that, if both Ed' and E q ' are constant, E' is also constant, both
as to its magnitude and as to its angle 'Y with the quadrature axis.
Since 'Y is constant, the angular acceleration of vector E' is equal to
CHANGE OF FLUX LINKAGES 113
that of the rotor of the machine and, in stability calculations, it is

correct to associate the angular inertia of the machine with E' instead
of with the quadrature axis.
In most round-rotor machines the transient reactances in the two
axes (xa,' and x q' ) are very nearly equal. If Xd' = x q' , then E' = E/
(see Fig. 35), and therefore assumption 2 is identical with assump-
tion 1. The usual association of angular inertia with E/ under as-
sumption 1 is then theoretically correct.
If Xd' and x q ' are unequal, a rigorous representation of the round-
rotor machine is somewhat more difficult than that of the salient-pole
machine, because vector Eqa, of a round-rotor machine does not lie on
the quadrature axis. The exact representation is described in the
second section below. Its use is seldom, if ever, justifiable, however,
because it has already been shown that, even for a salient-pole ma-
chine, the effect of saliency on transient stability is negligible; that is,
x q ' may be assumed equal to Xd' with very little error. It therefore
appears perfectly reasonable to make the same assumption for a
round-rotor machine, in which the difference between xa,' and xq' is
smaller than it is in a salient-pole machine. The assumption xa,' = xq'
permits the simple method of assumption 1.
Assumption 3 permits the effect of voltage-regulator action on the
direct-axis field winding to be taken into account, while the quadra-
ture-axis circuit is assumed to be absent, as it is in a salient-pole
machine. The calculating procedure is like that already described
and illustrated for a salient-pole machine. Assumption 3 would have
good theoretical justification if the quadrature-axis time constant T q'
were very short; however, T q ' is about half of Ta,'. Nevertheless, the
assumption of saliency in a nonsalient-pole machine should produce
no greater error than neglecting saliency in a salient-pole machine
does.
Assumption 4- leads to the most exact and least simple method of
calculation. The changes of flux linkages of the rotor circuits on both
axes due to decrement and voltage-regulator action are taken into
account. The procedure is explained in the next two sections of this
chapter.
Calculation of change of flux linkages of round-rotor machine.

If, for any reason, it is not accurate enough to assume that the flux
linkages of the main rotor circuits are constant, then the changes of
these linkages can be calculated in a manner similar to that previously
described for the salient-pole machine. The only difference is that a
separate calculation is made for each axis. By eq. 246, the rate of
change of flux linkage of the direct axis is

dEq ' E ez - Eq
--=
dt T dO
,
The corresponding equation for the quadrature axis is
dEd '
- - = --,
-e, [256]
dt T qo
where the exciter voltage has been set equal to zero, as there is none
on the quadrature axis. For point-by-point calculation, the average
rate of change in time interval At is, by analogy to eq. 255,
AEo,' Ed
t;t = - T'
qO+
(Ed)
Ed' 2
At
[257]
Point-by-point calculation of swing curves of multimachine system,

taking into account changes of flux linkages of rotor circuits on both
axes. Each machine of a multimachine system, whether of salient or
nonsalient-pole type, may be represented on the calculating board or
on a circuit diagram by a variable .e.m.I. Ell, = E hd + jEh,q in series
with any convenient value of constant reactance Xh. This e.m.f. is
related as follows to the voltage E' = Eo,' + jEq ' , which depends on
the flux linkages of the rotor circuits. Since (byeq. 210)
E' = V + rI + jXd'Id - xq'Iq [258]
and, similarly,
Ell, = V + rl + jXh1d - xh,Iq [259]
it follows by subtraction that
Ell, = E' + j(Xh, - xd')Id - (XII, - xq')/q [260]
In the equations given above, the complex quantities are referred to
the direct axis of the machine. In terms of components parallel to
the quadrature and direct axes, eq. 260 becomes:
Ehd = Ed' - (XII, - xq')Iq [261]
Ehq = E q' + (XII, - xd')Id [262]
These relations are shown in Fig. 56.
The procedure will now be described for calculating the swing curves
point by point when this general representation is used. It will be
shown later how the procedure can be simplified by appropriate choice
of reactance XII,.
POINT-BY-POINT CALCULATION 115
1. The network must first be adjusted or solved for normal, or

prefault, conditions. When this has been done, the magnitude and
phase of the internal voltage Eh and of the current I of each gen-
erator are known. Terminal voltage V, power Pll' and reactive
power Q can also be found if desired. 'The angular position 0 of the
quadrature axis of every machine can be found as the phase of E qd ,
which can be measured or calculated by
[263]
In the steady state, Eqd lies on the quadrature axis. Angle 0 and
the phase angles of the complex quantities in this paragraph and in
the subsequent paragraphs of the procedure are expressed with re-
~ I
I rI
d
1\
\ "",/'
II Direct ',90° //
I
axis
,v"'"
~///
",
FIG. 56. Vector diagram of a synchronous machine in the transient state, showing
voltage Eh behind an arbitrary reactance Xh.
spect to an arbitrary reference axis which is common to the entire

system. Angle 0 being known, the current components I d and I q
and voltage components E hd and E hq can be measured or calculated,
thus making possible the computation of Ed' and E q ' by eqs. 261
and 262. In a·ddition,
Eq = E hq + (Xd - xh)I d
= E qd + (Xd - xq)I d [264]
Ed = 0 [265]
The initial value of exciter voltage E ez is equal to Ed. Thus all
quantities can be computed for the prefault condition.
2. Upon the occurrence of the fault, Ed', E q ' , E ez , and 0 are un-
changed at the first instant, but all the other quantities mentioned
116 RYNCHRONOUS l\IACHINES
in paragraph 1 are changed, and their new values must be found,

as follows: New values of La and I q should be assumed and, from
them and the old values of Ed' and E q', Ehd and Ehq are calculated
by eqs. 261 and 262 for every machine. The complex value of ap-
plied voltage Eh, referred to the system reference axis, is
[266]
The network is now solved for the currents. Since the positions of
the quadrature axes are known, the currents can be resolved into
components I d and I q, which are compared with the assumed values.
If they do not agree, the new values of I d and I q are used to calcu-
late a new value of Ek • The process is continued until agreement is
obtained. Thus the network is solved by a cut-and-try method.
'The power output P 11 of each generator is measured or calculated.
New values of Ed and E q are calculated from
Ed = E kd - (xq - xh,)Iq [267]
Eq = Ek q + (Xd - xh,)Id [268]
Conditions immediately after occurrence of the fault are now known.
3. A time interval ~t is chosen and the changes of E q ' and Ed' of
each machine in this interval are calculated by eqs. 255 and 257,
respectively. New values of Ed' and E q ' are found by adding the
increments ~Ed' and t!,.E q' to the old values.
4. The rotor position of every machine at the end of the first
interval is computed in the usual ,yay from the inertia constant M
and the acceleratng power P a. (The average of fault and prefault
values of Pais used for the first interval.)
5. The values of Ed' and E q ' are now known from step 3, and the
values of 0 are known from step 4. This enables us to solve the
network again by means of the cut-and-try process described in
paragraph 2. Thus new values of I a and I q , E ka and Ehq, and P u
are obtained.
6. Steps 3, 4, 5, and 2 are repeated as many times as necessary,
depending upon how far the swing curves are to be carried out. In
step 3 the same value of E ex is used repeatedly if there is no voltage
regulator; or, if a regulator is assumed to act, values of Ee:r; are read
from the curves of exciter-voltage build-up or build-down.
7. If there is any sudden change in I a and I q , due, for example,
to the clearing of a fault or the reclosing of a breaker, then the net-
work must be solved for both the old and the new conditions, using
in both cases the same values of Ed', E q ' , Ee:r;, and o. The incre-
POINT-BY-POINT CALCULATION 117
ments of E q ' and Ed' during the interval after the discontinuity are
calculated by using in eqs. 255 and 257 the values of E q and Ed
after the discontinuity. The increment of 5, however, is calculated
by using the average accelerating power before and after the dis-
continuity.
Some special cases will now be considered.
Salient-pole machine. Choose Xh = X q; then Eh = Eqd , E hq = Eqd"
and Ehd, = O.
In step 1 of the general procedure, eq. 263 is unnecessary, and 5 is
the angle of Eqa = Eh not only in the steady state but also in the
transient state. Equation 262 becomes
E qa = E q' + (xq - xd')Id [269]
and eq. 261 becomes
[270]
In step 2 of the procedure, a new value of Ed' does not need to be
calculated, because Ea' is always zero. Therefore, only Ia need be
assumed, not [q. Equations 266,267, and 268 become:
Eqa = Eqd,/a [271]
s, = 0 [272]
E q = E qa + (Xd, - xq)Ia [273]
Equation 272 holds for both transient and steady states because there
is no quadrature-axis rotor circuit.
In step 3 E a' is still zero.
Round-rotor machine with laminated rotor. The procedure for this
unusual type of machine is the same as that for the salient-pole ma-
chine except that, since Xd = X q = Xh" eq, 268 becomes simply
e, = Eqa, [274]
Round-rotor machine with solid rotor and xa,' ~ xq'. This case is
perfectly general. Nevertheless, the general procedure described above
can be simplified somewhat by choosing Xh, equal either to Xd,' or xq' .
On the assumption of the former alternative, eqs. 260, 261, and 262
become:
Eh = E' - (xa,' - xq')Iq [275]
Eh,d = Ed' - (xa,' - xq')Iq [276]
E hq = E q' [277]
In step 1 of the procedure,eqs. 263 and 264 become:

Eqa = Eh + j(xq - Xd')! [278]
E q = E q' + (Xd - xd')Id [279]
In step 2, E hq need not be calculated since it is equal to E q ' . There-
fore, I d need not be assumed, but only I q. Equations 267 and 268
become:
Ed = Ed' - (xq - xq')I q [280]
s, = E q' + xd')Id
(Xd - [281]
The procedure for the assumption of Xh = x q ' will not be described
because it offers no particular advantage over the procedure for the
assumption of Xh = Xd'.
Round-rotor machine with solid rotor and Xd' = xq'. This is the usual
type of round-rotor machine. The following simplifications apply in
addition to those stated for the preceding case: Equation 260 becomes
Eh = E' = E/ [282]
and eq. 262 becomes
E h q = E q' [283]
In step 2 neither Ehd nor Ehq need be calculated because they are
equal to Ed' and E q ' , respectively. Therefore, neither I d nor I q need
be assumed and the network can be solved without the cut-and-try
procedure for finding the values of Eh.
SATURATION
The effects of saliency and of, change of field flux linkage have now
been treated, but so far the effect of saturation of the iron has been
neglected. Saturation will now be considered.
Saturation in a transformer. In a transformer having a closed iron

core the fluxes linking the windings can be divided into the mutual
flux, the preponderant part of which is wholly in the core, and the
several leakage fluxes, which follow paths largely in air and whose
reluctance is determined principally by the reluctance of the air paths.
Under normal operating conditions the mutual flux constitutes a very
large part, and the leakage fluxes a small part, of the flux linking any
winding. Consequently, the flux in the core is substantially the same
all around the core and is substantially equal to the mutual flux.
Saturation of all parts of the core then depends only on the mutual
flux. The mutual flux justifiably can be assumed to be a function of
the sum of the ampere-turns of all windings on the core, regardless
SATURATED NONSALIENT-POLE MACHINE 119
of the positions of the windings. This sum is the exciting m.m.f. The
mutual flux is a nonlinear function of the exciting m.m.f, as found
from an open-circuit test. The leakage fluxes, on the other hand, are
very nearly linear functions of the currents.
Leakage reactances
/ (constant) ~
~I Exciting~
reactance
(variable)
FIG. 57. Equivalent T circuit of an iron-core transformer.
Currents and voltages in a two-winding iron-cored transformer can

be calculated from the equivalent T circuit of Fig. 57, which consists
of two linear leakage reactances and one nonlinear (iron-cored) excit-
ing reactance. The upshot of this discussion is that saturation in a
transformer can be considered as a function of a single variable, the
internal e.m.f. E induced in a winding by the mutual flux. *
Saturation in an induction motor can be treated by assuming it to be

a function of the air-gap flux (the flux that crosses the air gap from
stator iron to rotor iron) and by using an equivalent T circuit similar
to that of the transformer.'
Saturation in a synchronous machine. In a synchronous machine

the air gap ordinarily is longer than in an induction motor, and the
leakage fluxes constitute a greater portion of the total flux. There is
less justification in the case of a synchronous machine than in the case
of a transformer, or even of an induction machine, for assuming that
the saturation of all parts of the core varies as a function of a single
variable when the load and the power factor change. Nevertheless,
the assumption is usually made in texts on a-c. machinery that the
saturation is a function of the air-gap flux. Let us see to what con-
clusions this assumption leads and whether those conclusions are in
accord with experimental facts.
Saturated nonsalient-pole machine in the steady state. Based upon

the assumption that saturation is a function of the air-gap flux only,
the vector diagram of a nonsalient-pole synchronous generator in the
*This treatment is not applicable to some types of current transformers.
steady state is as shown in Fig. 58. The air-gap flux cPa induces in the
armature winding a positive-sequence e.m.f. Ea , which is proportional
in magnitude to cPa and lags 90 0 behind it. It is assumed that the
generator is delivering a lagging current I, which, flowing through the
FIG. 58. Vector diagram of a nonsalient-pole synchronous generator operating

in the steady state with lagging current.
armature leakage reactance xi, produces a voltage drop jxzI 900 ahead
of I. The armature terminal voltage is V = E a - jxll. To produce
air-gap flux cPa, an m.m.f, njI T is required, where nj is the number of
turns in the field winding and IT is the current read from the no-load
saturation curve (Fig. 59) corresponding to armature voltage Ea.
Thus saturation is assumed to depend upon Ea. The required m.m.f.
n/IT is the vector sum of two components: field m.m.f, nlll and arma-
ture m.m.f. nal. The latter m.m.f, is also called armature reaction.
Upon dividing by nl, we find that the fictitious field current IT is the
vector sum of the actual field current II and the field equivalent of
armature reaction (na/nl)1 = kl, as shown in Fig. 58. Given the
armature terminal voltage V, armature current I, and terminal power-
factor angle cP, one can find the required field current I I as follows:
Find air-gap voltage E a by vector addition of V + jxlI. From the
no-load saturation curve read IT corresponding to Ea. Then find II
by vector subtraction IT -kI, where IT leads E a by 900 and kI is in
phase with I.
At zero power factor, lagging current, vector triangles V, XlI, E a
and IT, kl, I I collapse to collinearity and the following relations are
true arithmetically:
V = E a - XlI [284]
II = IT + kI [285]
ZERO-POWER-FACTOR SATURATION CURVE 121
The zero-power-factor rated-current saturation curve (Fig. 59) is a

graph of armature terminal voltage V against field current I I with
armature current I held constant at its full-load value and power-
factor angle cP constant at, gOo lag. According to eqs. 284 and 285,
any point B' (II, V) on the zero-power-factor curve lies to the right
by a constant distance kT and downward by a constant distance xlI
Eo or Ep
FIG. 59. No-load and zero-power-factor rated-current saturation curves and the
Potier triangle.
from the corresponding point A' (I T, E a ) on the no-load saturation

curve. The latter curve portrays V against II for I = 0, but eqs. 284
and 285 show that at no load V = E a and II = IT. Hence the no-
load saturation curve shows E a versus IT for any load, in accordance
with our previous assumptions. Given the no-load saturation curve
and constants xi and k, one can construct a zero-power-factor satura-
tion curve for any constant armature current I by displacing the given
curve downward by distance xLI volts and to the right by kI amperes.
Saturation curves for other power factors can be constructed through
use of the vector diagram (Fig. 58). From these curves the field cur-
rent required for any load current and terminal voltage can be found.
By experiment it is found that the zero-power-factor saturation

curve is indeed parallel to the no-load saturation curve. That is, one
curve can be made to coincide with the other by shifting every point
of it through the same vectorial distance AB = A'B', Fig. 59. This
fact furnishes a way of determining armature leakage reactance Xl and
effective turns-ratio k. One curve can be drawn on tracing paper, and
distances AC and CB can be found by trial so as to make this curve
coincide with the other. The following alternative construction is
available which requires only two given points on the zero-power-
factor saturation curve: the short-circuit point B and some other
point B' near rated voltage. Measure line segment BO and layoff
the equal distance B'0' horizontally to the left from point B' on the
zero-power-factor curve above the knee. Through point 0' draw line
0'A' parallel to the lower, straight portion OA of the no-load satura-
tion curve and intersecting that curve at point A'. Triangle A'B'C' =
triangle ABC is called the Potier triangle. 1 According to our hypoth-
esis, vertical leg A'C' of this triangle equals XlI and horizontal leg
B' C' equals kI. Since I is known, Xl and k can be found.
In practice the value of Xl thus found exceeds the value of Xl calcu-
lated by the designer. It appears reasonable, because of this circum-
stance, to suppose that saturation is not a function of air-gap flux
(or the corresponding voltage E a ) . Nevertheless, the fact that the
no-load and zero-power-factor saturation curves are parallel, or at
least very nearly so, enables us to define and to use a new reactance,
the Potier reactance X p , determined from the vertical side of the Potier
triangle, xpI. For typical values of X p see Table 2. The Potier react-
ance replaces the leakage reactance Xl in the vector diagram, Fig. 58,
and the voltage behind Potier reactance, E p , replaces the air-gap volt-
age Ea. Saturation is assumed now to be a function of Potier voltage
E p instead of !Sa. The use of Potier reactance and voltage is justified
by the fact that through their use one can construct not only the
zero-power-factor saturation curve but also saturation curves for other
power factors which agree well with curves determined by load tests.
If there were no saturation, the no-load and zero-power-factor con-
stant-current "saturation curves" would be straight parallel lines
(Fig. 60), and there would be no unique way of resolving the constant
separation between those parallel lines into vertical and horizontal
components representing Potier reactance drop and armature reaction,
respectively. Indeed, instead of assuming some of each (QR and RP),
one could go to either extreme: (1) assume that the horizontal dis-
tance MP between the curves was due to armature reaction and that
the Potier reactance was zero, or (2) assume that the vertical distance
ZERO-POWER-FACTOR SATURATION CURVE 123
NP between the curves was due to reactance drop and that the arma-
ture reaction was zero. In the first case Iv is the field current required
to generate terminal voltage V on open circuit, and AlP is the field
current required to overcome armature reaction. The actual field
current required to generate terminal voltage V at rated-current zero-
power-factor load is If, which is the sum of I ; and MP. In the second
v IT
Field current
FIG. 60. No-load and zero-power-factor "saturation curves" of unsaturated
synchronous generator.
case, V is the terminal voltage, NP is the synchronous reactance drop

xa,I, and E q = V + xa,I is the internal voltage behind synchronous
reactance, or the excitation voltage. Note the proportionalities be-
tween armature voltages and field currents represented by the same
points on the no-load line:
V NP
-- = -
E
= - q = - = wMf
s, [286]
Iv MP If 17,
Indeed, in the usual per-unit system, wMj = 1, so that each compo-
nent, of armature voltage is numerically equal to the corresponding
component of field current. Vector diagrams showing conditions at
any power factor can be drawn either entirely with current vectors or
entirely with voltage vectors. The second alternative is generally
used in practice, with the result that the fictitious internal voltage E q ,
which, in an unsaturated machine, is proportional to the field current,

is the vector sum of terminal voltage V and reactive drop xdI.
Voltage vector diagram of saturated nonsalient-pole machine. It

is desirable to extend the voltage vector diagram to cover the case of
a saturated synchronous machine. For this purpose, the field current
~ ------------------
ET - - - - - - - - - - - - - - - -
Ep Is u{'1 e
~ I ?>~o~ C I
~ ~ ~~i. I
J! b?'a: I
'0
>
!?
'llf
~i
~o I
0"6
,I
::J I
10 I I
E
~
J I
< I I
I I
I I
I I
I I
I J
I I
I I
I I
I I
I I
I I
I I
I I
Ip IT If
Field current
FIG. 61. Graphical construction for finding the components of field current and
armature voltage due to saturation corresponding to Potier voltage E p • The com-
ponents are Is and E s, respectively.
and its components in the previous vector diagram (Fig. 58) are con-
verted into corresponding armature voltages by means of the "air-gap
line," which is an extension of the lower, straight portion of the no-load
saturation curve (Fig. 61). The vector current triangle IT, kI, I, in
Fig. 58 is replaced by the corresponding vector voltage triangle E T ,
jxaI, E/. This voltage triangle is then swung clockwise 90° so that
E T lies along Ea. Furthermore, Ea and Xl in Fig. 58 are replaced by
VECTOR DIAGRAl\I OF SATURATED MACHINE 125
E p and X p , respectively. The resulting vector diagram is shown in

Fig. 62.
In Fig. 61 it will be noted that the field current IT corresponding to
any particular Potier voltage E p consists of two components : I p , the
field current required to drive the flux across the air gap, and I., the
additional field current required to drive the flux through the saturated
iron. The corresponding armature voltages, read on the air-gap line,
are E p and E s , the sum of which is E T . In the absence of saturation,
FIG. 62. Vector voltage diagram of a nonsalient-pole synchronous generator.

This diagram is derived from that of Fig. 58 by replacing Ea and XI by E p and Xp,
respectively, and by replacing the field-current vectors by corresponding voltage
vectors, rotated 90° clockwise.
E. = 0 and E p = E T . In this case vector ET would not jut beyond

vector E p , as in Fig. 62, but would coincide with E p , and vector jxaI
would lie on the same line with jXpI, as shown in Fig. 63. The sum of
X p and Xa is Xd, the unsaturated synchronous reactance. The vector
sum of V and jXdI, or of E p and jxaI, is E q , corresponding to the field
current that would be required if there were no saturation. Even if
there is saturation, vectors jxaI and E. may be added to E p in the
opposite order from that shown in Fig. 62. This is done in Fig. 63.
The vector sum of Eq and E. is Ej, the voltage corresponding to the
field current of the saturated machine.
The procedure for finding the field current of a saturated nonsalient-
pole machine may be summarized as follows: Find Eq = V + jXdIjust
as it would be found for an unsaturated machine. Then, as a correction
for saturation, add to it E., in phase with E p = V + jxpI. The magni-
tude E. is found from the no-load saturation curve at ordinate E p as
shown in Fig. 61. The magnitude of the sum, EI = Eq + E., equals
the field current in per unit. The phase of Ej, and not of Eq , indicates
the position of the quadrature axis.
Ordinarily very little error in Ej is produced by adding E, in phase

with E q instead of in phase with E p • Then both Eq and Ej represent
the position of the quadrature axis, However, this procedure causes
some error in ~.
FIG. 63. Vector voltage diagram of a nonsalient-pole synchronous generator,

derived from that of Fig. 62 by adding E, and jxaI to Ep in reverse order, E q is
proportional to the field current of an unsaturated machine; E" to that of a
saturated machine. E" the correction for saturation, is parallel to E p and has a
magnitude found from the construction of Fig. 61.
Vector diagram of saturated salient-pole machine. If the same

assumptions and reasoning regarding saturation that were applied to
the nonsalient-pole machine are applied to the salient-pole machine,
4i:'-----------...------ Ef ------+------~
~----------Eq---------~~
FIG. 64. Vector voltage diagram of a saturated salient-pole synchronous generator

in the steady state.
the vector diagram of Fig. 64 is obtained. Here voltage E q is calcu-

lated in the same way as for an unsaturated salient-pole machine, and
then E, is added to it as a correction for saturation. E 8 is a function
SATURATED MACHINE IN THE TRANSIENT STATE 127
of E p as shown in Fig. 61. For convenience, E s is taken parallel to

E q rather than to Ep • The theoretical justification for doing so is
possibly greater with the salient-pole machine than with the non-
salient-pole machine. The sum of E q and E, is E/, a fictitious arma-
ture voltage proportional to the field current.
Saturated synchronous machine in the transient state. In the

transient state, as well as in the steady state, saturation is assumed to
depend upon E p , the voltage behind Potier reactance. (The relation
between E p and E s , the correction to E q for saturation, was given in
Fig. 61.)
The correction for saturation affects the calculation of change of
field-flux linkage; that is, of E q ' . Equation 241 is still valid, but,
upon changing it into a voltage equation, we must use
[287]
instead of E q (eq. 242). As a result, the voltage equation replacing

eq. 246 is
dE q ' _ E ex - Ef E ex - Eq - Es
[288]
dt - T dO' T dO'
For point-by-point calculation, eq. 255 is replaced by
[289]
To find E s , it is necessary to know E p • The latter can be measured

easily on a calculating board if the machine is represented in the net-
work as shown in Fig. 65. If the machine is represented, as previously
discussed, by any reactance Xli, and the voltage Eh, behind this react-
ance, Xli, may be divided into two parts, X p and Xli, - X p , as shown in
Fig. 65a. . Then E p is the voltage of the common terminal of the two
reactors. If the machine is represented by X q and E qd , the reactances
are X p and X q - X p , as shown in Fig. 65b. If E p cannot be measured,
it can be calculated thus:
[290]
Except for the addition of the correction for saturation, the point-by-
point calculation of E q ' and the swing curves is carried out as previ-
ously described. During a sudden change of current, E q ' is constant,
but E p and E h change.
Saturation has little effect upon transient stability and it is usually
neglected, especially when E q ' or E/ is assumed constant. Actually
Xd' varies somewhat with saturation.t! though not nearly to the extent
that Xd changes. Any change in Xd' affects the relation between termi-
nal voltage and Ed', but such changes are usually neglected.
In steady-state stability studies (treated in Chapter XV), saturation
is an important factor. It will be discussed further in that chapter.
:h
Xh-Xp Xp
I
(v (a)
~q
; Xq - Xp Xp ,
I
jEqd (EP Iv (b)
FIG.65. Representation of a saturated synchronous machine as part of a net-

work. The use of two reactors in series makes possible the measurement of Ep ,
the voltage behind Potier reactance.
Short-circuit ratio. The short-circuit ratio of a synchronous machine

is defined as the ratio of the field current required to produce rated
voltage at rated speed and no load to the field current required to
produce rated armature current under a sustained three-phase short
circuit. This definition is illustrated by Fig. 66, where OAjOB is the
short-circuit ratio.
If there were no saturation, the short-circuit ratio would be equal
to the reciprocal of the synchronous reactance. Saturation has the
effect of increasing the short-circuit ratio.
The short-circuit ratio has significance with respect to both the
performance of the machine and its cost.6 9 The lower the short-
circuit ratio, the greater is the change in field current required to main-
tain constant terminal voltage for a given change in load (armature
current) and the lower is the steady-state stability limit. Thus, a
machine of low short-circuit ratio requires an excitation system that
SHORT-CIRCUIT RATIO 129
is able to provide large changes of field current quickly and reliably.

It is more dependent upon its excitation system than is a machine of
high short-circuit ratio, but, given a suitable excitation system, can
give as satisfactory performance under load as does the machine of
high short-circuit ratio. An appropriate excitation system can also
increase the steady-state stability limit.
2.5
1.0 2.0
.....
's
1.0 '(3~
I
I
~
o
.r::
...... 1 en
-0 1
sl 0.5
~I
J21
~I
I
~---""""--~Io-.-_-----''---_--'''O
0.5 A 1.0 B 1.5 2.0
Field current If
FIG. 66. Typical saturation curves of a steam-turbine-driven generator. The

short-circuit ratio is OAjOB.
The lower the short-circuit ratio, the lower are the size, weight, and
cost of the machine. Hence, as a result of improvements which have
been made in excitation systems, there has been a trend toward the
use of generators of lower short-circuit ratio and, consequently, of
lower cost.
It is practical to build steam-turbine-driven generators with short-
circuit ratios anywhere in the range from 0.5 to 1.1. Most modern
generators of this type have ratios in the range from 0.8 to 1.0, but
it appears likely that 0.7 will become more generally used. Water-
wheel-driven generators usually have higher short-circuit ratios, up
to 2.0, whereas synchronous condensers may have ratios as low as 0.4.
REFERENCES
1. A. POTIER, II Sur la reaction d'induit des alternateurs" (Armature Reaction
of Alternators), L'Ecla1:rage Electrique (Paris), vol. 24, pp. 133-41, 1900.
2. R. E. DOHERTY, Cl A Simplified Method of Analyzing Short Circuit. Problems,"
A.l.E.E. Trans., vol. 42, pp. 841-9, 1923.
3. R. E. DOHERTY and C. A. NICKLE, "Synchronous Machines - I and II,"
A.I.E.E. Trans., vol. 45, pp. 912-47, 1926.
4. R. E. DOHERTY and C. A. NICKLE, "Synchronous Machines - III, Torque-
Angle Characteristics Under Transient Conditions," A.I.E.E. Trans., vol. 46,
pp. 1-18, 1927.
5. R. W. WIESEMAN, "Graphical Determination of Magnetic Fields, Practical
Applications to Salient-Pole Synchronous Machine Design," A.I.E.E. Trans.,
vol. 46, pp. 141-54, 1927.
6. R. E. DOHERTY and C. A. NICKLE, "Synchronous Machines - IV, Single-
Phase Short Circuits," A.I.E.E. Trans., vol. 47, pp. 457-87, April, 1928. Disc.,
pp.487-92.
7. P. L. ALGER, "The Calculation of the Armature Reactance of Synchronous
Machines," A.I.E.E. Trans., vol. 47, pp. 493-512, April, 1928. Disc., pp. 512-3.
8. R. H. PARK and B. L. ROBERTSON, "The Reactances of Synchronous Ma-
chines," A.I.E.E. Troms., vol. 47, pp. 514-35, April, 1928. Disc., pp. 535-6.
9. R. H. PARK, "Definition of an Ideal Synchronous Machine and Formula for
the Armature Flux Linkages," Gen. Elec. Reo., vol. 31, pp. 332-4, June, 1928.
10. R. H. PARK, "Two-Reaction Theory of Synchronous Machines - Part I,
Generalized Method of Analysis," A.I.E.E. Trans., vol. 48, pp. 716-30, July, 1929.
11. R. E. DOHERTY and C. A. NICKLE, "Synchronous Machines - V, Three-
Phase Short Circuit," A.I.E.E. Trans., vol. 49, pp. 700-14, April, 1930.
12. F. C. LINDVALL, "The Educational Value of the Theorem of Constant
Linkages," Gen. Elec. Rev., vol. 33, pp. 273-8, May, 1930.
13. C. F. WAGNER, "Damper Windings for Water Wheel Generators," A.I.E.E.
Trans., vol. 50, pp. 140-51, March, 1931.
14. L. A. KILGORE, "Calculation of Synchronous Machine Constants - Re-
actances and Time Constants Affecting Transient Characteristics," A.I.E.E.
Trans., vol. 50, pp. 1201-14, December, 1931.
15. S. H. WRIGHT, "Determination of Synchronous Machine Constants by Test
- Reactances, Resistances, and Time Constants," A.I.E.E. Trans., vol. 50,
pp. 1331-51, December, 1931.
16. L. P. SHILDNECK, "Synchronous Machine Reactances - A Fundamental
and Physical Viewpoint," Gen. Elec. Rev., vol. 35, pp. 560-5, November, 1932.
17. R. H. PARK, "Two-Reaction Theory of Synchronous Machines - II," A.I.E.E.
Trans., vol. 52, pp. 352-5, June, 1933.
18. S. B. CRARY, L. P. SHILDNECK, and L. A. MARCH, "Equivalent Reactances
of Synchronous Machines," A.I.E.E. Trans., vol. 53, pp. 124-32, 1934.
19. L. A. MARCH and S. B. CRARY, "Armature Leakage Reactance of Synchro-
nous Machines," A.I.E.E. Trans., vol. 54, pp. 378-81, April, 1935. Disc., pp.
1116-8.
20. CHARLES KINGSLEY, JR., "Saturated Synchronous Reactance," A.I.E.E.
Trans., vol. 54, pp. 300-5, March, 1935. Disc., pp. 1111-3, October, 1935.
21. T. A. ROGERS, "Test Values of Armature Leakage Reactance," Elec. Eng.,
vol. 54, pp. 700-5, July, 1935.
REFERENCES 131
22. B. L. ROBERTSON, "Tests on Armature Resistance of Synchronous Machines,"

Elec. Eng., vol. 54, pp. 705-9, July, 1935.
23. L. A. KILGORE, "Effects of Saturation on Machine Reactances," A.I.E.E.
Trans., vol. 54, pp. 545-50, May, 1935. Disc., pp. 1113-6, October, 1935.
24. W. A. THOMAS, "Negative-Sequence Reactance of Synchronous Machines,"
A.I.E.E. Trans., vol. 55, pp, 1378-85, December, 1936. Dise., vol. 56, pp. 903-4,
July, 1937.
25.A. S. LANGSDORli', "Contributions to Synchronous-Machine Theory,"
A.I.E.E. Truns., vol. 56, pp. 41-8, January, 1937. Disc., pp. 907-9, July, 1937.
26. C. CONCORDIA and H. PORITSKY, "Synchronous Machine with Solid Cylin-
drical Rotor," A.I.E.E. Trons., vol. 56, pp. 49-58 and 179, January, 1937. Disc.,
pp. 906-7, July, 1937.
27. S. B. CRARY, "Two-Reaction Theory of Synchronous Machines," A..I.E.E.
Trans., vol. 56, pp. 27-31 and 36, January, 1937. Extension to machines having
capacitance in the armature circuit. Disc., pp. 905-6, July, 1937.
28. STERLING BECKWITH, ccApproximating Potier Reactance," A.I.E.E. Trans.,
vol. 56, pp. 813-8, July, 1937. Disc., pp. 1318-9.
29. B. L. ROBERTSON, T. A. ROGERS, and C. F. DALZIEL, "The Saturated Syn-
chronous Machine," A.I.E.E. Trans., vol. 56, pp. 858-63, July, 1937. Disc., pp.
1502-3, December, 1937.
30. C. CONCORDIA, "Two-Reaction Theory of Synchronous Machines With Any
Balanced Terminal Impedance," A.I.E.E. Trans., vol. 56, pp. 1124-7, September,
1937.
31. JAMES B. SMITH and CORNELIUS N. WEYGANDT, "Double-Line-to-Neutral
Short Circuit of an Alternator," A.I.E.E. Trane., vol. 56, pp. 1149-55, September,
1937.
32. A. R. MILLER and W. S. 'VEIL, JR., "Alternator Short-Circuit Currents
Under Unsymmetrical Terminal Conditions," A.I.E.E. Trans., vol. 56, pp. 1268-78,
October, 1937.
33. C. F. WAGNER, "Unsymmetrical Short Circuits on Water-Wheel Generators
Under Capacitive Loading," A.I.E.E. Trtms., vol. 56, pp. 1385-95, November,
1937. Disc., vol. 57, pp. 406-7, July, 1938.
34. B. R. PRENTICE, "Fundamental Concepts of Synchronous Machine Re-
actances," A.I.E.E. Trans., vol. 56, supplement, pp. 1-21, 1937.
35. EDITH CLARKE, C. N. WEYGANDT, and C. CONCORDIA, "Overvoltages
Caused by Unbalanced ShortCircuits - Effect of Amortisseur Windings," A.I.E.E.
Trans., vol. 57, pp. 453-66, August, 1938. Disc., pp. 466-8.
36. C. CONCORDIA, S. B. CRARY, and J. M. LYONS, "Stability Characteristics
of Turbine Generators," A.I.E.E. Trans., vol. 57, pp. 732-41, 1938. Disc., pp.
741-4.
37. R. B. GEORGE and B. B. BESSESEN, "Generator Damper Windings at Wilson
Dam," A.I.E.E. Trans., vol. 58, pp. 166-71, April, 1939. Disc., pp. 171-2.
38. W. W. KUYPER, "Analysis of Short-Circuit Oscillograms," A.I.E.E. Trans.,
vol. 60, pp. 151-3, April, 1941. Disc., pp. 752-3.
39. THOMAS C. McFARLAND, "Correction for Saturation," A.I.E.E. Trans.,
vol. 61, pp. 233-5, May, 1942.
40. REINHOLD RttDENBERG, "Saturated Synchronous Machines Under Transient
Conditions in the Pole Axis," A.I.E.E. Trans., vol. 61, pp. 297-306, June, 1942.
Disc., pp. 444-5.
41. CHAS. A. WAGNER, lC Machine Characteristics," Chapter 7 of Electrical
Transmission and Distribution Reference Book by Central Station Engineers of

the Westinghouse Elec. & Mfg. Co., East Pittsburgh, Pa., 1st edition, 1942.
42. R. V. SHEPHERD and C. E. KILBOURNE, "The Quadrature Synchronous
Reactance of Salient-Pole Synchronous Machines," A.l.E.E. Trans., vol. 62,
pp. 684-9, November, 1943. Disc., p. 926.
43. E. L. HARDER and R. C. CHEEK, "Regulation of A-C Generators With
Suddenly Applied Loads," A.l.E.E. Trans., vol. 63, pp. 310-8, June, 1944.
44. A. W. RANKIN, "The Equations of the Idealized Synchronous Machine,"
Gen. Elec. Reu., vol. 47, pp. 31-6, June, 1944.
45. GORDON F. TRACY and WILLIAM F. TICE,"Measurement of the Subtransient
Impedances of Synchronous Machines," A.I.E.E. Trane., vol. 64, pp. 70-6, Febru-
ary, 1945. Disc., pp. 429-31.
46. C. CONCORDIA and F. J. MAGINNISS, "Inherent Errors in the Determination
of Synchronous-Machine Reactances by Test," A.I.E.E. Trans., vol. 64, pp. 288-94,
June, 1945. Disc., pp. 491-2.
47. A. W. RANKIN, "Per-Unit Impedances of Synchronous Machines," A.I.E.E.
Troms., vol. 64, pp. 569-73, August, 1945.
48. C. CONCORDIA, S. B. CRARY, C. E. KILBOURNE, and C. N. WEYGANDT, JR.,
"Synchronous Starting of Generator and Motor," A.l.E.E. Trans., vol. 64, pp.
629-34, September, 1945. Disc., p. 956.
49. A. W. RANKIN, "Per-Unit Impedances of Synchronous Machines - II,"
A.l.E.E. Trans., vol. 64, pp. 839-41, December, 1945.
50. A. W. RANKIN, "The Direct- and Quadrature-Axis Equivalent Circuits of
the Synchronous Machine," A.l.E.E. Trans., vol. 64, pp. 861-8, December, 1945.
51. A. W. RANKIN, H Asynchronous and Single-Phase Operation of Synchronous
Machines," A.l.E.E. Trans., vol. 65, pp. 1092-102, 1946.
52. C. CONCORDIA and M. TEMOSHOK, "Resynchronizing of Generators,"
A.l.E.E. Trans., vol. 66, pp. 1512-8, 1947.
53. A. M. ANGELINI," General Topogram for Synchronous Machines," C.l.G.R.E.,
1948, Report 114.
54. G. D. FLOYD and H. R. SILLS, "Generator Design to Meet Long Distance
Transmission Requirements in Canada," C.l.G.R.E., 1948, Report 131.
55. GABRIEL KRON, "Steady-State Equivalent Circuits of Synchronous and
Induction Machines," A.l.E.E. Trans., vol. 67, part I, pp. 175-81, 1948.
56. L. T. ROSENBERG, "A Semiempirical Approach to Voltage Dip - with
Suddenly Applied Loads on A-C Generators," A.I.E.E. Trans., vol. 68, part I,
pp. 160-6, 1949. Disc., pp. 166-7.
57. WILLIAM C. DUESTERHOEFT, "The Negative-Sequence Reactances of an
Ideal Synchronous Machine," A.l.E.E. Trans., vol. 68, part I, pp. 510-3, 1949.
Disc., pp. 513-4.
58. N. E. WILSON, "Transients in 2-Phase Synchronous Machines," A.I.E.E.
Trans., vol. 68, part II, pp. 1360-7, 1949.
59. C. CONCORDIA, "The Differential Analyser as an Aid in Power System
Analysis," C.l.G.R.E., 1950, Report 311.
60. H. S. KIRSCHBAUM, "Per-Unit Inductances of Synchronous Machines,"
A.I.E.E. Trans., vol. 69, part I, pp. 231-4, 1950.
61. SAAD L. MIKHAIL, "Potier Reactance for Salient-Pole Synchronous Ma-
chines," A.I.E.E. Trans., vol. 69, part I, pp. 235-8, 1950.
62. ROBERT D. CAMBURN and ERIC T. B. GROSS, "Analysis of Synchronous
Machine Short Circuits," A.l.E.E. Trans., vol. 69, part II, pp. 671-8,1950. Disc.,
pp.678-9.
PROBLEl\IS 133
63. GABRIEL KRON, ccClassification of the Reference Frames of a Synchronous

Machine," A.l.E.E. Trans., vol. 69, part II, pp. 720-7, 1950.
64. SAAD L. MIKHAIL, ccSymmetrical Short Circuits on Saturated Alternators,"
A.l.E.E. Trans., vol. 69, part II, pp. 1554-62, 1950.
65. CHARLES CONCORDIA, "Synchronous Machine Damping and Synchronizing
Torques," A.I.E.E. Trans., vol. 70, part I, pp. 731-7, 1951.
66. CHARLES CONCORDIA, Synchronous Jl.{achine« - Theory and Performance,
John Wiley & Sons, Inc., New York, 1951.
67. JOH. GRABSCHEID, "The Transient E.mJ. Induced in the Armature Winding
and the Transient Active Power of a Synchronous Alternator Connected to an
Infinite Bus through External Impedances," C.I.G.R.E., 1952, Report 335.
68. F. K. DALTON and A. W. W. CAMERON, "Simplified Measurement of Sub-
transient and Negative-Sequence Reactances in Salient-Pole Synchronous Ma-
chines," A.l.E.E. Trans., vol. 71, part III, pp. 752-6, October, 1952. Disc., pp.
756-7.
69. C. M. LAFFOON, "The Significance of Generator Short-Circuit Ratio,"
Westinghouse Engineer, vol. 12, pp. 207-8, November, 1952.
PROBLEMS ON CHAPTER XII

1.. Discuss the energy relations in the following two cases considered in
the text as examples of applications of the law of constant flux linkages:
(a) copper ring suddenly removed from a permanent magnet, (b) relay
armature suddenly released.
2. Show whether and how the theorem of constant flux linkages applies
to closed circuits having series capacitance, initially charged. Does it hold
if the capacitance is suddenly changed by switching or by mechanical
motion of the plates?
3. A transformer having self-inductances L 1 and L 2 and mutual induct-
ance M is operating in the steady state with voltage e = Em sin wt applied
to the primary terminals and with the secondary terminals open. At time t1
the secondary terminals are suddenly short-circuited. Derive expressions
for the primary and secondary currents, neglecting resistance. Assume
the primary voltage to be unaffected by the short circuit.
4. Plot the currents of the preceding problem for a transformer having
k = 0.9 (a) if the short circuit is applied at tl = 0, (b) if it is applied at
tl = Tr/2w.
6. A single-phase generator, having armature self-inductance La, field
inductance Ls, and mutual inductance M = M m cos wt, is initially operating
at no load and at field current 10• At time t = tl the armature is suddenly
short-circuited. Derive expressions for the armature and field currents,
neglecting resistance.
6. Plot the currents of the preceding problem for a generator for which
Mm/ VLaLb = 0.8, short-circuited at the instant when the armature flux
linkage is a maximum.
7. Prove that the magnetic energy stored in t\VO coupled circuits is
W = ! (1/! aia + 1/!bib ) .
8. By use of the expression for stored magnetic energy just given and the
relation that torque is the derivative of energy with respect to angle, plot
the torque of the single-phase generator for which M m/ v' LaLb = 0.8, short-
circuited at the instant when the armature flux linkage is zero.
9. Prove, by the method suggested in the footnote on 1). 0, that mutual
inductance is the sa.me in both directions.
10. A circuit of inductance /.11 is carrying current 11 when suddenly
additional inductance L 2 is switched into the circuit in series with Lt.
What is the current in the circuit right after the switching? Discuss energy
relations in the circuit.
11. A circuit of inductance £1 carrying current 1 1 and another circuit of
inductance L 2 carrying current I 2 are initially so located with respect to
one another that their mutual inductance is zero. How much mechanical
work must be done on the system to move one coil with respect to the
other so that the mutual inductance becomes M, (a) if the motion is very
sudden? (b) if the motion is very gradual and the tV{O currents remain con-
stant at their original values?
+
12. An e.m.f. e = 1 3t volts is applied at t = 0 to a circuit having
a resistance of 1 ohm and an inductance of 1 henry. The initial current is
1 amp. Find the current in the circuit (a) analytically and (b) by point-
by-point computation from t = 0 to 3 sec.
13. Repeat the last problem with an initial current of 4 amp.
14. What initial current in Probe 12 would lead to a linear increase
of current?
15. Show both analytically and by the graphical "follow-up" method
what wave form of voltage must be applied to an R-L circuit to obtain a
linearly increasing current, which begins at zero at zero time.
16. It is known that the differential equation
dx +~ = 0 (a)
dt T
has the solution
x = Ae- tl T (b)
Show that the corresponding difference equations
X n+l - Xn + Xn = 0 (e)
~t T
and
Xn+l - Xn Xn - 0 (d)
~t + T + (~t/2) -
have the solutions
(e)
and
(j)
respectively. Find T 1 and T 2 as functions of T and dt. Find the percentage
errors in T 1 and T 2 in terms of the number of points per time constant
(T / at).
PROBLEl\lS 135
17. Find the negative-sequence inductance L 2 of a salient-pole syn-

chronous machine without amortisseur windings in terms of the actual
self- and mutual inductances of the windings, Neglect field resistance.
18. Prove that if sinusoida.l negative-sequence voltages are impressed on
the armature windings of a synchronous machine the ne~a.t,lvP-8p.quence
reactance is the "harmonic mean' of ;td" and x;",
19. A turbogenerator having the reactances and time constants listed
below is initially operating with no load and with an armature voltage of
1.2 per unit. Then a three-phase short circuit is applied. Calculate and
plot d-e, and a-c. components of armature current and resultant armature
current versus time, assuming a fully offset wavee
Xd = 1.15 per unit Ta = 0.15 sec.
Xd' = 0.25 per unit 7 d'
1
= 1.50 sec.
Xd" = 0.10 per unit Td " = 0.03 sec.
20. The measurements given in Table 14 were made on the ordinates of
the envelopes of the oscillogram of one armature current taken when a three-
phase short circuit was suddenly applied to an unloaded turbogenerator
TABLE 14
MEASUREMENTS OF SHORT-CIRCUIT OSCILLOGRAM (PROBLEM 20)
Ordinates of Envelopes Ordinates of Envelopes

Time (inches) Time (inches)
(cycles) (cycles)
Upper Lower Upper Lower
0 2.63 -0.63 15 0.86 -0.62
1 2.24 -0.50 20 0.75 -0.63
2 1.96 -0.44 25 0.67 -0.61
3 1.73 -0.41 30 0.63 -0.59
4 1.56 -0.42 40 0.54 -0.54
5 1.44 -0.44 50 0.48 -0.48
6 1.33 -0.45 60 0.43 -0.43
7 1.24 -0.48 90 0.32 -0.32
8 1.16 -0.50 120 0.25 -0.25
10 1.05 -0.55 00 0.15 -0.15
rated at 10,000 kva., 11.8 kv., three-phase, 60 cycles, 1,800 r.p.m, The.
open-circuit voltage was 5.9 kv. The scale factor of the oscillogram is 1,930
amp. per inch (instantaneous value). Find the direct-axis reactances (in
per unit) and time constants (in seconds).
21. Draw to scale a vector diagram (similar to Fig. 34) of a salient-pole
synchronous generator having the following reactances (in per unit) :
Xd = 1.15 Xd' = 0.37
Xq = 0.75 X q' = 0.75
Neglect resistance. Take the load current as 1.00 per unit at a power factor
of 0.91 lag with respect to the terminal voltage, which is 1.00 per unit.
22. Draw to scale a vector diagram (similar to Fig. 35) for a turbogener-
ator having the following per-unit reactances:
:rd ~ 1.15 x(z' ~ 0.20
;t q =. 1.00
Neglect resistance. The machine is operating in the steady state with load
current and terminal voltage as specified in the preceding problem.
23. Work Example 4 with 50% external reactance between the generator
and the infinite bus. The terminal conditions stated in that example still
apply to the generator terminals, so the voltage of the infinite bus is different.
24. Would it be correct to represent a salient-pole machine on a calculating
board by a source of voltage E q ' in series with an adjustable reactance, rang-
ing .from xa,' to xq? If so, what would be the proper value of reactance as a
function of angle 0 + cP between E q ' and I?
CHAPTER XIII
EXCITATION SYSTEMS
In the last chapter, methods were given for calculating the field
current and field flux linkage of a synchronous machine when the
voltage applied to its field circuit was either constant or varying as a
known function of time. In this chapter, the methods of supplying
field voltage, the methods of making it vary, the methods of calculat-
ing its variation as a function of time, and the effect of such variation
on power-system stability are presented.
Description of excitation systems. The field windings of synchro-

nous machines are nearly always supplied with direct current from
d-e, generators called exciters. Former practice, now almost obsolete,
was for a station to have an excitation bus fed by a number of exciters
operating in parallel and supplying power to the fields of all the a-c.
generators in the station. The present practice is for each a-c. gen-
erator to have its own exciter, which is usually direct-connected to the
main generator but sometimes is driven by a motor or a small prime
mover or both. The advantages of the unit-exciter scheme over the
common-excitation-bus scheme are:
1. Greater reliability. Trouble on one exciter affects only one a-c.
generator. Reliability of the excitation system is of utmost impor-
tance. It is better to disconnect a generator from the bus than to
let it run for any length of time without excitation because in the
latter case the remaining generators have to supply not only the
load but also the reactive power for magnetizing the generator that
has lost its own excitation. *
2. Saving in rheostat losses. No rheostats are needed in the field
circuits of the main generators for controlling the voltage or reactive
power because the exciter-field rheostats can be used instead, and
these operate at a much lower power level than do the main-gen-
erator field rheostats.
3. Simpler layout of the station. The excitation bus, much of the
*Loss-of-field relay protection is.described in Chapter X, Vol. II.
137
138 EXCITATION SYSTEMS
switchgear associated with that bus, and main-field rheostats are

eliminated. All exciters need not have the same rated voltage.
4. Better sulitability to the use of automatic voltage regulators. The
division of reactive power between generators can be controlled
automatically by the voltage regulators. (See section on "Parallel
operation.")
In what follows the unit-exciter scheme is assumed. 'Though with
this scheme the excitation of the main a-c. generator could be regu-
lated by means of either a main-field rheostat or an exciter-field rhe-
ostat, the latter is generally used, both because rheostat losses are less
and because the control equipment is less cumbersome.
Exciters .are shunt- or compound-wound* and are rated at either
125 or 250 volts. They normally deliver 0.5% to 3% of the rated
power of the main machine.
Some exciters operate self-excited, others are separately excited,
and still others use a combination of self- and separate excitation.
The excitation of a separately excited exciter is sometimes furnished
by a rectifier or storage battery, but is supplied ordinarily from a
smaller d-c. generator called a pilot exciter. The conventional pilot
exciter is operated at approximately constant voltage, the excitation
of the a-c. generator being controlled, as before, by the field rheostat
of the main exciter. The pilot exciter is flat-compounded, and is rated
at 1.5% to 5% of the kilowatt rating of the main exciter. In some of
the newer methods of excitation, the pilot exciter is a rotating ampli-
fier, operated at varying voltage.
The advantages of separately excited exciters are:
1. Quicker response of the exciter voltage to a change in the re-
sistance of the regulating rheostat. The reason for, and the advan-
tage of, quicker response are explained later in this .chapter.
2. Stable operation at low voltage. Some synchronous machines
are required occasionally to operate with low excitation. This is
true of a synchronous condenser when taking lagging volt-amperes
from a bus in order to hold down the bus voltage. It is also true of
a generator required to charge a long, lightly loaded, high-voltage
line having much capacitance. If the excitation of the a-e. machine
is reduced to a low value by decreasing the voltage of a self-excited
exciter, the exciter may become unstable; that is, its voltage is
likely to drift away from the desired value. By special design
(having a reduced section of the main poles which begins to saturate
*Shunt-wound exciters respond more uniformly to the control of voltage regu-
lators than compound-wound exciters do.
E~XCITATION SYSTEl\I8 139
at 15 to 20% of rated voltage), a self-excited exciter can be made to

operate stably at voltages down to 25% of its rated value. A sepa-
rately excited exciter, however, can he operated stably at. as Iowa
voltage as required.
The disadvantages of separate excitation from a. constant-voltage
pilot exciter are:
1. The complication of an additional rotating machine with its
maintenance of brushes, commutator, etc.
2. A larger field rheostat for the main exciter.
The first disadvantage holds for the rotating-amplifier type of pilot
exciter also, but the second does not.
Unless one of the advantages listed above is required, a self-excited
exciter is used. Self-excitation is commoner than separate excitation
on small a-c. generators (about 15 Mva. and below), whereas the con-
trary holds for large a-c. generators.
The combination of self- and separate excitation of the main exciter
is found in some of the newer excitation schemes using rotating regu-
lators. These schemes will be described later.
The excitation of the synchronous machine is usually controlled, as
already stated, by means of a regulating rheostat in the field circuit
of the main exciter. This rheostat may be varied either manually or
automatically. In the latter case, a generator-voltage regulator is
employed.
Manual control of the generator voltage is usually satisfactory for
generators at attended stations the load of which changes slowly, as
on turbogenerators of large metropolitan systems. In some such sys-
tems, the use of voltage regulators was formerly considered to be more
of a hazard than a benefit, for instances occurred in which a sticking
contact or an accidental ground in the regulator reduced the field cur-
rent to such a low value that the generator pulled out of step. 58 Re-
cently, voltage regulators have come into use on metropolitan systems
because of the availability of more reliable and more effective regu-
lators, because of the need for operation of generators at high power
factor at times of light load - a condition under which a good volt-
age regulator significantly raises the margin of steady-state stability,
and because of the use of generators of lower short-circuit ratio.
Voltage regulators are necessary on synchronous condensers. (the pur-
pose of which is to control the voltage), on water-wheel-driven gen-
erators (to hold down the voltage in case of loss of load and consequent
overspeed), and on steam-turbine-driven generators subject to sudden
changes in load. They aid stability of all synchronous machines.
Types of excitation system. The principal types of excitation

system now in use (having automatic voltage regulation) are the
following:
1. Self-excited exciter and direct-acting rheostatic type of voltage regu-
lator (Fig. 1). The voltage regulator changes the resistance in the
field circuit of the exciter. The voltage-sensitive element of the regu-
lator (for example, a solenoid) acts directly on the rheostat to vary
its resistance. This system is commonly used on the smaller a-c.
generators. The voltage regulators are described in more detail later
in this chapter.
r
Exciter
.. , I
A-c generator
,
'----1
I I
I I
II II~ 0"irect -acting
.
I I voltage regulator
I I
II '--+0 I .....
1
L --1
FIG. 1. Excitation of a-c. generator with self-excited exciter and direct-acting

rheostatic voltage regulator.
2. Main and pilot exciters and indirect-acting rheostatic type of voltage

regulator (Fig. 2). When a small deviation of voltage occurs, the
voltage-sensitive element of the regulator closes contacts which con-
trol a motor-driven switch which raises or lowers the resistance of the
field rheostat of the main exciter. If the voltage deviation is greater,
additional contacts are closed which short-circuit the field rheostat to
produce a rapid build-up of the alternating voltage or contacts are
opened which introduce the entire resistance of the rheostat to produce
rapid build-down. This regulator has the disadvantage of a "dead
band," that is, a small range of voltage in which no corrective action
is produced. This dead zone reduces the accuracy of voltage regula-
tion and prevents the regulator from increasing steady-state stability.
Until recently this has been the standard scheme of excitation for
large machines. The voltage regulators are described in a later sec-
tion of this chapter.
TYPES OF EXCITATION SYSTEM 141
3. Rotating main exciter with electronic pilot exciter and electronic

regulator. 24 The electronic pilot exciter is a polyphase rectifier using
thyratron tubes or the like.
The a-c. supply for the rectifier may be either the main power system
or a permanent-magnet generator direct-connected to the main ma-
chine. The electronic regulator controls the firing angle of the tubes
Pilot exciter
/ I
/ I
(
I ~ Indirect-acting voltage regulator
I I
I I
I I
I I
I
I
I
I
I
I
L .J
'---fjl t-------'
FIG. 2. Excitation of a-c. generator with main exciter, pilot. exciter, and indirect-
acting rheostatic voltage regulator.
and thus varies the output voltage of the rectifier which is impressed
on the field winding of the main exciter. No field rheostat is used.
This system has been applied principally to synchronous condensers
but also to some generators.
4. Electronic main exciter and electronic regulator. 20 ,24 ,40 ,41 A six-
phase rectifier of larger capacity, using ignitron tubes, supplies .direct
current to the field of the main generator. Again, the regulator con-
trols the firing angle of the rectifier tubes. The anodes of each pair of
tubes are connected to their load through a two-pole circuit breaker.
When this breaker is opened, the load can be carried by the remaining
four tubes. Thus, tubes can be replaced without shutting down the
unit. The power supply for the rectifier is either the main a-c. gen-
erator or an auxiliary polyphase generator, usually direct-driven. If
the supply is from the main a-c. generator, series compensation/" may
be employed to boost the voltage in proportion to the reactive current

fed from the generator.
Electronic main and pilot exciters have operated very satisfactorily
but are not widely used because both their first cost and their main-
tenance cost have been found to be greater than for other types of
excitation systems.
5. Main exciter, rotating amplifier, and static voltage regulator. The
main exciter is either an ordinary shunt-wound machine or a shunt-
Current
FIG. 3. Voltage-current characteristics of elements used in impedance-type
voltage regulator.
wound machine with additional field windings, The rotating amplifier

is a d-e, machine especially designed as a power amplifier. 'Such ma-
chines, known by the trade names of Amplidyne (General Electric
Company), Rototrol (Westinghouse Electric Corporation), or Regulex
(Allis-Chalmers Manufacturing Company), are described in a later
section of this chapter. The important feature of such machines is
that a large output may be controlled by a few watts' input, which can
be supplied by a static type of regulator having no dead band and no
moving parts. Such regulators are accurate and reliable. They are
of either of two types, electronic or impedance. The electronic regu-
lator compares the constant voltage drop across a glow tube or across
a resistor in the plate circuit of a pentode with the rectified alternating
voltage being controlled, the difference being applied to a control-field
winding of the amplifier, or each voltage being applied to one of a
pair of such windings, the m.m.f.'s of which are opposed. The im-
pedance type of regulator'" compares the current through a nonlinear
impedance (usually an iron-cored inductor) with that through a linear
impedance (usually a capacitor). See Fig. 3. When the alternating
TYPES OF EXCITATION S'YSTEl\i 143
voltage is correct, the currents through the two impedances are equal.
When the voltage is high or low, one current or the other predominates
and, after rectification, excites the control-field winding (or windings)
of the rotating amplifier. In some cases, where more power gain is
required, a. magnetic amplifier has heen placed between the voltage
regulator and the rotating amplifier."!
Rotating amplifier Main exciter A-c generator

~ ~ ~
FIG. 4. Excitation of a-c. generator by main exciter which is separately excited

from a rotating amplifier controlled by the voltage regulator.
Several different methods of exciting the main exciter are used.

One (Fig. 4) is to excite the main exciter separately from the output
of the rotating amplifier.i" However, it is generally better to use
some combination of self- and separate excitation of the main exciter
Rotating amplifier Main exciter A-c generator
I \ / \ I . \
FIG. 5. Excitation of a-c. generator by main exciter which is both self-excited

and separately excited by a rotating amplifier connected in series with the shunt-
field circuit of the main exciter.
144 EXCITATI()N SYS'rEIvIS
so that the rotating amplifier can be removed from service for main-
tenance without shutting down the generating unit.
One such method is to connect the rotating amplifier in series with
the self-excited shunt field (Fig. 5). 24,51 ,5~ The field rheostat is so
adjusted that when the amplifier voltage is zero the exciter operates
self-excited to supply the proper field current for average loud on the
a-c. generator. The voltage of the amplifier either bucks or boosts
that of the exciter armature, as required for proper control of the
alternating voltage. The amplifier may be disconnected by a transfer
switch, the main exciter then being manually controlled.
Rotating amplifier
~
Main exciter
~
, A-c generator
A
FIG. 6. Excitation of a-c. generator by main exciter with two field windings
one of which is self-excited, the other separately excited by a rotating amplifier.
Another method (Fig. 6) is for the main exciter to have one shunt-
connected field winding with rheostat adjusted to provide excitation
for average conditions and another field winding separately excited
by the rotating amplifier, which either bucks or boosts the self-excita-
tion as required. 31 This method has the advantage of having no
switching in the main field winding of the exciter, but has the disad-
vantage of requiring a special type of exciter.
These schemes using rotating amplifiers have been extensively em-
ployed in recent years.
6. Rotating amplifier as main exciter and static regulator. 32 A two-
stage Rototrol has been used as the main exciter. One advantage of
this scheme is its rapid response because of the use of a series field
winding on the Rototrol. (See the next section of this chapter for
further explanation.)
EXCITER WITH SERIES WINDING 145
7. Main exciter, magnetic amplifier, and static regulator. 55 ,57 This

scheme, which is one of the newest, is similar in general to scheme 5
with an exciter with two field windings; except that the rotating ampli-
fier is replaced by a magnetic amplifier.
Pilot exciter Bridge Main exciter A-c generator

~ rheostat ~ ~
Mutual inductor, - M
FIG. 7. Excitation of a-c. generator by main exciter with series and separate
excitation.
Exciter with series winding. 25 ,59 The increase of field current of the
a-c. generator which is induced by an increase of lagging reactive
armature current, due to a short circuit or to swinging of the machine,
can be utilized in a series-field winding on the exciter to raise the
exciter armature e.m.f. This e.m.f. can be raised much more quickly
by means of the series field than it can be by suddenly reducing the
resistance in the shunt or separately excited field winding of the
exciter or even by suddenly raising the voltage applied to such wind-
ing by a rotating amplifier. The increased e.m.f. tends to maintain
the increased field current of the a-c. generator and so diminishes the
rate of decay of this current.
Figure 7 shows an exciter with a series-field winding and a separately
excited field winding. The latter is supplied with current from a pilot
exciter through a rheostat controlled by the voltage regulator. There
appears also in the figure a mutual inductor (transformer) the purpose
of which is as follows. A sudden increase of current in the series-field
winding tends to induce a decrease of current in the separately excited
field circuit, thus decreasing the effectiveness of the series field in
raising the armature flux and e.m.f. To prevent this, the two circuits
are decoupled by the addition of an external mutual inductance equal
146 F~xcrrATION SYSTEMS
and opposite to the mutual inductance between the two field windings
in the exciter itself.
The provision of series excitation on the exciter increases the time
constant of the alternator field circuit by decreasing the equivalent
resistance of that circuit. The RI drop, which is proportional to the
current in that circuit, is compensated, wholly or partly, by a com-
ponent of exciter armature e.m.f. which is also proportional to the
current (saturation being avoided). The increased time constant
means a slower decay of alternator field current and field flux linkage.
Overcompensation can provide an increasing flux linkage.
Series excitation is beneficial both to transient stability and to
steady-etate stability.
On the other hand, the inclusion of the inductance of the series-field
winding of the exciter in the field circuit of the a-c. generator raises
the transient reactance of the generator slightly and thus diminishes
proportionally the increase of field current required to maintain con-
stant flux linkage of the field circuit. There also occurs a redistribu-
tion of this flux linkage, the linkage of the exciter series field being
increased while the linkage of the field winding of the a-c. generator
is correspondingly decreased. These disadvantageous effects are be-
lieved to be small.
Rotating amplifiers. 37 ,44 ,56 Since special d-e, generators of several

types play such an important part in modern schemes of excitation, it
seems worth while to give a short description of these machines.
The Rototrol is similar in construction to an ordinary d-e, generator
except that the magnetic circuit is designed so that it does not saturate
in the working range. The machine has a self-exciting series-field
winding (connected in series with the armature, the load, and a
"tuning" resistor) and also has one or more control-field windings for
separate excitation. The tuning resistor is adjusted so that the "resist-
ance line" coincides as nearly as possible with the straight part of the
saturation curve. Thus the series field can supply all or very nearly
all the m.m.f. required at any voltage. The control-field winding or
windings supply the small difference needed in the steady state be-
cause of noncoincidence of the resistance line with the saturation curve,
as well as the m.m.f. needed under transient conditions to force the
voltage up or down.
The Regulex differs from the Rototrol in that the self-excitation is
supplied by a shunt-connected field winding instead of a series-con-
nected one. Therefore, the transient performance is characterized by
an additional time constant.
DRIVES FOR EXCI1'ERR 147
Both the Rototrol and the Regulex are single-stage amplifiers with
positive feedback. They have power gains of the order of 25,000 to
50,000.
The Amplidyne is ad-c. machine of special construction having
two sets of brushes 90 elec. deg. apart, one set of brushes being short-
circuited. The control-field winding or windings are located on what
we may call the direct axis and produce a speed voltage on the quad-
rature axis. The brushes on the quadrature axis are short-circuited;
hence a small m.m.f. in the control-field winding produces a large
current through the short circuit and a large m.m.f. (cross-magnetizing
armature reaction) in the quadrature axis. Rotation of the armature
with respect to the quadrature-axis flux induces a speed voltage be-
tween the direct-axis brushes, which are the output terminals. When
load is connected to these terminals, the armature reaction resulting
from the load current would oppose the m.m.f. of the control-field
winding, thus constituting a negative feedback which would reduce
the gain were not this armature reaction counteracted by a compensat-
ing winding, distributed in slots in the pole faces and carrying the load
current. The Amplidyne is inherently a two-stage amplifier, and gives
a power gain of from 10,000 to 250,000. It has a short time constant,
of the order of 0.05 to 0.25 sec.
A Rototrol having two stages of amplification in one 4-pole machine
has been built. 28 ,29 The armature is lap-wound without equalizers.
Two diametrically opposite poles, which in a conventional machine
would have the same polarity, are magnetized as the north and south
poles of a two-pole machine by the control windings of the 'first stage.
The output of this stage is taken from t\VO diametrically opposite
brushes and is used to energize the control-field winding of the second
stage, wound on all four poles in the normal manner for a four-pole
mac.hine. The output of the second stage is taken from all four brush
arms. Self-excited series or shunt windings, as already described for
the single-stage Rototrol or Regulex, may be connected to this stage.
Additional windings to compensate for armature reaction and to aid
in commutation are connected as described in Ref. 28.
Drives for exciters. The excitation system of a large synchronous

machine, and especially that of a generator, must have the utmost
reliability. The exciters should require but little maintenance and
they should be driven by a means which is reliable even during low
voltage or short circuits on the main power system and which will
permit the generator to be started up even when the main power sys-
tem is dead. If the main exciter is separately excited, the pilot exciter
148 EXCI'I'ATION SYSTEMS
also requires an equally reliable drive. If the main exciter is both

self- and separately excited, with separate excitation controlled by a
rotating amplifier, the requirements for reliable drive of the latter are
not so stringent, for the main exciter can be operated, if necessary, on
hand control with self-excitation.
Direct-connected exciters are the most widely used and may be con-
sidered as the standard type. Direct connection gives the most reli-
able drive, not being. affected by troubles on the a-c. system. The
record of reliability is very good.52 Some maintenance, such as chang-
ing brushes, must be done while the exciter is under load, or at least
while it is rotating, and this involves some hazard. The speed of the
exciter must be equal to that of the main generator rather than that
which gives the best design for the exciter. The limitations of the
direct-connected exciter have been felt in t"\VO different directions, on
slow-speed water-wheel-driven generators and on very large high-
speed turbine-driven generators. On slow-speed generators, direct-
connected exciters are larger and more expensive and have a slower
response than higher speed separately driven exciters. Turbine-gen-
erator units are being manufactured in larger capacities than before,
thus requiring exciters of larger rating. The new units are usually
built for 3,600 r.p.m., whereas, in the past, lower speeds were common.
The turbines, generators, boilers, and auxiliaries have become more
reliable, thus permitting longer times of operation between shutdowns
for maintenance. Thus there is a desire for greater reliability of the
exciter, while, at the same time, the larger ratings and speeds increase
the difficulties of commutation. A lower speed facilitates the design
of a reliable exciter of large rating.
Gear-driven exciters permit the exciter speed to be different from the
speed of the main generator, while retaining the reliability of the
direct drive.
Motor-driven exciters21 ,39 ,49 present advantages in addition to free-
dom in the choice of speed. The exciter may be placed in a less ex-
pensive and more convenient location than at the end of the generator
shaft. There commutation is better because of the absence of vibra-
tion transmitted from the turbine foundation. Moreover, if a spare
source of excitation is available, maintenance work on the exciter can
be done with the exciter shut down but without shutting down the
main generator.
The motor-driven exciter requires a reliable source of electric power.
In some stations the motor may be fed from a house generator, which
also supplies power to other station auxiliaries. In other stations the
motor is fed through a transformer from the main a-c. generator or
TIRRILL VOLTAGE REGULATORS 149
the main a-c. bus, in which case the alternating voltage is subject to
reduction or interruption during system disturbances. In order for
the excitation system to be reliable, it should be able to provide suit-
able ceiling voltage and rate of response when the motor is operated
on sustained low voltage equal to 70% of normal voltage or during
total absence of voltage (due to severe faults) of 12 to 15 cycles' dura-
tion. This may be accomplished by providing the following features:
1. Inertia of the motor-generator set increased by addition of a
flywheel to a value H = 5 (based on exciter .name-plate kilowatt
rating).
2. Use of an induction motor having maximum torque of about
five times its full-load torque.
3. An exciter ceiling voltage of at least 120% of its rated voltage
when the driving motor is operating at 70% of its rated voltage.
Hydrogen cooling. 24 ,52 Hydrogen-cooled synchronous condensers
operating at speeds up to 900 r.p.m. have been furnished with direct-
connected main exciters located in a hydrogen-filled compartment
which can be isolated from the main condenser compartment when
maintenance is to be done on the exciter. It has been found that hy-
drogen cooling of the exciter not only reduces the temperature rise of
the exciter but also decreases the wear of the commutator and brushes.
Hydrogen coolinghas been proposed for the exciters of turbogenerators.
Tirrill voltage regulators. In the past, the commonest type of

generator-voltage regulator was the vibrating type, or Tirrill regulator.
This regulator has contacts which open and close continually several
times per second. When closed, they short-circuit the regulating rhe-
ostat, and the exciter-field current and armature voltage build up.
When the contacts are open, the rheostat is reinserted in the field cir-
cuit, causing the field current and armature voltage to decrease. Be-
cause the period of vibration of the contacts is much shorter than the
time constant of the exciter-field circuit, the percentage fluctuation in
the field current and armature voltage is not great. The average values
of these quantities depend upon the time of engagement of the con-
tacts, which may be defined as the ratio of the time that the contacts
are closed to the period of vibration. A change in the time of engage-
ment has the same average effect as changing the setting of an exciter-
field rheostat.
One type of vibrating regulator, having a d-e. vibrating magnet, is
shown in Fig. 8. The main contacts are mounted on two pivoted arms,
each of which is actuated by an electromagnet. The winding of one
electromagnet is supplied with direct voltage from the exciter arma-

ture; the other, with alternating voltage from the main generator
(through a potential transformer). A decrease of either the alternat-
ing or the direct voltage tends to close the main contacts, and an in-
crease of either voltage tends to open them.
When the alternating voltage is correct, the upward force of the
a-c. magnet is in equilibrium with the unbalanced weight of the mag-
net core and counterweight. Furthermore, the force of this magnet is
A-c generator
~
D-c magnet
(vibrating)
Main contacts
A-c magnet
1"------------'
Exciter-field Differential I
rheostat relay
~ashpot
FIG. 8. Tirrill voltage regulator with d-e. vibrating magnet.
independent of position throughout a certain range; therefore the mag-

net core is in equilibrium in any position in this range if the alternating
voltage is at its normal value. This magnet is heavily damped by
means of a dashpot. Therefore, if the alternating voltage is wrong,
the magnet slowly moves the main contacts which it controls.
The d-e. magnet is not damped. Its force is opposed by the ten-
sion of a spring. This magnet and the contact actuated by it are in
continual vibration, caused by the following chain of events. If the
exciter voltage is high, the upward pull of the d-e, magnet on its core
overcomes the force of the spring and opens the main contacts. Open-
ing of the main contacts causes a differentially wound relay to open its
TIRRILL VOLTAGE REGULATORS 151
contacts (which can handle more current than the main contacts) and
thereby to insert resistance in the exciter-field circuit. Thereupon the
exciter armature voltage decreases, weakening the d-c. magnet suffi-
ciently to allow the spring to close the main contacts again. Closing
of the main contacts closes the relay contacts, which short-circuit the
exciter-field rheostat and make the exciter voltage build up. The
process is repeated over and over again, with the result that the ex-
citer armature voltage constantly increases and decreases through a
small range, as shown in Fig. 9.
Time
FIG. 9. Effect of vibrating regulator on exciter voltage during steady conditions

of main generator. The ratio tc/to is constant.
The time of engagement of the vibrating main contacts is such that

the average force exerted by the magnet during a cycle of vibration
equals the average force of the spring. This relation follows from the
fact that the average momentum of the magnet core is zero and the
fact that the force transmitted through the contacts is negligible. As
a consequence of this relation, the average exciter voltage depends
upon the position of the main contact that is actuated by the a-c.
magnet, because the position of this contact controls the average posi-
tion of the vibrating contact and hence the average spring tension.
If the alternating voltage is too low, the a-c. magnet slowly raises
the main contact which it controls. The higher this contact is raised,
the greater is the average exciter voltage, and when the exciter voltage
has been increased enough to raise the alternating voltage to the cor-
rect value, the a-c. magnet comes to rest. Similarly, if the alternating
voltage is too high, the a-c. magnet lowers its main contact, lowers
the exciter voltage, and lowers the alternating voltage to the correct
value. Because of the design of the a-c. magnet, the alternating volt-
age is restored to the same value regardless of the amount of load on

the main generator.
It may be asked why the motion of an a-c. magnet alone could not
be used to operate the main contacts and thus to cut resistance in and
out of the exciter-field circuit. The reason for adding the d-c. magnet
is to prevent overshooting of the alternating voltage. The inductance
of the field circuit of the a-c. generator causes changes of the alternat-
ing voltage to lag behind changes of the exciter voltage. If the exciter
voltage should continue to build up until the alternating voltage had
risen to the desired value, the exciter voltage would then be too high
for the steady state and the alternating voltage would continue to
rise even if the exciter voltage should begin to decrease because of
regulator action. The exciter voltage would continue to decrease
until the alternating voltage had gone through a maximum and de-
creased to normal. At that time the exciter voltage would be too low
for the steady state and the alternating voltage would continue to
decrease. Thus large oscillations of the alternating voltage would
occur, which might or might not die out. Such oscillations are pre-
vented by the d-e, magnet operating the main contacts in anticipation
of the action of the a-c. magnet.
Although the Tirrill regulator gives good voltage regulation, it re-
quires more maintenance than the more modern types, it is noisy,
and there is a possibility that a sticking contact may cause the volt-
age to go too high.
Rheostatic voltage regulators. The Tirrill regulator was superseded

by regulators of the rheostatic type in which the regulating resistance
is varied continuously or in small steps instead of being first completely
cut in, then completely cut out. Under steady conditions, all parts of
the regulator are at rest; therefore the wear is small.
Rheostatic regulators can be classified into direct-acting and indirect-
acting types. In the direct-acting type, the voltage-sensitive element
of the regulator (for example, a magnet or a torque motor) controls
the rheostat through a direct mechanical connection. Special types of
rheostats must be used, so that only a small motion and small amount
of energy are required to vary the resistance from one extreme to the
other. In the indirect-acting type of rheostatic regulator, the voltage-
sensitive element operates contacts which, in turn, control a motor to
drive the rheostat. This action suffices to correct small or gradual
changes of voltage, but the motion of the rheostat is too slow for
rapid correction of sudden, large changes. When such changes occur,
additional contacts on the voltage-sensitive element actuate relays
DIRECT-ACTING REGULATORS 153
which cut in or out the entire regulating resistance in order to correct

the abnormal voltage as rapidly as the response of the exciter and
of the main generator will permit.
Direct-acting voltage regulators. Typical of these are the General

Electric "Diactor," the Westinghouse "Silverstat," and the Allis-
Chalmers (Brown-Boveri) rocking-contact regulator. The first two
_ _ _ _ _ Regul.tor ,I,
...... Apparatus edern.11o regul.tor - - - - - '
I ReClifier.,.;'
I
I
I
I
II currentt"nsf~rmer (ifused)
I
I
Voll.ge- I
I
.djusting ,
rheost.l
I -\
' I "
-,potenti.l)
transformer
Auxiliary .pparatus furnished as
I
one sell-contained unit with its
I,
own terminal board
I,
I
FIG. 10. "Diaetor" direct-acting exciter-rheostatic voltage regulator for a-c. gen-
erator. (By courtesy of General Electric Company.)
are shown in Figs. 10 and 11, respectively. In both these regulators

the voltage-sensitive element is an electromagnet, the winding of
which is energized from one phase of the a-c. circuit whose voltage is
to be controlled, through a potential transformer, a voltage-adjusting
rheostat (manually operated), and a dry rectifier. A small motion of
the armature of the electromagnet acts to cut the required amount of
resistance in or out of the regulating rheostat in the exciter-field
circuit.
The two regulators differ principally in the form of the regulating
rheostat.
154 EXCITATION SYSTEMR
The rheostat of the "Diador" consists of two or four stacks of non-

metallic resistance plates. Each resistance plate has a silver-button
insert, slightly thicker than the plate, passing through the plate at the
front end. Insulating spacers separate the plates at the rear of the
stack. At the center of the stack there are metal contact plates be-
tween the resistance plates. As the rontact plates are thicker than
I'
I r-------i----:-::-;c-:-:-p r<>;---vNW>JVO
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I Polenli.1
I transformer
FIG. 11. "Silverstat" direct-acting exciter-rheostatic voltage regulator for a-c.

generator . (By courtesy of Westinghouse Electric Corporation.)
either the protruding portion of the silver buttons or the insulating

spacers, they act as fulcrums on which the resistance plates may rock.
They also form electric contacts between the resistance plates. The
stack is operated by a rod, which, under the tension of a spring, pulls
down the front end of the top plate. The pull of the spring is counter-
acted by the force of the electromagnet.
When the resistance plates are tilted back so that the silver buttons
are separated, the rheostat is in the position of maximum resistance,
as shown in Fig. 12a. The current path is then through the alternate
resistance plates and metal contact plates. If the stack is gradually
tilted forward, its resistance is gradually reduced, both because of a
decrease in contact resistance between the resistance plates and metal
Current path
(a)
r-1
Currentnpath
A
....
F=
F=
F
f-
""
u =' F
~
~
~
.r:
15
0
I-
t
'"
(e)
U ~
FIG. 12. Paths of current through rheostatic stack of "Diactor" generator-

voltage regulator. (a) High-resistance position, (b) medium-resistance position,
(c) low-resistance position. (By courtesy of General El ectric Compa ny.)
155
contact plates and because the resistance plates are progressively

short-circuited by the silver buttons as they come together. Figure 12b
shows the current path for an intermediate position. In the position
of minimum resistance, shown in Fig. 12c, all the resistance plates are
short-circuited by the silver buttons.
Resistance
Driving
element-
Silver button"
FIG. 13. The "Silverstat" rheostat. (By courtesy of Westinghou,qe Electric

Corporation.)
The"Silverstat" differs from the rheostat just described in that the

resistors are stationary and only the contact assembly is movable
(see Fig. 13). Taps from the stationary rheostat are brought by
wires to a set of leaf springs, clamped together at one end, with in-
sulation between adjacent springs. In the maximum-resistance posi-
tion the free ends of the springs rest upon an insulating block, and the
springs do not touch one another. A silver button is set into each
spring near the free end. The driving element presses on the springs
and causes the silver buttons to come together and successively short-
circuit sections of resistance.
DIRECT-ACTING REGUL.~TORS 157
Both the "Diactor" and the "Silverstat" employ as a damping

device a transformer, the ·primary winding of which is connected
across the exciter armature, the secondary winding in series with the
rectifier output terminals and the coil of the control magnet. A volt-
age appears across the secondary terminals of this transformer only
when the d-e, primary voltage is rising or falling. When the voltage
of the exciter is rising as a result of regulator action, the voltage of
the a-c. generator rises with considerable lag. However, the secondary
Resistors
FIG. 14. Rocking-contact rheostat, used in Allis-Chalmers (Brown-Boveri) gen-

erator-voltage regulators.
voltage of the damping transformer under these circumstances is of

such polarity that it appears to the regulator coil that the alternating
voltage is rising faster than it actually is. In this manner overshoot-
ing of the exciter and generator voltages is prevented, or is limited to
a small amount, depending upon the amount of damping feedback
introduced.
It should be remarked that excitation systems containing voltage
regulators are closed-cycle control systems. Concerning such systems,
various investigators have worked out a mathematical theory for
explaining and predicting such characteristics as accuracy, speed,
hunting, and damping. References 15 and 16 discuss the "Silverstat"
and "Diaotor" regulators, respectively, from this viewpoint.
The Allis-Chalmers (Broun-Boveri) voltage regulator has a rocking-
contact rheostat, shown in Fig. 14. This rheostat has a stationary
commutator consisting of a large number of segments insulated from
one another and arranged in a circular arc. These stationary seg-
ments are connected to resistors. The moving contact is sector-shaped.
158 EXCITATION SYSTEl\IS
Its contact surface, which rides in a V-shaped groove on the inner

surface of the commutator, is a curved strip of carbon of somewhat
smaller radius than that of the commutator. A rocking motion of the
sector causes it to make contact with various segments of the commu-
tator, thus cutting resistance in or out. One or more of these sectors
is actuated either hy a solenoid or by a split-phase torque motor, op-
posed by a spring. A damping device, consisting of either a dashpot
or an eddy-current brake, is coupled to the main moving parts through
a "recall spring." When a change of voltage occurs, the initial motion
of the sectors is greater than required, but, as the deviation of voltage
from normal decreases, the recall spring pulls the sector back fast
enough to prevent the voltage from overshooting.
Indirect-acting voltage regulators. The use of the direct-acting

regulators described above is confined to small- and medium-sized a-c.
generators (up to about 25 Mva, at 1,200 r.p.m. or higher) having
self-excited exciters. For larger generators, and for those with sepa-
rately excited exciters, the exciter-field rheostats become too large to
be directly actuated by the voltage-sensitive element. Therefore
motor-driven face-plate rheostats are resorted to, and the rheostat
motor is controlled by the normal contacts of the voltage regulator for
correcting gradual changes of voltage. Quick-response contacts also
are provided for raising or lowering the voltage as rapidly as possible.
These contacts actuate high-speed contactors which cut the regulating
resistance all in or all out, as required. As already stated, the quick-
response contacts are actuated by voltage deviations larger than those
actuating the normal contacts. Great voltage deviations are caused
by large, sudden changes of load, by short circuits, and by generator
swings.
Use of polyphase voltage. In the past it was customary to furnish
the voltage-sensitive element of the regulator with one line-to-line
voltage. This practice is still followed on the regulators for small
generators. If, however, it is important that the regulator always
raise the excitation when a fault occurs, it is preferable to use all
three line-to-line voltages. The reason is that, if a single line-to-line
voltage is used, it may increase, rather than decrease, during a fault on
another phase, thereby causing the generator excitation to be lowered
instead of raised. The voltages of the three-phase circuit, usually ob-
tained through two potential transformers connected V-V, can be
utilized in any of the following ways:
1. The output of a positive-sequence filter network is applied to
the regulator coil.
INDIRECT-ACTING REGULATORS 159
2. The d-e, output of a three-phase rectifier is applied to the coil.

3. The voltage-sensitive element is a three-phase induction torque
motor.
Westinghouse Type B J indirect-acting exciter-rheostatic generator-
voltage regulator. The circuits of this regulator are shown in Fig. 15.
The voltage-sensitive element is an electromagnet supplied with direct
current from a three-phase rectifier. This magnet actuates two sets
of contacts, the normal-response contacts Land R and the quick-
response contacts AL and AR.
Voltage deviations of ±O.5% .cause one of the normal-response con-
tacts to close. If the voltage is low, contact R closes, energizing the
coil of contactor NR, which runs the main-exciter rheostat motor in
the direction to cut resistance out of the rheostat. Similarly, if the
voltage is high, contact L closes, energizing contactor NL, which runs
the motor in the opposite direction. The rheostat motor has two series-
field windings, one for each direction of rotation.
Voltage deviations of ±1.5 to 5% (depending upon the setting)
cause one of the quick-response contacts to close in addition to one
of the normal-response contacts. If the voltage is low, contact AR
closes, energizing contactor QR, which short-circuits the main-exciter
field rheostat and an additional field-forcing-up resistor. This action
gives maximum speed of build-up of the exciter voltage. At the same
time the rheostat motor is cutting resistance out of the rheostat. If
the voltage is high, contact ilL closes, energizing contactor QL which
inserts the field-forcing-down resistor into the exciter-field circuit.
This causes the exciter voltage to build down rapidly. At the same
time the rheostat motor is cutting resistance into the field rheostat.
Each set of regulator contacts is equipped with an antihunt device
in the form of a coil (NH and QH) which, when energized, increases
the gap distance between contacts Rand L (or AR and AL) so as
temporarily to decrease the sensitivity of the regulator.
Suppose that a small deviation from normal voltage occurs. Either
contact L or contact R closes and energizes NR or NL. These con-
tactors not only operate the rheostat motor, as already described, but
also through additional contacts operate antihunt device NH, and
through a third set of contacts shunt the coil of the contactor with a
capacitor-resistor timing circuit. The antihunt device opens contact
R or Land would de-energize the corresponding contactor were its
coil not shunted with the timing circuit. The contactor is de-energized
after a short time delay, which permits the rheostat arm to be driven
a definite distance - for example, froIn one button to the next on the
rheostat face-plate. After the rheostat motor stops, it is desirable to
provide a time delay to allow the a-c. machine voltage to reach a

steady value. Such delay is obtained by means of a dashpot on the
antihunt device, which prevents the regulator contacts from immedi-
ately returning to their normal position. After expiration of this
Pilot exciter A-c generator
~ ~
Differential
field
resistor
Field-forting-down
c resistor
Voltage -adjusting
rheostat
D-c control bus
D-c control bus

FIG. 15. Simplified circuits of Westinghouse Type BJ indirect-acting exciter-
rheostatic generator-voltage regulator. (By courtesy of Westinghouse Electric
Corporation.)
INDIRECT-ACTING REGULATORS 161
delay, the normal-response contacts will again close if the alternating

voltage has not returned to normal. This will start another cycle of
operation such as just described, and these cycles will continue until
normal alternating voltage has been restored.
When the original deviation is large enough, the regulator contacts
will remain closed even though the antihunt device changes the con-
tact setting. The rheostat motor will then run continuously until the
alternating voltage gets within the zone for which the antihunt device
is set, whereupon intermittent action, as previously described, restores
the voltage to normal.
The antihunt device on the quick-response contacts works in a
manner similar to that just .described for the device on the normal-
response contacts.
General Electric 'llype GIlA-4 indirect-acting generator-voltage regu-
lator. See Fig. 16 for the circuit diagram. The voltage-sensitive
element is a three-phase torque motor (an induction motor with solid
cylindrical steel rotor) whose motion is opposed by a spring. Mounted
on the shaft of the torque motor is a contact assembly comprising four
contact arms, arranged in two pairs: the accelerating contacts (R2
and £2) in the rear and the notching contacts CRl and £1) in the
front. Each pair consists of a raising contact (R1 or R2) on the left
and a lowering contact (L1 or £2) on the right. Between the two
notching contacts is a metal star wheel, and between the two acceler-
ating contacts is a smooth wheel. Both wheels are driven at 4 r.p.m.
by the contact motor. When the voltage is normal, none of the four
contacts touches a wheel. When the voltage deviates from normal,
the torque motor tilts the contact assembly, causing one or two of the
contacts to touch the rotating wheels and complete the circuit to a
contactor. If the voltage is slightly too low, the raising notching con-
tact R1 intermittently touches the star wheel, causing the rheostat
pilot motor to run intermittently in the direction to cut out resistance.
Likewise, if the voltage is slightly too high, the lowering notching
contact £1 touches the star wheel, causing the pilot motor to cut
resistance in. Time is allowed between points on the star wheel for
the alternating voltage to approach a steady value. In addition, as
each point on the star wheel engages the contact R1 or £1, it gives the
latter a mechanical impulse which tends to restore the contact assembly
to the neutral position. This action, in conjunction with the inter-
mittent operation of the rheostat, prevents overtravel of the latter
and consequent hunting. There is also a magnetic damping device to
prevent excessive oscillation of the contact assembly.
The greater the voltage deviation from normal, the greater is the
162 EXCITATION SY'S'rElVIR
time of engagement of the notching contacts, causing the rheostat

motor to run faster.
If the voltage deviation is great enough, one of the accelerating
contacts touches the smooth wheel and operates a high-speed relay
which cuts in or out all the regulating resistance, At the same time
the rheostat motor runs at, full speed.
Control bus Front view

of regulator
j
Potential
transformers
Damping
magnet Generator
Rheostat-motor Rheostat-motor Exciter

raise contactor lower contactor shunt
field
FIG. 16. Elementary connection diagram of General Electric Type GFA-4

indirect-acting exciter-rheostatic voltage regulator with face-plate-type series
rheostat and self-excited exciter. (By courtesy of General Electric Company.)
Allis-Chalmers Type J indirect-acting generator-voltage regulator.

This regulator has a rocking-contact rheostat similar to those used on
the direct-acting regulators made by the same company. In the
indirect-acting regulator, however, the rheostat is motor-driven. The
rheostat motor is controlled through relays by "rheostat-inching"
PARALLEL OPERATION 163
contacts driven by the voltage-sensitive element, a torque motor with

a hollow cylindrical rotor. The engagement of the rheostat-inching
contacts is not continuous, since the relays which control the motor
also introduce antihunting rheostats into the circuit of the torque
motor, thereby changing its voltage setting. This causes the torque
motor to open the contacts again, de-energizing the rheostat relays
and motor. As a result the rheostat is moved in short steps and over-
travel is prevented.
When larger voltage deviations occur, the control element closes
another set of contacts, which cause field-forcing contactors to cut
in or out large blocks of regulating resistance.
Parallel operation. When a-c. generators are operated in parallel on

the same bus, the division of reactive power between generators is
controlled by their excitation. If voltage regulators are used, they
must be modified so that they will not only hold the bus. voltage
constant but also give the proper division of reactive power. Modern
voltage regulators are intended for use with individual nonparalleled
exciters. When so used, the regulator is adapted most simply for
parallel operation by the provision of cross-current compensation. In
this method the voltage furnished to the voltage-sensitive element of
each regulator is increased in proportion to the lagging reactive current
supplied by its own machine. Then, if a machine is supplying more
than its share of reactive current, its regulator will act as though the
bus voltage were too high: that is, it will decrease the excitation
and thereby decrease the reactive current supplied by its machine.
The result of cross-current compensation is to give a slightly drooping
curve of voltage versus load. The amount of droop is usually negli-
gible, however. If each machine is given the same droop at rated
current, zero power factor, then the machines will divide the reactive
current in proportion to their kilovolt-ampere ratings.
Figures 10 and 11 show, in dashed lines, how cross-current compen-
sation is added to the circuits of direct-acting regulators. A biasing
voltage proportional to the current in one phase wire is added vecto-
rially to the voltage between the other two wires. At zero power factor
lagging, the biasing voltage augments the main voltage; at zero power
factor leading, it subtracts from the main voltage; and at unity power
factor, it adds ill quadrature and thus has little effect.
If the drooping voltage characteristic is objectionable, a flat charac-
teristic can be obtained by differential cross-current compensation. In
this method each regulator is biased with the difference of the reactive
current in its own machine and the average of such currents in the
other machines instead of with the reactive current of its own ma-
chine alone.
If there is reactance between regulated generators or stations, not
less than 6% based on the rating of the smaller machine or station,
the regulators can be operated successfully without cross-current
compensation. Because of the prevalent practices of operating gener-
ators as units with their step-up transformers (paralleled on the high-
voltage side) and of using bus-sectionalizing reactors in metropolitan
systems where step-up transformers are not used, the required amount
of reactance is usually present between machines in the same station,
and therefore cross-current compensation is seldom required. How-
ever, compensation is still recommended unless the reactance is some-
what greater than 6%.
""'-",'''''~'''- ±
+ ....----1
Pilot - exciter
Yfield rheostat
Shunt field
(a) (b) (c)
FIG. 17. Wheatstone-bridge rheostat for wide-range control of excitation.

(a) Full positive excitation applied to main-exciter field, (b) approximately zero
excitation applied to main-exciter field, (c) slightly negative excitation applied to
main-exciter field. (By courtesy of General Electric Company.)
Wide-range excitation. As previously "mentioned, synchronous

condensers or generators with line-charging functions must sometimes
operate at very low excitation. Because a self-excited exciter becomes
unstable if operated at too Iowa voltage, it is necessary either to use a
main-generator field rheostat or to separately excite the exciter. The
latter alternative is preferable from the standpoint of ease of control.
RELATION '1'0 THE STABILITY PROBLE~1 165
Sometimes synchronous condensers are required to operate with

reversed excitation in order to increase the lagging reactive power
drawn by the condenser. Three methods of accomplishing this wide-
range control are:
1. Wheatstone-bridge field rheostat (Fig. .17).
4~
2. A differential-field winding on the main exciter, operated at
constant current (Fig. 15).
3. Use of a rotating amplifier.
Excitation limits. 22,31,58 In some installations it may be desirable to

provide upper or lower limits to the excitation. An upper limit may
be provided to prevent overheating of the field winding or to limit
the armature current of synchronous condensers even though such
limitation allows the voltage to drop below the desired value. A
lower limit is sometimes used on generators to insure against loss of
synchronism due to insufficient excitation. The lower limit may be
fixed or it may be a function of the load on the machine. The variable
lower limit is preferable to the fixed limit, for it allows better control
of the voltage at light load while insuring adequate excitation for
stability at heavy load. In the earliest installations of a variable
lower excitation limit, the limit was a function of the position of the
throttle or gate of the prime mover. More recently it has been a
function of active power output of the generator.
Excitation limits are most easily provided in the excitation systems
using static regulators.
Relation of excitation system to the stability problem. Increased

speed of exciter response was one of the first means suggested and
applied for improving power-system stability. The excitation system
affects stability under both transient and steady conditions. One can
understand this by recalling that the power transmitted in a two-
machine system is, according to the approximate equation, propor-
tional to the product of the internal voltages of the two machines,
divided by the reactance. Even when calculated more rigorously, the
power is increased if either internal voltage is increased. These state-
ments hold regarding the power at any particular value of angular
separation, and hence also for the maximum power. It is also true
on a multimachine system that raising the internal voltages increases
the power that can be transmitted between any two machines or
groups of machines. Therefore, it is apparent that raising the internal
voltages increases the stability limits.
166 EXCITATION SYSTElVIS
Under transient conditions power is calculated with the use of the

transient reactances of the synchronous machines and the voltages
behind transient reactance (which, it will be recalled, are proportional
to field flux linkage). Upon the occurrence of a fault, the flux linkages
are initially the same as they were just before the fault occurred.
During the fault, however, the flux linkages decay at a rate described
by the short-circuit time constant T d', which is least in the event of a
three-phase short circuit at the terminals of the machine and is some-
what greater for less severe types and locations of fault.
If the fault is sustained for a long time, a machine may survive the
first swing of its rotor, but, because of the continued decrease of its
field flux linkage, it may pull out of step on the second swing or on
subsequent swings, An excitation system controlled by an automatic
voltage regulator causes the flux linkage first to decrease more slowly
and then to increase (see curves of E q ' in Fig. 42 of Chapter XII).
As a result, a machine which does not go out of step on the first few
swings will not go out of step on subsequent swings of the same dis-
turbance. On a more severe disturbance, however, it may pull out of
step on one of the first few swings, Park and Bancker'! state that a
moderate exciter response (150 to 250 volts per second for 250-volt
fields) is usually sufficient to prevent loss of synchronism on other
than the first swing. They further state that the gain in stability
limit due to a higher response than this is small in comparison with
the gain obtainable with this value of response. In substantiation of
these statements they report the results of calculations made on a
water-wheel generator feeding a load center over transmission lines of
three different lengths, namely, 60, 120, and 250 miles. The power
which could be carried without loss of synchronism on the first swing,
after the occurrence of a line-to-ground fault on the generator bus,
with fast enough exciter response (about 600 volts per second) to
maintain constant field linkages was found to be only from 4% to 6%
greater than the power that could be carried with exciter response
(150 to 200 volts per second) sufficient to prevent pullout on the
second swing. Nevertheless, it is still true that the higher the speed
of exciter response, the greater is the power limit.
Floyd and Sills34 express similar conclusions regarding the necessary
speed of response of excitation systems for water-wheel-driven gener-
ators, saying that, if a rate of response of 1.5 or 2.0 per unit is specified,
there is a very little gain in stability over that obtained with a rate of
1.0. (See page 170 for the definition of nominal exciter response
expressed in per unit.)
Back in 1928, clearing times of 30 to 60 cycles were not uncommon.
RELATION TO THE STABILIT)" PROBLEM 167
Under those conditions the action of the excitation system during a

fault was an important factor in power-system stability. However,
high-speed clearing of faults has become the standard practice wherever
stability is critical. With short fault duration, the decrease of flux
linkages during the fault is small and the effect of the excitation system
on the linkages during this period is likewise small.
After the fault has been cleared and the synchronous machines are
swinging apart, the armature current still has a demagnetizing effect,
though a smaller effect than that during the fault. (Recall that, when
the machines have swung 180° apart, the currents are the same as
those caused by a three-phase fault about half-way between machines.)
Therefore, even when high-speed clearing is used, the excitation system
can lessen the decrease of flux linkages or - better still - cause an
increase of linkages. This it can do more effectively after the fault
has been cleared than before because the demagnetizing effect of the
armature current is smaller. Thus the excitation system can aid
transient stability appreciably even though high-speed clearing of
faults is employed. The faster the excitation system responds to
correct low voltage, the more effective it is in improving stability.
Concordia'" has shown that the exciter response required to give
the same stability limit as that obtained with constant flux linkage is
not constant but increases with increasing switching time. In other
words, the required exciter time constant decreases with the increase
of switching time, as shown by the following results calculated for a
200-mile transmission system:
Switching Time Exciter Time Constant
(see.) (sec.)
0.02 2.7
0.06 0.7
0.10 0.3
These results indicate that the effect of fault-clearing time is even

more important than is indicated by the usual transient stability
studies in which constant flux linkages are assumed.
In the steady state, power is calculated with the use of saturated
synchronous reactances and the voltages behind these reactances
(which are proportional to the respective field currents). If the volt-
ages are constant, the power limit is reached when the phase angle
between the voltages becomes 90°. An automatic voltage regulator
tends to hold the terminal voltages constant. If it were entirely
successful in doing so under' all conditions, the power limit would be
reached when the angle between terminal voltages reached 90°. In
168 EXCITATION S\TSTEMS
other words, the power limit would depend upon the reactance of the
circuits between machines, instead of upon this (external) reactance
plus the internal (synchronous) reactances. The power limit would
be greatly raised by such ideal voltage regulation, particularly if the
internal reactance were the major part of the total reactance.
Unfortunately, voltage regulators cannot raise the power limit to
anything near the theoretical value just discussed. Even though the
regulator itself acts without delay, the field circuits of the exciter and
of the main generator are sluggish. Consequently, the desired restora-
tion of voltage cannot always be obtained rapidly enough to increase
the electric power in a way to match the mechanical power and thus to
prevent power differentials that pull the machines out of synchronism.
It has been found by tests, however, that the use of quick-response
excitation systems does raise the steady-state stability limit a sub-
stantial amount. The limit is approximately equal to that calculated
on the assumption of constant voltage behind a reactance intermediate
between saturated synchronous reactance and transient reactance.
In other words, quick-response excitation counteracts approximately
half of the armature reaction.
Stable operation with automatic voltage regulators at values of
power beyond the stability limit obtainable with hand control of the
excitation is termed dynamic stability. The excitation system, by
increasing the internal voltage, must increase the synchronizing power
more rapidly than the increase of angle decreases the synchronizing
power. The gain in stability limit obtainable by this means is relatively
great in the case of machines connected without external reactance,
but is relatively small whan the external reactance is large.
Although the use of quick-response excitation and voltage regu-
lators for increasing steady-state stability was proposed as long ago as
1928,8,9,10 they have been little used for this purpose. The reasons
probably are:
1. In most cases the prudent loading of lines is determined by the
transient stability limit, which is lower than the steady-state power
limit.
2. The gain in steady-state stability limit ascribable to the exci-
tation system is small if the external impedance is large, as it usually
is where stability is an important factor in limiting the power
transmitted.
3. Operators have been hesitant to depend upon automatic
voltage regulators for stable operation at rated load.
An application of quick-response excitation for increasing the steady-
DEFINITIONS OF EXCITER RESPONSE 169
state stability limit has been made to ship-propulsion systems having

a synchronous generator and a synchronous motor connected through
negligible external reactance.l" By this means it was possible to
use much smaller synchronous machines than could have been used
otherwise.
The recent trend toward the use of turbogenerators of lower short-
circuit ratio, and hence of lower inherent steady-state stability, has
aroused interest in the use of quick-response excitation. Even though
the generators are normally stable when operated with hand control
of excitation, the use of suitable voltage regulators increases the mar-
gin of stability.
To be effective in increasing the steady-state stability limit, the
excitation system should have a voltage regulator with no dead band.
Quick-response excitation, in addition to raising stability limits,
diminishes the disturbance to voltage caused by such events as a short
circuit, the opening of a transmission line, swinging of machines, or
the disconnection of a generator; and thus it improves the quality of
electric power service.
Dynamic stability is discussed further in Chapter XV.
Definitions of exciter response. In the foregoing discussion of how

system stability is affected by excitation systems, the term "exciter
response" was used without definition. Definitions of this term
adopted by the A.I.E.E. 17 will now be quoted and commented upon.
"Exciter response is the rate of increase or decrease of the main-
exciter voltage when resistance is suddenly removed from or inserted in
the main-exciter field circuit," as by the action of the voltage regulator.
The response may be expressed either in volts per second or in per-unit
voltage per second. In the latter case, the base voltage is not usually
the rated voltage of the exciter, as might be expected, but .rather the
nominal collector-ring voltage of the main generator, a quantitywhich
is defined below. (N ote that this base voltage differs from that used
in the last chapter, where it was the voltage required to send through
the field winding of the main generator the current required to give
rated voltage on the air-gap line.)
The actual rate of response of a particular exciter varies with such
circumstances as (1) speed of rotation, (2) whether the exciter is self-
excited or separately excited, (3) if separately excited, its initial field-
circuit voltage and resistance, (4) the initial armature voltage and
current, (5) the amount of resistance cut in or out of the field circuit,
and (6) the load on the exciter during build-up or build-down, a load
determined not only by the characteristics of the field circuit of the
main generator but also by transient field current induced by changes

of the armature current of the main generator. Moreover, the rate of
response during build-up is usually different from that during build-
down. Even with all the foregoing conditions fixed, the response
varies from instant to instant during the process of build-up or build-
down.
Since the exciter response depends upon so many different factors it
is necessary to define a "nominal exciter response" which is a definite
quantity for a particular excitation system and thus is suitable as a
basis for specifications, guarantees, and tests, as .well as for approximate
comparison of the effectiveness of different excitation systems.
The nominal exciter response is defined as the numerical value
obtained when the nominal collector-ring voltage is divided into the
slope, expressed in volts per second, of that straight-line voltage-
time curve which begins at nominal collector-ring voltage and
continues for one-half second, under which the area is the same as
the area under the no-load voltage-increase-time curve of the exciter
starting at the same initial voltage and continuing for the same
length of time.
Nominal collector-ring voltage is the voltage required across the
collector rings to generate rated kilovolt-amperes in the main machine
at rated voltage, speed, frequency, and power factor, with the field
winding at a temperature of 75 degrees Centigrade.!?
When nominal exciter response is determined by test, the exciter
shall be operating at normal speed and with normal rheostat and field
connections. In applying the short circuit to the rheostat, only the
regulating resistance, and not any permanent resistance supplied to
limit the field current, shall be short-circuited. * If the exciter is
separately excited, the voltage of the pilot exciter or other source of
excitation shall be the same as will be used in actual operation.
The definition of nominal exciter response may b~ understood better
by perusal of Fig. 18. The actual build-up of voltage, starting at
nominal collector-ring voltage Oa, will follow a curve somewhat like
aed. This curve may be determined either by test or by calculation,
as explained later in this chapter. Straight line ac is drawn so that
the area under it is equal to that under the actual build-up curve
during the first half-second. The nominal exciter response is then the
slope of line acin volts per second, divided by Oain volts.
The justification for this particular way of defining nominal exciter
response is somewhat as follows: The time of 0.5 sec. is chosen because
"In the excitation system of Fig. 15, for example, both the "main-exciter field
rheostat" and the "field-forcing-up resistor" would be short-circuited when testing
for exciter response, but the "permanent resistance" would not be.
VALlJES OF EXCITER RESPONSE 171
the time during which quick-response excitation is applied in practice

is of this order of magnitude, The area nnder the curve during this
time is used instead of the increase of voltage during the same time,
because, when the exciter voltage is applied to an inductive load of
negligible resistance, the change in flux produced thereby is equal to
r
this area. In symbols,
Ilep = 0 e dt
The nominal response is defined for build-up, rather than build-down,

because the former action is required during and immediately after
d
I Area abc=
I area abde
1
I
Slope I,
-oa = nominal response
'0 I
Wo I
J9 I
g Nominal collector-ring voltage
..., a --------~n;------
'~ t
'? ,
~ I
g I
I
I
I
0.5 sec, ~
I
o Time
FIG. 18. Illustrating the definition of "nominal exciter response."
the presence of a fault in order to improve system stability. The

definition specifies that the exciter be unloaded, because change of
voltage caused by load is not great and because both test and calcu-
lation are simplified immensely by the assumption of no load.
Ceiling voltage is the value that the open-circuit voltage of the
exciter will attain ultimately if the regulating rheostat remains short-
circuited.
Typical values of exciter response and of other exciter quantities.

The nominal collector-ring voltage of the a-c. generator is usually between
72% and 95% of the rated voltage of the exciter (between 90 and 120
volts for a 125-volt exciter, or twice those values for a 250-volt exciter).
As an average value, 80% (100 volts) may be taken.
The ceiling voltage varies considerably, but typical values are from
120% to 135% of rated voltage (1.50 to 165 volts for a 125-volt exciter).
A considerable difference between ceiling voltage and nominal collector-
ring voltage is necessary to obtain rapid build-up of the exciter voltage.
The exciter voltage required to drive enough current in thefield winding
of the a-c. generator to generate rated voltage of thelatter on theair-gap line
is about 37% of the rated voltage of the exciter (45 volts for a 125-volt
exciter). This quantity was used as the unit of exciter voltage in the
preceding chapter.
The standard value of exciter response is about 1.0 unit (100 volts per
second for a 125-volt exciter). More rapid rates up to 2.0 units (200
volts per second) are obtainable at a small additional cost. Still
faster rates up to about 5.0 units (500 volts per second) can be provided.
"Superexcitation," with response as high as 6,000 to 7,000 volts per
second on a 250-volt field (30 to 35 units), has been attained by em-
ploying an exciter rated at 600 volts with a ceiling of about 1,000 volts.
This very high rate of response has been employed on synchronous
condensers.
Quick-response excitation systems are usually characterized by the

following features:
1. An exciter having a high value of nominal response and a
high ceiling voltage.
2. A reliable mechanical drive for the exciter.
3. A quick-acting voltage regulator.
These features will be discussed in turn.
1. Fast response of the exciter can be achieved (a) by subdividing
the field winding into several parallel paths with external series resist-
ance in each path in order to obtain a short time constant, and (b) by
using separate excitation (pilot exciter). High rotational speed is
helpful in achieving a short time constant. A high ceiling voltage
requires an exciter of larger size than usual.
2. The importance of reliable mechanical drive for the exciter can
be appreciated by considering the conditions that prevail during a
fault on the main a-c. circuit. The field current of the generator under-
goes an increase, induced by the sudden increase of lagging armature
current. At the same time, the exciter is called upon to increase its
voltage as rapidly as possible. Thus the exciter must deliver current
CALCULATION OF EXCITER RESPONSE 173
higher than normal at a voltage higher than normal, which means that
both electrical output and mechanical input are much higher than
normal. If the exciter is motor-driven, it should be noted that the
motor voltage may be subnormal during the fault, thereby reducing
the pullout torque of the motor. Direct drive of the exciter from the
main prime mover, of course, is very reliable. *
3. If an indirect-acting exciter-rheostatic voltage regulator with
high-speed field-forcing contacts is used, the time lag of the regulator
in the event of a fault on the a-c. system consists of the time required
for the voltage-sensitive element to close its contacts plus the time for
the high-speed relay or contactor to close its contacts.. This lag should
be as small as possible. In present regulators it is about 3 cycles (0.05
sec.). After the regulator has short-circuited the regulating resistance
in the field circuit of the exciter, there is nothing more that the regulator
can do to increase the exciter response.
To insure that the regulator will act correctly for all types of short
circuit, the voltage-sensitive element of the regulator should be con-
trolled by the voltage of all phases of the three-phase circuit. Methods
of accomplishing this have already been described.
If the exciter has a differential-field winding, the regulator should
open it when rapid build-up is required, for otherwise it decreases the
exciter response.
Calculation of exciter response - general remarks. The definition

of exciter response, the importance of quick response, and some of the
means of attaining it have been discussed. There remain to be treated
the quantitative calculation of exciter response and whatever conclu-
sions can be drawn from such calculations.
Exciter response is shown most completely by curves of exciter
armature voltage as a function of time. Such curves for an existing
exciter can be determined most easily and accurately by test. A two-
element oscillograph is used, one element to record the armature
voltage, the other to trace a timing wave. For an exciter not yet built,
or one which it is proposed to alter, the voltage-time curve must be
calculated. Thus it may be determined whether a proposed design
will meet requirements or whether an existing machine can be modified
to meet them. Such calculations will now be considered.
The voltage-time curves may be for either increasing or decreasing
*If a fault pereists, the speed of a motor-driven exciter will drop. 'fhel'efore,
in order to obtain equal actual response, a higher nominal exciter response is needed
for a motor-driven exciter than for a direct-driven exciter, the speed of which is
maintained.
voltage ("build-up" or "build-down"), but the former is more impor-

tant than the latter for power-system stability. If an exciter-rheostatic
voltage regulator is used, build-up is initiated by the sudden removal
of resistance from the field circuit of the exciter; build-down, by the
sudden insertion of resistance. The voltage-time curves may be calcu-
lated for a loaded or an unloaded exciter. Though the result for the
loaded exciter is what is actually wanted, both calculation and test are
much simpler for the unloaded than for the loaded exciter. Further-
more, the voltage-time curves of separately excited machines for the
two conditions differ by only a few per cent. For these reasons, the
build-up of voltage of unloaded exciters will be analyzed first, and that
of loaded exciters will 'be considered later.
Response of unloaded exciter. The circuit diagrams for separate

excitation and self-excitation of the main exciter are shown in Figs. 19
and 20, respectively. These diagrams also show some of the notation
A-c generator
Pilotexciter Main exciter
if
Contacts of regulator
FIG. 19. Circuit diagram of separately excited exciter.
that will be used. The characteristics of the two excitation systems

are considerably different from one another, particularly in their speed
of response, as will appear from the analysis.
The equation of voltage around the field circuit of the main exciter,
whether separately or self-excited, is:
N ~f + Ri = E or e [1]
where N = number of field turns in series.

cPf = flux in each field core, in webers.
R = field-circuit resistance, in ohms.
i = field current, in amperes.
E = pilot-exciter armature voltage.
e = main-exciter armature e.m.f., in volts.
RESPONSE OF lTNLO.ADED EXCITER 175
The equation applies to separate excitation if E is used, and to self-

excitation if e is used.
If the pilot exciter is flat-compounded, its voltage E may be taken
as constant. If the pilot excite! is undercompounded or shunt-wound
and is conservatively applied, its armature voltage may be assumed to
decrease linearly with load. This assumption is handled conveniently
by taking E as a constant equal to the no-load.voltage of the pilot
exciter and increasing R by the proper amount to account for the linear
decrease of voltage.
A-c generator
cPt
Contacts of regulator
FIG. 20. Circuit diagram of self-excited exciter.
If the field winding is arranged in several parallel paths, Rand i may

be taken either per path or for all paths in parallel, the product Ri
being the same in either case.
Equation 1 takes no cognizanceof the effect of eddy currents induced
in solid parts of the magnetic circuit, Eddy currents oppose, and hence
retard, any change of flux. Their effect will be discussed again in the
section on "Calculation of nominal exciter response."
For our present purpose it is advisable to replace cPl bye, the arma-
ture voltage, because the magnetization curve of the exciter gives e
versus i, rather than cPl versus i. If the exciter is running at constant
speed, which may be safely assumed, then the armature voltage e is
proportional to the flux cPa which crosses the air gap and links the
armature conductors; that is,
e = kcPa [2]
where
k == Znp [3]
60a
where Z = total number of conductors on armature.
n = speed of rotation of exciter (r.p.m.).
p = number of poles.
a = number of parallel paths through the armature.
a is equal to 2 for a simplexwave winding, and is equal to p for a simplex

lap winding; for a multiplex winding the foregoingvalues are multiplied
by the degree of multiplicity, that is, hy 2 for a duplex winding, by 3
for a triplex winding, and so on.
Now cPj, the flux linking the field winding, exceeds cPa, the flux linking
the armature winding, hy (PI, the leakage flux. The leakage flux passes
from pole to pole without entering the armature iron, a large part of it
passing between pole tips as shown in Fig. 21. It is feasible to treat
the leakage flux according to either of two different assumptions:
first. that it is proportional to the armature flux; or, second, that it is
FIG. 21. Armature or air-gap flux "'a, leakage flux "'I, and field flux "'I = "'a + </>,.
proportional to the field current. These two assumptions would be
equivalent to one another if the armature flux were proportional to
the field current, that is, if there were no saturation. Saturation is
present, however, as shown by the curvature of the magnetization
curve; and the two different assumptions stated above differ as to the
assumed effect of saturation on the leakage flux. The first assumption,
namely, that the leakage flux is proportional to the armature flux,
assumes that the leakage flux is affected by saturation to the same
degree as is the armature flux. This probably overemphasizes the
effect of saturation on the leakage flux. The leakage fluxis less affected
by saturation than is the armature flux because the former passes over
a longer air gap than the latter does and because the highest saturation
is usually found in the teeth. The second assumption, namely, that
the leakage flux is proportional to the field current, assumes that the
leakage flux is not affected by saturation at all. It is equivalent to
assuming a constant leakage inductance. This assumption probably
underestimates the effect of saturation on the leakage flux. No doubt
the truth lies somewhere between the solutions obtained from these
two assumptions. Because both assumptions lead to results which
RESPONSE OF UNLOADED EXCITER 177
differ by a negligible amount, it does not matter greatly which assump-

tion is used.
The two assumptions may be formulated algebraically as follows:
According to the first assumption,
cPz = ClcPa [4]
and since
[5]
it follows that
[6]
where C1 is a constant known as the coefficient of dispersion. The value
of a must be greater than 1, and usually is in the range from 1.1 to 1.2.
In other words, the leakage flux is from 10% to 20% of the useful or
armature flux. According to the second assumption,
cPz = C2i [7]
and
cPJ = cPa + C2~. = rPa + LI.

N~ [8]
where LI is the leakage inductance.

If the ratio (u - 1) of leakage flux rPz to useful flux cPa is known for
some point on the air-gap line of the magnetization curve having co-
ordinates ea g and i a g , the leakage inductance can be computed from
the relation:
L, = (cPz) ~aa ~ [8a]
cPa ~ag k
Using eqs. 2 and 6, the differential equation (1) becomes

N« de .
k °dt + R~ = E or e [9]
for the assumption of constant coefficient of dispersion. Using eqs. 2

and 8, it becomes
N de di .
- -+
k dt
Lz -
dt
+
R~ = E or e [10]
for the assumption of constant leakage inductance.

In eqs. 9 and 10, e is a nonlinear function of i, as given by the magnet-
ization curve; hence eqs. 9 and 10 are differential equations with
variable coefficients. For solving such equations at least four different
methods are available, namely:
1. Formal integration.
2. Graphical integration.
3. Point-by-point calculation.
4. Use of differential analyzer or of other numerical computers.
These methods will now be discussed briefly.
A 8olution by formal integration! requires that the variable coefficients
can be expressed in terms of one of the variables and that the resulting
expression can be integrated by formal methods. In the case of exciter
transients, this means that the magnetization curve must be repre-
sented by an empirical equation, such as the Frohlich equation,
ai
e=b+i [11]
or the modified Frohlich equation,
e = -ai- + c~· (121

b+i
in which a, b, and c are constants chosen to make the equation best fit
the test data. If eq. 11 or 12 is substituted into eq. 9 or 10, the equa-
tions may be integrated by separation of variables i and t, and an
explicit solution for t may be obtained.
In the method of graphical integration, the equation is rearranged so
as to express time t as an integral. The integrand is determined by a
graphical construction involving the use of the magnetization curve.
It is then plotted against the independent variable, and the integral is
evaluated by measuring the area under the ·curve. The particular
methods presented in this chapter are those of R. Rudenberg.l-"
In the point-by-point calculations, one or more of the time derivatives
are usually assumed to remain constant throughout a short interval of
time, their values during the interval being determined from conditions
at the beginning, or preferably at the middle, of the interval. The value
of a dependent variable at the end of an interval is taken as its value
at the beginning of the interval plus its time derivative during the
interval multiplied by the length of the interval. A variation of the
point-by-point method consists in assuming an increment in one of
the dependent variables and computing the corresponding increment of
time. The point-by-point method is widely applicable to the solution
of all kinds of transient problems - in fact, to the solution of all kinds
of differential equations. The reader has already encountered applica-
tions of it to the solution of the swing equation of a machine and to the
calculation of field transients in the main generator.
A differential analyzer is a machine for integrating and for solving
differential equations.P Other types of numerical computers which
SEPARATELY EXCITED EXCITER 179
have been developed recently, such as the relay and electronic types,
can be used for the same purpose. Obviously the application of the
machine-integration method is limited by the fact that in most cases
no suitable computing machine is available.
Of the four methods listed for solving the differential equations of
the exciter, the most useful and convenient are the solution by graphical
integration and the point-by-point solution. These methods will now
be described in detail, as applied to the solution of eqs. 9 and 10.
Voltage-time curves of unloaded separately excited exciter by the

method of graphical integration. First, a constant coefficient of dis-
persion will be assumed, making eq. 9 applicable:
--+
qN de R'
k dt
1,=
E [13]
qN de = E _ Ri [14]
k dt
Build-up. Let subscripts 1 and 2 be used to denote the initial and

final conditions, respectively. Thus, the armature voltage and field
current build up from a point on the magnetization curve having co-
ordinates it and el to a point having coordinates i 2 and e2. Now
multiplyeq. 14 by e2/E.
qNe2 de Re2.
---=e2--1, [15]
kE dt E
or
de
T- = ~e [16]
dt
where
T = uNe2 [17]
kE
and
.
se = e2 - lle2 e2 .
- 1, = e2 - -:- 1, [18]
E ?,~
T is a constant having the dimensions of time and, therefore, plausibly

may be called a "time constant." It should be remembered, however,
that, as the build-up of neither e nor i is exponential, T cannot be given
the same interpretation as the time constant of a linear circuit. T is
equal to the field flux linkage divided by the field Ri drop, both being
180 EXCITATION SYSTEl\IS
taken for the final condition. For this condition the flux linkage is
e2
N cP12 = N (fcPa2 = N k (T [19]
and the Ri drop is Ri 2 = E. The final field flux linkage may be written
also as Li2 , where IJ is the nominal exciter field inductance, equal to
the field flux linkage per unit field current at the final condition. The
time constant then takes the familiar-appearing form
T = Li2 = ~ [2 ]
R~ R 0
fie (eq. 18) may be given the following geometric interpretation (see
Fig. 22a): On the same coordinates with the magnetization curve
e e
....-----~e
1
(a) (b) (c)
FIG.22. Build-up of voltage of separately excited exciter, obtained by the

method of graphical integration. Constant coefficient of dispersion is assumed.
draw a straight line through the origin and the final operating point
i 2, e2. The slope of this line is e2/i2 = Re2/E, and for any given
abscissa i the ordinate of the line is
[21]
Ae is then the vertical distance from this line up to the horizontal line
drawn through the intersection of this line and the magnetization
curve. Byeq. 16, the rate of change of armature voltage e is propor-
tional to fie.
t:
Solving eq. 16 for t, we obtain
t = T - [22]
el lie
SEPARATELY EXCITED EXCITER 181
which gives the time required for the voltage to build up from its
initial value el to any value e. The integral may be evaluated graphi-
cally by plotting 1/ jj.e as a function of e and measuring the area between
the curve and the e axis between the lower limit el and a running upper
limit e.
The whole process of graphical solution is illustrated in Fig. 22. In
Fig.. 22a the magnetization curve is drawn, and on the same axes there
is drawn a sloping straight line 02 through the origin and the final
operating point 2 on the magnetization curve, and also a horizontal.
e
.i,
ae-----'
(a) (6) (c)
FIG.23. Build-down of voltage of separately excited exciter, obtained by the
method of graphical integration. Constant coefficient of dispersion is assumed.
straight line A2 through the final operating point. The value of ~e

corresponding to any given value of e is then found as the distance
FDB from the sloping line 02 to the horizontal line A2 along a vertical
line through point D which lies on the magnetization curve at ordinate
e = OC. Corresponding values of e and Lle are tabulated. The recip-
rocal of L\e is calculated and plotted versus e, as shown in Fig. 22b.
The shaded area is measured for various values of c. This area, multi-
plied by T, gives the value of t corresponding to e, and e is plotted
against t in Fig. 22c.
In order to allow for the retarding effect of eddy currents in the iron
upon build-up, T may be increased by an empirical amount.
Build-down. Build-down is handled in exactly the same manner as
build-up. The process is illustrated in Fig. 23. The sloping line 02
is again drawn through the final operating point, but is steeper than
before. ~e is now a negative quantity. The time constant T is smaller
than it was for build-up, because e2 is smaller (see eq. 17).
Constant leakage inductance. If constant leakage inductance is
assumed instead of constant coefficient of dispersion, the analysis is
slightly different. Equation 10, which holds for this case, may be
written thus:
- -d ( e +-~
N Lzk -) = E - R'~ [23]
k dt N
N !! (e + e') = E - Ri [24]
k dt
where
, Lzk.
e =-t [25]
N
e+e'
e
e2
----sf
el
& ...&-1'-------- Ee1

tan- 1 ~k
(b) (c)
FIG.24. Build-up of voltage of separately excited exciter, obtained by the

method of graphical integration with the assumption of constant leakage inductance.
Multiply eq. 24 by e2/E, obtaining

Ne2 d
- - - (e
kE dt
+ e, ) = Re2 .
e2 - - 1,
E
[26]
T itd (e + e ) = de
I
(27)
t = TIe el
d(e + e')
ae
[28]
where
T = N e2 [29]
kE
[30]
EXAMPLE 1 183
T is now defined differently than it was for the analysis based on

constant coefficient of dispersion, but lie is the same.
Equation 28 is solved graphically by the process shown in Fig. 24.
As shown in Fig. 24a, e' is represented by the distance between the
i axis and the straight line having a slope - Llk/N. In Fig. 24b,
1/ de is plotted against e + e'. In all other respects the process is the
same as that used for constant coefficient of dispersion.
~ 200 t - - - - l - - t - - - - - I i - - - - + - - - - - - I
(5
~
-...:
E
Q) 150 t---t--T--~-+---4---~
e
.anJ
~
~ 100 t - - t - - - - 1 ' 1 - - - - J - - - - J - - - - - 1
50 t - I ' - - - r - t - - - + - - - - I - - - - - - 1
5 10 15 20
i, Field current (amperes)
:FIG. 25. Magnetization curve of exciter (Examples 1 and 2).
EXAMPLE 1
The following data pertain to an exciter:
Rated voltage 250 volts
Rated output 100 kw.
Rated speed (n) 1,800 r.p.m.
Number of poles (p) a
Type of armature winding simplexlap
Number of armature slots 108
Number of conductors per slot 4
Coefficient of dispersion (0-) 1.15
Shunt wound
Number of field turns per pole 1,500
Number of field circuits 1
Field resistance 14.6 ohms
The magnetization curve is given in Fig. 25.
The exciter is operating without load and is separately excited from a
250-volt source. Initially the field circuit is open. It is then closed through
such resistance as to give an ultimate voltage of 275. Plot armature voltage
and field current as functions of time. Increase the time constant by 5%
to allow for the effect of eddy currents in the yoke.
TABLE 1
CALCULATION OF BUILD-UP OF VOLTAGE OF
SEPARATELY EXCITED EXCITER (EXAMPLE 1)
e2 •
e -:-~ Ae
1,2 Ae
(volts) (volts)
(volts) (kv.- 1)
7 0 275 3.64
50 20 255 3.92
100 38 237 4.22
150 59 216 4.63
200 87 188 5.31
230 119 156 6.41
250 159 116 8.62
260 187 88 11.35
270 240 35 28.6
273 260 15 67
275 275 0 co
Solution. The time constant is given by eq. 17, in which k is given by eq. 3.
Here
Z = total number of conductors on armature = 108 slots X 4 conductors
per slot = 432.
a = number of parallel paths in armature, which for simplex lap winding is
equal to the number of poles. Hence a = 6.
k = Znp = 432 X 1,800 X 6 = 12 960
60a 60X6 '
The time constant, without correction for eddy currents, is (byeq. 17)
T' = qNe2 = 1.15 X (1,500 X 6) X 275 = 0.879 sec.
kE 12,960 X 250
Increase this value by 5% to allow for the effect of eddy currents.
T = 1.05 X 0.879 = 0.922 sec.
EXAMPLE 185
TABLE 2
CALCULATION OF BUILD-UP OF VOLTAGE OF
SEPARATELY EXCITED EXCITER (EXAMPLE 1), CONT'D.
Area
e
(volts) fde Lle
t
(sec.)
i
(amp.)
7 0 0 0
20 0.05 0.05 0.4
50 0.16 0.15 1.0
80 0.28 0.26 1.5
100 0.36 0.34 1.9
130 0.49 0.45 2.5
150 0.58 0.54 3.0
180 0.73 0.67 3.6
200 0.83 0.77 4.3
220 0.94 0.87 5.2
230 1.00 0.92 5.8
240 1.07 0.99 6.5
250 1.15 1.06 7.7
260 1.24 1.15 9.0
270 1.41 1.30 11.7
273 1.52 1.41 12.5
275 00 00 13.2
In Fig. 25 the final operating point is indicated by a small circle on the

magnetization curve at the specified ceiling voltage of 275. The correspond-
ing field current is 13.2 amp. A straight line is drawn through this point and
the origin. The field-circuit resistance is 250/13.2 = 18.9 ohms, of which
18.9 - 14.6 = 4.3 ohms is added externally.
Corresponding to arbitrarily chosen values of e ranging from 0 to 275 volts,
the values of Lle are read from Fig. 25 by the construction indicated in
Fig. 22a. These values of ~e are entered in Table 1, and the values of 1/Lle
are computed and tabulated there. In Fig. 26, 1/Lle is plotted versus e and
the graphical determination of
j
e
de
81 ~e
is carried out by counting squares. The areas so determined are entered in

Table 2. These values, multiplied by T = 0.922 sec., give the value of t
corresponding to e. The corresponding value of i is read from the magnetiza-
tion curve (Fig. 25) and entered in Table 2. In Fig. 27, e and i are plotted
as functions of t. Note that the curve of e versus t is almost straight until the
ceiling voltage is nearly reached.
300r---,----,----,---.,..-----,
250
200
1 small square = 10 x 0.001 = 0.01

~
~ 150
'"
100
50
0.010 0.015 0.020 0.025

to (volts-I)
FIG. 26. Graphical integration (Example 1).
300 15
250
200 10
'iii'
~ e
Ql
~ 150 c.
E
'" .e
.~
100 5
0.5 1.0 1.5

t (seconds)
FIG. 27. Build-up of armature volt age and field current of separately excited
exciter (Example 1).
186
SELF-EXCITED EXCITER 187
Voltage-time curves of unleaded self-excited exciter by the method

of graphical integration. Assuming a constant coefficient of dispersion,
eq. 9 holds:
uN de + Ri = e [31]
k dt
uN de
- - = e - R1,. [32]
k dt
de
T - = ile [33]
dt
t= T . - [deile
61
[34]
where
T= uN [35]
k
ile = e - Ri [36]
- - - - - - l : 1 .J...
e
(a) (b) (c)
FIG. 28. Build-up of voltage of self-excited exciter, obtained by the method of

graphical integration. Constant coefficient of dispersion is assumed.
Time constant T for the self-excited machine is different from those

previously used for the separately excited machine (eqs. 17 and 29) in
that it is independent of the final value of armature voltage. Hence
the same value of T applies both to build-up and to build-down.
The graphical interpretation of t:J.e is shown in Fig. 28a. It is the
vertical distance from point F on the "field-resistance line" 02 (drawn
from the .origin to the final operating point) to point D on the mag-
netization curve. The abscissa of both points represents the field
current i, while the ordinate of point D represents the armature volt-

age e for the particular operating condition.
Figure 28 shows the whole process of finding the build-up of voltage
graphically. Except for the values of T and Lie, this process is like that
used for the separately excited machine. Note that Lie is small at first,
increases to a maximum, and then decreases to zero. This is reflected
in the slope of the voltage-time curve (Fig. 28c), which has a different
shape from that for separate excitation (Fig. 22c).
e e
I
I
----~---------~~~
JD I I
I I 1
I I I
I t I
I. I I
_1I
-T----r------------
IB
1 I
I
I
-
J I I
I I I
I
: I i. -.. I
!:J.e
(a) (b) (c)
FIG. 29. Build-down of voltage of self-excited exciter, obtained by the method

of graphical integration. Constant coefficient of dispersion is assumed.
Figure 29 illustrates the procedure for finding the build-down curve.

~e is again represented by the distance FD from the field-resistance
line to the magnetization curve, but is now negative.
Constant leakage inductance. With this assumption, eq. 24 holds if
E is replaced bye.
!ik !!.dt (e + e') = e - Ri [37]
T dtd (e + e ) I
= ~e [38]
t =T Ie
61
dee + e')
Lie
[39]
where
N
T=- [40]
k
~e = e - Ri [41]
EXAMPLE ~ 189
The graphical solution is performed with e' represented as in Fig. 24,

but /1e determined as in Figs. 28 and 29.
EXAMPI,E 2
The exciter described in Example I is operated at no load with self-excita-
tion. Plot curves showing armature voltage and field current versus time.
The initial and final values are to be the same as in Example 1.
250
200
1 small square =10 X 0.005 =0.05
~
..
~ 150
100
50
QI0 Q14
FIG. 30. Graphical integration (Example 2).
Solution. For self-excitation the time constant is given by eq. 35. As in

Example 1, this value will be increased by 5% to allow for the retarding
effect of eddy currents,
T = 1.05 uN = 1.05 X 1.15 X 9,000 = 0.840 sec. (a)

k 12,960
Values of lie are read from Fig. 25 according to the construction shown in
Fig. 28a and are listed in Table 3. In Fig. 30 1/lie is plotted against e and
the areas are measured by counting squares and are entered in Table 3.
These areas, multiplied by T = 0.840 sec., give the value of t corresponding
to e. Values of i corresponding to e are read from the magnetization curve
190 EXCITATION S'YSTEMS
(Fig. 25). In Fig. 31 e and i are plotted against t. The values of e and i for
separate excitation are plotted in the same figure for comparison. It is
evident that separate excitation gives much faster build-up of e and i than
self-excitation does, particularly at the low values of e.
TABLE 3
CALCULATION OF BUILD-UP OF ARMATURE VOLTAGE AND FIELD CURRENT
OF SELF-ExCITED EXCITER (EXAMPLE 2)
1
e Ri ~e Area t i
Ae
(volts) (volts) (volts) (sec. ) (amp.)
(volt")
7 0 7 0.143 0 0 0
10 2 8 0.125 0.36 0.30
20 9 11 0.091 1.39 1.17 0.4
30 13 17 0.059 2.12 1.78
40 18 22 0.045 2.62 2.20
50 20 30 0.033 3.00 2.52 1.0
70 26 44 0.023 3.56 2.99
100 38 62 0.016 4.12 3.46 1.9
150 59 91 0.011 4.78 4.01 3.0
200 87 113 0.009 5.25 4.40 4.3
230 119 111 0.009 5.52 4.64 5.8
250 159 91 0.011 5.72 4.80 7.7
260 187 73 0.014 5.85 4.91 9.0
270 240 30 0.033 6.05 5.08 11.7
273 260 13 0.077
275 275 0 00 13.2
Comparison of response of separately and self-excited machines.

Tire method of graphical integration which has just been described
affords an excellent means of comparing the exciter response for the
two different kinds of excitation. If, in making the comparison, we
assume constant coefficient of dispersion, the response is given, for
either kind of excitation, according to eqs. 16 and 33, by
[42]
In this expression both de and T are defined differently for separate

excitation than for self-excitation.
The time constant is defined as
T = uN for self-excitation (eq. 35)

k
COMPARISON OF RESPONSE 191
and as
T = oN. e2 for separate excitation (eq. 17)
k E
The former is fixed by the design of the exciter, whereas the latter
can be controlled by the external circuit. It is possible to raise the
pilot-exciter voltage E without changing the ceiling voltage (which,
for build-up, is e2), provided the field-circuit resistance R is increased
300_--.....---...,..------y----..-------r--, 15
200 10
-g 150
en
.-..
en
~
Q)
...... E
~
....
~
~
100 5
50
0
2 3 4 5
t (seconds)
FIG. 31. Build-up of armature voltage e. and field current i of'separately and
self-excited exciters (Examples 1 and 2).
in proportion to E. This can be done by the addition of external series

resistance. By this means the resistance of the field circuit is increased
without changing its inductance; hence the time constant LIR is
decreased. During build-down the difference between the two time
constants is still more pronounced, as then, for separate excitation,
e21E is very small.
It is apparent that separate excitation has an advantage over self-
excitation in the possibility of decreasing the time constant by external
means. The chief difference between the two kinds of excitation,
however, is not in the factor T, but in de of eq. 42. The latter quantity
is defined as
for separate excitation (eq. 18)
and as
for self-excitation (eq. 36)
During build-up e2 is always greater than e, since it is the final value

of e; and during build-down e2 is always less than e. In either case the
absolute value of ~e is greater for separate excitation than for self-
excitation, if the same magnetization curve and same final operating
point are used. The difference in the values of ~e for the two kinds of
excitation is shown more vividly by the graphs than by the equations;
see Figs. 22 and 28 for build-up, Figs. 23 and 29 for build-down.
Line FB, which represents ~e for separate excitation, is greater than
FD, which represents ~e for self-excitation. The difference is most
pronounced at the beginning of the transient.
It is now obvious that, even for equal time constants, separate
excitation gives much faster exciter response than self-excitation does.
Voltage-time curves of unloaded exciter by point-by-point calcula-

tion. 13 Constant leakage inductance will be assumed. The differential
equation for this assumption is eq. 10. In order to put this equation
into suitable form for point-by-point solution, replace the time de-
rivatives, de/dt and di/dl, by ratios of finite increments, ~e/~t and
~i/~t, respectively, and solve the equation for ~t. The result is as
follows:
~t = _(N_/_K_)_~e_+_L_,A_i
[43]
E (or e) - Ri
It is used in the following way. The ordinates of the magnetization
curve are divided into increments ~e, which may be taken as 10 volts
each for a 250-volt exciter. * For each value of e, the corresponding
value of i is read from the magnetization curve and tabulated. The
difference betw~en successive values gives ~i. For i in the denominator,
and also for e in the case of self-excitation, the average values for the
interval (half the sum of the two end values) are used. With all
quantities in the right-hand member of eq. 43 known, ~t is calculated.
It denotes the time required for the armature voltage to change by an
*~e now signifies the increment of e during a time interval ~t. It should not be
confused with the ~e used in the method of graphical integration.
NOMINAL EXCITER RESPONSE 193
amount Se. The curve of voltage versus time can be plotted from the
increments of voltage and time, starting at zero time with the known
initial value of voltage. The first and last intervals may have different
values of /:ie from the others; there is no necessity for using equal values
throughout.
The advantage of assuming ile and calculating ill, instead of the
reverse, is that by so doing the average values of e and i for the interval
are easily determined. The use of average values, instead of values
at the beginning of an interval, greatly reduces the cumulative error.
If constant coefficient of dispersion is assumed instead of constant
leakage inductance, eq. 9 yields
~t = _(_(1N~/K--.;.)_~_e
[44]
E (ore) - Ri
which is solved in the manner just described except that ili need
not be found.
Calculation of nominal exciter response. The methods of obtaining

a voltage-time curve of an unloaded exciter by test and by two kinds of
calculation have been described. After such a curve has been obtained
for voltage build-up starting at the nominal collector-ring voltage of
the main machine, the nominal exciter response is easily calculated by
reference to the definition. First it is necessary to obtain the area
under the curve during the first half-second. This area may be meas-
ured by a planimeter or found by counting squares on the graph paper.
If the curve has been calculated by the point-by-point method, the
area under it is readily calculated by summing the products of each at
by the average value of voltage e for the interval.
It is necessary next to find the slope of that straight line under which
the area during 0.5 sec. is equal to the area under the voltage-time
curve during the same time. In Fig. 18 this line is ac. Let its slope
be called m.
be be
m=-=-=2bc [45]
ab 0.5
since ab = 0.5 sec. If the areas under the line and under the curve are
equal, they will still be equal.if the area of the rectangle below nominal
collector-ring voltage is subtracted from each. Let the remaining
area abdea under the curve and above this voltage be denoted by A.
Line ac must be determined so that the area of triangle abc is equal to
A. The area of the triangle is half the product of base and altitude.
194 EXCITATION S'YSTEl\1S
Hence
A = !(ab) (be) [46]
Since ab = 0.5 sec.,
1 be
A :::II 2(0.5) (be) = --- [47]
4
be = 4A [48]
When this value of be is substituted into eq. 45, the slope is given in
terms of the area as
m = 8A [49]
It is not necessary actually to draw line ac, because its slope can be
found by measuring area abdea under the exciter build-up curve, and
multiplying it by 8.
Finally, the nominal exciter response is the slope thus found divided
by nominal collector-ring voltage. In Fig. 18,
. 1 · 8 X area abdea in volt-seconds
Nomina exciter response = I
engt
h O· I
a In vo ts
[50]
1.00.-:::---r----r-----,r-------,r----~
...
.s
u
J2
c
:8 0.95t---t----I---~-t--------1I--~
~
8
oS
EO.90t---t---I------1I---~~---I
.~
C.
C.
<
234 5
Calculated nominal response
FIG. 32. Correction factors for the effect of eddy currents on nominal exciter
response. (W. A. Lewis, Elec. Jour. 13)
Correction for the screening effect of eddy currents. I 3 If the voltage-

time curve was found by calculation in which the effect of eddy cur-
rents was neglected, the calculated nominal response will be too high.
It should be corrected for the effect of eddy currents by multiplying
the uncorrected value of nominal response by a correction factor read
from Fig. 32. The correction factors plotted there were found by
EXAMPLE 3 195
comparison of calculated values with values found by test on exciters

with rolled steelframes and laminated pole pieces. The effect of eddy
currents on the response of such exciters is seen to be small except
at high values of response. If the no-load saturation curve is obtained
by test, the accuracy of the calculated nominal response after appli-
cation of the correction factor can be expected to be within ±5%
of the value found by test. Note that the correction factors apply
only to the over-all effect of eddy currents on nominal response, not
to individual ordinates of the voltage-time curve. The shape of this
curve as found by test may differ greatly from the corresponding
calculated curve.
EXAMPLE 3
Calculate the nominal response of the exciter described as follows:
Rated output 167 kw.

Rated voltage 250
Rated speed 1,200 r.p.m,
Number of poles 6
Number of field turns per pole 1,450
Number of parallel paths of field 3
Field resistance per path at 75 C. 27 ohms
k = e/cPa 10,080
Leakage flux/useful flux on the air-gap line = (1 - 1 0.15
No-load saturation curve see Fig. 33
when used with an a-c. generator having a
Nominal collector-ring voltage 190
The following data pertain to external apparatus in the field circuit of the
exciter:
External resistance (for all paths) 11.6 ohms
Separate excitation
Voltage of pilot exciter 250
NOTE: This example is taken from an article by W. A. Lewis in the Electric

J ournazt 3 by permission.
Solution. In contrast with Examples 1 and 2, this example will be com-
puted by the point-by-point method and under the assumption of constant
leakage inductance.
To find the leakage inductance L 1 use eq. Sa and the fact that the leakage
flux is 15% of the useful flux at any point on the air-gap line (straight part
of the saturation curve). Take the point (marked in Fig. 33) ea g = 100 volts,
i aa = 0.82 amp. per path. The number of turns in series irs

1,450 turns per pole X 6 poles 2 900 t
N = =, urns per path (a)
3 parallel paths
~ = 2,900 = 0.288 sec. (b)

k 10,800
I I
Ceiling voltag~ Ie- ~
320
/"
280 /
1/ Q)
J ::s r - -
t '~
;S I - -
VI
1~1~ 6.1 } Sixth .2 r - -

IT Mean I interval c
~ f-
/I Nominal coUector- ~ :s
u
ring voltage t----' ':2 r - -
I ~
~ r-
/ ~ I---
~
Q)
lig
of! I---
~! --
VI
Eagl
flag
80 Q)
V II~
J ~~
I---
40
/ Ii
oo~ 1 234 5
1# field current per circuit (amperes)
FIG. 33. No-load saturation curve of exciter (Example 3). (Reprinted from
The leakage inductance is
L z = ( -cPz) -:- - = 0.15 -100 X 0.288 = 5.27 henrys per path

eag N (c)
cPa ''tag k 0,82
The field-circuit resistance per path is

R = 27 + (3 X 11.6) = 27 + 34.8 = 61.8 ohms (d)
Point-by-point computation is performed as indicated by eq, 43, which

becomes:
~t = (N/k)de + Lzdi = 0.288 de 5.27 di (e) +
E - Ri 250 - 61.8i
EXAMPLE 3 197
The response curve is computed in Table 4, columns 1 through 11. Figure

33 shows how t1i and the mean i are found for the sixth interval (e, 240 to
250 volts). The computed response curve is drawn in Fig. 34.
The area under the response curve and above 190 volts for 0.5 sec. is
computed in columns 12 and 13 of Table 4. Since the last interval (310 to
360
c
/V
320
~:..~~ ~
280
Calculated ~
response curve:1-- e ~~btained-· ~
by computation ~
columns 12, 13
,",~'I
~V' ~t ; ~
~~ ~~
~
a b
I-- Nominal response = O.Sbtoa) =1.61 -
80
40
o0 g
o 0.1 0.2 0.3 0.4 0.5
t, Time (seconds)
FIG. 34. Computed response curve of exciter (Example 3). (Reprinted [rom
320 volts) runs from t = 0.488 to 0.666 sec., it is necessary to divide it into
two parts at t = 0.500 sec. From the response curve (Fig. 34), the corre-
sponding value of e is 311 volts. The area is A = 38.3 volt-sec.
The slope of straight line ac (Fig. 34) under which the area is equal to that
under the response curve is, by eq. 49,
m = SA = 8 X 38.3 = 306 volts per second (f)
The nominal exciter response is
306
- = 1.61 units
· (g)
190
Application of the correction factor 0.97 read from Fig. 32 gives:

Nominal response = 1.61 X 0.97 = 1.57 units (h)
~
(;/:)
00
TABLE 4
CoMPUTATION OF EXCITER RESPONSE. EXCITER: 167 Kw., 250 VOLTS, 1,200 R.P.M., SIX POLES (EXAMPLE 3)
Calculation of Response Curve Calculation of Area

1 4 5 6 7 8 9 10 11 12 13
- - -2- -3- LzAi Numerator RX I I1A
E-Ri At t ~1:ean
e i Ai 5.27 X (N /k)Ae Col. 4 + Meani (Mean i) Col. lOX
0.288 X 10 250-Col. 8 Col. 6 + Col. 9 2: Col. 10
001.3 Col. 5 61.8XCol. 7 Ordinate Col. 12
tx:1
~
,~
-190- -1.57- - -
200 1.66 0.09 0.47 2.88 3.35 1.62 100 150 0.022 I
0.022 5 0.1 :3
210 1.76 0.10 0.53 2.88 3.41 1.71 106 144 0.024 0.046 15 0.4 >
"'"3
.......
220 1.87 0.11 0.58 2.88 3.46 1.82 113 137 0.025 0.071 25 0.6 c
230 1.97 0.10 0.53 2.88 3.41 1.92 119 131 0.026 I 0.097 35 0.9 Z
240 2.08 0.11 0.58 2.88 3.46 2.03 125 125 0.028 0.125 45 1.3 :n.
250 2.21 0.13 0.69 2.88 3.57 2.15 133 117 0.031 I 0.156 55 1.7 ~
Jl
260 2.36 0.15 0.79 2.88 3.67 2.29 142 108 0.034 0.190 65 2.2 -3
270 2.52 0.16 0.84 2.88 3.72 2.44 151 99 0.038 l 0.228 75 2.9 t?j
280 2.70 0.18 0.95 2.88 3.83 2.61 161 89 0.043 0.271 85 3.7 ~
-.n
290 2.91 0.21 1.11 2.88 3.99 2.81 174 76 0.053 I 0.324 95 5.0
300 3.16 0.25 1.32 2.88 4.20 3.04 188 62 0.068 0.392 105 7.1
310 3.45 0.29 1.53 2.88 4.41 3.31 204 46 0.096 I 0.488 115 11.0
320 3.78 0.33 1.74 2.88 4.62 3.62 224 26 0.178 I 0.666 ~25
~
311 0.012 I
\
Final interval (310 to 320 volts) divided into two parts I 0.500 120 1.4
320 0.166 f
} {
I 38.3
f Total area
ROTATING AMPLIFIER 199
Voltage-time curve of unloaded exciter having rotating-amplifier

e.m.f, in series with its shunt-field circuit. The circuit diagram for
this type of excitation is shown in Fig. 35. The equation of voltages
around the closed path is
N
d~J
dt + Ri = c +E [ ]
51
Ej-
-----,
Armature of
mainexciter
~
Armature
of rotating
amplifier + e Field windingof
+ ; a-c generator
I
r - - -... ------...J
Main-exciter
field winding
FIG. 35. Circuit diagram of rotating-amplifier pilot exciter in series with shunt-
field circuit of main exciter.
E, the armature e.m.f. of the rotating amplifier, is positive when

boosting e, the armature e.m.f. of the main exciter. If constant
coefficient of dispersion is assumed, this equation becomes:
uN . de = e ~ (Ri - E) [52]
k dt
or, abbreviating as in eq. 33,
de
T - == Ile [53)
dt
where
T = uN [54]
k
as before (eq. 35), but
/le = e - (Ri - E) [55]
which differs from the previous expressions for Se. The graphical
representation of the new Ile is given in Fig. 36a. Here the curve of
e versus i is the magnetization curve of the main exciter, as before.
The curves of Ri - E versus i are straight lines parallel to the field
resistance line Ri but displaced vertically a distance E representing
the e.m.f. of the rotating-amplifier pilot exciter. This displacement
is downward if the pilot exciter is boosting; upward, if bucking.
200 ExcrrATION SY~rr:EIVlS
8e is the vertical distance between the magnetization curve and the

straight line Ri - E.
Assume that the field rheostat is adjusted so that, with no e.m.f.
in the pilot exciter, the e.m.f. of the main exciter is eo. Then the
straight line through tho origin and intersecting the magnetization
Ri-E
"
/ Buck Neutral Boost "
E<O E=O E>O
e e e
--- -, -::;-::..---
1
&
(b) (c)
...
g
~
1
FIG. 36. Build-up of voltage of exciter having armature of pilot exciter in series
with the shunt-field circuit of the main exciter (Fig. 35). Method of graphical
integration, with coefficient of dispersion assumed constant.
curve at ordinate eo is the field resistance line. Assume, further, that

the initial main-exciter voltage is el, the maintenance of which re-
quires a small bucking voltage from the pilot exciter. Suppose that
some disturbance, such as a short-circuit, lowers the voltage of the a-c.
generator so much that the pilot-exciter voltage goes to full boost in
order to build up the voltage of the main exciter as rapidly as possible,
and remains at full boost until ceiling voltage e2 is reached. According
to this assumption, E remains constant during the build-up process
and, hence, the line Ri - E is fixed. The process of graphical integra-
tion is conducted as already described except for the manner of
ROTATING AMPLIFIER 201
determining de. That is, 1/ de is plotted against e, as shown in Fig.

36b, and the area multiplied by T represents the time of build-up.
In the case shown, the field resistance line is almost parallel to the
straight part of the magnetization curve. As a result, ~e is almost
constant, and the curve of e versus t is almost linear until saturation
is reached.
i
FIG. 37. Comparison of Lle for build-up caused by insertion of series voltage in
shunt-field circuit of main exciter with Lle for build-up caused by reduction of the
resistance of that circuit.
Figure 37 compares de for the excitation system being considered

with ~e for a system previously considered, in which build-up was
caused by reduction of the resistance of the shunt-field circuit of a self-
excited exciter, the ceiling voltage being the same in both cases. It
is clear that de is greater for insertion of series boosting voltage from
the rotating amplifier than it is for reduction of the field resistance.
Since the time constant is the same in both cases, the build-up is more
rapid with the rotating amplifier.
Rapid build-down can also be obtained with the rotating amplifier,
the line Ri - E of Fig. 36a being elevated initially to maximum
buck. However, the bucking voltage decreases as the alternating
voltage comes near the correct value. A similar decrease of boost
voltage occurs during build-ups in which correct alternating voltage
is approached.
The calculation of exciter voltage versus time by the method of
graphical integration has been outlined. Point-by-point calculation
may be used instead, in which case eq. 52 is rewritten in the form
dt = (uNjk) de [56]
e +E - Ri
in which ~e is the change of e in time interval At. The point-by-point

calculation is similar to that already described and illustrated.
Exciter with two or more field windings. We have discussed the

calculation of build-up of exciter voltage in an exciter having combined
self- and separate excitation voltages applied to the same field winding.
Other types of exciters have two or more field windings, one of which
is self-excited, while one or more others are separately excited. We
will now discuss the two-field exciter with the aid of certain simplifying
assumptions.t'' First, saturation will be neglected, with the result that
the armature e.m.f. e is a linear function of the two field currents i a
and i b, thus:
[57]
where A a and A b are the steady-state voltage gains and R; and R b are
the resistances of the field circuits. Second, it will be assumed that
the same flux ef> links both field windings. This is very nearly true
because both are wound on the same pole cores. The applied voltages
ea and eb are consumed in the resistive and inductive drops:
· N de/>
Ca = Ra~a + a dt [58]
eb = R b~b· + N de/>
b dt [59]
Solution of eqs. 57, 58, and 59 for dc/>/dt yields:

def> Aaea + Abeb - e
[60]
dt - AaNa AbNb +
Assuming a constant coefficient of dispersion,
c kef>
:=3 _.-
[61]
a
we obtain:
de
[62j
dt
The rate of increase of exciter armature voltage is, as before, equal

to an accelerating voltage ~e divided by a time constant. The ac-
celerating voltage is the difference between a linear combination of
the voltages applied to the two field windings and the instantaneous
armature voltage. The time constant is the sum of the time constants
RESPONSE ()F LOADED EXCITER 203
of the two field circuits separately* (including external resistance but

with external inductance assumed negligible). It is shown in Ref. 46
that the time constant of each field winding is approximately propor-
tional to the volume of that winding; hence the sum of the time con-
stants of two windings will be almost equal to the time constant of a
single field winding occupying the same spaee.F~quation62 also shows
that two applied field voltages are additive, very much as if they were
connected in series. Thus we see that it makes little difference in the
response of the machine whether the two voltages are applied to
separate windings or, in series, to the same winding.
If a more exact analysis is required, the build-up of exciter volt.age
can be calculated point by point without the necessity of the two
simplifying assumptions which were made in the foregoing approximate
analysis.
If the self-excitation is furnished by a series field instead of a shunt
field, it is clear that the exciter cannot be considered to be unloaded:
the field circuit of the a-c. generator must be taken into account.
Response of loaded exciter. The terminal voltage of a loaded

exciter differs from the voltage of the unloaded exciter with the same
field current because of the effects of armature resistance (including
brush drop), armature inductance, and armature reaction. Of these,
resistance and reaction are the most important. The effect of armature
inductance is negligible at the low rates of change of armature current
which are encountered during build-up and build-down, Modern
exciters have interpoles and the brushes are set on neutral. Conse-
quently there is no magnetizing or demagnetizing armature reaction.
There is, however, a cross-magnetizing armature reaction, which has
a demagnetizing effect unless the machine has a compensating winding,
The demagnetizing effect of cross-magnetizing armature reaction
can be explained as follows: The combined m.m.f. of field and armature
currents varies linearly from one side of a pole to the other. Over one
half of the width of the pole the armature m.m.f, augments the field
m.m.f., while over the other half the armature m.m.f, subtracts from
the field m.m.I, As a result, the flux density is increased on half the
area of each pole face and is decreased on the other half. However,
because of saturation, the decrease on one half exceeds the increase on
the other half, and there is a net decrease of flux.
The effect of both armature resistance and armature reaction is to
decrease the exciter response during build-up, especially that of a self-
"It was shown in Chapter XII that, for two coupled circuits having a little leak-
age, there is also a second, very short, time constant.
excited exciter. This fact will be demonstrated by the analysis which

follows:
When a fault occurs on the armature circuit of the main (a-c.)
generator, the field current of this generator is increased suddenly to
an abnormally high value, which, flowing through the exciter armature,
produces considerable resistance and reaction drop. This high value
Field current
FIG. 38. Saturation curves for constant armature current.
of armature current may be assumed to remain substantially constant

during most of the period of exciter build-up, not only because of the
long time constant of the main-generator field but also because of the
bolstering effect of the exciter build-up. By assumption of constant
armature current during build-up, the analysis is facilitated.
If the armature current of the exciter is constant, the inductive drop
is zero, the armature resistance drop and brush drop are constant, and
the effect of armature reaction on the armature voltage is a function
RESPONSE OF LOADED EXCITER 205
of saturation - hence of the field current. The combined effect of the

several drops can be found by test. The results are plotted as a load
saturation curve, or curve of armature terminal voltage versus field
current under the condition of constant armature current (see Fig. 38).
By addition of the constant resistance and brush drops to the ordinates
of the load saturation curve, another curve is derived showing the
internal voltage, or voltage generated by the air-gap flux, as a function
of field current. The last curve is called the "distortion curve." Its
abscissas exceed those of the no-load saturation curve by the amount
of armature reaction. This amount is not constant, even though the
armature current is constant, but varies with the curvature of the
no-load saturation curve and is greatest in the vicinity of the knee.
Under the assumption of constant armature current, the voltage-
time curve of a loaded exciter can be calculated by the method of
graphical integration in a manner differing only in detail from that
employed for an unloaded exciter. The load saturation curve and the
distortion curve are used instead of the no-load saturation curve (Fig.
39). Specifically stated, the differences in procedure are as follows:
1. The variable of integration e is the internal armature voltage,
which is now the ordinate of the distortion curve instead of that of
the no-load saturation curve.
2. For a self-excited machine, the forcing voltage de is the differ-
ence bet,veen the armature terminal voltage (read on the load
saturation curve) and the field Ri drop. For a value of e represented
by DG in Fig. 39, the corresponding value of de is represented by EF.
For a separately excited machine, Lle is represented by AF, as before.
(It is assumed here that the no-load ceiling voltage AG is made the
same for both kinds of excitation.)
3. Usually a curve of e versus t suffices. However, if a curve of
exciter armature terminal voltage versus tis wanted it can be obtained
by subtracting the constant armature-resistance and brush drops
from e.
We are now able to deduce the effect of load upon exciter response
during build-up. Let us consider first the self-excited machine. For
the same field current i in Fig. 39, the values of de are CF at no load
and EF at load. However, in using the method of graphical integra-
tion the comparison should be made for the same value of internal
voltage e. On this basis de is D'F' at no load and EF at load. On
either basis, the effect of load is .to reduce de noticeably. Since t1e
appears in the denominator of the integrand, the integral (representing
the time required for e to build up between specified limits) is increased
by load. In a separately excited machine de is A' F' at no load and

AF at load. Because of armature reaction, load slightly reduces ae
and increases the time of build-up.
Field
resistance
line
~I
I
I
Gl G
Field current
FIG. 39. Quantities required for calculating the response of a loaded exciter by
the method of graphical integration.
Thus, with either kind of excitation, the exciter response is decreased

by load. The effect is much greater in the self-excited exciter than in
the separately excited one, however.
Besides reducing the response, load also reduces the ceiling voltage.
That of the self-excited exciter is reduced from AG to BG (Fig. 39);
that of the separately excited exciter, from AG to HG.
Calculation of field current of main generator during build-up of

exciter. Methods of calculating the field current of the main (a-c.)
generator by point-by-point methods have been described in the last
INCREASING TI-IE RESPONSE OF AN EXCITER 207
chapter. These methods assume that the exciter voltage is a known

function of time. This function (the exciter voltage-time curve) can
be found by methods given in this chapter. The simplest method is
to calculate the voltage-time curve for no load and to assume that the
result is valid also for the load condition. This assumption causes
little error when applied to separately excited exciters. In the case of
self-excited exciters it is advisahle at least to take armature resistance
into account while calculating the build-up curve. It is more accurate
to include the effect of armature reaction also. The armature current
of the exciter can be assumed constant, as already mentioned.
Regardless of which of these methods is used in calculating exciter
build-up, the resistance of the armature of the exciter should be included
in the value of total resistance of the field circuit of the main generator
if the internal voltage of the exciter is used in this circuit. (An alterna-
tive way is to omit armature resistance and to use armature terminal
voltage.)
The armature current of the exciter is not actually constant during
build-up. The effect of its variation upon build-up can be taken into
account, if desired, by means of a point-by-point solution in which
both the exciter voltage and the field current of the main generator
are calculated together. However, as such refinement is seldom, if ever,
needed, the procedure will not be described. The assumption of
constant exciter-armature current or the neglecting of armature
resistance and reaction entirely has the advantage that it is possible
to calculate the voltage-time curve of the exciter prior to, and inde-
pendently of, the point-by-point calculation of main-generator field
current.
Methods of increasing the response of an exciter.P If the response

of an existing exciter is not great enough, the following methods of
increasing it should be considered:
1. Providing separate excitation for a self-excited machine. In

what follows, separate excitation is assumed.
2. Increasing the ceiling voltage by reducing the permanent external
field-circuit resistance, if any, or by increasing the excitation voltage.
The attainable ceiling voltage may be limited either by saturation
or by commutation.
3. Reducing the number of field turns in series either by increasing,
the number of parallel paths or by providing a new field winding
withlarger conductor and fewer turns. If it is not desired also to
increase the ceiling voltage, enough permanent external resistance
208 EXCITA1'ION SYSTEMS
should be added to the field circuit to give the same field resistance
per series turn, and hence the same total ampere-turns, as before.
4. Increasing both the excitation voltage and theficld resistance in the
same ratio so that the ceiling voltage is unchanged.
5. Providing a new armature winding or a new armature wound for
a higher voltage. This is feasible only if the exciter' has a greater
capacity than required, because a higher voltage armature has a
lower current rating and the collector-ring voltage of the main
generator is not changed.
Methods 3 and 4 are similar in that they decrease the time constant
of the exciter-field circuit without raising the ceiling voltage. The time
constant of the field circuit of a separately excited exciter, as given by
eqs. 17 and 29, is directly proportional to the number of turns Nand
inversely proportional to the excitation voltage E. Method 3 decreases
N, whereas method 4 increases E.
Although the time required for the exciter voltage to build up from
one given value of voltage to another varies inversely as the time
constant, the nominal response of the exciter does not vary in the same
'vVay. If, for example, the time constant is halved, the exciter voltage
,vBI build up in 0.25 sec. to the value which it previously reached in
0.5 sec. During the second quarter-second, however, the voltage is
nearer the ceiling value and rises more slowly than during the first
quarter-second. Therefore, the nominal response, being defined for a
time of 0.5 see., is less than doubled.
In order to increase the nominal response to the desired value it may
be necessary to increase the ceiling voltage. The required ceiling
voltage can be estimated from eq. 65, which is derived from the follow-
ing considerations: The maximum conceivable value of nominal
response for a given ceiling voltage would be attained if the exciter
voltage could rise instantly from the nominal collector-ring voltage to
the ceiling voltage and then remain at the ceiling voltage. If E;
denotes the ceiling voltage and Eo the nominal collector-ring voltage,
the area under the voltage-time curve and above Eo for 0.5 sec. is
A = 0.5(E c - Eo). By eq. 50, the nominal response is
Ae) = SA = 4(Ec - Eo) = 4 (Ec _ 1) [63]
( At nom Eo Eo Eo
In practice, the greatest nominal response that can be attained is from
0.4 to 0.65 of this theoretical value. Use of the higher value gives
(dtde) ~ 2.6 (E c _ 1) [64]

nom Eo
REFERENCES 209
whence
e, : : : Eo [1 + 0.38 (t:,.e)
At nom
] [65]
This gives the required ceiling voltage approximately as a function of

the desired nominal response.
REFERENCES
1. O. G. C. DAHL, Electric Power Circuits: 'Theory and Applications, Vol. II,
Power System Stability. McGraw-Hill Book: Co., New York, 1st ed., 1938. Chap.
XV, U Analysis of Unloaded Exciters." Chap. XVI, ccAnalysis of Loaded Exciters."
Chap. XVII, " Action of Voltage Regulators and Excitation Systems."
2. Electrical Transmission and Distribution Reference Book, by Central Station
Engineers of the Westinghouse Electric and Manufacturing Company, East Pitts-
burgh, Pa., 1st edition, 1942. Chap. 7, "Machine Characteristics," by C. F.
WAGNER, Part XIII, "Exciter Response," pp. 168-74. Chap. 8, "Power-System
Stability," by R. D. EVANS, "Excitation Systems," pp. 201-4.
3. REINHOLD RttDENBERG, "Fremd- und Selbsterregung von magnetisch gesat-
tigten Gleichstromkreisen," Wiss. Veroffentlichungen aU8 dem Siemens-Konzem,
vol. I, p. 179, 1920; also Elektri8che Schaltvorgange, Julius Springer, Berlin, 1923 and
1933. Solution by graphical integration.
4. F. L. MOON, "When May Alternator Field Rheostats be Omitted?" Elec.
Jour., vol. 14, pp, 47-50, February, 1917.
5. C. A. BODDIE, ,t Automatic Voltage Regulators," Elec. Jour., vol. 14, pp. 75-
84, February, 1917. Tirrill regulator with a-c. vibrating magnet.
6. J. H. ASHBAUGH, "The Extended Broad-Range Voltage Regulator," Elec.
Jour., vol. 22, pp. 559-61, November, 1925.
7. C. A. POWEL, "Quick-Response Excitation for Alternating-Current Synchro-
nous Machinery," Elec. Jour., vol. 24, pp. 157-62, April, 1927.
8. R. E. DOHERTY, "Excitation Systems: Their Influence on Short Circuits and
Maximum Power," A.I.E.E. Trans., vol. 47, pp. 944-56, July, 1928. Disc., pp.
969-74. Dynamic stability.
9. C. A. NICKLE and R. M. CAROTHERS, "Automatic Voltage Regulators:
Application to Power Transmission Systems," A.I.E.E. Trans., vol. 47, pp. 957-69,
1928. Disc., pp. 969-74. Contains an analysis of the vibrating type of regulator.
10. J. H. ASHBAUGH and H. C. NYCUM, "System Stability with Quick-Response
Excitation and Voltage Regulators," Elec. Jour., vol. 25, pp. 504-9, October, 1928.
11. R. H. PARK and E. H. BANCKER, "System Stability as a Design Problem,"
A.I.E.E. Tran8., vol. 48, pp. 170-94, January, 1929.
12. V. BUSH, "The Differential Analyzer: A New Machine for Solving Differ-
ential Equations," Jour. Franklin Insi., vol. 212, pp. 447-88, October, 1931.
13. W. A.. LEWIS, "Quick-Response Excitation," Elec. Jour., vol. 31, pp. 308-12,
August, 1934.
14. A. VAN NIEKERK, "Determining the Ratio of Exciter Response," Elec.
Jour., vol. 31, pp, 361-4, September, 1934.
15. C. R. HANNA, K. A. OPLINGER, and C. E. VALENTINE, "Recent Develop-
ments in Generator Voltage Regulation," A.I.E.E. Troms., vol, 58, pp. 838-44,
1939. Diso., p. 844.
16. W. K. BOICE, S. B. CRARY, GABRIEL KRON~ and L. W. THOMPSON, "The
Direct-Acting Generator Voltage Regulator," A.I.E.E. Trons., vol. 59, pp. 149-56,
March, 1940. Disc., pp. 156-7.
17. American Standard Definitions of Electrical Terms, American Institute of
Electrical Engineers, New York, 1942. Definitions 10.95.440, 10.95.450, and
10.95.460.
18. C. CONCORDIA, "Steady-State Stability of Synchronous Machines as Affected
by Voltage-Regulator Characteristics," A.I.E.E. Trans., vol. 63, pp. 215-20,
May, 1944. Disc., pp. 489-90.
19. E. L. HARDER and C. E. VALENTINE, "Static Voltage Regulator for Rototrol
Exciter," A.I.E.E. Trans. (Elec. Eng.), vol. 64, pp. 601-6, August, 1945.
20. H. A. P. LANGSTAFF, H. R. VAUGHAN, and R. F. LAWRENCE, "Application
and Performance of Electronic Exciters for Large A-C. Generators," A.I.E.E.
Trans. (Elec. Eng.), vol. 65, pp. 246-54, May, 1946. Disc., pp. 515-20.
21. R. B. BODINE, S. B. CRARY, and A. W. RANKIN, "Motor-Driven Exciters for
Turbine Alternators," A.I.E.E. Trans. (Elec. Eng.), vol. 65, pp. 290-7, May, 1946.
Disc., pp. 515-20.
22. J. B. MCCLURE, S. 1. WHITTLESEY, and M. E. HARTMAN, "Modern Exci-
tation Systems for Large Synchronous Machines," A.I.E.E. Trans., vol. 65, pp.
939-45, 1946. Disc., pp. 1148-50. Summary of practice.
23. C. CONCORDIA, S. B. CRARY, and F. J. l\1AGINNISS, "Long-Distance Power
Transmission as Influenced by Excitation Systems," A.I.E.E. Trons., vol. 65,
pp. 974-87, 1946. Disc., pp. 1175-7.
24. F. M. PORTER and J. H. I{INGHORN, "The Development of Modern Exci-
tation Systems for Synchronous Condensers and Generators," A.I.E.E. Trans.,
vol. 65, pp. 1020-8, 1946. Disc., pp. 1156-9. Electronic pilot exciter, electronic
main exciter, and Amplidyne pilot exciter.
25. G. DARRIEUS, "Long-Distance Transmission of Energy and the Artificial
Stabilization of Alternating Current Systems," C.l.G.R.E., 1946, Report 110.
26. E. L. HARDER, "Solution of the General Voltage Regulator Problem by
Electrical Analogy," A~I.E.E. Trans., vol. 66, pp. 815-25, 1947. Disc., p. 825.
27. C. L. KILLGORE,. "Excitation Problems in Hydroelectric Generators Supply-
ing Long Transmission Lines," A.l.E.E. Trans., vol. 66, pp. 1277-82, 1947. Disc.,
pp.1282-4.
28. A. W. KIMBALL, "Two-Stage Rototrol for I . ow-Energy Regulating Systems,'
A.l.E.E. Trans., vol. 66, pp. 1507-11, 1947.
29. M. M. LIWSCHITZ, "The Multistage Rototrol," A .I.E.E. Trans., vol. 66,
pp. 564-8, 1947.
30. C. STEWART, "Automatic Voltage Control of Generators," Jour. I.E.E.,
vol. 94, part IIA, pp. 39-48, May, 1947. Disc., pp. 60-5.
31. J. E. BARKLE and C. E. VALENTINE, "Rototrol Excitation Systems," A.I.E.E.
Trans., vol. 67, part I, pp. 529-34, 1948. Disc., p. 534.
32. C. LYNN and C. E. VALENTINE, ((Main Exciter Rototrol Excitation for
Turbine Generators," A.I.E.E. Trans., vol. 67, part I, pp. 535-9, 1948. Disc.,
p.539.
33. C. CONCORDIA, ((Steady-State Stability of Synchronous Machines as Affected
by Angle-Regulator Characteristics," A.I.E.E. Trans., vol. 67, part I, pp. 687-
90,1948.
34. G. D. FLOYD and H. R. SILLS, ((Generator Design to Meet Long Distance
Transmission Requirements in Canada," C.I.G.R.E., 1948, Report 13l.
35. H. DAVID and .1. FAVEREAU, "Stability of Alternators with Series Excitation
Connected by a Long Line to a High-Power System,' C.I.G.R.E., 1948, Report 305.
REFERENCES 211
36. W. A. LEWIS, "Excitation Requirements and Control of Reactive Power,"

Proc. Midwest Power Conf. (Chicago), vol. 10, pp. 145-56, 1948.
37. C. LYNN, "Rotating Regulator Exciters," Proc. Midviee; Pouier Conf,
(Chicago), vol. 10, pp. 157-67, 1948."
38. JAMES T. CARLETON, "The Transient Behavior of the Two-Stage Rototrol
Main Exciter Voltage Regulating System as Determined by Electrical Analogy,"
A.I.E.E. Trans., vol. 68, part I, pp. 59--R3, 1949.
39. A. G. lVIELLOR and IV1. TEMOSHOK, "Excitation Hystem Performance with
Motor-Driven Exciters," A.I.E.E. Trans., vol. 69, part I, pp. 321-7, 1950.
40. A. P. COLAIACO, A. A. JOHNSON, and J. E. REILLY, "Design and Test on
Electronic Exciter Supplied from Common Shaft-Driven Generator," A.I.E.E.
Trans., vol. 69, part I, pp. 328-35, 1950. Disc., pp. 335-7.
"T.
41. A. H. PHILLIPS, II. LAMBERT, and D. R. PATTISON, "Excitation Improve-
merit - Electronic Excitation and Regulation of Electric Generators as Compared
to Conventional Methods," A.I.E.E. Trans., vol. 69, part I, pp. 338-40, 1950.
Disc., pp. 340-1.
42. E. L. HARDER, R. C. CHEEK, and J. M. CLAYTON, "Regulation of A-C
Generators with Suddenly Applied Loads - Part II," A.I.E.E. Trans., vol. 69,
part I, pp. 395-405, 1950. Disc., pp. 405-6.
43. C. J-J. KILLGORE and N. G. HOLMDAHL, "Hydroelectric Generator Excita-
tion Experience at Grand Coulee," A.I.E.ltJ. Trans., vol. 69, part II, pp. 855-60,
1950.
44. R. -M. SAUNDERS, "Dynamoelectric Amplifiers,". Elec. Eng., vol. 69, pp.
711-6, August, 1950.
45. FRANK E. BOTHWELL, "Stability of Voltage Regulators," A.I.E.E. Trans.,
vol. 69, part II, pp. 1430-3, 1950. Disc., p. 1433.
46. DIDIER JOURNEAUX, "Transient Operation of D-C Generators," Allis-
Chalmers Elec. Rev., vol. 15, no. 3, pp. 27-34, 3rd quarter, 1950.
47. C. CONCORDIA, "The Differential Analyzer as an Aid in Power System
Analysis," C.I.G.R.E., 1950, Report 311.
48. H. F. STORM, "Static Magnetic Exciter for Synchronous Alternators,"
A.I.E.E. Trans., vol. 70, part I, pp. 1014-7, 1951.
49. H. G. FRUS,F. N. MCCLURE, and W. H. FERGUSON, "Selection of Character-
istics for Turbine-Generator Motor-Driven Exciters," A.I.E.E. Trans., vol. 71,
part III, pp. 176-82, January, 1952. Disc., pp. 182-3.
50. H. A. CORNELIUS, W. F. CAWSON, and H. W. CORY, "Experience with
Automatic Voltage Regulation on a Llfi-Megawatt Turbogenerator," A.I.E.E.
Trans., vol. 71, part III, pp. 184-7, January, 1952.
51. G. K. KALLENBACH, F. S. ROTHE, H. F. STORM, and P. L. DANDENO, "Per-
formance of New Magnetic .Amplifier Type Voltage Regulator for Large Hydro-
electric Generators," A.I.E.E. Trans., vol. 71, part III, pp. 201-5, January, 1952.
Disc., pp. 205-6.
52. F". M. PORTER and H. B. MARGOLIS, Operating Experience with Shaft-
C(
Driven Exciters in Air and Hydrogen," A.I.E.E. Trans., vol. 71, part III, pp. 758-
67, October, 1952. Disc., pp. 767-8.
53. W. A. HUNTER and M. TEMOSHOK, "Development of a Modern Amplidyne
Voltage Regulator for Large Turbine Generators," A.I.E.E. Trans., vol. 71, part
III, pp. 894-900, October, 1952. Disc., pp. 9OO-I.
54. CH. LAVANCHY, "A New Solution for the Regulation of Counter-Excitation
Synchronous Compensators," C.I.G.R.E., 1952, Report 331.
55. J. E. BARKLE, C. E. VALENTINE, and J. T. CARLETON, H Magamp Regulation
212 EXCITA'fION SYSTEMS
for Synchronous Machines," Westinghouse Engineer, vol. 12, pp. 204-7, November,
1952.
56. A. TUSTIN, Direct Current Machines for Control Systel1UJ, The Macmillan
Company, New York, 1952.
57. J. T. CARLETON, P. O. BOBO, and "V. F. HORTON, "A New Regulator and
Excitation System," A..I.E.E. Trans., vol. 72, part III, pp. 175-81, April, 1953.
Disc., pp. 181-3.
58. L. F. LISCHER and R. B. GEAR, "Rotating Regulators Improve Light-Load
Stability," Electric Light and Power. Part I, vol. 31, no. 12, pp. 106-11, October,
1953. Part II, vol. 31, no. 13, pp. 133-7, November, 1953.
59. MILAN VIDMAR, JR., "Comparison of Four Main Types of Rotary Exci-
ters Employed in Modern Practice, with Regard to the Exciting Effectiveness,"
C.I.G.R.E., 1954, Report 138.
60. R. DAVID and L. GATESOUPE, "Results of Recent Tests Regarding Different
Systems of High Speed Excitation and De-excitation of Alternators," C.I.G.R.E.)
1954, Report 147.
PROBLEMS ON CHAPTER XIII

1. Represent the magnetization curve of Fig. 25 by the modified Frohlich
equation (eq. 12).
2. Obtain a formal solution of eq. 9 for separate excitation by substituting
eq. 12 into it and integrating.
3. Having solved Probs. 1 and 2, work Example 1 by use of the formal
solution obtained in Probe 2. Compare results.
4. Repeat Probe 2' for self-excitation.
6. Having solved Probs. 1 and 4, work Example 2 by use of the formal
solution obtained in Probe 4.
6. Find the voltage-time curve for build-down of the separately excited
exciter of Example 1 from. ceiling voltage to an ultimate value of one-third
the ceiling voltage.
7. Repeat Probe 6 for self-excitation.
8. Work Example 1 using the assumption of constant leakage inductance
instead of constant coefficient of dispersion. The coefficient of dispersion is
1.15 only on the straight part of the magnetization curve. How much
difference is there between the two assumptions in the time required for the
armature voltage to build up to 250?
9. Find the nominal response of the separately excited exciter of Example
1. The nominal collector-ring voltage of the a-c. generator is 180 volts.
10. Find the nominal response of the self-excited exciter of Example 2.
The nominal collector-ring voltage of the a-c. generator is 180 volts.
11. Work Example 1 by point-by-point calculation.
12. Work Example 2 by point-by-point calculation.
13. Work Example 3 by graphical integration.
14. Plot the voltage-time curves of the self-excited exciter of Example 2
from an initial voltage of 180 to the ceiling voltage obtained if all external
resistance is removed from the field circuit. Compare the following two
conditions: (a) no load, (b) constant load current of 150% of rated value.
PROBLEMS 213
The armature reaction at the knee of the no-load saturation curve amounts
to 5% of the field current there, and elsewhere it is proportional to the curva-
ture of the saturation curve. The full-load armature I R drop is 5% of rated
voltage, and the brush drop is 2 volts.
16. In Example 3, what is the value of the constant corresponding to
2.6 in eq. 64?
16. It is proposed to reconnect the field winding of the separately excited
exciter of Example 1 in two parallel circuits. How much permanent external
resistance should be used in order to keep the ceiling voltage at 275 volts?
What would the new value of nominal response be? Compare it with the
original value calculated in Probe 9. Nominal collector-ring voltage is 180.
17. Work Prob. 16 for three parallel circuits.
18. In order to obtain the same improvement in nominal response as in
Probe 16, but without reconnecting the field winding, what excitation voltage
and what external resistance should be used? The ceiling voltage is to
remain at 275 volts.
19. What voltage and current must be supplied by the pilot exciter when
the main exciter is generating its ceiling voltage (a) in Probe 16, (b) in
Probe 18?
20. It is proposed to reconnect the field winding of the separately excited
exciter of Example 1 into two parallel circuits and, at the same time, to
increase the ceiling voltage to 350. How much permanent external resistance
should be used? What would the new value of nominal response be? (Com-
pare it with the values obtained in Probs. 9 and 16.) Nominal collector-ring
voltage is 180.
21. It is proposed to rewind the armature of the separately excited exciter
of Example 1 in duplex wave, The field circuit would not be changed.
What would the new ceiling voltage be? What would the nominal response
be in building up from 180 volts?
22. Explain why a slow-speed exciter has a longer time constant than a
high-speed exciter.
23. The exciter described in Example 1 is to be operated with combined
self- and separate excitation, a rotating amplifier being connected in series
with the shunt-field circuit. The resistance of this circuit is adjusted so that
the no-load exciter armature voltage is 180 when the e.m.f. of the rotating
amplifier is zero. Then the e.m.f, of the amplifier is changed to a value which
will cause the no-load exciter voltage to build up to a ceiling value of 275
volts, this amplifier e.m.f. being held constant. Calculate and plot the curve
of build-up of exciter voltage. Also calculate the nominal exciter response
under these conditions. Compare the maximum power output of the rotating
amplifier with that of the pilot exciter in Example 1.
CHAPTER XIV
DAMPER WINDINGS AND DAMPING
Almost all synchronous motors, synchronous condensers, and

synchronous converters and many salient-pole synchronous generators
are equipped with damper windings (also called amortisseur windings).
In a few installations damper windings have been designed with regard
to their effect upon system stability, but in most installations they have
been provided for other purposes. Regardless of the reasons for their
use, they affect stability to a greater or lesser extent.
Types of damper windings. There are two principal types of damper

windings: (1) complete, or connected; (2) incomplete, nonconnected,
or open. Both types consist of bars placed in slots in the pole faces and
connected together at each end. The end conductors in the complete
damper winding (Fig. la) are closed rings connecting the bars of all
poles; whereas, in the incomplete damper winding (Fig. Ib), they are
broken between poles. Thus, in the latter type, the damper winding
of each pole is an independent grid. Complete dampers are similar to
the squirrel-cage winding of an induction motor, except that the bars
are not uniformly spaced, being omitted between the poles. The end
rings sometimes have bolted joints between poles to facilitate the
removal of a pole. Generator designers prefer incomplete dampers for
mechanical reasons, particularly where the peripheral speed is high,
because of the danger of damage to the unsupported connections
between poles by high centrifugal force. Complete dampers, however,
are electrically superior and are commoner now than incomplete
dampers.
Damper windings may be classified also according to their resistance.
Low-resistance damper windings produce their greatest torque at small
slips; high-resistance damper windings, at large slips.
Occasionally a double-deck damper windi11g is used. This consists
of a low-resistance, high-reactance damper, which is effective at small
slips, and a high-resistance, low-reactance damper, which is effective at
large slips. Double-deck dampers are used (like double squirrel-cage
windings on induction motors) to improve the starting and pullin char-
acteristics of synchronous motors. They have been used on a few
214
PURPOSES OF DAMPER WINDINGS 215
generators to improve stability by giving both positive-sequence

damping and negative-sequence braking (both of which are discussed
later in this chapterl .l"
Field collars have sometimes been used as substitutes for damper
windings. A field collar is a heavy short-circuited copper loop placed
(b)
FIG. 1. Amortisseur windings or damper windings, (a) connected and (b) non-
connected.
around the pole core near the outer end. Field collars can be used
together with a single damper winding to give results similar to those
given by a double-deck damper.
Damper windings are not used on turbogenerators, but the solid
steel rotor cores of such machines provide paths for eddy currents and
thus produce the same effects as dampers. Even in machines with
laminated salient poles and no dampers, eddy currents in the pole faces
have a very small damping effect.
Purposes of damper windings. The chief reasons for providing

damper windings on salient-pole synchronous machines are the
following:
1. To provide starting torque for synchronous motors, condensers,
and converters.
216 DAMPER WINDINGS .AND DAMPING
2. To suppress hunting. This was the first application of dampers

and the one for which they were named. Generators driven by
reciprocating engines tend to hunt because of the pulsating torque
of the engines. Motors driving loads of pulsating torque, such as
compressors, likewise tend to hunt. Hunting from these causes is
"forced" hunting. There is also a spontaneous hunting that occurs
when synchronous machines are connected together by circuits
having a high ratio of resistance to reaotance.v'' This condition is
likely to occur if the frequency is low, if the conductor size is small,
or if the inductive reactance is partially cancelled by use of series
capacitors. Both forced and spontaneous types of hunting are
greatly decreased in amplitude by low-resistance dampers.
3. To damp oscillations started by aperiodic shocks, such as short
circuits or switching, A possible cause of oscillations between
generators in the same station is inequality of speeds of response of
their several exciters. Oscillations from any of these causes are
damped more rapidly if damper windings are added.
4. To prevent distortion of voltage wave shape caused by unbalanced
loading; in other words, to suppress harmonics. Single-phase gener-
ators, phase balancers, and polyphase generators carrying poorly
balanced load require heavy damper windings not only in order to
suppress harmonics but also for other reasons (listed as items 5 and
6 below). Water-wheel generators without dampers may produce
dangerous overvoltage during single-phase short circuits because of
resonance between the capacitance of transmission lines and the
inductance of the machines and lines at harmonic frequencies.vP'"
Dampers have been installed in some generators which did not
originally have them in order to decrease the harmonic voltages
to safe values."
5. To balance the terminal voltages of the several phases during
unbalanced loading, that is, to decrease the negative-sequence
voltage. Damper windings decrease the negative-sequence re-
actance. The lower the negative-sequence impedance, the lower is
the negative-sequence voltage.
6. To prevent overheating of thepole pieces of single-phase generators
and phase balancers from eddy currents induced by sustained
negative-sequence armature currents. The flux linkages of low-
resistance dampers are nearly constant. Hence the dampers shield
the pole pieces from variations of flux.
7. To provide a braking torque on a generator during an unsym-
metrical fault, and thus to reduce the accelerating torque during
the fault.
EFFEC'!' ON MAOHINE CONS'rAN'fS 217
8. To provide additional torque for synchronizing generators, espe-

cially those which are synchronized automatically. The dampers
are particularly helpful in case the generator is synchronized out of
phase or somewhat off speed. They also help to pull the generator
back into step after synchronism has been lost because of a fault.
9. To-reduce circuit-breaker recovery-voltage rates.
10. To reduce the stress on-the insulation of the field winding during
current surges in the armature circuit, especially those which do not
affect all poles alike, such as those due to internal faults.
Effect of damper windings on synchronous-machine constants.

Damper windings affect the following synchronous-machine quanti-
ties: Xd", x q " , X2, rs, T d " , and r;". l'he effects are discussed below
and are illustrated numerically by Tables 1, 2, 3, and 4.
l'ABLE 1
CONS'l'ANTS OF A SYNCHUONOUS GENERA'rOR AS Ali'li'ECTED BY TYPE
4
OF DAMPER WINDING
(100 kva., 2,300 volts, 25.2 amp.)
Type Xd " Xq " v'Xd"x qi i x(/' /Xd" Tad' I r'aq "
No damper 0.260 0.577 0.388 2.22 0.028 0.105

Connected copper 0.157 0.146 0.151 0.93 0.036 0.047
Connected Everdur 0.171 0.157 0.164 0.92 0.063 0.111
Nonconnected copper 0.154 0.390 0.245 2.53 0.037 0.113
The constants are in per unit. They are test results, obtained by applying single-
phase voltage to two armature terminals with the rotor stationary.
The conductivity of Everdur is approximately 6% of that of copper.
TABLE 2
CONSTANTS OF A SYNCHRONOUS CONDENSER AS AFFECTED BY 'fYPE
OF DAMPER WINDING4
(5,000 kva., 4,COO volts, 721 amp.)
r2 X2 = ! (Xd" + x q" )
Type
Test Calculated Test Calculated
No damper 0.0';1:5 0.040 0.75 0.69
Connected copper 0.026 0.029 0.195 0.215
Connected brass 0.045 0.044 0.195 0.215
Connected Everdur 0.12 0.125 0.20 0.215
--------- ---~-.-
The constants are in per unit.

218 DAMPER WINDINGS AND DAl\1PING
TABLE 3
REACTANCES OF LARGE WATER-WHEEL GENERATOR
AS AFFECTED BY CONNECTED COPPER DAMPER WINDING'
(32.5 Mva., 0.8 power factor, 12 kv., three-phase,
60 cycles, too r.p.m., vertical shaft)
Calculated values
Per-Unit Without With Per-Cent

Reactance Dampers Dampers Reduction
Xl 0.22 0.22 0
Xd 1.15 1.15 0
xq 0.73 0.73 0
Xd' 0.40 0.40 0
xq' 0.73 0.73 0
Xd" 0.32 0.265 17.2
x"
q 0.73 0.29 60.3
X2 0.48 0.28 41.7
Xo 0.14 0.123 12.1
TABLE 4
CALCULATED CONSTANTS OF WATER-\VHEEL GENERATOR
AS AFFECTED BY INCOMPLETE DAMPER WINDING 12
(9 Mva., 0.8 power factor, 13.8 kv., 60 cycles, 180 r.p.m.)
Reactances are in per unit, time constants in seconds.
Incomplete Damper and Top-Field Collar Only,

Constant
Top-Field Collar No Damper
0.91 0.91
0.26 0.26
0.14 0.177
0.524 0.524
0.524 0.524
0.368 0.524
2.86 2.86
0.055 0.0136
0.044
Subtransient reactances in both axes of the rotor are decreased by

damper windings (see Tables 1, 3, and 4). A connected damper
decreases the quadrature-axis subtransient reactance xq " more than a
nonconnected damper does (see Table 1). Furthermore, the connected
damper makes x q " more nearly equal to the direct-axis subtransient
reactance Xd" than the nonconnected damper can. The ratio xq " /xa,"
can be made as low as 1.35 by a nonconnected damper, and much lower
by a connected damper.f
EFFECT UPON STABILITY 219
.
N egat~ve-sequence reactance X2, b emg
· t h e mean 0 f Xd II an d Xq I I, ·IS
decreased by damper windings.
Negative-sequence resistance r2 may be high or low, depending upon
the resistance of the damper windings (see Table 2). Negative-
sequence impedance is discussed more fully in a later section of this
chapter.
Subtransient time constants T d' I and T q" depend upon the resistance
and equivalent inductance of the dampers.
Effect of damper windings upon stability. The principal effects of

damper windings upon power-system stability may be listed as follows:
1. Positive-sequence damping.
2. Negative-sequence braking during an unsymmetrical fault.
3. Effect of negative-sequence impedance upon positive-sequence
electric power output of the machine during an unsymmetrical fault.
A lesser effect sometimes worth considering is:
4. D-c. braking.
Positive-sequence damping results from the torque caused by inter-
action of the damper currents with the positive-sequence (forward-
rotating) magnetic field in the air gap. Except during starting or
pulling out of step, the slip of the rotor with respect to the positive-
sequence field is low and oscillatory. Positive-sequence damping
causes the oscillations of the machine rotors, after an aperiodic shock
that does not cause loss of synchronism, to decrease in amplitude and
ultimately to die out entirely. Low-resistance amortisseurs greatly
increase the amount of positive-sequence damping. Positive-sequence
damping is present both during and after a fault. However, it is much
more effective after clearing of a fault than during the fault because
the positive-sequence voltage and flux are greater after clearing.
By absorbing energy from the oscillation, positive-sequence damping
may prevent a machine which has survived the first swing from going
out of step on the second or subsequent swings because of reduction
of field flux linkage. However, the damping is not great enough to
increase significantly the power that can be carried through the first
swing. Its effect is almost always neglected in calculating power limits.
Negative-sequence braking results from the torque caused by inter-
action of the damper currents with the negative-sequence (backward-
rotating) magnetic field in the air gap. Such a field is present in a
polyphase machine only when an unbalance - usually an unsym-
metrical short circuit - is present on the polyphase circuit to which
220 DAMPER WINDINGS AND DAMPING
the armature winding is connected. The slip of the rotor with respect
to the negative-sequence field is 2 - s, where e is the positive-sequence
slip. Since 8 is small (except, as already noted, during starting and
out-of-step operation), the negative-sequence slip is very nearly equal
to 2. It never reverses, nor does the resulting torque reverse. The
torque always retards the rotor, and hence may be called a "braking"
torque. Its effect is therefore equivalent to a reduction of mechanical
input torque. In a generator, which tends to speed up during a fault,
the negative-sequence torque decreases the .accelerating torque. On
the other hand, in a motor, which tends to slow down during a fault,
the negative-sequence torque increases the absolute value of accelerat-
ing torque (now a negative, or retarding, torque). It is apparent, then,
that this braking effect is desirable in a generator but undesirable in a
motor, a condenser, or a generator which acts as an equivalent motor.
The braking effect is present only as long as an unsymmetrical fault is
on the system. The greatest braking torque is afforded by high-
resistance dampers, and very little, by low-resistance dampers. Thus
the effect may be limited to generators by using high-resistance dampers
on generators only.
In many respects negative-sequence braking is analogous to resist-
ance grounding, which reduces the accelerating torque of a generator
during ground faults. Both of these means of improving transient
stability are, less important now that high-speed fault clearing is
generally used than they were when slower speeds were common. The
higher the speed of clearing, the less important is the shock caused by
the fault itself in comparison with the shock of opening the faulted line
to clear the fault. This is true even if high-speed reclosing is employed.
Negative-sequence impedance. As already mentioned, a damper
winding decreases the negative-sequence reactance of the machine on
which it is installed and may either increase or decrease the negative-
sequence resistance. In general, the negative-sequence impedance is
lowered by a damper winding, especially by a low-resistance damper.
Lowering the negative-sequence impedance of any machine on a net-
work lowers the negative-sequence impedance of the network viewed
from the point of fault, and thus also lowers the impedance of the fault
shunt representing any type of short circuit except a three-phase short
circuit. The effect is greater for a line-to-line short circuit than for
any other type, and is greatest for a fault located near the machine
equipped with the dampers. Usually a fault location near the principal
equivalent generators is taken as the most severe one and the one used
as a criterion of system stability. Lowering the impedance of the fault
shunt weakens the tie between machines during the time that the fault
INDUC'fION-NIO'rOR 'fHEC)RY 221
is on the system, as indicated by an increase of the transfer reactance

and a decrease of the amplitude of the sine terms in the power-angle
equations; also, in the case of a two-machine system, hy a decrease of
the amplitude of the power-angle curve.
'rhus, while an unsymmetrical fault. is 011 the system the damper
winding on a generator gives rise to two opposing effects: (1) braking
torque, which is beneficial to stability, and (2) decrease of negative-
sequence impedance, which is detrimental to stability. Which of these
two effects predominates depends largely on the resistance of the
damper winding. With a low-resistance damper, the decrease of nega-
tive-sequence impedance usually predominates, while with a high-
resistance damper the braking torque usually predominates. This is
true because increasing the resistance of the damper both increases the
negative-sequence impedance and increases the braking torque.
The more rapidly faults are cleared, the less important are both of
these effects. The use of high-resistance dampers to increase the sta-
bility limits of generators is therefore less important now than it was
when slower clearing speeds were in general use.
Damper action explained by induction-motor theory. Inasmuch as

a damper winding is similar to the squirrel-cage rotor winding of an
induction motor, and the armature winding of a synchronous machine
is like the stator winding of an induction motor, the various effects of
the damper windings on stability (discussed above) can be analyzed
- at least approximately - by induction-motor theory. This theory
therefore will be briefly reviewed.
The induction motor is represented by an equivalent T circuit similar
to that used for the transformer carrying a resistive load. See Fig. 2.
The notation is explained in the caption of the figure. All quantities
are for positive phase sequence. All rotor quantities are referred to the
stator in magnitude, phase, and frequency. For example, I, is the
current which, if flowing in a stationary rotor winding having the same
number of phases and the same effective number of turns as the stator
winding and placed so that each rotor phase were exactly opposite the
corresponding stator phase, would react magnetically on the stator in
the same manner as does the actual rotor current flowing in the actual
rotor winding.
A motor with stationary rotor is merely a polyphase transformer and
can be represented by the equivalent circuit of a transformer. If the
rotor is turning, it can be replaced by an equivalent stationary rotor,
as will be explained.
The positive-sequence air-gap flux rotates at synchronous speed with
222 DAMPER \VINDINGS AND DAl\1PING
respect to the stator and induces in the stator winding an e.m.f. of

applied frequency f. The flux rotates at slip speed with respect to the
rotor and induces in the rotor winding an e.m.f. of slip frequency sf.
Rrls
~
R, x, s, 1-,
Rr rr:
v E
FIG. 2. Equivalent T circuit of induction motor for positive phase sequence.

R, = stator resistance.
X, = stator leakage reactance.
X,. = rotor leakage reactance at standstill (referred to stator).
R; = rotor resistance (referred to stator).
8 = slip.
B, = exciting susceptance.
Ge = exciting conductance.
I, = stator current.
I,. = rotor current (referred to stator).
Ie ::: exciting current.
E ::: e.m.f. induced in stator by air-gap flux
::: e.m.f. induced in rotor at standstill by air-gap flux (referred to stator).
V = stator terminal voltage.
The magnitude of the rotor e.m.f, is also proportional to the slip. At

standstill (8 = 1) it is (by definition) E; at slip 8 it is sEt The rotor
leakage reactance is also proportional to the slip. At standstill it is
X r ; at slip e it is sX r• The rotor current is
I _ sE
[1]
r - R; +jsXr
Division of numerator and denominator by 8 puts eq. 1 in the form
E
Ir = - - - - [2]
Rr/s +jXr
Equation 2 is algebraically identical to eq. 1, but is open to a new and
important physical interpretation, as follows: The current in the short-
circuited rotor turning at slip 8 Ceq. 1) is equal to the current which
would flow in a stationary rotor if the rotor-circuit resistance were
increased from R, to Rr/s Ceq. 2). The equivalent circuit of Fig. 2
INDUCTION-MOTOR THEORY 223
accords with eq. 2 in that the rotor circuit of impedance R r/8 jX, +
is connected across voltage E.
In the normal operation of an induction motor the slip is a small
positive quantity, making resistance Rr/s in the equivalent rotor
circuit much greater than the actual rotor resistance R; The addi-
tional resistance is Rr (1 - 8)/8, shown in Fig. 2 as a separate resistor.
The rotor input, which is received by induction from the stator, is
I r 2R r
- - watts per phase [3]
8
but the rotor copper loss is only I r 2R; watts per phase. The difference,
I r2R r 1 - 8 watts per phase [4]

8
equal to the fictitious copper loss in the fictitious added resistor, is

converted to mechanical power. It equals the mechanical output plus
rotational losses.
Torque. Mechanical power is torque T times speed N, and the actual
speed N of the motor is (1 - 8) times its synchronous speed N,.
Therefore,
1- 8
P = NT = (1 - 8)N BT = I r 2R r - - ' watts per phase [5]
8
Hence the torque is

__ Ir2Rr
T watts per phase per unit of speed [6]
N s8
If this same torque were developed at synchronous speed, the mechani-
cal power would be
r
N 8T = I r2 R watts per phase [7J
8
The torque at any speed may be expressed in terms of the power it

would produce at synchronous speed. The unit of torque so defined
is the "synchronous watt."
T = Ir
2 n.
-._- synchronous watts per phase [8]
8
The torque in synchronous watts is equal to the rotor input in watts.

Equation 8 also gives the torque in per unit if I, and R.). are in per unit.
In order to find an equation for torque as a function of slip, suppose
that either the stator terminal voltage V or other voltages in the net-
work to which the motor is connected are constant. The network to

the left of the E in Fig. 2 can then be replaced by the equivalent circuit
of Thevenin's theorem. Let the constant e.m.f. of this circuit be V',
Torque
Speed
-1
Backward
o Forward 2 Sync. speed
2 o -1 Slip
FIG. 3. Torque of an induction motor as a function of speed or slip.
and the impedance, R s' + jX B ' . These combine with the rotor im-
pedance, Rr / s + [X», to form a simple series circuit in which the rotor
current I, flows. Accordingly, the scalar value of the rotor current is
given by
2 _ (V')2
[9]
t, - (Ra' + Rr / s)2 + (Xa' + X r )2
This value, when substituted in eq. 8, gives the following expression
for torque:
T _ __ (V')2 Rr/s
[10]
- (R s ' + R r / S)2 + (Xs ' + X r )2
A curve of T versus slip 8 is shown in Fig. 3. By taking the derivative
of T with respect to Rr/s, we find that maximum torque occurs at
the slip
[11]
Substitution of eq. 11 into eq. 10 gives the value of maximum torque as

. (V')2
T = [12]
m 2[Rs' + V(Rs ' )2 + (X s' + X r )2]
At speeds above synchronous, the slip is negative. Under this
condition, both the mechanical power (given by eq. 5) and the torque
INDITCTION-MOTOR 1'HEORY 225
(given by eq. 6 or 8) become negative, that is, a retarding torque. The

machine is then an induction generator.
If a synchronous machine with dampers hunts or swings, the slip is
alternately positive and negative. The torque is likewise alternately
positive and negative, and always tends to bring the speed back to
synchronous speed.
Negative-sequence applied voltage causes a wave of air-gap flux to
rotate backward at synchronous speed. If the slip of the rotor with
respect to the positive-sequence field is 8, the slip with respect to the
negative-sequence field is 2 - 8. The positive-sequence slip of a
synchronous machine is normally so small that the negative-sequence
slip may be taken as 2 without appreciable error. Equation 10 and
Fig. 3 show that for this condition the torque is positive though the
speed is negative. In other words, the torque opposes the rotation;
it is a braking torque. Equation 4 shows that for 8 = 2 the mechanical
power output,
I r 2Rr 1 -2 2 -- -!.2I r2R r watts per phase [13]
is negative (that is, it is actually input) and is equal to half the rotor
copper loss, I r 2R r • The other half of the rotor copper loss is supplied
electrically via the stator terminals. Thus there is at the same time
both electric and mechanical power input, all of which goes into losses,
principally copper losses. The mechanical power input is equal to the
electric power input to the rotor (air-gap power) and is a definite
fraction of the input to the stator.
The negative-sequence rotor currents have a frequency (2 - 8 )f, or
very nearly 2f, where f is the stator frequency. In a synchronous
machine such currents flow both in the field winding and in the damper
windings if damper windings are provided, otherwise, in the field wind-
ing and in the field pole faces.
The influence of rotor-circuit (damper) resistance R; on damping
torque can be shown by plotting curves of torque versus slip for various
values of R; It is apparent from eq. 10 that, if R; and 8 are changed
proportionally, the torque is unchanged. Therefore the effect of in-
creasing the rotor resistance is to increase the abscissas of the torque-
slip curve in direct proportion to the rotor resistance. See Fig. 4. The
slip at which maximum torque occurs is directly proportional to rotor
resistance (eq. 11), but the maximum torque itself is independent of
rotor resistance (eq. 12).
In order to give high damping torque at small values of slip, the
dampers should have as Iowa resistance as possible. In order to give
high negative-sequence braking torque, the dampers should have a

much higher resistance, such that maximum torque occurs at or near
8 = 2. A damper which is designed for one of these purposes is poorly
suited for the other purpose.
Torque
FIG. 4. Effect of rotor-circuit resistance on torque-slip curve,
Negative-sequence impedance of a machine may be found by

putting 8 = 2 in the equivalent circuit of Fig. 2 and reducing this
circuit to a single impedance r2 + jX2 at the stator terminals. Obvi-
ously, the negative-sequence resistance T2 depends greatly upon the
rotor resistance R; (also see Table 2). If the exciting admittance is
neglected in comparison with the rotor admittance, as it can be at
s = 2, then
[14]
Calculation of damping. Although induction-motor theory has been

used above in a qualitative explanation of damping in synchronous
machines, it does not necessarily follow that the same theory is accurate
enough to be used for the calculation of damping. Induction-motor
theory assumes constant slip and a symmetrical rotor with only one
set of windings. (A symmetrical rotor is one in which there is no
difference between the direct- and quadrature-axis circuits.) A syn-
chronous machine, even though equipped with damper windings, has an
unsymmetrical rotor, with a field winding on only one axis, and the slip
of the machine varies during the disturbances considered in stability
studies. Nevertheless, induction-motor theory is considered applicable
to the calculation of negative-sequence braking. For the calculation
of positive-sequence damping, however, synchronous-machine theory
is invoked.
CALCULA'fION OF NEGATIVE-SEQUENCE BRAKING 227
Calculation of negative-sequence braking. The slip of the rotor with

respect to the negative-sequence field is very nearly 2. Its value may
be taken as 2 with little error, even though the machine is actually
swinging or starting to pull out of step, because the torque is nearly
independent of slip in this vicinity (see Fig. 3).
If the rotor circuits in the two axes are unlike (that is, if Xd" ¢ xq "
or if T dO" ~ T qO"), then the negative-sequence torque pulsates at twice
the rated frequency. This fluctuation is so rapid compared with the
frequency of electromechanical swings that the average torque, as
calculated by induction-motor theory, can be used with negligible error
in stability calculations. Byeq. 13, the mechanical power correspond-
ing to this torque is
[15]
(Since the speed is substantially constant, the torque is directly pro-
portional to the mechanical power. At normal speed, power and torque
are numerically equal if expressed in watts and synchronous watts,
respectively, or if both are expressed in per unit, as is more usual in
stability work).
In eq. 15 l r2 is the negative-sequence rotor current, and R; is the
rotor resistance. Since the negative-sequence rotor current is directly
proportional to the negative-sequence stator (armature) current 1,2,
we may write:
[16]
where K b is a constant. If exciting admittance, represented by the
shunt branch of Fig. 2, is neglected, as is permissible for the negative-
sequence circuit, then l r2 ~ 182 and also, from eq. 14, R; ~ 2 (r2 - rt),
where Tl is the positive-sequence resistance, equal to the armature
resistance R s• With these substitutions made, eq. 15 becomes:
r, ~ I s2 2 (r2 - rl) [17]
This form is convenient for computing the negative-sequence braking
power of synchronous machines because the positive- and negative-
sequence resistances of the machines are usually given rather than the
rotor resistance. The negative-sequence armature current, which is
present only during unsymmetrical faults, can be calculated by the
method of symmetrical components, as explained in Chapter VI,
Vol. I. I t can be measured on a calculating board if the negative-
sequence network is set up. An expression, such as eq. 16 or 17, giving
the torque in terms of negative-sequence current, is preferable to one
like eq. 10, giving the torque in terms of the voltage, because the former
is simpler and because the negative-sequence current of the machine
is just as easy to compute or to measure as the negative-sequence

voltage is.
The negative-sequence torque is always a retarding or braking
torque. It therefore has the same effect on the angular motion of the
rotor as a decrease of mechanical input torque. In the calculation of
swing curves, negative-sequence damping may be handled as a reduc-
tion of the input.
When damper action was neglected, the accelerating power was
taken as the difference between the mechanical power input and the
electric power output (both being internal values):
[18]
When negative-sequence braking is taken into account, another term

is included. Thus
[19)
where Pi, is the braking power. The term P u, as before, represents only
the positive-sequence power output.
The reason that negative-sequence electric power does not appear
in eq. 19 will be shown by the following "balance sheet" of power.
The mechanical power input at the shaft is consumed in the follow-
ing ways:
1. Rate of increase of kinetic energy.
2. Positive-sequence power output at terminals.
3. Negative-sequence power output at terminals.
4. Positive-sequence stator copper loss.
5. Negative-sequence stator copper loss.
6. Half of negative-sequence rotor copper loss.
7. Other half of negative-sequence rotor copper loss.
8. Rotational loss.
Item 3 is actually a negative quantity; that is, there is a negative-
sequence input at the armature terminals. This input supplies the
losses of items 5 and 6. Hence, items 3, 5, and 6 add to zero and may
be dropped from the list. Item 1 is P a . Item 2 plus item 4 is P u ,
the power supplied by the machine to the positive-sequence network,
the latter including the armature resistance. Item 7 is Pb, the braking
power. The shaft input less item 8 is Pi. Thus eq. 19 is justified.
Negative-sequence braking can be neglected except in machines
having a high-resistance damper winding. It exists, even in such
machines, only during the existenceof an unbalanced fault.
EXAMPLE 1 229
EXAMPLE 1
A hydroelectric station is delivering rated load at unity power factor and.
at normal voltage to a, receiving system which may be considered as an
infinite bus. Between the hydroelectric station and the receiving system is
a double-circuit high-voltage transmission 'line having a positive-sequence
reactance of 0.40 per unit and a zero-sequence reactance of 0.75 per unit on
0.10
O-1I~
0.40
A~
F~u't
(a)
~f-6
~~
0.30 0.10 0.40 0.10
(b)
~FIG. 5. (a) One-line diagram and (b) sequence networks of the power system
of Example 1.
a base equal to the rating of the station. (See Fig. 5a.) The transformers
at each end of the line have a reactance of 0.10 per unit on the same base.
They are connected in .1 on the low-voltage side and in Y, with solidly
grounded neutral, on the high-voltage side. The generators have the follow-
ing per-unit constants:
Xd' = 0.30 rl = 0.005
Xa = 0.20 H = 3.0
With low-resistance dampers,
T2 = 0.03
but, with high-resistance dampers,

T2 = 0.12
230 DAMPER \VINDINGS AND DAl\1PING
Calculate the initial accelerating power of the generators for each type of
damper during a two-line-to-ground fault adjacent to the high-voltage
sending-end bus, and again for a line-to-line fault. at the same place, Neglect
the effect of all resistance except the negative-sequence resistance of the.
generators,
Solution. Prefa.ult condition, Assume that the voltage behind Xd' is
constant. The initial value of this voltage is
E' = V + jxI (a)
= 1.00 + jO.9 X 1.00
= 1.35 /42.0°
Two-line-to-ground fault. The sequence networks are shown in Fig. 5b,

connected to represent a two-line-to-ground fault. The reduction of the
network is shown in Fig. 6. In Fig. 6a the identity of the negative-sequence
path through the generators is preserved. The current in this branch must
0.40 0.50
(a)
0.40 0.50 4.20
(b) (c)
FIG. 6. Reduction of the network of Fig. 5 (Example 1).
be found in order to calculate the negative-sequence braking torque. The

resistance of this branch has two values, one for each type of damper. How-
ever, as this resistance has a negligible effect on the positive-sequence power,
it is omitted in the further reduction of the fault shunt, as shown in Fig. 6b.
In Fig. 6c the T circuit has been changed into a 1r, whose architrave reactance
is 4.20 per unit. The initial positive-sequence power output of the generators
while the fault is present is, accordingly,
E'V. ~ = 1.35 X 1.00 SIn
P u = --SInu · 42 00
. (b)
X 4.20
= 0.321 X 0.669 = 0.215 per unit
EXAl\1PLE 1 231
The initial mechanical input is the same as before the fault, or 1.000 per unit.
The initial accelerating power, with negative-sequence braking neglected, is
therefore (by eq. 18):
P a = Pi- F; = 1.000- 0.215 = 0.785 per unit (c)
The negative-sequence braking torque will now be calculated for each
type of damper. The initial voltage across the fault shunt is almost inde-
pendent of the negative-sequence resistance. Let it be called V.f. Then
application of Kirchhoff's current law to Fig. 6b yields:
1.35/42.0° - VI 1.00/0 - VI VJ
jOAO + jO.50 - jO.0606 = 0 (d)
( 1+
VJ 0.40
1
0.50 + 0.0606
1) =
1.35/42.0°
oA()- +
1.00/0
0.50
21.0VI = 3.37 /42.0° + 2.00/0

= 2.00 + 2.50 + j2.26
21.0VI = vi (4.50)2 + (2.26)2 = 5.04
VI = 0.240 per unit
The negative-sequence current In the generators with high-resistance
dampers is:
0.240 0.240 ·
I2 = = - - = 0.743 per unit
+
vi (0.12)2 (0.30)2 0.323
With low-resistance dampers it is:
12 = 0.240 = 0.240 = 0.797 per unit

vi (0.03)2 + (0.30)2 0.301
The negative-sequence braking torque with high-resistance dampers is
(by eq. 17):
Ph = 1 22 (1'2 - fl) = (0.743)2(0.120 - 0.005)
= 0.063 per unit
and with low-resistance dampers:
Ps = (0.797)2(0.030 - 0.005) = 0.016 per unit
The accelerating power with high-resistance dampers is (byeq. 19):
P« = (Pi - P u ) - Pb = 0.785 - 0.063 = 0.722 per unit
With low-resistance dampers, it is:
P a = 0.785 - 0.016 = 0.769 per unit
Line-to-line fault. The connections of the sequence networks for repre-
senting this type of fault are as shown in Fig. 5, except that the zero-sequence
network is omitted. The reduction of the networks proceeds as in Fig. 6,

except that the shunt reactance shown as 0.076 per unit in Fig. 6a becomes
0.50 per unit; that shown as 0.0606, Fig. 6b, becomes 0.187 with resistance
neglected; and the architrave reactance shown as 4.20 per unit, Fig. 6c,
becomes 1.97 per unit. The last substitution, when made in eq. b, gives the
positive-sequence electrical output as 1.35 X 0.669/1.97 = 0.458 per unit.
Therefore the accelerating power with dampers neglected is 1.000 - 0.458 =
0.542 per unit. The second substitution, made in eq. d, yields:
9.8:3 V I = 5.04
VI = 0.513 per unit
With high-resistance dampers,
0.513 .
12 = - - = 1.59 per unit
0.323
P b = (1.59)2 X 0.115 = 0.290 per unit
Pa = 0.542 - 0.290 = 0.252 per unit
With low-resistance dampers
0.513
12 = - - = 1.70 per unit·
0.301
Pb = (1.70)2 X 0.025 = 0.073 per unit
P a = 0.542 - 0.073 = 0.469 per unit

A summary of results is given in Table 5. From these results we may con-
clude that the high-resistance dampers are more effective than the low-
TABLE 5
SUMMARY OF RESULTS, EXAMPLE 1
NEGATIVE-SF;QUENCE BRAKING POWER
Accelerating Power in Per Unit

A
Type of Fault Low- High-

Dampers
Resistance Resistance
Neglected
Dampers Dampers
Two-line-to-ground 0.785 0.769 0.722
Line-to-line 0.542 0.469 0.252
resistance dampers in producing negative-sequence braking and that such

braking is present to a much greater extent during a line-to-line fault than
during a two-line-to-ground fault. However, as faults not involving ground
are of infrequent occurrence on steel-tower, high-voltage transmission lines,
and as two-line-to-ground faults are usually assumed in calculations of sta-
CALCULATION OF D-C BRAI{ING 233
bility limits, not much advantage can be taken of the effectiveness of high-
resistance dampers on line-to-line faults. The use of high-speed switching,
moreover, reduces the benefits of dampers in producing negative-sequence
braking.
Calculation of d-c. braking. For three-phase faults at or very near

the armature terminals of a machine, the d-e, components of short-
circuit armature current persist long enough to have an appreciable
effect. The d-e, components of armature current induce in the rotor
circuits currents of fundamental frequency, which give rise to a braking
torque similar to that caused by negative-sequence armature currents.
The d-e. braking torque (or power), like the negative-sequence braking
torque, decreases the accelerating torque, and may be considered as
all or part of Pi, in eq. 19.
Although the rotor currents induced by direct armature currents are
of fundamental frequency, whereas those induced by negative-sequence
armature currents are of twice fundamental frequency, nevertheless
for a given magnitude of armature current the magnitudes of the
various rotor currents are substantially the same in the two cases,
being determined mainly by the reactances. There is, however, this
important difference: that whereas in the negative-sequence case half
the rotor copper loss was supplied mechanically and therefore produced
braking torque, in the d-e. case all the rotor copper loss is supplied
mechanically. This loss, expressed in per unit, is I r 2R r , or twice the
value given by eq. 15. Using the same value of R; as before, we get
for the d-e, braking power twice the value given by eq. 17, namely:
Ps ~ 218 2(r2 - r1) [20]
Here 18 is the effective value of the armature currents, which we can
imagine to be polyphase currents of very low frequency, whose instan-
taneous values equal the direct armature currents. The crest' value of
these polyphase currents corresponds to the total direct current, that is,
to the value of d-e, component that would exist in one phase of the
armature if the switching angle were such as to give this component its
maximum value. As stated in Chapter XII, for an unloaded machine
the total direct current is initially equal to the initial crest value of the
a-c. component of armature short-circuit current and decays exponen-
tially with time constant T a. If the total instantaneous d-e, component
of armature current is denoted by ide, the expression for d-e, braking
power becomes
Pb ~ ide2(r2 - r1) [21]
Note that as ide decreases exponentially P b also decreases.
For types of fault other than three-phase and for faults not very
near the armature terminals, ide decays so rapidly that its braking
effect is negligible.
Calculation of positive-sequence damping. The slip of the rotor

with respect to the positive-sequence armature voltage is small and
varies both in magnitude and sign. Equation 10 and Fig. 4 show that
at small slips the torque of an induction machine is almost directly
proportional to the slip. In the practical calculation of positive-
sequence damping, therefore, the damping torque - and likewise the
damping power - are assumed to be proportional to the slip. However,
because of the difference between the direct and quadrature axes, the
constant of proportionality depends upon the angular position of the
rotor. These two features appear in an equation, derived by Park'?
and quoted in modified form by Dahl, l for the damping power of a
synchronous machine connected to an infinite bus through series
reactance. In the derivation, the following assumptions were made:
1. No resistance in armature circuit.
2. No resistance in field circuit.
3. Small slip.
4. Damping action caused by only one set of rotor windings
(the damper windings).
The equation for damping power is:

'
P D -_ E 2SW [(Xd ( - Xd ")T"
dO
')2
• 2 ~
sin
+ (X q'( - X q")T"
qO
')2
2
cos
~] [22]
Xe + Xd u
z, + Xq U
where PD = instantaneous damping power in per unit.

E = voltage of infinite bus in per unit.
s = slip in per unit (see discussion in next paragraph).
w = 21r X rated frequency.
Xd', Xd", x q ' , x q " = machine reactances in per unit, as previously
defined (Chapter XII).
~te = external reactance in per unit in series with armature.
T dO'" T qo/ = open-circuit subtransient time constants in the direct
and quadrature axes, respectively, both expressed in
seconds.
a = angle by which the machine leads the angle of the in-
finite bus.
In stability calculations it is preferable to take P D and s as positive

when there is induction generator action, even though to do so is .con-
CALCULATION OF POSITIVE-SEQUENCE DAMPING 235
trary to the usual conventions of induction-motor theory. Then,

in eq. 22,
1. do
s = 360j' dt [23]
where f is the frequency of the infinite bus in cycles per second and
do/ dt is in electrical degrees per second.
Derivation of equatio·n. The rigorous derivation of eq. 22 is too long
and complicated to be given here, but a derivation based on induction-
machine theory and leading to the same result will be presented. It
must first be assumed that the rotor is symmetrical (that is, alike on
both axes) and that the field windings are closed but not excited. *
FIG. 7. Direct-axis equivalent circuit of synchronous machine, for calculation

of positive-sequence damping. The quadrature-axis circuit is similar.
In order to apply induction-machine theory, we must next establish

an equivalent circuit similar to Fig. 2 but with its constants expressed
in terms of the usual synchronous-machine reactances and time con-
stants. The three sets of windings (armature, field, and damper) are
inductively coupled, somewhat like the windings of a three-circuit
transformer. However, since we wish to find the damper current I,
due to the armature voltage E, we can set up an equivalent T circuit
in which the identity of the armature and damper circuits only is
preserved, while the identity of the field circuit is lost, even though
the presence of the field circuit affects the impedance of the common
branch of the T, much as the eddy-current paths do in the equivalent
T circuit of a two-winding transformer or machine. Such a circuit is
shown in Fig. 7. As discussed in Chapter XII, the reactance of either
series arm ("leakage reactance") can be made zero by proper choice of
the arbitrary ratio of transformation. Let the armature leakage re-
actance of the machine itself be set equal to zero in this manner, leaving
only the external reactance x, in the series arm. The impedance seen
from the armature terminals, with the field winding short-circuited
and the damper winding open, is the transient reactance Xd'. Conse-
"It can be shown that the excitation does not affect the damping torque if the
armature resistance is negligible.
quently, the shunt arm of the equivalent T circuit is Xd'. By analogy

with the equivalent circuit of the induction motor, the damper branch
of the equivalent circuit has a resistance Rrdls, where Rrd is the damper
resistance (referred to the armature) and s is the slip. At large slips
Rrdls is negligibly small, and the impedance seen from the armature
terminals is the subtransient reactance Xd". Therefore, the damper
leakage reactance Xrd must have a value such that, in parallel with Xd',
It
• gives
• a reactance Xd.
II Th us,
/
" XrdXd
= -----, [24]
+
Xd
Xrd Xd
whence
[25]
N ow that the equivalent circuit has been established, expressions

will be found for the stator and rotor currents. The former is
. E
[26]
f. =. + (Rrd/s + jXrd) (jXd')
+(J Xrd + Xd')
JX e R rd /S .
which, at small slip, becomes

E
Is ~ -- , [27]
j(x e +Xd)
The latter is
.
JXd,
,
I r = Is .
R rd / 8 + J(Xrd + »« ) I
. ,
__ I JXd [28]
-- 8 Rrd/ s
which, upon substitution of eq. 27, becomes

,-- EXd'
Ir [29]
+ Xd
,-- ,
(z, )R r d / 8
The damping power (byeq. 5) is
. _ I 2R 1 - 8 _ E 2 (xd' )2(1 - 8)8
PD r rd - [30]
(z, + Xd ) Rrd
- I 2
S
The factor (1 - 8) may be discarded because, by assumption, 8 is very

small. The damper resistance Rrd will be replaced by an expression
involving the direct-axis subtransient open-circuit time constant T dO'/.
Since the resistance of the field circuit is neglected, this time constant
CA·LCULATION OF POSITIVE-SEQUENCE DAMPING 237
is the ratio of the equivalent inductance of the damper winding to the

resistance of the damper winding, the equivalent inductance being
determined with the field winding closed and the armature winding
open (because we want the open-circuit time constant). According to
the equivalent circuit (Fig. 7), the reactance corresponding to the
equivalent inductance of the damper is Xrd + Xd'. Byeqs. 24 and 25,
it may be written also as
[31]
Hence the time constant is

" L X (Xa,')2
Ta,o = --- = ---- = , II [32]
Rra, WRrd W(Xd - »« )Rrd
from which
Rra, = w(xa,' - [33]

Xd") T dO"
By use of this expression for Rrd, eq. 30 becomes:

_ E 2 ( Xd I - Xd " ) W T dO" S
PD [34]
-
(X e + Xd12)
If the rotor is unsymmetrical, the damping power fluctuates at
twice the slip frequency between the value given by eq. 34 and that
given by a similar equation containing the corresponding quadrature-
axis constants. The resultant fluctuating damping power is found by
substituting for E in eq. 34 and in the corresponding quadrature-axis
equation its direct- and quadrature-axis components, namely:
Ed = E sin 0 [35a]
Eq = E cos 0 [35b]
Thus, the direct-axis damping power is:
_ E 2 sin 2 0 (Xd' - Xd")wT dO" S
P Dd - (Xe+ Xd ')2 [36a]
The quadrature-axis damping power is:

P _ E 2 cos2 ~ (
X q' - X q") W T qOII S
+ X q')2
(J
[36b]
Dq - (z,
The total damping power is the sum of eqs. 36, which is eq. 22.
Effect of armature resistance. In large machines with effective damper
windings, the effect of armature resistance is usually negligible. When
it is not negligible, its effect is to decrease the damping torque,

and in extreme cases the net damping may be negative, leading to
sustained hunting. Approximate equations for damping torque,
taking both saliency and armature resistance into account, are given
by Liwschitz.P
Extension to multimachine system. Although eq. 22 was derived for
one machine connected to an infinite bus, its use can be extended
first to a system of two finite machines and then to a multimachine
system.
First consider two finite machines. In calculating the damping
power of one machine, the other machine can be represented (at
least approximately) by a constant e.m.f, in series with a constant
reactance. According to Thevenin's theorem, the latter machine and
the intervening network, viewed from the terminals of the former
machine, can be replaced by an equivalent e.m.f. and impedance in
series. The equivalent e.m.f. then takes the place of the voltage E
of the infinite bus in eq. 22, while the equivalent impedance (in which
the resistance is usually small compared with the reactance) takes
the place of the external reactance X e• The damping power of the
other machine can be computed in similar fashion, using new values
of E and X e•
In a multimachine system a similar procedure can be followed. In
calculating the damping of anyone machine, the remainder of the
system, viewed from the terminals of that machine, is replaced by
an equivalent e.m.f. and impedance, according to Thevenin's theorem,
and these are used as E and X e, respectively, in eq. 22. The equivalent
e.m.f. and impedance may be found on an a-c. calculating board, as
follows: The machine whose damping is being calculated is discon-
nected from the rest of the system, and the open-circuit voltage of
the system is measured in both magnitude and phase. It is the equiva-
lent e.m.f. E. The equivalent impedance can be found by dividing E
by the short-circuit current of the system (not including the machine
in question).
In the course of a point-by-point determination of swing curves,
E is read at each step. It varies both in phase and in magnitude.
The slip s is calculated by eq. 23, taking 0 as the difference of angles
of the machine in question and of E, and replacing do/dt by ~o/ ~t.
In order that dO/ dt may represent the slip at the particular instant
of time for which P D is being calculated, .10 should be taken as the
average for the intervals immediately preceding and immediately
following the instant. However, as ~o for the latter interval depends
upon P o, and therefore cannot be calculated until after P n has been
CALCULATION OF POSITIVE-SEQUENCE DAMPING 239
calculated, it must be estimated by extrapolation. Then, if the

estimated value of ~o does not agree with the value calculated from
the accelerating power and inertia constant, successive approximations
can be made until satisfactory agreement is reached.
The calculation of damping power in a multimachine system is
probably less rigorous .than in a two-machine system because in the
multimachine system the magnitude of E may vary, because of the
relative swinging of the several machines comprising the remainder
of the system. This variation of E is a departure from the idea of an
infinite bus, as assumed in the derivation of eq. 22.
Accelerating power. After the damping power P D has been com-
puted, as just described, it is taken into account as an additional
component of positive-sequence electric output power: hence it
diminishes the accelerating power. The expression for accelerating
power (replacing eqs. 18 and 19) is:
[37]
where, as before, Pi is the mechanical power input; Pi, the negative-

sequence braking power; P u , the positive-sequence synchronous
power output; and, in addition, P D is the positive-sequence damping
power output.
Flow of damping power in the positive-sequence network. When
damping occurs, the positive-sequence power output of the machine
is the algebraic sum of the damping power and the synchronous power
(calculated as though the machine had no damper windings). If the
former is much smaller than the latter, it maybe sufficiently accurate
to solve the positive-sequence network as though only the synchronous
components of power existed. Then cognizance of the damping power
would be taken only in the calculation of accelerating power (eq. 37).
From the values of accelerating power of every machine of the
system, the new angular positions of the machines would be calcu-
lated, and the network again solved taking only synchronous power
into account.
If, however, the damping power should be as great as 5% or 10%
of the synchronous power, it might affect conditions in the net-
work sufficiently to warrant taking it into account in the network
solution.
The circuit of Fig. 7 gives a clue as to how damping power can be
represented on an a-c. calculating board if saliency is neglected. To
the usual representation of a synchronous machine, consisting of the
transient reactance in series with the voltage behind transient re-
240 DAMPER WINDINGS AND DAl\IPING
actance, there is added a parallel branch. When the machine is

running slow, the added branch may consist merely of a resistance
Rrd/S, adjusted to absorb an amount of power equal to the calculated
negative (accelerating) damping power.
However, when the machine is running
fast, a source of power is required to repre-
sent the damping power; and the only
practical source is one of the phase shifters
ordinarily used to represent generators.
Then two phase shifters are used to rep-
resent one synchronous machine, as shown
in Fig. 8. The phase shifter that supplies
FIG. 8. Representation of a
(or absorbs) synchronous poweris adjusted,
synchronous machine by two as usual, to have the proper voltage (usually
phase shifters: one to supply constant) behind transient reactance at the
or absorb synchronous power proper angular position 0, as calculated for
P tt, as usual; the other to sup- each step. The phase shifter that supplies
ply or absorb damping power
PD. Wattmeter WI reads
(or absorbs) damping power is adjusted
r; W2 reads PD. to have a power output equal to the cal-
culated damping power and to have unity
power factor, or no reactive power. These adjustments are readily
made with the aid of a wattmeter-varmeter. Unity power factor is
assumed because, in Fig. 7, Rrd/S » Xrd.
EXAMPLE 2
A 60-cycle salient-pole synchronous generator having the following
constants:
Reactances in per unit
Xd = 1.15 Xd' = 0.37 x(/' = 0.24
x q = 0.75 x q' = 0.75 x q" = 0.34
Time constants in seconds
T dO' = 5.0 T d o" = 0.035 T qo" = 0.035
is delivering a current of 1.00 per unit at a power factor of 0.91, lagging, to
an infinite bus having a voltage of 1.00 per unit. A three-phase short circuit
occurring at the terminals of the generator is cleared in 0.20 sec. without
disconnecting the generator from the bus. Calculate and plot a swing curve,
taking into account both field decrement and damping. Compare the curve
with those obtained in Example 8, Chapter XII, in which damping was
neglected.
NOTE: The generator constants, the prefault conditions, the fault, and the
time of clearing are the same as in Example 8.
EXAMPLE 2 241
Solution. There is no damping during the fault. Consequently, when the

fault is cleared, 0.20 sec. after its occurrence, conditions are the same as in
Example 8 with field decrement taken into account; namely,
o = 106.0°
E q ' = 1.04 per unit
The swing curve after 0.20 sec. is calculated, as before, from the relation
(eq. e of Example 8):
(a)
where now, taking damping into account,
Pa = Pi - Pu - PD = 0.91 - (P u + PD) (b)
The synchronous power is calculated as before, by eq, g of Example 8:
F; = 2.70E q' sin 0 - 0.685 sin 20 (c)
Herein E q ' may be assumed to be the same as it was in Example 8, Table 12,
since it was found to decrease at a fairly constant rate (Fig. 44, Chapter XII),
which will not be affected by the damping. The damping power is calculated
from eqs. 22 and 23:
P D= E2X21rf~o [(Xd' -Xd'.') T " SIn

dO · 2 ~+ (xq'-xq")T qo" 2 ~J
u cos u
360f~t (X,+Xd') 2 (x,+X q ' ) 2
(1.00)2X21r~O [(0.37-0.24)0.035 · ~+ (0.75-0.34)0.035 ~J
= 2
SIn u
2
cos u
360XO.05 (0+0.37)2 (0+0.75)2
= (0.0116 sin" 0 + 0.0089 cos 2
o)do
= (0.0102 + 0.0014 cos 2o)~o (d)
Since the fluctuation of damping power with 0 is small, it will be sufficiently

accurate to use the average value,
PD = 0.0102~o (e)
taking ~o as the average value for the time intervals immediately preceding
and following the instant for which P D and P a are to be computed.
The swing calculations are carried out in Table 6.
The swing curve is plotted as curve c in Fig. 9. Curves a and b are plotted
from the results of Example 8, Chapter XII, in which damping is neglected.
Comparison of the three curves shows that the beneficial effect of damping
almost compensates for the detrimental effect of field decrement even though
the fault lasts for more than two-thirds of the time required for the machine
to swing to its maximum angular displacement from the bus. Consequently,
the assumptions of constant flux linkage and negligible damping lead to but
little error. If the fault had been cleared sooner, damping probably would
have been more effective in reducing the amplitude of swing.
~
~
TABLE 6 t:1
COMPUTATION OF SWING CURVE AFTER CLEARING OF FAULT- >-
DAMPING AND FIELD DECREMENT TAKEN INTO ACCOUNT (EXAMPLE 2)
s=
~
~
E' 2.70Eq ' -0.685 Pu ~a PD Pa, ~2a ~a a
t q
sin a sin 20 sin 2a (p.u.) (est.) (p.u.) (p.u.) (deg.) (deg.) (deg.) ~
(sec.) (p.u.) sin a ...-c
Z
0.20- 1.04 0 0 0.91 34.4 tj
~
0.20+ 1.04 0.961 -0.530 2.70 +0.36 3.06 30.8 0.31 -2.46 Z
0.20 avg. 1.04 -0.78 - 8.4 106.0 o00
26.0
0.743 -0.994 2.04 +0.68 2.72 15.2 0.15 -1.96 -21.2 132.0
>-
0.25 1.02 Z
4.8 e
0.30 0.99 0.686 -0.998 1.83 +0.68 2.51 - 4.0 -0.04 -1.56 -16.8 136.8 e
-12.0 >-
0.96 0.823 -0.935 2.13 +0.64 2.77 -21.0 -0.21 -1.65 -17.8 124.8 s=
0.35 ~
~
-29.8 Z
0.40 95.0 o
i
POWER-ANGLE CURVE \VITH DAMPING INCLUDED 243
140
130
i
j 120
~
'0
i
~ 110
.!!
i
~
100
Q3 OA
Time (seconds) afteroccurrence of fault
FIG. 9. Effect of damping upon swing curves (Example 2). (a) Damping and
field decrement neglected (constant flux linkage). (b) Damping neglected, but
decrement taken into account. (c) Damping and decrernent taken into account.
Power-angle curve with damping included. The electric power out-

put of a synchronous machine consists, as has already been pointed
out, of two components: synchronous power P u, which is a function
of the angular position of the machine relative to the other machines,
and induction-generator or damping power P D, which is a function
of the relative angular velocity. The ordinary power-angle curve of
a two-machine system is a curve of P u versus angular difference a
and is a single-valued function of a. The operating point moves along
this curve, its position being a function of time, and, if the motion is
oscillatory, the two extreme points, reached in each oscillation, are so
placed that the net area between the curve of P u and the horizontal
line Pi is zero. The effect of damping may he shown in the power-
angle curve by adding P D to the ordinates, thus showing the total
electric power output P u + P D as a function of a. The new curve
lies above the old one when a is increasing and below it when a is
decreasing, and forms a spiral of approximately elliptical form, as
shown in Fig. 10. Each successive oscillation is smaller, and the

curve converges on the point of equilibrium, which is the intersection
of the curve of P u and the horizontal line of Pi. The principle of
equal areas still applies: the net area between the curve of P u + P D
and the line Pi is zero at the extreme values of 0, where the velocity
is zero - points B, D, G, J, and M in Fig. 10. Area ABC = area CDE;
area DEF = area FGH; area GHI = area IJK; and area JKL = area
LMN.
Angle. 6
FIG. 10. Power-angle curve with damping included . (Plotted from data of
Ref. ll. The marked points are for equal intervals of time.)
Heating effect of rotor currents. 20 - 24 When a synchronous machine

is subjected to an unbalanced short circuit, negative-sequence currents
are induced in the rotor. Because of the high frequency of these
currents, they are confined to paths near the surface of the rotor,
such as the damper windings and pole faces of a salient-pole machine
or the slot wedges of a round-rotor machine, with the retaining rings
acting as end connections between wedges. These paths have a
relatively high resistance and much heat is developed in them. If the
currents are sustained for more than a few seconds, the heat may
cause damage: the slot wedges of turbogenerators may be heated to a
temperature where they lose their mechanical strength to resist the
REFERENCES 245
high centrifugal forces and fail in shear, or the retaining rings may
expand from heat and part contact with the rotor body, thus giving
rise to arcing at the shrink fits between rings and rotor body. Experi-
ence has shown that the machine will not be injured if f 12 2dt does
not exceed 30, 12 being the negative-sequence current in per-unit
and t, the time in seconds.
REFERENCES
1. O. G. C. DAHL, Electric Power Circuits: Theory and Applications, Vol. II,
"Power System Stability," McGraw-Hill Book Co., Inc., New York, 1938. Chap.
XIX, "Damper Windings and Their Effect."
Engineers of the Westinghouse Electric & Manufacturing Co., East Pittsburgh,
Pa., 1st edition, 1942, pp. 141-5, 160-3, 200-1.
3. C. F. "TAGNER, "Damper Windings for Water-Wheel Generators," A.I.E.E.
Trtms., vol. 50, pp. 140-51, March, 1931. Disc., pp. 151-2.
4. C. F. \VAGNER, "Unsymmetrical Short Circuits on Water-Wheel Generators
Under Capacitive Loading," Elec. Eng., vol. 56, pp. 1385-95, November, 1937.
5. C. F. WAGNER, "Overvoltages on Water-Wheel Generators," Elec. Jour.,
vol. 35, pp. 321-5 and 351-4, August and September, 1938.
6. EDITH CLARKE, C. N. WEYGANDT, and C. CONCORDIA, "Overvoltages Caused
by Unbalanced Short Circuits - Effect of Amortisseur Windings," A.I.E.E.
Trans., vol. 57, pp. 453-66, August, 1938. Disc., pp. 466-8.
7. R. B. GEORGE and B. B. BESSESEN, "Generator Damper Windings at Wilson
Dam," A.I.E.E. Trons., vol. 58, pp.166-71, April, 1939. Disc., pp. 171-2. Instal-
lation of dampers on generator which had been in service 12 years without dampers
in order to decrease harmonic overvoltages during unsymmetrical short circuits
with capacitive loading.
8. C. A. NICKLE and C. A. PIERCE, u Stability of Synchronous Machines: Effect
of Armature Circuit Resistance," A .1.E.E. Trans., vol. 49, pp. 338-50, January,
1930. Disc., pp. 350-1.
9.. C. F. WAGNER, "Effect of Armature Resistance Upon Hunting of Synchronous
Machines," A .1.E.E. Trans., vol. 49, pp. 1011-24, July, 1930. Disc., pp. 1024-6.
10. R. H. PARK, "Two-Reaction Theory of Synchronous Machines," Part I,
A.I.E.E. Trtms., vol. 48, pp. 716-27, July, 1929. Disc., pp. 727-30. Part II,
A.l.E.E. Trans., vol. 52, pp. 352-4, June, 1933. Disc., pp. 354-5.
11. S. B. CRARY and M. L. 'VARING, "Torque-Angle Characteristics of Synchro-
nous Machines Following System Disturbances," A.I.E.E. Trans., vol. 51, pp. 764-
73, September, 1932. Disc., pp. 773-4.
12. F. A. HAMILTON, JR., "Field Tests to Determine the Damping Character-
istics of Synchronous Generators," A.I.E.E. Trans., vol. 51, pp. 775-9, September,
1932.
13. M. M. LIWSCHITZ, "Positive and Negative Damping in Synchronous Ma-
chines," A.I.E.E. Trans., vol. 60, pp. 210-3, May, 1941.
14. ROGER COE and H. R. STEWART, "Generator Stability Features Fifteen
Mile Falls Development," Elec. Jour., vol. 28, pp. 139-43, March, 1931.
15. C. CONCORDIA and G. K. CARTER, "Negative Damping of Electrical Machin-
ery," A.l.E.E. Trans., vol. 60, pp. 116-9, March, 1941. Disc., p. 752.
16. WILLIAM L. RINGLAND, "Damper Windings for Waterwheel and Diesel

Engine Driven Generators," Proc. Midwest Power Conference (Chicago), vol. 8,
pp. 89-94, 1946.
17. LEEA. KILGORE and EUGENE C. WHITNEY, "Spring and Damping Coeffi-
cients of Synchronous Machines and Their Application," A.I.E.E. Trans., vol. 69,
part I, pp. 226-9, 1950. Disc., pp. 229-30.
18. CHARLES CONCORDIA, "Synchronous Machine Damping Torque at Low
Speeds," A.I.E.E. Trans., vol. 69, part II, pp. 1550-3, 1950.
19. CHARLES CONCORDIA, "Synchronous Machine Damping and Synchronizing
Torques," A.I.E.E. Trans., vol. 70, part I, pp. 731-7, 1951.
20. A. WUST and J. DISPAUX, "The Operation of Turbo-Alternators under
Unbalanced or Out-of-Step Conditions," C.I.G.R.E., 1952, Report 106.
21. R. F. LA'VRENCE and R. W. FERGUSON, "Generator Negative-Sequence
Currents for Line-to-Line Faults," A.l.E.E. Trans., vol. 72, part III, pp. 9-16,
February, 1953.
22. M. D. Ross and E. I. KING, "Turbine-Generator Rotor Heating During
Single-Phase Short Circuits," A.I.E.E. Trans., vol. 72, part III, pp. 40-5, February,
1953.
23. P. L. ALGER, R. F. FRANKLIN, C. E. KILBOURNE, and J. B. MCCLURE,
"Short-Circuit Capabilities of Synchronous Machines for Unbalanced Faults,"
A .I.E.E. Trans., vol. 72, part III, pp. 394-403, June, 1953. Disc., pp. 403-4.
24. E. I. POLLARD, "Effects of Negative-Sequence Currents on Turbine-Generator
Rotors," A.l.E.E. Trans., vol. 72, part III, pp. 404-6, June, 1953. Disc., p. 406.
25. TH. LAIBLE, "The Effect of Solid Poles and of Different Forms of Amortisseur
on the Characteristics of Salient-Pole Alternators," C.l.G.R.E., 1950, Report 111.
PROBLEMS ON CHAPTER XIV

1. Work Example 1 for a line-to-ground fault.
2. In Example 1 find the critical clearing time of a two-line-to-ground
fault with and without negative-sequence braking of the generator due to
high-resistance dampers.
3. Find the transient stability limit of the system of Example 1 with and
without negative-sequence braking if a two-line-to-ground fault near the
sending end is cleared in 0.1 sec.
4. In Example 1 find the value of r2 that gives maximum negative-
sequence braking power during (a) a two-line-to-ground fault at the sending
end of the transmission line and (b) a line-to-line fault at the same location.
Can the accelerating power be made zero in either case? Assume that
X2 is not affected by varying r2.
5. Find the initial value in per unit of the d-e. braking torque produced
in a typical turbogenerator due to a three-phase short-circuit on the arma-
ture terminals. Also find the torque 0.05 sec. later.
6. Work Example 2 taking into account the variation of positive-sequence
damping due to saliency.
7. Work Example 2 neglecting saliency and field decrement. Carry out
the swing curve for several cycles of swing and find the percentage decrease
in amplitude of swing per cycle.
CHAPTER XV
STEADY-STATE STABILITY
Preceding chapters of this book have been devoted almost entirely

to transient stability, which, as will be recalled, is the ability of the
synchronous machines of a power system to remain in synchronism
after a specified sudden, severe, unrepeated shock (usually a short
circuit). This chapter discusses steady-state stability, which is the
ability of synchronous machines to remain in synchronism after a very
minute disturbance. Small disturbances are always present in a
power system; for example, there are gradual changes in load, manual
or automatic changes of excitation, irregularities in prime-mover
input, and so forth. Obviously these small disturbances cannot cause
loss of synchronism unless the system is operating at, or very near, its
steady-state stability limit. This limit is the greatest power that can be
transmitted on a specified circuit, under specified operating conditions
in the steady state, without loss of synchronism. The steady-state
stability limit 'of a circuit is greater than the transient stability limit
of the same circuit under the same operating conditions.
Importance of steady-state stability. A knowledge of steady-state

stability limits is important for the following reasons:
1. A system can be operated above its transient stability limit but not
above its steady-state limit. When so operated it is stable as long as
no fault occurs, or even if a fault does occur provided the fault ·is
sufficiently less severe than that for which the transient limit was
calculated. Since most faults on long overhead transmission lines
are caused by lightning, it might well be considered prudent to
operate such a line above its transient stability limit during fair
weather and to reduce the load below the transient limit before the
onset of a thunderstorm. Moreover, lines can be built that are
virtually lightningproof, * and such lines can be operated prudently
above their transient limits even during storms.
*While no line is completely immune to lightning, the severity of the lightning
stroke (as measured by its current) required to flash over the line insulation depends
upon such features of the line design as shielding due to ground wires, tower-footing
resistance, spacing of line and ground conductors, and flashover voltage of line
insulators. As the severity of lightning strokes varies greatly, the probability of
flashover can be greatly reduced by improvement in these features of design.
247
248 STEADY-STATE STABILITY
Another instance of circuits operated above their transient

stability limits is the following: Some generating stations feeding
into metropolitan areas over long transmission lines were formerly
operated sectionalized, each generator, or group of generators,
being connected to the load area over a separate transmission
circuit. 6 In the event of a fault on any main circuit, that circuit
and the associated generators were disconnected. As automatic
reclosing was not used, the transient stability limit was zero. This
mode of operation was dictated by slow fault clearing, which would
have caused loss of synchronism between the metropolitan system
and the entire remote station if the latter were not sectionalized.
The advent of faster relays and breakers has rendered this type of
operation unnecessary and undesirable. (See Study 3, Chapter
VII, Vol. I.)
In the several cases just described where lines are operated above
the transient limits for a fault on the main line, the practical stability
limit is largely determined by faults on other, more remote, perhaps
lower voltage circuits which, in the aggregate, are faulted much
more frequently than the main circuits. In many cases, however,
the limits so determined are fairly close to the steady-state limits of
the main circuits and may be approximated by subtracting a reason-
able margin from the steady-state limit.
Tie lines interconnecting two large power systems are often
designed for a normal power transfer which is small compared with
the capacity of the systems. Unless the power flow on such lines
is carefully controlled, steady-state pullout is likely to occur. Auto-
matic control for tie-line loads is available and is widely used. This
control utilizes impulses transmitted over communication circuits
to the governors of the prime movers.
2. The transient-stability limit can be raised by increasing the speed
of clearing faults but cannot exceed the steady-state stability limit of
the final circuit (that is, with the faulted circuit disconnected unless
high-speed reclosing is employed). The great improvements which
have been made in relaying and circuit breaking (discussed in Vol.
II) have already greatly raised the transient stability limits of many
power systems and are available for raising those of others. Further
improvements, particularly in the speeds of circuit breakers, may
confidently be expected. Nevertheless, the steady-state stability
limit of any system represents a "ceiling" which cannot be exceeded
through improvements in relays and breakers but which is fixed
primarily by the impedances of lines, generators, and other apparatus.
When very short clearing times are used, the transient stability
STEADY-STATE STABILITY LIl\fIT 249
limit, calculated on the basis of constant field flux (constant voltage

behind transient reactance), may even exceed the steady-state
limit with the faulted circuit switched out, calculated on the basis
of constant field current (constant voltage behind saturated syn-
chronous reactance) of value determined by the prefault condition.
This is possible because the transient reactance of a synchronous
machine is considerably smaller than its saturated synchronous
reactance (discussed in a later section of this chapter). In such a
case the system would actually be unstable - though probably
not on the first swing - unless voltage regulators were used to
prevent the transient field current from decaying too much. If, on
the other hand, the calculated transient limit were below the steady-
state limit, it would be concluded that the system was stable. This
conclusion rests on the supposition that the amplitude of swings
is decreased by damping fast enough to make up for the field decre-
ment. The use of voltage regulators gives assurance of stability.
When fast clearing is used (say 6 cycles or under) it is advisable
to check the steady-state stability limits as well as the transient
limits, lest an erroneous conclusion be reached on the basis of the
latter alone. The same reasoning applies to the switching out of an
unfaulted line.
Historically, steady-state stability was studied first. Then, as it was
recognized that short circuits were the most severe hazard to stable
operation, attention was directed to transient stability and its improve-
ment. More recently, progress of the art of power-system protection
has again directed attention to steady-state stability, though not to
the disregard of transient stability. As mentioned above, the principal
factors which again direct attention to steady-state stability are:
(1) increased fault-clearing speeds, which make the transient limits
closely approach the steady-state limits, and (2) increased reliability
of transmission lines and other equipment, which makes faults less
frequent and therefore makes it less dangerous to risk instability in
the event of a fault. To this list might be added (3) improved methods
of system operation and out-of-step relaying, which make the dis-
turbance due-to loss of synchronism less severe than formerly.
Transmission lines not exceeding 200 miles in length usually have
rated loads well below the steady-state stability limit. At lengths of
300 miles or more steady-state stability may be a limiting factor. 16
The steady-state stability limit, as defined above, is the maximum
power that can be transmitted in the steady state without loss of
synchronism. This definition needs to be made more precise.
In the first place, where should the power be measured? If a line

has any loss, the maximum power at the sending end of the line is
greater than the maximum power at the receiving end and occurs at a
different angular separation. From a practical standpoint, however,
the received power is more important than the sent power and, therefore,
is to be understood even though not stated explicitly. There is no
advantage in sending more power into the line if so doing does not
increase the received power but only increases the losses.
In the second place, what is the nature of the disturbance that
ultimately causes loss of synchronism if the stability limit is exceeded?
The disturbance is taken as a slow, sustained increase in load. The
rate of increase of load should be slow compared with the natural
frequencies of mechanical oscillation of the machines and also compared
with the rate of change of field flux in the machines. As a result, the
slow change in load is accompanied by a slow adjustment of the angular
displacements of the machine rotors, without oscillation, but there
is no change in the field currents.
In the third place, what cognizance should be taken of the manual
or automatic control of frequency and voltage? Any increase of load
changes the system frequency and decreases the terminal voltages of
the machines. After each small increment of load the generator inputs
are assumed to be adjusted so as to maintain constant frequency and
the field currents are,assumed to be increased so as to maintain constant
terminal voltages. * It is important to note that the increase of field
currents occurs after the increase of load, as it would with manual
control of the voltage. If the increase of field current should occur
simultaneously with the increase of load, as it could if suitable voltage
regulators were used, the stability limit would be increased signifi-
cantly. The' limit under this condition is called the dynamic stability
limit or the steady-state stability limit with automatic devices; this is
discussed briefly in Chapter XIII and also later in:the present chapter.
Unless otherwise specified, the steady-state stability limit is under-
stood to be the static limit, that is, the limit with hand control.
To summarize the conditions just discussed: The steady-state stability
limit of a particular circuit of a power system may be defined as the
maximum power at the receiving end of the circuit that can be transmitted
without loss of synchronism if the load is increased in very small steps
"In some cases other operating conditions may be specified than constant terminal
voltage; for example, constant voltage at a point other than the machine terminals,
or terminal voltage varying with load in a specified manner, or constant field current
during large changes in load. The important point is that the operating conditions
of the power system assumed in the analysis should agree with those that are likely
to be used in actual operation of the system.
TWO-MACHINE SYSTEM WITH NEGLIGIBLE· LOSSES 251
and if the field currents are changed after each increment of load so as to
restore the normal operating conditions (usually constant terminal voltages).
This definition suffices for a two-machine system. For a multi-
machine system, additional stipulations are necessary as to how the
power is to be supplied and absorbed. Such stipulations are not
included in any general definition but are decided upon in each individ-
ual stability study in accordance with the contemplated operating
conditions.
Two-machine system with negligible losses. The type of power

system whose steady-state stability limit can be found most simply
consists of two synchronous machines, a generator and a motor,
connected through a network of pure reactances. Because no power
is lost in pure reactances, the electrical output of the generator is
always equal to the electrical input of the motor. One power-angle
curve can therefore be used to show how the power transmitted from
the generator to the motor depends upon the angular separation of
the rotors of the two machines, or upon the phase difference of the two
internal voltages (behind saturated or equivalent synchronous re-
actances), under the condition of constant field currents (constant
internal voltages). The equation of this curve is
E •
P = -1E2
- SIn 0 [1]
x
where P = power transmitted from machine 1 to machine 2.
E1 = internal voltage of machine 1.
E2 = internal voltage of machine 2.
x = transfer reactance between the internal voltages; that is,
the reactance of the architrave of the equivalent 11' circuit
to which the reactive network, including the machine
reactances, can be reduced.
o = angle by which E 1 leads E 2 •
If the internal voltages are constant, the maximum power occurs
at 6 = 90° and is
P« = E 1E2 [2]
X
The system is stable at any value of power less than Pm, provided
o < 90°. This may be shown as follows: Suppose that the shaft load
on the motor is increased, thus causing the mechanical output to exceed
the electrical input. There is then a net retarding torque, causing
the motor to slow down and thereby to increase the angle o. As a
increases, the electrical input of the motor increases until it equals

the mechanical output. At the same time the electrical output of the
generator increases and for a while exceeds the mechanical input.
There is then a net retarding torque on the generator. The frequency
drops until it is corrected by the governor increasing the mechanical
input of the generator. The system then operates in the steady state
at the new load. The load can be increased in this manner in small
steps until the maximum power given by eq. 2 is reached at 0 = 90°.
Any further increase in mechanical power of either machine cannot be
balanced by a corresponding increase in electric power. There is,
therefore, an accelerating torque on the generator or a retarding
torque on the motor, either of which increases o. An increase of 0
beyond 90° decreases the electric power. Both the accelerating torque
of the generator and the retarding torque of the motor increase, 0
increases rapidly, and synchronism is lost. The steady-state stability
limit evidently is reached at 0 = 90°.
In the case just considered, the field currents of the machines were
assumed constant. If the field currents are constant while the load
is varied over a wide range, the terminal voltages of the machines will
vary considerably. Usually it is desired that the terminal voltages
have specified values, which are held as the load is increased, so that
when the stability limit is reached the voltages still have the specified
values. Nevertheless, as already explained, the voltage regulation
is assumed to be so slow that pullout occurs under the condition of
constant field current instead of constant terminal voltage. Accord-
ingly, the angle 0 between the internal voltages is 90° at pullout.
Calculation is required to find the magnitudes of the internal voltages
when the angle between them is 90° and the terminal voltages have
their specified values. This calculation may be carried out by a cut-
and-try process, which will now be illustrated, or graphically, as
described later.
EXAMPLE 1
Find the steady-state stability limit of a system (Fig. 1) consisting of a
generator of equivalent reactance 0.60 unit connected to an infinite bus
through a series reactance of 1.00 unit. The terminal voltage of the genera-
tor is held at 1.10 units and the voltage of the infinite bus, at 1.00 unit.
Solution. The vector diagram is that of Fig. 2. If the voltage of the
infinite bus is taken as reference, the terminal voltages are
VI = 1.10/8
V 2 = 1.00 / 0
where 8 is the angle between them.
EXAMPLE 253
Assume 8 = 30°. Then the current is
I = VI - V2 = 1.10~ - 1.00LQ = 0.95 + jO.55 - 1.00
jXe jl.OO .i1.00
= jO.55 - 0.05 == 0.55 + 1·0.OF)
jl.OO ·
Xl = 0.60 Xe =1.00
FIG. 1. Circuit of two-machine reactance system the steady-state stability limit

of which is found in Example 1.
The internal voltage of the generator is

E 1 = VI + jx1I = 0.95 + jO.55 + jO.60(O.55 + jO.05)
= 0.95 + jO.55 +
jO.33 - 0.03
= 0.92 +jO.88 = 1.27 /44 0
E2 = V2 = 1.00
FIG. 2. Vector diagram of two-machine reactance system the steady-state
stability limit of which is found in Example 1.
The angle between internal voltages is

0=44°
and the transmitted power is
2 • ~
P = -E}E
-SInu = 1.27 X 1.00 SIn
· 440 0 55
=.
x 1.60
The power can he computed also from the reference voltage and the in-phase
current:
P = 1.00 X 0.55 = 0.55
The power-angle curve (power versus () is shown as A in Fig. 3. The
maximum power with this value of excitation is 1.27 X 1.00/1.60 = 0.79.
However, as the power is increased the excitation must also be increased.
O·
1.0 f----f-------1--r""'-F-t----'s;.--+~~-_l_---_1
--
:t::
C
::I
.~
.e
..,
i)
~
£c 0.5 f-------,IIiHL-#----j----'-----+--~Afl..__\,.__-_1
30· 60· 90· 120· 150· 180·

Angle Ii between internal voltages
FIG. 3. Power-angle curves of the two-machine reactance system the steady-state

stability limit of which is found in Example 1. Curves A, B, C, D are for constant
excitation; curve E, for constant terminal voltage.
Next assume 0 = 45°.

1= 1.10~ - 1.00[2 = 0.78 + jO.78 - 1.00
j1.00 j1.00
-0.22 + jO.78 = 0.78 + '0.22

j1.00 J
E1 = 0.78 + jO.78 + jO.60(0.78 + jO.22)

= 0.78 + jO.78 + j0.47 - 0.13
= 0.65 + j1.25 = 1.41 /62.5°
0=62.5°
P = 1.00 X 0.78 = 0.78
EXAMPLE 1 255
The power-angle curve is shown as B in Fig. 3. The maximum power with

this value of excitation is 1.41 X 1.00/1.60 = 0.88. However, the excita-
tion should be further increased.
N ext assume () = 60°.
I = 1.10 ~ - 1.00 L.Q = 0.55 + jO.95 - 1.00

j1.00 j1.00
= -0.45 + JO.95 = 0.95 + '0.45

J1.00 J
E 1 = 0.55 + jO.95 + jO.60(0.95 + jO.45)

= 0.55 + jO.95 + jO.57 - 0.27
= 0.28 + jl.52 = 1.54/79.5°
5 = 79.5°
P = 1.00 X 0.95 = 0.95
The power-angle curve is shown as C in Fig. 3. The maximum power with
this value of excitation is 1.54 X 1.00/1.60 = 0.96.
Next assume 0 = 75°.
I = 1.10!J.!!.. - 1.00 L.Q = 0.28 + j1.06 - 1.00

jl.OO jl.OO
= -0.72 +jl.06 = 1.06 + '0.72
j1.00 J
E1 = 0.28 + j1.06 + jO.60(1.06 + jO.72)

= 0.28 + j1.06 + jO.64 - 0.43
= -0.15 + jl.70 = 1.70 /95°
5 = 95°
P = 1.00 X 1.06 = 1.06
The power-angle curve is shown as D in Fig. 3. The maximum power with
this value of excitation is 1.70 X 1.00/1.60 = 1.06.
The system would be unstable with 8 = 75°, because a > 90°.
In Fig. 3, curve E is the locus of constant terminal voltages of the specified
value. The power at 0 = 90°, read from this curve, is the steady-state
stability limit. It is 1.03 per unit.
The value of () at the stability limit is 70°. If 8 could be increased to 90°
without instability, the maximum power at the specified terminal voltages
would be
P VIV2
=- - = 1.10 X 1.00
= 110
. per unl
it
x; 1.00
This is the crest value of curve E -of Fig. 3. If curve E had been plotted
against 8, the angle between terminal voltages, it would have been sinus-
oidal; but, being plotted against 0, the angle between internal voltages, it
jR distorted.
FIG. 4. Two synchronous machines connected through external reactance x..
Clarke diagram for two-machine series reactance system. The

cut-and-try process illustrated in Example 1 can be avoided through
use of a graphical construction devised by Miss Clarke. 1 The con-
struction will be discussed now for two machines connected through
a
FIG. 5. Vector diagram of the circuit of Fig. 4 (Clarke diagram).
series external reactance, as in Fig. 4. In the next section the diagram

will be extended to the case of two machines connected through any
reactance network, and, later, to the case of series impedance having
a resistance component.
Figure 5 is a vector diagram of the circuit of Fig. 4. The current
vector is drawn horizontally; consequently, the I» drops are drawn
vertically. The several voltage vectors radiating from point 0 differ
CLARKE DIAGRAM 257
by the appropriate I x drops. In order to represent conditions at the

steady-state stability limit, the diagram must be drawn so that Viand
V 2 have their assigned values and so that E I and E 2 are 90° apart.
Furthermore, the several I x drops must be proportional to the corre-
sponding reactances since the current I is the same in all reactances
of the series circuit.
The construction is carried out in the following way:
1. Layoff segments [XI, Ix., and IX2 along a straight line, making
them proportional to the given values of Xl, X e, and X2, respectively,
to any convenient scale.
2. With l(XI +xe + X2) = Ix as diameter, construct a semi-
circle (shown by a dotted line in Fig. 5). Any point 0 lying on this
semicircle meets the requirement that E I lead E 2 by 90°. This fact
arises from the theorem of geometry that any angle inscribed in a
semicircle is a right angle.
3. Construct a portion of the locus of constant ratio V l/V 2 (also
shown by a dotted line in Fig. 5). If VI = V 2 , this locus is the per-
pendicular bisector of Iic; If VI ~ V 2 , the locus can be proved to be
a circle whose center is on the vertical line of I x drops or an exten-
sion thereof. However, the easiest way to construct it probably is
to locate several points on it by intersections of arcs struck with
centers at the heads of vectors VI and V2 and radii in the specified
ratio V I/V2.
4. The intersection of this locus with the semicircle is point O.
Vectors EI , VI, V2, and E2 may now be drawn.
5. The scale of the diagram is set by the specified values of VI
and V 2 •
6. The scale having been found, the power (which is the same at
all points of the circuit) can be calculated. Two simple ways of
calculating it are the following:
a. Measure the lengths of vectors E I and E 2 • Then
P« = E 1E2 = E 1E2 [3]

Xl + Xe + X2 X
b. Measure Em, the perpendicular from 0 to Ix. It represents

the in-phase component of each of the four voltages. It also
represents the lowest voltage at any point on the line between
machines, at which point the power factor is unity. Also measure
+
the diameter of the circle, Ix = I (Xl .Xe + X2) and divide it by
the total reactance x in order to find the current I. Then
Pm = EmI [4}
258 STE.ADY-STATE ST.ABILITY'P
For values of V 1/V2 differing greatly from 1.00, the locus of con-
stant V l/V2 ,vill not intersect the semicircle. The lack of an inter-
section indicates that no power can be transmitted stably with the
assumed values of terminal voltage.
EXAMPLE 2
Solve Example 1 by use of the Clarke diagram.
Solution. As shown in Fig. 6, layoff AB = IXI = 0.60 unit and BF =
IX2 = 1.00 unit. With the midpoint C of line A.F (0.80 unit from each end)
as center, draw a semicircle AOF of radius 0.80 unit. Construct the locus
DGHOK by locating points G, H, K by intersections of arcs having centers
at Band F. For point K, the radii are 1.10 from Band 1.00 from F; for
point H, they are 0.83 and 0.75; for point
A G, 0.55 and 0.50. The intersection of this
1
locus with the semicircle is point O. Draw
OA,OB, and OF, representing E I , VI, and
V 2 = E 2 , respectively. Measure OF. It
is 0.83 unit of length and represents 1.00
unit of voltage. Therefore, the scale is
1 unit of length = 1.21 units of voltage.
Measure OAt It is 1.38 units of length
and therefore represents a voltage of
1.38 X 1.21 = 1.67 units.
The steady-state stability limit is there-
fore
Pm -_ 1.67 X 1.00 -_ 1•04 UDl'ts
1.6
This agrees well with the value (1.03)
found in Example 1.
Extension of Clarke diagram to cover

anyreactancenetwork. Any network of
pure reactances, through which power
FIG. 6. Clarke diagram for deter-
mination of the steady-state sta- is transmitted from one synchronous
machine to another, can be reduced to an
bility limit of the system of Fig. 1
(Example 2). equivalent series-reactance circuit simi-
lar to the one considered in the last sec-
tion. First the portion of the network outside the machines is reduced to
an equivalent 1r, which will be composed of pure reactances if the original
network was. As the machines themselves can be represented by their
internal voltages and reactances (their resistances being negligible),
we now have a pure-reactance network of the form shown in Fig. 7.
(The shunt branches are shown as capacitive, though they may equally
well be inductive.) This circuit could be further reduced, say, to an
EXTENSION OF CLARKE DIAGRAM 259
equivalent 7r connecting the internal voltages. However, this will not

be done because it is necessary to preserve the identity of the terminals
of the machines, where the voltages are specified. The internal points,
on the other hand, need not be preserved. Further- reduction is there-
fore accomplished by an application, from Miss Clarke,' of Thevenin's
theorem to the portions of the network to the left and right of archi-
trave reactance X e• Thevenin's theorem states that a two-terminal
x,.
FIG. 7. Reactance network between the terminals of two machines reduced to an

equivalent 'Ir.
linear network can be replaced by an e.m.f. and an impedance in series

without affecting conditions outside the network. The impedance of
Thevenin's equivalent circuit is equal to the input impedance of the
original network with its internal e.m.f.'s short-circuited. Therefore,
in our case, the Thevenin's impedance at the generator end is the
reactance of the parallel combination of the generator reactance Xl
and the shunt reactance X s , or
, XIX s
= [5]
XI
Xl + X,
Similarly, the Thevenin reactance at the motor end is
I X2 X r
X2 = [6]
X2 + Xr
The e.m.f, of Thevenin's theorem is the open-circuit voltage at the
terminals of the network. At the generator. end, this is the voltage
across X, with X e disconnected, or
, Xs Xl
,
EI = [7]
Xl + z, EI = - E1
Xl
Similarly, at the motor end the equivalent e.m.f, is
[8]
The circuit of Fig. 7 is thus reduced to the equivalent series circuit of

Fig. 8. If E I and E 2 are constant, E I ' and E 2 ' are also constant. The
maximum power that can be transmitted over the circuit of Fig. 8 for
constant values of E I ' and E 2 ' (corresponding to constant field cur-
rents) occurs when EI'leads E 2 ' by 90°. This value of power is also the
maximum value for the circuit of Fig. 7. Since the system is stable
for all values of power up to the maximum, the value so found is the
steady-state stability limit of the system.
Xs Xr
E'1 E'
2
.FIG. 8. The circuit of Fig. 7 reduced to a series circuit by

use of Thevenin's theorem.
The Clarke diagram therefore can be used as before except that Xl,
X2, E I , and E 2 are replaced by Xl" X2', E I ' , and E 2 ' , respectively
(eqs, 5 through 8), and X e now represents the architrave reactance of
the equivalent 7r circuit of the external network. The actual internal
voltages and currents of the machines can be calculated if desired.
However, they are not needed for calculating the stability limit, for it
can be calculated in terms of the equivalent quantities, thus:
_ E 1'E2 '
Pm [9]
+ X e + X2
- , ,
Xl
where E I ' and E 2' at 90° phase difference are determined by means of
the Clarke diagram.
EXAMPLE 3
A generator having an internal reactance (equivalent synchronous re-
actance) of 0.60 per unit feeds power to an infinite bus over a transmission
line, the equivalent 1r of which has an architrave impedance of jl.0 and
pillars of -j5.0 each. The terminal voltage of the generator and the
voltage of the infinite bus are both held at 1.0 per unit. Find the steady-
state stability limit.
Solution. The Thevenin reactance of the generator (eq, 5) is
Xl' = 0.60( -5.0) = -3.0 = 0.68
0.60 - 5.00 -4.4
EQUATION FOR STABILITY LIMIT 261
From a Clarke diagram, similar in form to Fig. 6, but with AB = 0.68,

Be = 0.16, and CF = 0.84, and with OD the perpendicular bisector of
BF, E I ' is found to be 1.53 per unit. The stability limit is (eq. 9)
1.53 X 1.00 .
Pm = 0.91 per unit
0.08 + 1.00
::=
Equation for steady-state stability limit of two-machine reactance

system. As an alternative to the graphical construction described in
an earlier section, an algebraic equation based on the Clarke diagram
can be used to find the steady-state stability limit. This equation will
now be derived.
The general plan of the derivation is to express the internal voltages,
E 1 and E 2 , in terms of the given terminal voltages, V l and V 2 , and the
reactances, and then to use eq. 3 for maximum power,
Refer to the vector diagram, Fig. 5. From the relations between the
sides of a right triangle, the in-phase component Em of the several
voltage vectors can be written in the following four ways, using succes-
sively triangles Oac, Obc, Ode, and Oec:
Em 2 = E l 2 -C2 [10]
= V l 2 - (C - IXl)2 [11]
= V 22 - (B - [X2)2 [12]
= E 22 - B 2 [13]
From eqs. 10 and 11,
V l 2 = E 12 - C2 + (C - [Xl)2
= E 12 - C2 + C2 - 2C1xl + [2Xl2
= E 12 - 2C1xl + [2X12 [14]
Similarly, from eqs. 12 and 13,
V 22 = E 22 - 2B1x2 + [2 X22 [15]
Now B[, CI, and [2 in eqs. 14 and 15 must be expressed in terms of
voltages and reactances. From right triangle Oae,
E 12 + E 22 = [2X2 = (C + B)(C + B) [16]
From eqs. 10 and 13,
E 12 - E 22 = C2 - B2 = (C + B)(C - B) [17]
Add eqs. 16 and 17 and divide by 2, obtaining
E 12 = C(C + B) = Clx [18]
Subtract eq. 17 from eq. 16 and divide by 2, obtaining

E2 2 = B(G + B) = Blx [19]
From eq. 18,
2
CI = E 1 [20]
x
From eq. 19,
BI=Ei [21]
x
From eq. 16,
[22]
Substitution of eqs. 20, 21, and 22 into eq. 14 gives:
Vt
2
= E1
2
- 2E1
2
~ + E (~Y + Ei (~Y
1
2
= E 12 (1 - :lY + E 2
2
(~Y
V 1
2X 2
= E 12(X - Xl)2 + E22X 12
= E1 2
(Xe + X2)2 + E22X 12 [23]
A similar substitution into eq. 15 gives:
V 2
2X2
= E12X22 + E 2 2(X e + Xl)2 [24]
Simultaneous solution of eqs. 23 and 24 yields:
E 2 _ + Xl)2 - V22X12]x2
[V12(X e
[25]
1 - (z, + Xl)2(X e + X2)2 - X1 2X22
e + X2)2 - V12X22]X2
2(X
E 2 _ [V2
[26]
2 - (z, + Xl)2(X e + X2)2 - X1 2X22
Hence
V12V22[(Xe + Xl)2(X e + X2)2 + X1 2X22]
- V14X22(Xe + Xl)2 - V24Xl~(Xe + X2)2
[27]
(z, + Xl)2(X e + X2)2 - X1 2X2 2
Substitution of eq. 27 into eq. 3 gives:
+ Xl)2(X e + X2)2 + X1 2X22]
V12V22[(xe
- V14x22(Xe + Xl)2 - V24X12(Xe + X2)2

~=~~=X----~--~------ [28]
x (x e + Xl)2(X e + X2)2 - X1 2X22
EXAMPLE 4 268
The denominator may be rewritten as

[(x e + Xl)(X e + X2) + XIX2][(Xe + XI)(X e + X2) - XIX2]
= + XIX e + X2 Xe + 2XIX2][Xe2 + XIX, + x2xel
[xe2
= [xe(x e + Xl + X2) + 2XIX2][Xe(X, + Xl + X2)]
= (XeX + 2XIX2)XeX
giving
V12V22[(Xe + XI)2(X e + X2)2 + Xl 2X2 j 2
- V14X22(Xe + Xl)2 - V24X12(Xe + X2)2 [29]

Pm =
Xe(XeX + 2XIX2)
This is the desired equation. Some values of V I and V 2 which differ

too greatly make the radicand zero or negative, showing that no power
can be transmitted stably at such voltages.
Considerable simplification results if the voltages are equal. Putting
VI = V 2 = V and simplifying, we get
2
Pm = V V (xe + 2x1 )(x e + 2X2) [30)
XeX + 2XIX2
Equations 29 and 30 can be applied to any reactance network by
use of the method given in the last section.
EXAMPLE 4
Check the results of Examples 1 (or 2) and 3 by means of eqs. 29 and 30.
Solution. Example 1 or 2. Data:
VI = 1.10, V2 = 1.00, Xl = 0.60, x, = 1.00, %2 = o.
Use eq. 29.
V1
2
V 2 2 = 1.21, V 14 = 1.46, V 2 4 = 1.00
x, + Xl = 1.60, x, + %2 = 1.00
(z, + Xl)2(X, + X2)2 + X1 2X 22 = (1.60)2 X (1.00)2 - 0 = 2.56
X2 2(X, + Xl)2 = 0
+ X2)2 = (0.60)2 X (l.OOra = 0.36
XI 2(X,
x = Xl + X, + X2 = 1.60
x,(x,x + 2XIX2) = 1.00(1.00 X 1.60 + 0) = 1.60
P _ V1.21 X 2.56 - 1.46 X 0 - 1.00 X 0.36
m - 1.60
= V3.10 - 0.36 V2.74

=- - =
1 04
. per unllt
1.60 1.60
This agrees with the previous results.
264 RTEADY-RTATE STABILITY
Example 3. Data:
V l = V2 =V = 1.00, Xt' = 0.68, Xe = 1.00, X2 = O.
Use eq, 30.
r; = (1.00)2V(1.00 + 2 X 0.68)1.00 = V2.36
1.00 X 1.68 +0 1.68
= 0.92 per unit
This value agrees well with the previous result (0.91).
Two-machine system with losses. Power-angle curves. Although

every actual circuit has losses, the methods developed in the preceding
sections of this chapter for circuits without losses furnish useful and
relatively simple approximations, the accuracy of which depends upon
the amount of loss. Nevertheless, methods of finding the steady-state
stability limits, taking losses into consideration, should be developed
not only for use where losses are high but also for establishing the
accuracy of the approximate methods.
The theory of steady-state stability of circuits with resistance is
complicated by the following facts: (1) Maximum power at the sending
end exceeds the maximum power at the receiving end and occurs at a
different angular separation of the two machines. (2) The maximum
stable angle depends upon the relative inertias of the machines and
upon governor action.
Let the power-angle curves of the two machines be considered
first. The network, including the reactances of the machines, can be
reduced to an equivalent 1r connecting the two internal points of the
machines and the neutral point. The method of doing this was de-
scribed in Chapter III. Let the following notation be used:
Y12 = Y12/ e12 = Y 21 = Y21 / e21 = mutual admittance of termi-
nals 1 and 2
Y11 = Y~l/ell = self-admittance of terminal 1 (sending end)
Y22 = Y 22 / e22 = self-admittance of terminal 2 (receiving end)
Y12 = Y12/-812 = admittance of architrave of 7r, also called the
transfer admittance
YOI = admittance of pillar of 1f' at sending end
Y02 = admittance of pillar of 7r at receiving end
812 = angle of transfer, or architrave, impedance
E 1 = E 1 & = internal voltage of machine 1 (the generator)
E 2 = E 2 & = internal voltage of machine 2 (the motor)
a = 01 - 02 = the angle by which E I leads E 2
POWER-ANGLE CURVES 265
The terminal admittances have the following values in terms of

the admittances of the elements of the equivalent 1r circuit:
Y12 = Y 21 = -Y12 = Y12/ 180° - 812
Y11 = YOI + Y12

Y22 = Y02 + Y12
Hence
812 = 8 21 = 180° - 812
The power input to the network at the sending end is, by eqs. 17
of Chapter III, Vol. I,
PI = E 12 Y l l
cos 811 + E 1E2 Y 12 cos (812 - 01 + 02)
2
= E 1 Y 11 cos 811 + E 1E2 Y 12 cos (180° - 812 - 0)
= E 12Y11 cos 811 +E 1E2Y12 cos (0 + 812 - 180°) [31]
and the input at the receiving end (output if negative) is
P2 = E2E1Y21COS (8 2 1 - 02 +01) + E 2 2Y22 cos 8 22
= E 22Y22 cos 8 22 + E 1E2Y12 cos (180° - 812 + 0)
= E 22 Y 22 cos 8 2 2 - E 1E2 Y 12 cos (0 - 812 ) [32]
With fixed magnitudes of internal voltages E 1 and E 2 , the first term of
the last line of each equation is constant while the second term is a
cosine function of o. Byeq. 31, maximum output of machine 1 (the
generator) occurs at o· = 180° - 812 = Oml. By eq. 32, maximum
input of machine 2 (the motor) occurs at 0 = 812 = Om2.
Here 812 is the angle of the architrave impedance. If this impedance
consists of pure inductive reactance (as in the case previously con-
sidered), 812 = 90° and both maxima occur at 0 = 90°. If the archi-
trave impedance consists of positive resistance and inductive reactance,
812 is less than 90°; then maximum sent power occurs at a value of 0
greater than 90°, and maximum received power, at a value of 0 less
than 90°. Power-angle curves for this case are plotted in Fig. 9.' The
loss in the line is also plotted versus o. Another possibility is that the
architrave impedance has negative resistance. This can occur, for
example, if there is a shunt resistance, as shown on the circuit diagram
of Fig. 10. When the T circuit shown there is converted to a 'lr, the
architrave impedance is
+ jX + (jX)(jX) + j2X
2
jX = _ X
R R
the real part of which is negative. With negative resistance and
inductive reactance, 812 lies between 90° and 180°. Consequently,
FIG. 9. Power-angle curves of two machines connected through series resistance

and inductive reactance.
FIG. 10. Power-angle curves of two machines connected through reactance with
an intermediate shunt resistance load.
EFFECT OF INERTIA 267
maximum sent power occurs at 0 less than 90°, and maximum received
power, at 0 greater than 90°. The power-angle curves for this case are
given in Fig. 10. The power consumed by the load is also shown.
With either series or shunt resistance, the maximum received power
(-P2) is less than the maximum sent, power (PI)'
Two-machine system with losses. Effect of inertia. Consider a

generator of finite inertia feeding power to an infinite bus through series
impedance with resistance. The power-angle curves for constant excita-
tion are shown in Fig. 9. The output of the generator is shown by
curve PI; the power received at the infinite bus, by - P 2 • If the power
input to the generator is slowly increased, the generator voltage is
advanced in phase, thus increasing the angular separation o. The out-
put of the generator increases so as to equal its input until PI reaches
its maximum value at 0 = Oml = 180° - 812 > 90°. Therefore, the
system is stable for all positive values of 0 less than Omi. However,
maximum power is received by the infinite bus at 0 = Om2 = 812 <
90° < Omi. Therefore, by definition, the steady-state stability limit is
the maximum value of received power, - P2, and this occurs at 0 = 0'm2.
The system is stable at this angle.
Next consider a motor offinite inertia receiving power from an infinite
bus through series impedance. The curves of Fig. 9 again apply. Now
PI is the output of the infinite bus, and - P2 is the input of the motor.
Let the shaft load on the motor be slowly increased. The motor
voltage is retarded in phase, thus increasing 0, and the input of the
motor increases so as to equal its output, until -P2 reaches its maxi-
mum value at 0 = Om2' If the motor load is further increased, syn-
chronism is lost.
In both of the preceding cases the stability limit is the maximum
value of - P2, which occurs at 0 = Om2 = 812 , which is less than 90°.
In the case of the finite motor synchronism is lost if 0 exceeds this angle,
but in the case of the finite generator 0 can increase from Om2 to Oml,
with an accompanying increase in generator output Pl, before syn-
chronism is lost. The generator therefore is more stable when operat-
ing at the stability limit than the motor is.
Next consider a generator and a motor both offinite, and not necessarily
equal, inertias. It would be expected that the maximum stable value
of ~ would be greater than Om2 but less than Oml, and that the exact
value would depend 'upon the relative inertias of the two machines,
being near Om2 if the inertia of the generator were high compared with
that of the motor, and near 0ml if the reverse were true. This expecta-
tion is borne out by the equivalent power-angle curve of a two-machine
system, derived in Chapter IV, Vol. I. By eq. 26 of that chapter, the

maximum of this curve occurs at
1 + M2
li M;=-
= lie = tan -1 (lltf M tan 812)
2
[33]
where ill 1 and M'2 are the inertia constants of the two machines. The
critical angle ac is always between the angles of maximum sent and
received power:
[34]
It approaches am2 if M I »M2 , and approaches aml if M I «M2 •
In each of the three cases considered above, all of which are for posi-
tive series resistance, the steady-state stability limit is reached at
o = Om2. However, it appears that 0 can increase beyond Om2 up to a
value ac before synchronism is lost. Thus at maximum received power
there is a margin of stability which depends upon the relative inertias,
being greatest when machine 2 (the motor) is an infinite bus.
The situation is somewhat different if the series resistance is negative.
The power-angle curves are given in Fig. 10. Here the maximum sent
power is reached at a smaller value of 0 than that for maximum received
power (Omi < Om2)' It is therefore necessary to investigate whether the
system is stable at values of 0 between Omi and Om2.
If the motor is an infinite bus, the generator pulls out of step at Om},
before -P2 reaches its maximum value. The stability limit is then
the value of - P2 at 0 = Ont!. If, however, the generator is an infinite
bus, the motor pulls out of step at Om2, where - P 2 is greatest. The
stability limit is greater for the infinite generator than for the infinite
motor.
If both machines are finite, the system would appear to be stable
for all values of 0 up to Oc computed from eq. 33 and lying between Oml
and Om2o The stability limit would then be the value of received
power at 0 = oc. The received power would be less than its maximum
value but greater than its value at 0 = Omi'
Two-machine system with losses. Effect of governor action. The

foregoing discussion has shown that a two-machine system is stable
without doubt for all positive values of 0 less than OmlOr 0m2, whichever
is smaller. Furthermore, the discussion of the effect of inertia has
indicated that the system would be stable at even larger values of 0,
up to Oc given by eq. 33. This increased range of stable angle increases
the stability limit if the series resistance is negative, but not if it is
positive, though in the latter case the margin of stability is increased.
EFFECT OF GOVERNOR ACTION 269
These conclusions were drawn from the equivalent power-angle curve,

which was based upon the assumption that the mechanical power of
both machines remained constant. For consistency with this assump-
tion, the disturbance used as a test of stability would consist of a small
angular displacement, which might be caused by a momentary small
increase of mechanical power of either machine. A sustained increase
of mechanical power, however, can lead to different results, depending
upon the manner in which the increase is applied and upqn the resulting
governor action tending to maintain the frequency of the system.
Let us investigate the stability of a two-machine system with nega-
tive transfer resistance (due to positive shunt resistance) when the
angular separation ~ is between ~ml and ~c. Instead of using the
equivalent power-angle curve of the system, we shall use the individual
acceleration curves of the two machines, which are plots of acceleration
versus a. The acceleration of machine 1 (the generator) is
[35]
where Pil = mechanical power input.

P u1 = electric power output.
Pal = accelerating power.
M 1 = inertia constant.
A similar relation holds for machine 2 (the motor). Hence each of the
two acceleration curves is derived from the corresponding power-angle
curve of Fig. 10 by first subtracting it from the mechanical input, which
is assumed constant, and then dividing the result by the inertia con-
stant. The acceleration curves so obtained are shown in Fig. 11 for
M 1 » M 2. Only a small range of 0 near 90° is shown, In the steady
state neither machine is accelerating. Therefore the two acceleration
curves intersect on the horizontal axis (point a). For stability without
governor action, curve 2 must have a greater slope than curve 1. At
values of a < Oml, the slope of curve 2 is positive and that of curve 1 is
negative. For ~ between ~ml and ~m2, both curves have positive slopes,
but that of curve 2 is greater than curve 1 up to ac, ,vhere the slopes
become equal. In Fig. 11, point a represents stable steady-state
operation at an angle between OmI and ~c. Now let the shaft load of the
motor be increased by a small amount. This lowers the motor char-
acteristic from 2 to 2'. .Initially the motor has a retardation ab., which
tends to increase O. When ~ reaches point c, the intersection of curves
1 and 2', both machines have small and equal accelerations cd. There-
fore 8 does not change immediately, but the system frequency slowly
increases until the governor decreases the mechanical input of the

generator. This lowers curve 1, causing the operating point of both
machines to reach stable equilibrium at point e. In this instance,
governor action tends to decrease 0 and therefore does not cause
instability.
Po
M
FIG. 11. Acceleration curves of two-machine system with

negative transfer resistance.
In practice, most "two-machine" systems consist not of a generator

and a motor but rather of two generators (or groups of generators) one
of which (no. 2) has a local load supplied partly from the adjacent
generator and partly over a transmission line from the distant generator
(no. 1). Machine 2, together with its local load, constitutes an "equiv-
alent motor." Both generators have governors. The power trans-
mitted from one to the other can be increased either as described before
or in the opposite order, thus: The governor of machine 1 is first
adjusted to increase the mechanical input, which we will assume to be
held constant thereafter. This shifts the acceleration curve upward
from 1 to 1'. Machine 1 has an initial acceleration af. Angle 0 in-
creases to the abscissa of point g, the intersection of curves l' and 2.
Here both machines have equal accelerations (lk. The frequency
increases. The governor of machine 2 thereupon decreases the me-
chanical input, an action which is equivalent to increasing the me-
chanical output of the equivalent motor and which lowers the motor
characteristic from 2 to 2'. If there were no further governor action
the machines would reach equilibrium at point i, the intersection of
curves I' and 2'. However, here the acceleration of both machines is
even greater than before. The governor of machine 2 therefore con-
CONSERVATIVE CRITERION FOR STABILITY 271
tinues to lower curve 2' until it no longer intersects curve 1', whereupon
synchronism is lost. With this method of increasing the load, governor
action tends to increase 0 and causes instability.
The conclusion is that with 0 in the range between Omi and Oc the
system may be either stable or unstable under the condition of small
sustained increase of load, taking governor action into consideration,
depending upon how the increase of load is accomplished.
The same kind of reasoning leads to a similar conclusion regarding
the stability of a two-machine system with positive transfer resistance.
In the range of 0 from Om2 to oc, the system may be either stable or
unstable. It is unstable if the shaft load of the motor is increased first,
but is stable if the input of the generator is increased first. As already
noted, however, these facts do not alter the stability limit which occurs
at 0 = Om2.
It has been assumed that the governor of one machine is set for
constant mechanical power and that the governor of the other machine
then maintains frequency. If both governors tend to maintain fre-
quency, the maximum stable angle would appear to depend upon the
relative sensitivity of the two governors.
Two-machine system with losses. Conservative criterion for

stability. To summarize the conclusions reached in the last two
sections: If Om1 is greater than Om2 (as it is when the transfer resistance
is positive), the steady-state stability limit is the maximum received
power and occurs at 0 = Om2. If Om1 is less than Om2 (as it is when the
transfer resistance is negative), the stability limit is less than the
maximum received power and occurs at an angle between Om1 and oc,
depending upon governor action. The larger of these angles, oc, lies
between Om1 and Om2, the exact value depending upon the relative
inertias of the two machines.
Because of the uncertainty that usually exists regarding whether
load will always be increased in the same manner, it is advisable to
take the steady-state stability limit of a two-machine system as the value
of received power at 0 = Om1 or Om2, whichever is smaller. (This value of
ois either 8 12 or 1800
- whichever is smaller, 812 being the angle
812,
of the transfer impedance.) This limit is a conservative .one.. In some
cases there will be a small margin either in the value of 0 or in the values
of both 0 and received power. Nevertheless, it may be considered
accurate enough for all practical purposes because a fairly large
variation in 0 in the range between Om1 and 0m2 produces a relatively
small percentage change in received power. Furthermore, the practical
operating power limit is usually set considerably below the calculated
steady-state stability limit (say at 75% of the latter) to allow for power
variations caused by changes of loads or by remote faults.
At values of ~ smaller than both ~ml and ~m2, the curve of PI versus
o has positive slope, while that of P2 has negative slope. Since
~ = ~l - ~2 by definition, the operation is stable if hoth
aPt
- >0 and
aP2
- >0 [36]
aO I a~2
These derivatives are called synchronizing-power coefficients. A con-

servative criterion for stability, applicable to multimachine systems
as well as two-machine systems, is that the synchronizing-power co-
efficients of all machines be positive. In other words, if anyone
machine is advanced in phase while the others are constant in phase,
its electrical output should increase (or its electrical input should de-
crease); conversely, if any machine is retarded in phase, its electrical
output should decrease (or its electrical input should increase). If
each machine meets this test, the system is stable. In making the
test, one keeps the internal voltages constant (representing constant
field currents). If the system is found "to be stable, the angular separa-
tion is increased and the internal voltages are increased so as to keep
the terminal voltages at their specified values; as a result the power
is increased. The test for stability is made again, and the procedure
is repeated until instability is indicated by one of the synchronizing-
power coefficients becoming negative. In this manner the steady-
state stability limit is found. The stability limit of a system of any
number" of machines can be determined advantageously by following
the described procedure on an a-c. calculating board. The stability
limit of some two-machine systems can also be found graphically as
described in the next section.
Clarke diagram for two-machine system" with resistance. In an

earlier section the Clarke diagram was described for a two-machine
system having inductive reactance without resistance. The stability
limit of such a system was reached when the angular separation 0
between internal voltages reached 90°, which angle was inscribed in a
semicircle. According to the conservative stability criterion discussed
in the last section, the effect of resistance is to change the value of 5
at which the stability limit is reached from 90° to a smaller value,
equal to either 812 or 180° - 812, whichever is smaller, where 812 is the
angle of the transfer impedance between internal voltages. In what
follows this angle will be called simply 812 , which value holds when
the transfer resistance is positive. Use will be made of the theorem of
CLARKE DIAGRAM WITH RESISTANCE 273
geometry that any angle inscribed in a circular arc of central angle

3600 - 2812 is 812,
The vector diagram of the system of Fig. 12, in which the two
machines are connected through external series impedance re + jx e,
is shown in Fig. 13. At the stability limit the angle between E 1 and
Vi.
FIG. 12. Two synchronous machines connected through

external impedance r6 jx e• +
,-'"
... ------- ......... -....,
a
,,'-
;I"
Locus of /1'/
O=812~/
I
I
I
I
/ b
I
I
I
0'
..... xw::-O""'O';'-+---_-f'>o~iiIIl' IXe
! \
Locus of \
specified \
l't /V2 \ ,e
\ I
\ ~ I
" I
" I
<, ~6' I
<, I ~ I
<,,-..
----e 1"\
_J_-Jh
I
FIG. 13. Vector diagram of the system of Fig. 12 (Clarke diagram).
E 2 is 812 = tan-1(x12/rI2) = tan-1[(xl + X e + x2)/r e]. The diagram

is constructed so as to meet both this requirement and the required
values of V I and V 2 by means of the following procedure:
1. With vector I as reference, layoff lXI, Ix e, Ire, and IX2 to any
convenient scale.
2. Draw the vector sum ae.
3. Draw!g, the perpendicular bisector of 00.
4. Extend ed until it intersects fg at g.

5. With g as center, draw circular arc aOe.
6. Construct part of the locus of constant ratio V t/V2 as described
for the Clarke diagram of a reactance system. The intersection of
this locus and circular arc aOe is point O.
7. Draw vectors E l , VI, V2 , and E 2 •
8. Determine the scale of the diagram from the measured length
and the specified value of VI or V 2.
9. Determine I by measuring the length of one of the Ix vectors,
converting the length to voltage, and dividing by the known value
of X.
10. Measure the receiving-end power-factor angle cP2.
11. Calculate the steady-state stability limit as
P2 = V 21 cos cP2 or Vm I [37]
The method of locating the center g of arc aOe so that the length of
the arc is 3600 - 2812 can be verified as follows: Angle aeh is 812 =
tan-l[(Ixl + IXe + [x2)/Ir e]. Angle eg! equals angle aeh because
they are corresponding parts of similar right triangles eg! and aeh.
Likewise angle agf equals angle egf. Hence angle ega is 2812 , arc aOe
is 3600 - 2812, and angle aOe is 812, the desired angle.
If the network between terminals of the machines has shunt branches,
it can be reduced to an equivalent 1r. If the shunt branches of the 7r
are purely reactive, they can be paralleled with the adjacent machine
reactances, as described previously, and the Clarke diagram can then
be constructed with the resulting parallel reactances Xl' and X2' used
in place of the machine reactances Xl and X2. The equivalent circuit
is then like that of Fig. 8 except that r e is inserted in series with X e•
The Clarke diagram of this equivalent circuit gives the maximum
received power (stability limit) at point E 2 ' of this figure. Since no
power is lost in reactance, the power there is the same as at point V2.
The power at V2 of Fig. 8 is the same as at V2 of Fig. 7 because the
circuits are equivalent here; and, since no power is lost in the shunt
reactance X r , the power at the terminals of machine 2 (at the right of
Xr, Fig. 7) is the same as that at E 2 ' of Fig. 8. A circuit with shunt
capacitive reactances arises if a transmission line between the machines
is represented by a nominal or equivalent 1r.
If the shunt branches of the 1r have resistance, however, the Clarke
diagram is no longer applicable because the loss in X r of Fig. 7 does not
equal the loss in x,. of Fig. 8 for the same values of V2 and I, and hence
the power inside the equivalent machine (at E 2 ' , Fig. 8) is no longer
equal to the power of the actual machine (at E 2 , Fig. 7). In this case
EXAMPLE 5 275
the stability limit can Le calculated by successive trials. First the

network including machine reactances is reduced to a 1r in order to find
the angle 812 of the transfer impedance. Then, using the specified
values of Vt and V 2 and an assumed angle between them, the vector
values of E 1 and E 2 are calculated. By trial, the angle between V1
and V2 is found that gives angle 812 between E 1 and E 2 . The received
power under this condition is the steady-state stability limit.
No algebraic equation has been derived for the steady-state stability
limit of a two-machine system with resistance.
EXAMPLE 5
By use of the Clarke diagram, find the steady-state stability limit of a
system like that described in Example 1 except that the external circuit
has a resistance of 0.3 unit.
A
0.60
C B
F_18 I
I
I
I
1.00
I
, /
,
I
I
D
FIG. 14. Clarke diagram for solution of Example 5.
Solution. The diagram is constructed in Fig. 14, with AB = 0.60 unit,

BC = 0.30 unit, and CD= 1.00 unit. Line AD is drawn, and its perpen-
dicular bisector EF is constructed. With F as center, arc AOD is drawn.
Three points on the locus V 1/ V 2 = 1.1 are found as the intersections of
arcs struck with points Band D as centers. Thus point 0 is located. From
0, vectors E I , VI, and V2 are drawn to points A, B, and D, respectively,
and Vm is drawn perpendicular to CD. The scale of the diagram is estab-
lished by measuring V 2, which is found to be 1.05 units of length and repre-
sents 1.00 unit of voltage. Hence one unit of length represents 0.952 unit of
276 RTEADY-STATE RTABILITY
voltage. As a check, V 1 is also measured. It is 1.16 units of length, repre-

senting 1.10 units of voltage. 1.10/1.16 = 0.948. Use the average, 0.950.
Vm measured 0.815 unit of length; hence it represents 0.815 X 0.950 =
0.775 unit of voltage. The current I is found as follows : CD = I X e = 1.00
unit of length = 0.95 unit of voltage. I = Ixe/x e = 0.95/1.00 = 0.95
unit of current. The steady-state stability limit is the received power,
VmI = 0.775 X 0.95 == 0.74 per unit.
Multimachine system. The conservative criterion for steady-state

stability, described earlier in this chapter for a two-machine system,
is also applicable to a system having any number of machines. Each
machine in turn is tested for stability by advancing the angular posi-
tion 0 of its internal voltage (the magnitude of this voltage being kept
constant) and noting whether the electric power output of the machine
increases or decreases. If it increases - that is, if
~~: > 0 [38]
machine n is stable. If every machine is stable, the system is stable.

In applying the test for stability to each machine of a multi-
machine system, one must make an assumption concerning the division
of incremental power among the remaining machines. The following
three different assumptions have been proposed:
1. That the relative angular positions of the remaining machines
be constant.
2. That the power outputs of all but two machines remain con-
stant while the angular displacement between these two is altered.
3. That, when the angular position of one machine is altered, the
accompanying change of power be divided among the remaining
machines in accordance with their prime-mover characteristics (or
load characteristics, in case of motors).
The choice between the three foregoing assumptions rests upon con-
venience and upon judgment as to which best represents the actual
or contemplated operating conditions.
Criteria based on assumptions 2 and 3 have been formulated mathe-
matically (assumption 2 in Ref. 8, assumption 3 in Ref. 9, both being
discussed also in Chapter 3 of Ref. 20 and in Chapter VIII of Ref. 14).
However, the amount of algebraic work increases rapidly with the
number of machines.
By far the simplest method of investigating a multimachine system
is to use an a-c. calculating board. In testing a machine for stability
by means of the a-c. board, the power source representing this machine
REPRESENTATION OF SYNCHRONOUS MACHINES 277
is advanced in phase. The treatment of the other machines during

the test depends upon which assumption is used. Assumption 1 (that
the other machines have fixed phase) is the easiest to apply on the
calculating board. If this assumption is used, the power sources repre-
senting the remaining machines need not be touched (unless their
inherent voltage regulation is poor). If assumption 2 or 3 is used,
the other power sources must be readjusted in phase so as either to
keep the power constant or to divide the power increment in the manner
decided upon. This may be done by trial without much difficulty.
In many studies it is sufficient to determine whether or not a system
is stable for a particular operating condition. In some studies, how-
ever, it is required to find the steady-state stability limit of a particular
machine or line. Before this can be done, the operating conditions
must be decided upon. If the stability limit of a machine (or station)
is "ranted, how are increments of power of this machine to be divided
among the remaining machines (or stations)? If the.stability limit of a
line is wanted, how are the increments of sending-end power to be sup-
plied by the various generators and how are the increments of receiving-
end power to be absorbed by the various motors or other loads?
The answers to these questions depend upon the expected operating
practice and are often dictated by other considerations than stability.
Among the considerations are: relative economy of generating stations,
expected extremes of water supply for hydro stations, best sites for new
generating plant or for enlargement and rehabilitation of old plants,
maintenance of proper voltage at substations, thermal load limit of
cables, expected growth of loads, provision of spinning reserve against
sectionalization of the system by faults or by instability, and so on.
Having decided upon how the increased amounts of power are to he
supplied and consumed, the stability limit of a line or machine is found
by increasing the power in steps. At each step the operating condi-
tions are adjusted appropriately (to give the right values of voltage, to
avoid overloading of circuits, and so on) and the system is then tested
for stability. The stability limit lies between the highest value of re-
ceived power for which the system is found to be stable and the lowest
value of received power for which the system is found to be unstable.
In testing for stability, it is not usually necessary to test every
machine at every step, but only those which appear to be weakest or
those whose angular separation is greatest.
Representation of synchronous machines. The methods of calcula-

tion presented in the preceding part of this chapter are based on the
assumption that the power system is a linear network, that is, one
representable by constant voltages and constant impedances. The

components of the power system which depart most from linearity are
the synchronous machines and the loads.
Let us first consider the synchronous machines. Their reactances
constitute a large part of the transfer impedances of the network, and
it is important, therefore, to use values that are at least approximately
correct. A synchronous machine in the steady state is represented by
its synchronous reactance and excitation voltage. These quantities
are affected by saliency and by saturation.
Effect of saliency. The steady-state power-angle equation of a
salient-pole machine was given in Chapter.XII as eq. 235 and, with
resistance neglected, as eq. 237. The power-angle curve was shown
in Fig. 36 of that chapter. The equations and curve contain a funda-
mental component and a second harmonic. The second harmonic,
called reluctance power, is proportional to the square of the terminal
voltage and to Xd - X q • When round-rotor theory is used for a salient-
pole machine, X q is assumed to be equal to Xd, whence it follows that the
reluctance power is zero. The effect of the reluctance power is to in-
crease the maximum power and to decrease the angle at which it occurs.
Usually pullout occurs with the excitation voltage much higher than
the terminal voltage, and under these conditions the effect of reluc-
tance power on the stability limit is small. However, in the case of an
underexcited synchronous condenser with normal terminal voltage,
the reluctance power considerably increases the stability limit. It is
even possible for such a machine to operate stably with reversed field
current.
With strong excitation, saliency can be neglected with little error -
and that on the safe side. In general, the effect of saturation is much
more important than that of saliency.
EXAMPLE 6
Find the stability limit of a salient-pole synchronous motor for which
Xd = 1.2 per unit and X q = 0.9 per unit when the motor is operated from
an infinite bus of voltage 1.00 per unit. The excitation voltage of the motor
is (a) 1.5, (b) 0.5 per unit. Neglect resistance and saturation. Find the
error which would be caused in each case by neglecting saliency.
Solution. (a) Use eq. 237 of Chapter XII, which gives power as a function
of angleo.
P = EqY
-Slnu
. ~ + y2 -Xd - X q · 2~
--SIn u
Xd 2XdXq
= 1.5 X 1.0 SIn

. u~ (1
.
+
0)2 1.2 - 0.9 · 2~u
SIn
1.2 2 X 1.2 X 0.9
= 1.25 sin 0 + 0.14 sin 20 (a)
EFFECT OF SATURATION 279
Find the maximum value of P by trial.
If 0 = 90°, P = 1.25 X 1.00 + 0.14 X 0 = 1.25.
If 0 = 80°, P = 1.25 X 0.985 + 0.14 X 0.34 = 1.27.

If 0 = 70°, P = 1.25 X 0.940 + 0.14 X 0.40 = 1.23.
The stability limit is approximately 1.27 per unit. Round-rotor theory gives
1.25 per unit. The error caused by neglecting saliency is -1.5%.
(b) With E q = 0.5, the first term of eq. a is one-third of its previous value.
P = 0.42 sin 0 + 0.14 sin 20 (b)
By trial, P111 = 0.49 at 0 =:: 650 • Round-rotor theory gives 0.42. The
error is -14%.
Effect of saturation. An unsaturated nonsalient-pole synchronous

machine is represented in the steady state by its unsaturated syn-
chronous reactance Xd, which is constant, and its excitation voltage Eq ,
which is constant if the field current is constant. A saturated syn-
chronous machine can be represented similarly by a reactance and an
e.m.f., but these are not constant. However, if properly chosen, they
may be assumed constant during small changes in load such as are
assumed to occur in the test for steady-state stability. The reactance
of a saturated machine has been variously called "adjusted," "equiva-
lent," or "saturated" synchronous reactance. The last two terms will
be used here with somewhat different significances.
The saturated and equivalent synchronous reactances are smaller
than the unsaturated reactance, and the stability limit computed with
saturation neglected and the excitation voltage assumed constant is
smaller than the true stability limit with saturation present and
constant field current. The values of saturated and equivalent syn-
chronous reactance vary with the saturation. The value of equivalent
reactance usually lies within the range 0.4 to 0.8 of the unsaturated
synchronous reactance.
Several methods of finding the values of saturated or equivalent
synchronous reactance have been advanced by various writers. 7 ,l O,12
These methods differ in their simplicity and in their accuracy,
An accurate way of representing a saturated nonsalient-pole machine
by means of a reactance and an e.m.f. was given by Putman. 3 Both
the reactance and the e.m.f. vary with saturation, being functions of
the voltage E z behind armature leakage reactance (or, more accurately,
of the voltage E p behind Potier reactance). The manner in which
these quantities vary with E p can be deduced from the vector diagram,
Fig. 15, which is much like that of Fig. 62 of Chapter XII. The
following notation is used:
I = armature current
Vt = armature terminal voltage
r a = armature resistance
X p = Potier reactance
Er =sEp
Ire
FIG. 15. Vector diagram of cylindrical-rotor synchronous machine with saturation.
Ep = voltage behind Potier reactance

ET = voltage that would exist behind Potier reactance if there were
no saturation and if field and armature currents were unchanged
. m.m.f. for iron /
8 = ETjE p = saturation factor* = 1 + f f' = AC AB
m.m.. or air gap
in Fig. 19
Xd = unsaturated synchronous reactance
Xm = Xd - xp = reactance equivalent to the armature reaction if
there were no saturation
EI = armature voltage which would be induced by field current I I if
there were no saturation
*In Refs. 3, 10, and 12 the saturation factor is denoted by k instead of by 8. How-
ever, k is deemed misleading for a quantity which is not constant.
.b~XAMPLE 7 281
It is clear from the vector diagram that the terminal conditions would
be the same if the saturated machine were replaced by an equivalent
unsaturated machine having an e.m.f.
E = Ef [39]
s
and a synchronous reactance
xp
Xs = xp +-sm =
X
xp + Xd -
s
[40]
Equations 39 and 40 give the saturated values of the excitation voltage

and of the synchronous reactance, respectively. Furthermore, the
load angle Ot between the excitation voltage and the terminal voltage
is the same for the equivalent machine of Fig. 15 as for the actual
machine.
When a saturated machine is represented according to eqs. 39 and
40, it must be borne in mind that s varies with E p and hence with load.
Therefore, in plotting a power-angle curve for constant field current,
variation of s must be taken into account. Kingsley 12 has devised
a method of obtaining the power-angle curve of a saturated cylindrical-
rotor machine connected through series impedance to an infinite bus.
His method, which is based on eqs. 39 and 40, together with means for
finding the values of s and related quantities, is illustrated in Example 7,
which follows. Example 7 is taken from Ref. 12 with permission.
EXAMPLE 7
A cylindrical-rotor synchronous generator whose open-circuit and short-
circuit characteristics are shown in Fig. 16 and whose armature resistance
and Potier reactance are, respectively, 0.022 per unit and 0.105 per unit
+
transmits power through external series impedance of 0.092 jO.436 per
unit to an infinite bus of voltage 1.00 per unit. The generator is operated
at constant excitation of 1.61 per unit. Calculate and plot the air-gap power
of the generator and the power received by the infinite bus as functions of
the angle 0 between the excitation voltage and the voltage of the infinite bus.
Solution. Let the following notation be used (see Fig. 15):
I = current
V = V /0 = voltage of infinite bus
E p = voltage behind Potier reactance
E, = E,/0 = unsaturated value of excitation voltage
E = E,/s = saturated value of excitation voltage
8 = saturation factor
P = air-gap power of the generator
(J
Pb = power delivered to the infinite bus

r, = external resistance
r a = armature resistance
r = r, +
r a = total series resistance
x, = external reactance
XI' = Potier reactance of armature
Xd = unsaturated synchronous reactance
{ ~
1.2
Air-gapline J ~
011
Open-circuit -
I /
i"'"
characteristic
1.0 2.0
I/
,
'"
~
C
::J
8. 0.8 1/ V 1.6~
......., J ~-Phase --
::J
&
i 0.6
/ short-circuit
characteristic - 1.2 1:
.......,
V
f--
/
(5
>
Q. ~
m.I
V V
::J
u
:i 0.4 0.8 =§
/ V e
'uI
0.2
VV 0.4~
1::
IV
o If/ 0
o 0.4 0.8 1.2 1.6 2.0 2.4
Excitation (per unit)
FIG. 16. Open-circuit and short-circuit characteristics of a cylindrical-rotor
synchronous machine (Example 7). (From Ref. 12, with permission.)
Xm = Xd
- x p = reactance equivalent to armature reaction when there is
no saturation
+
x, = xp xml8 = saturated synchronous reactance
+
x = x, x, = total series reactance, saturated value
Z = Z /90 0 - a = r + i» = (r e ra) + j(X, x,) + +
Zc = Zc/90° - a c = (r, + ra) + j(X e + Xp )
a = tan- 1 (rlx)
a c = tan- 1 [(r, + fa)/(X, + Xp ) ]
6 = angle by which E leads V
For the case considered in this example, the following circuit parameters
(all expressed in per unit) are given:
V = 1.00 fa = 0.022 r, = 0.092
E1 = 1.61 xp = 0.105 x, = 0.436
and several other quantities, which are independent of saturation, are
readily calculated.
EXAMPLE 7 283
From the curves of Fig. 16 at an excitation of 1.2 per unit, the open-circuit
voltage on the air-gap line is 1.2 per unit and the short-circuit current is
1.24 per unit. Hence
Xd = -1.20 = 0 .970
1.24
Xm = Xd - xp = 0.970 - 0.105 = 0.865
r = r, + fa = 0.092 + 0.022 = 0.114
Xe + Xl' = 0.436 + 0.105 = 0.541
Zc = V (r e + r a )2 + (x. +X p )2 = V (0.114)2 + (0.541)2 = 0.553
r, + fa 1 0.114
= tan- 1 - - - = tan- - - = 11.9
1
ae = tan- 0.210 0
z; + Xl' 0.541
In terms of the notation given above, the power-angle equations 31 and
32 for the case of series impedance may be rewritten as follows:
EV . E'lr
Pa = Z SIn (0 - a) + Z2 (a)
Ph = -
EV .
Z
sm (0 + a) - V 2r
-
Z2
(b)
Upon substitution of E = EJls (eq. 39), eqs, a ann b become:

EfV . E f 2r
pa = - SIn (0 - a) + 0 (c)
sZ (SZ)2
EIV . V2r
Ph = -- sin (0
sZ
+ a) - -
Z2
(d)
In eqs, c and d, sZ, a, and Z2 are variables dependent upon saturation, which
varies with 0 even though E J and V are constant. Hence, to determine the
values of these quantities to be used in eqs, c and d, it is necessary to express
the Potier voltage E p (which determines the saturation) as a function of o.
This may be done as follows:
From Fig. 15,
Ep = V + ZcI (e)
and
E-V
!=-- (J)
Z
Hence
Ep = V + Zc E - V = ZeE + (Z - Zc)V (g)
Z Z
Since
E = E, (h)
8
and
Z - Zc = jXm (i)
S
(j)
Hence
sZE p = ZcEJ + jXmV
= ZcEJ /0 + 90° - a c + XmV /90° (k)
The right-hand side of eq. k is the sum of two vectors differing in phase by
an angle (0 - ac). Hence, by the law of cosines, the magnitude of the vector
sum is given by
(SZE p)2 = (ZcE,)2 + (x mV)2 + 2Z cx mEJV cos (0 - a c) (l)
which, for the case under consideration, becomes:
(SZE p )2 = (0.553 X 1.61)2 + (0.865 X 1.00)2
+ 2 X 0.553 X 0.865 X 1.61 X 1.00 cos (0 - 11.9°)
= 1.54 +
1.54 cos (0 - 11.9°) (m)
From eq. l or m, for any assumed value of 0, the corresponding value of

(SZEp )2 can be calculated. This quantity is a function of E p , as are also
variables sZ, a, and Z2, which appear in eqs. c and d. By assuming various
values of E p , these values can be calculated, and curves can be drawn
showing sZ, a, and Z2 as functions of (sZE p ) 2. The values read from the
curves can then be substituted in eqs. c and d to find P a and P b for any
value of o.
The construction of the auxiliary curves will be illustrated for one assumed
value of E p , namely E p = 1.00.
The saturation factor 8 is defined as the abscissa of the open-circuit
characteristic divided by the abscissa of the air-gap line. For E p = 1.00,
from Fig. 16,
8 = 1.19 == 1.19
1.00
0.865
X8 + X-sm = 0.105 +--
= xp
1.19
= 0.105 + 0.726 = 0.831
x = X e + z, = 0.436 + 0.831 = 1.27

Z = vr 2 + x = V (0.114)2 + (1.27)2 = 1.27
2
Z2 = 1.61
sZ = 1.19 X 1.27 = 1.51
(SZE p )2 = (1.51 X 1.00)2 = 2.28
ct = tan- 1 (rlx) ;= tan- 1 (0.114/1.27) = tan- 1 0.090 = 5.1°
EXAMPLE 7 285
By carrying through similar calculations for several other assumed values
of E p , points can be calculated for plotting the auxiliary curves of Fig. 17.
By use of the auxiliary curves, points on the power-angle curves can be
calculated. The following calculations for one assumed value of 0 illustrate
the method:
Let 0 = 60°. From eq. m,
(SZE p )2 = 1.54 + 1.54 cos (60.0° - 11.9°)
= 1.54 + 1.54 cos 48.1° = 2.57
2.0 Z2
........... <,
~
1.8
-, / ~
\ ./
-:
"<~
./
~" '\
~ ~~ ~
7
"- -, Q!-
--- ~
......---~ K
1.2
...--
a:
~~
Z2 -,
1.0 3
o 2 345
(sZEp )2 (per unit)
FIG. 17. Auxiliary curves for Example 7. (From Ref. 12, with permission.)
From the curves of Fig. 17 at abscissa 2.57, the following quantities are read:
sZ = 1.54, Q! = 5.2°, Z2 = 1.55
Substitution of these values and of 0 = 60° into eqs. c and d gives:
P = 1.61 X 1.00 sin (600 _ 5.20) + (1.61)2 X 0.114 = 0.980
a 1.54 (1.54)2
Ph = 1.61 X 1.00 sin (600 + 5.20) _ (1.00)2 X 0.114 = 0.876
1.54 1.55
The complete power-angle curves are plotted in Fig. 18. This figure also
shows test results. The agreement is good, the calculated values of power
being, at the worst, about 0.02 per unit below the test results.
The maxima of the curves are:
Pa(max) = 1.24 per unit
Pb(max) = 1.02 per unit
These maxima can be found, if desired, without plotting the complete
power-angle curves. To find P a(max) assume an approximate value of ex,

+
substitute a = 90° a into eq.l, and compute (sZEp)'J. From the auxiliary
curve, read sZ and substitute it into eq. c, also putting sin (8 - ex) = 1.
Similarly, Pb(max) is found by putting 0 = 90° - a in eq. l, computing
(sZEp ) 2, reading sZ and Z2 from the auxiliary curves, and substituting
their values into eq. d, also putting sin (0 a) = 1. +
1.2
P~ "".- ---..
~
/
1.0 Y Pb
~~
t ~
V" <,
d ~1'
II
IJ"'
r
r
,J
0.4
)'
0.2 V
I
oV
o 20 40 60 80 100
Power angle 0 (elec. deg.)
FIG. 18. Power-angle curves for Example 7 by calculation and by test. Test
points are shown by small circles. (From Ref. 12, with permission.)
Equivalent synchronous reactance. For general use in stability

studies, it is preferable to use, instead of the saturated synchronous
reactance and excitation voltage, which vary with load as illustrated
in the foregoing example, a reactance and an e.m.f. which are constant
during small changes in load, even though their values depend upon the
initial operating condition. Such quantities are called equivalent syn-
chronous reactance and equivalent excitation voltage. If they are used,
it is possible to apply the test for stability, previously discussed, that
the synchronizing power dP/d8 be positive.
The problem may be thought of as that of finding the constant e.m.f,
and reactance of an unsaturated machine which, for small changes of
load, is equivalent to a given saturated machine. The two machines
are equivalent if, for given terminal conditions V and I, the same
change of V occurs for the same small change of I. In other words, the
EQUIVALENT REACTANCE 287
derivative dV/dI (or its reciprocal) should be the same for both
machines.
For present purposes consider the point behind Potier reactance as
the terminal. Then the external network (including Potier reactance)
is not affected by saturation. The relation between voltage and current
of the saturated machine is
E = EJ _ jXm I [41]
p 8 s
and of the equivalent unsaturated machine

Ep = E/' - jxm'I [42]
where EI , Ef ' , Xm , and xm ' are constant, while s is a function of E p ,
The derivative dI/dE p will be found for each machine, and the two
expressions will be equated to find reactance xm' . For the saturated
machine, from eq. 41,
sE p = E f - jxmI
I = Ef - sE p
jXm
dI 1 d(sE p )
-=---- [43]
dE p jXm dE p
For the equivalent unsaturated machine, from eq. 42,
I = E/. -,E p
JXm
dI 1
dE p - - jx:! [44]
Equate eqs. 43 and 44 and solve for xm ' , obtaining

I dE p
X m = Xm d(sE p ) [45]
The equivalent synchronous reactance is

dE p
= X p + Xm = X p + (Xd
I
:teq - :tp ) d(sE ) [46]
p
The real derivative, dEp/d(sEp), equals the slope of the no-load

saturation curve if the latter is regarded as a plot of E p against BEl'
(Fig. 19); or, for any other scale of abscissas, the derivative equals
the slope of the no-load saturation curve at ordinate E p divided by
the slope of the air-gap line.
However, in eq. 46 Ep is complex, and the value of the derivative

varies with the phase angle between Ep and dEp , as illustrated in
Fig. 20 for a finite increment ~Ep. In Fig. 20a, aEp is in phase with
Ep • Here the derivative
dEp s», [47]
d(sE p ) = d(sE p )
Ep /
Air-gap line, // /
slope 1 V =/
/ //~/
A .PL~C ~
/;1' No-load
Tangent, -' / saturation
dE~ "
slope d(sE
~ ". // curve
)p,,- /
"./ /
/ /
/",," / 'Secant,
/ Ep = J.
slope sE
/ p S
/
/
~/
~~
o s~
FIG. 19. The slopes of the tangent and the secant through the operating point of
the no-load saturation curve are the minimum and maximum values, respectively,
of the magnitude of the derivative dEp/d(sEp) which appears in eq. 46 for equivalent
synchronous reactance.
which is real and has a magnitude which, as already stated, is equal

to the slope of the saturation curve. In Fig. 20c, on the other hand,
~Ep is in quadrature with Ep • The saturation factor 8 is the same
before and after addition of the increment dEp , the two triangles are
similar, and the derivative is now
dE p s, 1
[48]
d(sE p ) = sE p = ~
It is again real, but its magnitude is greater than in Fig. 20a and is
represented geometrically by the slope of the secant through the
operating point on the saturation curve (Fig. 19). In Fig. 20b, aEp is
neither in phase with nor in quadrature with E p . There is some change
in saturation, but not so much as there was in Fig. 20a for the same
absolute value of I~Epl. The two triangles are not similar, ~(8Ep)
is not parallel to dE p , and hence the derivative in eq. 46 is complex.
EQUIVAIJENT REACTANCE 289
Consequently, the equivalent reactance given by that equation is

complex; or, in other words, the equivalent impedance is not a pure
reactance. Its resistance component is usually neglected, however.
The magnitude of the derivative in b lies between the two extremes
of a and c. Hence the equivalent synchronous reactance lies between
the extremes
slope of no-load saturation curve
Xeq(min) = X p + (Xd - X p ) f .. [49]*
slope 0 air-gap hne
and
Xd - x
xeq(max) = xp + ._---
s
p
[50"]
tS2Ep2
I ~(sEp)
I ~(8Ep)
E p2 ~slEpl SEplr-'1sE p2
~Ep I~E I
Epl Epl Ep2
(a) (b) (c)
~p dE p AE p 1
A(sE p ) == d(sE p ) A(sE p ) =~
FIG. 20. Showing the effect of phase on the derivative dEp/d(sEp).
For the same degree of saturation, the minimum value of equivalent

reactance occurs when the saturation is varying most rapidly; the
maximum, when the saturation is not varying. Both are smaller than
the unsaturated synchronous reactance Xd.
The angle between E p and dE p , and hence the value of equivalent
synchronous reactance, for given initial values of Ep and I depends
upon the external network, including all other machines, and upon
the nature of the small change of load (which, in testing for stability,
is usually an increase in angular displacement). This dependence may
*Equation 49 is similar to the expressions for equivalent reactance given by
Clarke 2 and Kilgore.l!
be readily appreciated if the external network is represented by an

impedance Ze~t and an e.m.f. Ee~t according to Thevenin's theorem,
as shown in Fig. 21.
A full investigation of the dependence of X eq upon the parameters
of this circuit is difficult. However, at or near pullout, a reactance
given by the average of eqs. 49 and 50 is approximately correct and is
recommended for general use in steady-state stability studies. This
average value is
Xeq(avg) ~ Xp +(
Xd - Xp )
-a:
S
S'BC [51]
ag
where Xp = Potier reactance.
Xd = unsaturated direct-axis synchronous reactance.
Stan = slope
of tangent to no-load saturation curve at point
whose ordinate is Ep •
S,ec = slope of secant through the same point and the origin.
Sag = slope of air-gap line.
The reciprocal of the short-circuit ratio (Chapter XII, p. 128) is often

used as an approximate value of the equivalent reactance of turbo-
generators, especially when operated
Zeq -jxp Ze,d +jxp underexcited.
The internal voltage E eq is simply
the voltage required behind X eq to
Ep give the actual initial conditions at
the terminals of the machine.
The angular position Oeq of voltage
E eq is not the same as the angular
Machine under I External position 0 of the rotor of the ma-
consideration I network
chine. However, if 0 is advanced,
Oeq is also advanced; and if 0 is re-
FIG. 21. Representation of a syn- tarded, Oeq is also retarded. Hence,
chronous machine connected to an
external network containing other the algebraic sign of dP/doeq is the
machines. same as that of dP/do, and a valid
criterion for stability is that dP/doeq
be positive. In other words, if a machine is represented by its equivalent
synchronous reactance and equivalent excitation voltage, it may be
tested for stability by advancing the phase of its equivalent excitation
voltage E eq and noting whether or not its power output increases.
EXAMPLE 8
Determine the correct value of equivalent synchronous reactance of the
generator of Example 7 when it is operating at its power limit under the
EXAMPLE 8 291
conditions of that example. Compare this value of reactance with the

minimum, maximum, and average values computed from eqs. 49, 50, and
51 and with the reciprocal of the short-circuit ratio.
Solution. Conditions at pullout. From Fig. 18 at the maximum of the
curve of P versus ~,
(J
o= 99°, P II = 1.24, and P b = 1.01

From eq. m of Example 7, with 0 = 99°,
(SZE1' )2 = 1.62
and from Fig. 17 at abscissa 1.62,
sZ = 1.46
Hence
s, = V (sZE = V1.62 p)
2
= 0.874
sZ 1.46
The power transmitted from a point where the voltage is E p through &
an impedance Zc/90° - a c to a point where the voltage is V /0 is, by
analogy to eq. a of Example 7,
EpV.
P (I =- - SIn
z,
(~
up - ac) + -Ep
Zc
2r
2
(a)
Substitution of numerical values gives

1.24 = 0.874 X 1.00 sin (~ _ 11.9~) + (0.874)2 X 0.114
0.553 p (0.553)2
= 1.58 sin (op - 11.9°) + 0.285
0.955 = 1.58 sin (01' - 11.9°)
0.604 = sin (01' - 11.9°)
01' - 11.9° = 37.2°
01'= 49.1°
Ep = E 1' & = 0.874/49.1° = 0.575 + jO.660
The current is
I = Ep - V = 0.575 + jO.660 - 1.000
z, 0.553 /78.1 °
_ -0.425 + jO.660 _ 0.783 ~
- 0.553 /78.1 ° - 0.553 /78.1 0
= 1.42/44.7° = 1.01 + jt.OO
The power delivered to the infinite bus is
Pb = V X Re (I) = 1.00 X 1.01 = 1.01
which agrees with the value read from the curve.
The copper loss in the circuit is

Pr = (1.42)2 X 0.114 = 0.23
Hence the air-gap power of the generator is
Pa = Pb + 1 2r = 1.01 + 0.23 = 1.24
This value agrees with the value read from the curve and used in the fore-
going calculations.
Determination oj equivalent reactance. At the point of pullout, dPa/d8 = O.
The equivalent reactance is the constant value of reactance which, backed
by the constant voltage Eeq, will give dPa/d8eq = 0 (where 8eq is the angle
by which E eqleads V) with the condi-
H tions deduced above. This reactance
may be found graphically by means
of a Clarke diagram, constructed
somewhat differently from the way de-
scribed before, because the power at
pullout is now known and the react-
ance is sought, whereas previously
the reverse was true. Another differ-
B ence to be taken into account is that
in the present example the condition
considered is that of maximum power
at the sending end of the circuit in-
~-==t=::+:=~~:.1!:""'_---:b I stead of at the receiving end. The
angular separation for this condition
is Oeq = 180 0 - 8 12 •
The construction of the Clarke dia-
gram is shown in Fig. 22. From point
o as origin, the following vectors are
drawn to scale:
G
FIG. 22.Determination of equivalent v = 1.000 +jO
reactance from Clarke diagram, E p
and the power limit being known E p = 0.575 + jO.660
(Example 8). I = 1.01 + jl.OO
From point F, the tip of vector E p , a line FH is drawn perpendicular to I.
Line AB, the perpendicular bisector of V, is then drawn, intersecting line
FH at point C. With C as center and radius CO, a circular arc is drawn
intersecting FH at point D. Length FD, which represents I(x eq - xJJ) , is
measured and is found to be 1.00 unit of voltage. Hence
I (x eq - x p)
X eq - Xp = = -1.00 = 0.705 per unit
.
I 1.42
x eq = 0.705 + xp = 0.705 + 0.105 = 0.810 per unit
EXAMPLE 8 293
/vI inimum. maximum, and ave1'age values of equivalent reactance. In
Fig. 16 at E p = 0.874, the slope of the tangent to the open-circuit charac-
teristic is
s= 1.30 - 0.27 = 1.03 = 0.62
tan 1.67 - 0 1.67
and the slope of the secant at the same point is

S = 0.874 = 0.91
,ee 0.96
The slope of the air-gap line is

Sag = 1.00
By eq. 49, the minimum value of equivalent synchronous reactance is
Xeq(min) = 0.105 + (0.970 - 0.105) X 0.62/1.00
= 0.105 + 0.865 X 0.62
= 0.105 + 0.535 = 0.64
By eq. 50, the maximum value is
Xeq(max) = 0.105 + 0.865 X 0.91/1.00
= 0.105 + 0.786 = 0.891
= 0.89
By eq. 51, the average value is
X,q(ava) = 0.105 + 0.865·VO.62 X 0.91/1.00
= 0.105 + 0.650 = 0.755
= 0.76
Equivalent reactance as reciprocal oj short-circuit ratio. From Fig. 16, the

excitation for unit short-circuit current is 0.96 per unit and that for unit
open-circuit voltage is 1.20 per unit. Hence
X eq = S.C1.R. = -0.96
1.20
=
0 ·
0.8 per unit
Conclusions. The following synchronous reactances were found:

Unsaturated value 0.97 per unit
Equivalent values
Minimum 0.64
Average 0.76
Maximum 0.89
1/S.C.R. 0.80
Actual 0.81
The actual value for the conditions of this problem is very close to the
reciprocal of the short-circuit ratio and does not differ greatly from the
average value.
Representation of loads. Loads, as well as synchronous machines,

have nonlinear characteristics although for convenience they are
usually represented by constant impedances. The impedances are
given such values that at normal voltage they consume the same
active and reactive power as the actual load does. The advantages
Pand Q Pand Q
100 100 \ \ \ 100
~
90
Q=O
~ 90 ty·"
, Q ,
\
\
\ 90~
I (lead) I P Q\ P
,,
.5 .5 .5
11f 1 1f 21 (lag),
~ 80 ~ 80 80~
I I I
I I
I I
70 70 I I 70
(a) Incandescent lamps (b) Synchronousmotors (c) Induction motors
Pand Q Pand Q
100 ,
V
\ O-c power
loadonly
/ \
100
\ 100
\ \
0,tl P g
~ 90 ,\~
" , >.
\ 90~ \ 90~
1,0 Q\ Q\
\"0 1/ ~
"0
.5 .5
\ \
I
,~.... ~
~ 80 I'~ .r,
,, ,
80~
\ I
'Ql~ i/ u \ I
70
I irJ: ctj Q
I
70 I 70 .
(d) Synchronousconverter (e) Rectifier supplying (f) Typical composite load
aluminum cells
FIG. 23. Active power P and reactive power Q as functions of voltage V for
different kinds of load.
of using constant impedances to represent loads are that on the calcu-

lating board the load units need not be readjusted while a machine is
being tested for stability, and in paper calculation the network, includ-
ing loads, can be reduced to the simplest form and the equations of
power previously derived can be used.
In the accurate determination of steady-state stability and stability
limits, the nonlinearity of loads should be taken into account under
some circumstances (see next section). The characteristics of loads
may be represented advantageously by curves of active power against
voltage and of reactive power against voltage. The shapes of these
curves vary with the kind of load.
For a constant-impedance load, both active and reactive power vary
as the square of the voltage; hence the curves are parabolas.
The power taken by incandescent lamps varies as a power of the
REPRESENTATION OF LOADS 295
voltage between 1.5 and 1.6. (The resistance of metal filaments

increases with temperature.) The reactive power is zero. See Fig. 23a.
The active power taken by synchronous motors is practically in-
dependent of voltage unless the voltage drops so low that the motor
pulls out of step. The reactive power varies with voltage, becoming
more leading as the voltage decreases. See Fig. 23b. The exact re-
active-power characteristic may be obtained from a synchronous-
machine chart. 5
The active power taken by induction motors is nearly independent
of voltage because the slip is small. The reactive power varies with
voltage, however. See Fig. 23c. The principal component of reactive
power, due to the exciting current, decreases with a decrease of voltage;
another component, due to load current, increases with a decrease
of voltage.
The power taken by synchronous converters and rectifiers depends
upon the nature of the d-e. load. For d-e, lighting load, the power
varies with the voltage as it does for a-c. lighting load. For d-e. motor
load, the power is nearly constant. The reactive power of a synchronous
converter varies like that of a synchronous motor. The reactive power
of a rectifier is more like that of a constant inductive impedance.
See Fig. 23d and e.
It is not feasible to take into account individually many small loads,
Usually the entire load on a substation or in a large load area is treated
as a unit. Such a load consists mainly of the several kinds of load
mentioned above, combined in proportions which differ according to
whether the load area is predominantly residential, commercial, or
industrial, and also according to the time of day. A typical segregation
of daytime loads might be as follows:
Induction motors 60%
Synchronous motors* 10%
Lighting and heating 30%
The characteristics of a composite load may be found by adding the
values of active power of the different kinds of load at a particular
value of voltage to obtain the total active power a.t this voltage, and
by similarly adding the components of reactive power to obtain the
total reactive power. The losses of the distribution feeders and trans-
formers may be included also. The characteristics of a typical composite
load are shown in Fig. 23f.
The most important feature of the curves of active and reactive
*Large synchronous motors can be treated as individual machines if that appears
desirable.
power against voltage are their slopes at normal voltage. In most

cases the initial load voltages may be assumed to be normal when a
test of stability is being made, regardless of the initial loads and of
the substation bus voltages. Most power systems have feeder-voltage
regulators which hold the voltages at the individual loads nearly
constant. However, when a test for stability is made by means of a
small increment of angle and load, the feeder-voltage regulators are
assumed not to work. This assumption is consistent with the usual
assumption that generator-voltage regulators work too slowly to
affect stability. Hence the slopes dP/dV and dQ/dV at V = 100%
suffice to determine the action of the load during the test for stability.
If the power of a load varies as the k'th power of the voltage
(P = KV k ) , then
dP = k~ [52]
dV V
For constant impedance, k = 2; for constant current, k = 1; and for

constant power, k = o. A similar relation holds for reactive power.
For any curves of P and Q against V, the slopes at normal value of
voltage may be expressed by a dimensionless quantity k as in eq. 52.
Effect of loads. Heffron31 has made an analytical study of the

effect of a shunt load on the steady-state stability of a system con-
sisting of a finite machine connected through a reactance of 0.4 per
unit to an infinite bus. The load is at unity power factor and has
anyone of three characteristics: k = 0.5, 1.0, or 2.0 (eq.. 52), and
either of two magnitudes: 1 or 5 times the rating of the finite machine.
It is connected at a point such that the reactance between the load
and the finite machine is either 0.1 or 0.3 per unit. The results of his
study show that the effect of the load on the steady-state stability
limit is negligible unless the load is large (5 per unit) compared to the
generator rating and electrically close to the generator (x = 0.1 per
unit). If the load is heavy and close, it has a stabilizing effect if
k = 1 (constant current), and an even greater stabilizing effect if
k = 2 (constant resistance), but it has the contrary effect if k = 0.5
(intermediate between constant current and constant power). The
effect of heavy, close load on the stability limit ranges from a 40%
increase to a 15% reduction. These results indicate that it is not
worth while to take load characteristics into account except for heavy
loads electrically close to the generator being tested for stability.
Crary has presented an analytical method of finding the synchro-
nizing power of any machine of a system of any number of machines
GRAPHICAL METHODS 297
and composite loads, taking into account the slopes of the load char-
acteristics (see Ref. 9 or Chapter 4 of Ref, 20).
On a calculating board the impedance of loads may be adjusted
manually as the voltage varies in order to give the proper variation
of P and Q.
Graphical methods are also available for taking into account load
characteristics as well as saturation of synchronous machines. Some
of these methods are described in the next section.
Graphical methods. By the use of graphical methods it is possible

to take into account the nonlinear characteristics of loads and of
synchronous machines while checking the steady-state stability of a
system or while finding its steady-state stability limit. In these
graphical methods the power system is figuratively cut in two at some
point at which the power flow is of interest. Each half of the system
has certain relations between active power, reactive power, and voltage
(all measured at the point of observation). These relations are de-
termined for appropriate initial conditions and constant field currents
in all synchronous machines.
If there is more than one synchronous machine in the same part of
the system, the division of power between the several machines can
be determined according to whichever of the three assumptions previ-
ously discussed under "Multimachine system" is deemed most appro-
priate. Synchronous condensers, however, should be assumed to take
no active power except when the stability of a condenser itself is being
checked.
The relations between voltage, power, and reactive power for each
half of the system can be plotted in any of three different ways:
1. Reactive power may be plotted against active power for con-
stant voltage, and a family of such curves may be drawn for different
values of voltage (see Fig. 24).
2. Voltage may be plotted against reactive power for constant
active power, and a family of such curves may be drawn for different
values of active power (see Fig. 25).
3. Voltage may be plotted against active power for constant
reactive power, and a family of such curves may be drawn for
different values of reactive power (see Fig. 26).
When the two parts of the system which were cut apart are rejoined,
the voltages of the two parts at the junction point are equal, and the
output of active and reactive power from one part equals the input of
active and reactive power, respectively, to the other part. Sets of
values of voltage, active power, and reactive power which satisfy both
parts of the system can be found by superposing the graphs for one
part upon the graphs of the same type for the other part. Thus, if
the first type of plot listed above (reactive power versus active power)
FIG. 24. Loci of constant voltage V in the complex power plane, P + jQ, for a
circuit consisting of constant e.m.f, E == 1 in series with constant impedance
Z = 1/75°.
is used, the intersection of the curve for the sending system for 100%
voltage and the curve for the receiving system for the same value of
voltage is one point which satisfies both parts. Another such point is
the intersection of the two curves for 90% voltage, and so on for other
values of voltage. See Fig. 27. A curve drawn through these points
(a curve of reactive power versus active power) is a characteristic of
the sending and receiving systems combined. Each point on the curve
is associated with a definite value of voltage. Hence the information
given by the curve could be replotted as active power against voltage

(Fig. 28) or as reactive power against voltage.
Similar information can be obtained by superposition of plots of the
second or third types. From plots of the second type, a composite
curve of voltage versus reactive power is obtained. The information
- - 2.0
~O .r.
y;;·~O
::::::::'0 z O.r.
I ... & ~ ~ :::::--::: 0 .0
.>
~
v .......
z ....... ~ ~
~
~ .........- ./
V
-:
~
~~ ~V/
6
/' ./ I ......- 10-
P~
/)V//V/V ,
\
---- V ......
f-
~.
~1I // -~
~
V ...
~~ ~
~
( V
v
...........:::: ,...
I p~ ~
~-~
\, "-
V ~
~
\ p~y.
- V~
-O~
~~
-,~ o
~9 1
1.0 2.0
Q
FIG. 25. Voltage Vasa function of reactive power Qfor constant active power P.
These curves are drawn for the same circuit as the curves of Fig. 24.
can be replotted as voltage versus active power or as reactive power

versus active power. From plots of the third type, a compositecurve
of voltage versus active power is obtained. The results can be re-
plotted as voltage versus reactive power or as reactive power versus
active power.
It is also possible to visualize (though not practical to plot) the
characteristics of each half of the system as a surface in three-dimen-
sional coordinates of active power, reactive power, and voltage. Each
of the three types of families of curves for each half of the system
results from intersections of the surface with a set of planes parallel
to one of the coordinate planes and the projection of the intersections
onto that plane. The common characteristic of the two halves of the
system is the three-dimensional curve resulting from the intersection
of the two surfaces. This curve can be projected onto anyone of the
three coordinate planes as a two-dimensional curve.
Of the three two-dimensional curves which satisfy both halves of the
system, those of voltage against active power (Fig. 28) and of reactive
,--..--.....---...,....-.-,.------2.,...-..-....----------
FIG.26. Voltage V as a function of active power P for constant reactive power Q.

These curves are drawn for the same circuit as the curves of Figs. 24 and 25.
power against active power (Fig. 27) are most useful. From either of
these curves the value of maximum power of either sign can be read. *
This is the power limit at the point of observation for constant field
currents in the synchronous machi-nes and for the assumed division of
power between machines of each group. This power limit mayor may
not be a stability limit. Suppose, for illustration, that the point
of observation lies between a synchronous generator and a static-
impedance load. In this case there will be a maximum power but no
*In practice only part of the curve need be constructed, depending upon the
assumed direction of power flow.
question of loss of synchronism. Again, suppose that the point of

observation lies between a synchronous generator and a combination
of a static-impedance or composite load and a large synchronous motor
or condenser. The critical point on the curve as regards stability may
lie on one side or the other of the point of maximum power.
Z=1.0~ Z=O.8m
-1
\
I
I I
I I
<t.\ I
I
0\
1\1 I
>1\ /
I
FIG. 27. Curve of reactive power Q versus active power P that satisfies the parts
of the circuit on both sides of the point of observation. Points on this curve are
obtained from the intersections of the loci of Fig. 24 for the left part of the circuit
with similar loci for the right part of the circuit and equal voltage.
Check for stability of synchronous machines. In order to check

whether a power system is stable or unstable at some particular oper-
ating condition, the point of observation should be taken in turn
between each synchronous machine and the remainder of the system.
According to the conservative criterion of steady-state stability, previ-
ously discussed, a machine is stable if its synchronizing power dPIda
is positive. In order that the curves of power P against voltage V or
reactive power Q may be used in this connection, an alternative
criterion for stability will be formulated. It is found that as the
angular separation between a machine and the remaining machines
is increased in either direction, the terminal voltage of the machine

decreases. At the condition of maximum power not only does dPIda =
o but also dP/dV = 0 and dP/dQ = O. If the machine is operating
stably, the voltage V is higher than it is at maximum power; whereas,
if the machine is unstable, the voltage is lower than at maximum power.
(The alternative criterion for stability is not expressed in terms of the
sign of the slope dP I dV because this depends upon whether the machine
being examined is operating as a motor or as a generator.) Thus it is
possible to check the stability of each synchronous machine by finding
~ ~
-/ ~
1.0
I( "\
v
.~
, }
~ J
1/
~,
~~
V
o
-1.0 o +1.0
p
FIG. 28. Curve of voltage V versus active power P that satisfies the parts of the
circuit on both sides of the point of observation.
a curve (such as that of Fig. 28) of active power against voltage at the
terminals of the machine (with constant field currents) and noting
whether at the initial operating condition the voltage is higher or lower
than it is at maximum power. A curve of active power versus reactive
power can be used in similar fashion if values of voltage are marked
on it (Fig. 27).
Synchronous condensers are checked for stability in the same manner
as synchronous motors.
Induction machines. The stability of an induction machine may be
checked by the same procedure as for synchronous machines. In case
of the induction machine there is no question regarding loss of synchro-
nism. Nevertheless, if an induction motor cannot develop as much
torque as required by its shaft load, it will "break down" and come to
rest. This condition is called instability. It may occur because of
low voltage. In checking for stability the assumption is made that
the induction motor operates stably at any slip less than the slip asso-
ciated with maximum power. Actually the motor may operate stably
at greater slips provided the slope of the curve of motor torque (or
PREPARATION OF CHARTS 303
power) against slip is greater than the slope of the curve of load torque
against slip. Hence the use of the same criterion for induction machines
as for synchronous machines is conservative.
If all machines of a system are stable, the system is stable.
Stability limit. In order to find the steady-state stability limit
graphically, the load carried by a particular machine or line is in-
creased, the values of bus voltage and the division of power among
machines being adjusted in accordance with established or proposed
operating procedure. Then, keeping the field currents constant, the
stability of 'each machine is checked as described above. These steps
are repeated as many times as necessary. The stability limit is reached
when, in the check for stability, maximum po\ver of any machine occurs
at the initial voltage (or reactive power).
Preparation of charts. The general principles of graphical methods

of examining steady-state stability have been discussed, but more
needs to be said on ways of preparing the charts that are used.
Charts are required for the separate components of the power system,
such as synchronous machines, transmission lines, transformers, and
loads. In addition, methods are needed for deriving the charac-
teristics of a part of the system consisting of two or more of these
components.
Transmission lines. Evans and Sels charta' are used in many
problems dealing with transmission lines with or without terminal
transformers. They are applicable to any linear circuit having a pair
of input terminals and a pair of output terminals. The sending chart
has rectangular coordinates Qs and P s, the reactive and active power
input to the sending end of the line. In these coordinates are drawn
circular loci representing constant magnitudes of sending-end voltage
V s and receiving-end voltage V R, with varying phase difference
between V s and V R. A family of such circles is drawn either for the
same value of V s and different values of V R or for the same value of
V R and different values of VB. In the sending chart for constant V 8
(shown in Fig. 29) the circles for various values of V R are concentric,
and radial lines may be drawn which are the loci of constant phase
angle between V B and V R. In the sending chart for constant V R
(shown in Fig. 30) the circles for various values of V s are eccentric
and intersect one another.
The receiving charts (Figs. 31 and 32) differ from the sending charts
in that the rectangular coordinates are QR and PR, the reactive and
active power output at the receiving end, instead of Qs and P B. The
circles may be drawn either for the same value of V R and different
304 STEADY·STATE STABILITY
til: 120%
110%
100%
o ___9_0%;,,;;;;,.D~-';;::-"':~~~~~-----PS
80%
70%
60%
Center
o
FIG. 29. Evans and Sels sending chart for constant sending voltage.
~=80%
~ __-_-_;;:_..._0lIIIIII:~~---------&
FIG. 30. Evans and Sels sending chart for constant receiving voltage.
PREPARATION OF CHARTS 305
values of VB, in which case they are concentric (Fig. 31), or they may
be drawn for the same value of V B and different values of V R, in which
case they are eccentric (Fig. 32).
o
Center
FIG. 31. Evans and Sels receiving chart for constant receiving voltage.
Briefly stated, the theory of these charts is as follows:

To derive the receiving charts, the sending voltage may be expressed
in terms of the receiving voltage and current and the general circuit
constants, thus:
[53]
which, solved for IR, becomes:
A 1
IR - VR +. B
= - B - Vs [54]
The receiving-end vector power is

- A 2 1 / [55]
PR+jQR=VRIR= -BVR +BVRVB~
Here 5 is the angle by which VB leads VR. For constant V R and V sand
varying 0, this equation represents a circle in the PR-QR plane. The
position of the center is given by the first term and the moving radius
by the second term. This is but one circle of the family of circles on
the receiving chart, others being obtained by giving new values to
either VR or Vs- In the latter case, the position of the center is un-
changed; in the former case, the centers lie on the same straight line at
various distances from the origin.
FIG. 32. Evans and Sels receiving chart for constant sending voltage.
The sending charts are derived in a similar manner. The receiving

voltage, in terms of the sending voltage and current and the general
circuit constants, is:
VB = DVs - BIs [56]
Solve for Is, obtaining:

D 1
Is = - Vs - -VB [57]
B B
The sending-end vector power is:
- D 1 ~
Ps+iQs=VsIs=-V s 2 --VRVs -6 [58]
B B
This equation represents a circle in the P s-Qs plane. Again the
position of the center is given by the first term. It is on the opposite
side of the origin from the center of the receiving-chart circle. The
second term represents the moving radius. As 0 increases, this sending
PREPARATION OJi' CHARTS 307
circle is traversed clockwise, whereas the receiving circle was traversed

counterclockwise.
The theory, construction, and use of the charts are explained in
greater detail in Ref. 4.
Synchronous machines. A useful form of synchronous-machine
chart is similar to the Evans and Sels receiving chart. The rectangular
coordinates of the generator chart are active and reactive power output.
In these coordinates are drawn loci for each of which both the field
current and either the terminal voltage or the voltage E p behind
Potier reactance are constant. The latter alternative has the ad-
vantage that for constant E p the saturation is constant; hence, as
previously shown for a nonsalient-pole machine, the saturated values
of excitation voltage Ejls and of reactance corresponding to armature
reaction xm/s are constant. Consequently, the locus for constant
values of E j and E p is a circle, which is easily drawn. By assigning
various values to E j on the one hand or to E p on the other hand, one
or the other of the two types of receiving charts is obtained. For
checking the stability of a synchronous machine graphically, the chart
for constant E j (corresponding to constant field current) is used. A
family of such charts may be prepared for different values of E p •
In the chart for a salient-pole machine, the loci of constant E j and
E p are not circles, but are curves that can be constructed from the
vector diagram (Fig. 64 of Chapter XII).
If a chart is desired having loci for constant terminal voltage in-
stead of for constant E p , it can be derived as explained below for
tandem combinations.
Tandem combination of synchronous machine and transmission line.
Suppose that a plot of QR versus PR is desired for various constant
values of V R at the receiving end of a transmission line, to the sending
end of which is connected a synchronous generator operated at constant
field current. The Potier reactance of the generator may be included
in either the generator chart or the transmission-line chart. In either
event the voltage at the junction point will be referred to as V 8 in
what follows.
In order to obtain the desired composite chart, the generator chart
for the given value of field current and the transmission-line sending
chart for some particular value of V R are superposed. The inter-
section of each curve on the generator chart for a particular value of
V 8 with the curve on the transmission-line chart for the same value
of V B is recorded. These points are then transferred from the sending
chart to the receiving chart for the same value of V R by noting the
values of V 8 and of angle between V 8 and V R. The result is a curve
of QR against P R for the particular value of V R. The process is then

repeated for other values of V R, resulting in a family of curves.
Parallel combination of machine, or machine and line, with load.
The load characteristics, discussed in an earlier section of this chapter,
are plotted as curves of P versus V and of Q versus V. For a given
voltage V, the active power of a parallel combination is equal to the
sum of the active powers of the two parallel circuits; likewise for the
reactive power. Hence each curve of Q versus P for a machine, or
machine and line, at a particular value of V is simply shifted vertically
by the Q of the load and horizontally by the P of the load at the same
value of V, with no change in the shape of the curve. Each curve
for a different value of V is shifted by different amounts.
Parallel combination of two or more machines with or without lines and
loads. As in any parallel combination, the P and Q of the combination
at any particular value of V equal the sums of the P's and Q's, re-
spectively, of the several parallel branches at the same value of V.
However, these sums are not uniquely determined by the voltage V
because for each machine there are an infinite number of pairs of
values of P and Q, represented by points on a curve, and any pair for
one machine can be added to any pair for another machine. Therefore,
one of the assumptions previously considered on division of power
must be chosen. If it is assumed that the phase angles between
excitation voltages of the various machines in the group are constant,
then corresponding to any point on the curve for one machine there
is a certain point on the curve for each other machine where the phase
difference between machines has the required value. If, on the other
hand, it is assumed that all machines but one in the group have con-
stant power outputs, then various points are taken on the curve for
the one machine whose output varies, but fixed points are used for
the others.
By means of tandem and parallel combinations, the characteristics
of almost any network on one side of the point of observation can be
derived. The superposition of the characteristics of one side upon
those of the other side leads, as already explained, to a composite
characteristic by means of which the steady-state stability of a system
can be examined.
In practice the entire characteristics need not be plotted, but only
that part of them in the vicinity of the operating point.
Resume. In order accurately to take into account the nonlinear

characteristics of machines and of loads, graphical methods consisting
of superposition of charts can be used for determining whether or
VOLTAUh: REUUl,ATORS 309
not a system is stable in the steady state. However, because of the

immense amount of labor required in the graphical method and be-
cause of the fact that steady-state stability limits usually are not
needed with exactness, approximate methods are generally used in
which loads are represented by constant impedances and synchronous
machines, by constant reactances and voltages. The reactances of the
machines theoretically should be the proper equivalent reactances for
the prevailing degree of saturation, external circuit, and change of
angle. Because of the difficulty of determining the correct value of
equivalent reactance for each condition, however, approximate values
are generally used. The steady-state stability of a multimachine
linear system is checked most easily by use of an a-c. calculating
board. The system is stable if dPIda is positive for each machine.
The steady-state stability limit of a two-machine linear system is
readily found by means of the Clarke diagram.
Steady-state stability with automatic voltage regulators. Whereas

automatic voltage regulators with dead band do not raise the steady-
state stability limit appreciably above that obtainable with close
manual control of excitation, properly designed excitation systems
with regulators of the continuously acting type can raise the limit
considerably.
A voltage regulator attempts to maintain the terminal voltage of
its machine constant, but it cannot succeed completely because of
insufficient amplification, limits or "ceilings," and delays in the exci-
tation system. It may be of interest, however, to consider the effect
of aperfect excitation system which would maintain constant terminal
voltage. We will consider three cases: (1) two equal finite round-
rotor machines, (2) one finite round-rotor machine connected directly
to an infinite bus, and (3) the same machine connected through
external reactance to the infinite bus.
Two finite machines. (See Fig. 33.) The power transferred from
one machine to the other is given by
[59]
where E 1 and E 2 are the excitation voltages of the two machines, 0is the
angle between these voltages, and Xl and X2 are the equivalent synchro-
nous reactances of the machines. As is well known, maximum power
occurs, for constant excitation, at 0 = 90°. If we take Xl = X2 = 1
and make the excitation voltages E 1 = E 2 = V2, so as to give rated
terminal voltage at rated current (Fig. 34), the power is

P = sin a [60]
and has the maximum value 1 at a = 90°, as shown in Fig. 35 by the
"unregulated" curve.
If the excitations are controlled so as always to maintain unit
terminal voltage, the required excitation voltages are (see Fig. 36) :
1
EI = E 2 = cos (a/2) [61]
and the power is
sin a a
p = 2 cos2 «(j/2) = tan 2 [62]
which is plotted as the "regulated" curve in Fig. 35. Here the power
increases beyond a = 90°, becoming infinite at a = 180°. Of course,
the excitation voltages and armature current would also become
infinite, which is obviously impossible.
FIG. 33. Two-machine system. FIG. 34. Voltage vector dia-

gram of two equal finite
machines at 8 == 90°. V t == 1
and I == 1.
One machine and infinite bus. If the machine is connected directly

to the infinite bus, the terminal voltage is constant, even without a
regulator, and the regulator may be used to keep the power factor
constant, say, at unity. Equation 59 then is valid with Xl = 1, X2 = 0,
and E 2 = 1. For the unregulated condition, let E 1 = V2, as before.
Then
v'2Xl. _J:-.
P = 0 sin a = v2smo [63]
1+
With regulation of power factor, E I = 1/C08 a, and
sin a
P=-=tana [64]
cos 0
These results are plotted in Fig. 37.
VOLTAGE REGULATORS 311
One machine, line, and infinite bus. Let the machine reactance = 1,
line reactance = 1, and voltage of infinite bus = 1. For the case
without regulation, let the excitation voltage E 1 be such as to produce
FIG. 35. Power-angle curves of two equal finite machines.
FIG. 36. Voltage vector diagram of two equal finite machines with V t maintained
constant at 8 > 90°.
unit terminal voltage at unit current. From the vector diagram

(Fig. 38), it is determined that E 1 = va.
Then, by eq. 59, the power
is
P = 2
vax
1. .
SIn 0 = 0.866 SIn 0 [65]
and has the maximum value of 0.866 at 0 = 90°, as shown by the

312 STEADY-8TATE STABILITY
"unregulated" curve of Fig. 39. For the case with regulation, let the
terminal voltage be kept at 1.
P = -I -XS1I n
.
a =
.
BID a [66]
1
2
1.4
P
6
FIG. 37. Power-angle curves of finite machine and infinite bus.
where a is the angle between the two fixed voltages V t and E 2 • Now
the power has the maximum value 1 at a = 90°. The corresponding
value of 0, found from the vector diagram of Fig. 40, is 116.6°. The
"regulated" curve of Fig. 39 shows this power as a function of o.
FIG. 38. Vector diagram of finite machine connected through reactance to an

infinite bus: 8 == 90°.
It is apparent from the foregoing calculations that, even with excita-

tion which could maintain the desired terminal voltage or power factor
exactly, the obtainable increase of power limit would vary considerably
VOLTAGE REGULATORS 313
with the type of power system, being infinite for two of the three cases
considered but only 15% for the third case.
The power systems which show more increase in steady-state power
limit due to an ideal excitation system also show more increase due to
an actual excitation system, although, of course, the increase is less
with an actual than with an ideal excitation system.
1
0.866
FIG. 39. Power-angle curves of finite machine connected through

reactance to infinite bus.
One of the most favorable cases, that of two machines (generator

and motor) connected together without external reactance, is encoun-
tered in electrical ship propulsion, and here the use of automatic
voltage regulators has effected a considerable economy in the size of
IXl =-/2
FIG. 40. Vector diagram of finite machine connected through reactance to an

infinite bus: a = 90°.
machines required. The cases encountered in long-distance power

transmission, in which there is a substantial amount of reactance be-
tween the points of controlled voltage, are less favorable as to the
amount of increase of power limit attainable by voltage regulators.
Actual excitation systems are unable to hold constant voltage at
the generator terminals or other external points; as a rule, they are
unable even to hold constant voltage behind transient reactance, in
attempting which they are aided by the long transient time constant.
Roughly speaking, the excitation system can increase the steady-
state power limit to a value corresponding to constant voltage behind

a reactance which is intermediate (usually about halfway) between
transient reactance and equivalent synchronous reactance. Hence
steady-state stability with automatic regulators, like transient stabil-
ity, is promoted by low transient reactance of the machines.
The use of series-wound main exciters can, as explained in Chap-
ter XIII, be made to maintain constant voltage behind transient react-
ance, this reactance being somewhat augmented by the reactance of
the series-field winding.
The steady-state stability of power systems having continuously
acting generator voltage regulators is analyzed by methods similar to
those used for analyzing servomechanisms. Such analysis is beyond
the scope of this book. The general theory of servomechanisms is
treated in many books. The application of this theory to the calcula-
tion of the steady-state stability limits of power systems with voltage
regulators is limited to a few articles by Concordia.19 ,24,26 His cal-
culations on excitation systems show increases in the steady-state
stability limit, resulting from the use of continuously acting voltage
regulators, of the order of 30% to 60% for two machines connected
without external reactance.
Rothe 32 states that the steady-state stability limit of a steam-turbine
generator of low short-circuit ratio, equipped with a continuously
acting voltage regulator, is roughly equal to that of a generator of
short-circuit ratio twice as great having a noncontinuously acting
voltage regulator.
REFERENCES
1. EDITH CLARKE, "Steady-State Stability in Transmission Systems: Calculation
by Means of Equivalent Circuits or Circle Diagrams," A.I.E.E. Trans., vol. 45,
pp. 22-41, February, 1926. Disc., pp. 80-94.
2. EDITH CLARKE, closing discussion of Ref. 1.
3. H. V. PUTMAN, "Synchronizing Power in Synchronous Machines Under Steady
and Transient Conditions," A.I.E.E. Trans., vol. 45, pp. 1116-24, September,
1926. Disc., pp. 1124-30.
4. O. G. C. DAHL, Electric Circuits - Theory and Applications, Vol. I, "Short-
Circuit Calculations and Steady-State Theory," McGraw-Hill Book Co., New
York, 1st edition, 1928. Chapter X, "Transmission-Line Charts,"
5. Same, Chapter XI, "Synchronous-Machine Charts."
6. R. A. HENTZ and J. W. JONES, "Stability Experiences with Conowingo Hydro-
electric Station," A.l.E.E. Trans., vol. 51, pp. 375-84, June, 1932. Disc., p. 384.
Abstract in Elec. Eng., vol. 51, pp. 95-102, February, 1932.
7. H. B. DWIGHT, "Adjusted Synchronous Reactance and Its Relation to
Stability," Gen. Elec. Rev., vol. 35, pp. 609-14, December, 1932.
8. EDITH CLARKE and R. G. LORRAINE, "Power Limits of Synchronous Ma-
chines," A.I.E.E. Trans., vol. 53, pp. 1802-9, 1934; also in Elec. Eng., vol. 52,
pp, 780-7, November, 1933.
REFERENCES 315
9. S. B. CRARY, "Steady-State Stability of Composite Systems," A.I.E.E.

Trans., vol. 53, pp. 1809-14, 1934; also in Elec. Eng., vol. 52, pp. 787-92, November,
1933. Appendix V. Disc., Elec. Eng., vol. 54, p. 477, March; p. 594, April, 1934.
10. S. B. CRARY, L. A. MARCH, and L. P. SHILDNECK, "Equivalent Reactance
of Synchronous Machines," Elec. Eng., vol. 53, pp. 124-32, January, 1934. Disc.,
pp, 484-8, March; 603-7, April.
11. L. A. KILGORE, discussion of Ref. 10, p. 488.
12. CHARLES KINGSLEY, JR., "Saturated Synchronous Reactance," Elec. Eng.,
vol. 54, pp. 300-5, March, 1935. Disc., pp. 1111-3, October, 1935.
13. STERLING BECKWITH, "Steady-State Solution of Saturated Circuits," Elec.
Eng., vol. 54, pp. 728-34, July, 1935.
14. O. G. C. DAHL, Electric Power Circuits - Theory and Applications, Vol. II,
"Power System Stability," McGraw-Hill Book Co., Inc., New York, 1938. Chap-
ters II through X.
15. CHARLES F. DALZIEL, "Static Power Limits of Synchronous Machines,"
A.I.E.E. Trans., vol. 58, pp. 93-101, March, 1939. Disc., pp. 101-2.
16. EDITH CLARKE and S. B. CRARY, "Stability Limitations of Long-Distance
A-C Power-Transmission Systems," A.I.E.E. Trans., vol. 60, pp. 1051-9, 1941.
Disc., pp. 1299-1303.
Engineers of the Westinghouse Electric & Manufacturing Co., East Pittsburgh,
Pa., 1st edition, 1942. Chapter 8, "Power-System Stability - Basic Elements of
Theory and Application," by R. D.. EVANS, pp. 175, 182-3, 185-8, 189-90, 195,
198, 200, 208.
18. JOHN GRZYBOWSKI HOLM, "Stability Study of A-C Power-Transmission
Systems," A.I.E.E. Trans., vol. 61, pp. 893-905, 1942. Disc., pp. 1046-8. .
by Voltage-Regulator Characteristics," A.I.E.E. Trans., vol. 63, pp. 215-20,
May, 1944. Disc., pp. 489-90.
20. SELDEN B. CRARY, Power System Stability, Vol. I, "Steady State Stability,"
John Wiley & Sons, Inc., New York, 1945.
21. C. CONCORDIA, S. B. CRARY, and F. J. MAGINNISS, "Long-Distance Power
Transmission as Influenced by Excitation Systems," A.l.E.E. Trans., vol. 65,
pp. 974-87, 1946. Disc., pp. 1175-6.
22. G. DARRIEUS, "Long-Distance Transmission of Energy and the Artificial
Stabilization of Alternating-Current Systems," C.I.G.R.E.,· 1946, Report 110.
23. W. FREY, "Artificial Stability of Synchronous Machines Employed for
Long-Distance Power Transmission," C.l.G.R.E., 1946, Report 317.
by Angle-Regulator Characteristics," A.I.E.E. Trans., vol. 67, part I, pp. 687-
90,1948.
25. A. J. GIBBONS and E. B. POWELL, "The Influence of Design and Operational
Practice on the Stability of Large Metropolitan Cable Systems," C.l.G.R.E.,
1948, Report 318.
26. C. CONCORDIA, ((Effect of Buck-Boost Voltage Regulator on Steady-State
Power Limit," A.l.E.E. Trans., vol. 69, part I, pp. 380-4, 1950. Disc., pp. 384-5.
27. W. G. HEFFRON and R. A. PHILLIPS, "Effect of a Modern Amplidyne Voltage
Regulator on Underexcited Operation of Large Turbine Generators," A.I.E.E.
Trans., vol. 71, part III, pp. 692-7, August, 1952. Disc., pp. 697 and 1134-5.
28. E. FRIEDLANDER, II Control Scheme for Improved Alternator Stability on
Low Excitation," C.I.G.R.E., 1952, Report 321.
316 STEADY-8TATE STABILITY
29. Cu. LAVANCHY, "A New Solution for the Regulation of Counter-Excitation
Synchronous Compensators," O.I.G.R.E., 1952, Report 331.
30. J. M. DYER, "Shunt Inductors Solve Problems of Unstable Operation,"
Elec. Light and Power, vol. 31, no. 2, pp. 92-4, February, 1953.
31. W. G. HEFFRON, JR., U A Simplified Approach to Steady-State Stability
Limits," A.I.E.E. Trtms., vol. 73, part III, pp. 39-44, February, 1954.
32. F. S. ROTHE, "The Effect of Generator Voltage Regulators on Stability and
Line Charging Capacity," O.I.G.R.E., 1954, Report 321.
PROBLEMS ON CHAPTER XV
1. It has been proposed to raise the steady-state stability limit of a two-
machine power system by interposing between the machines a bank of
quadrature-boost transformers which would introduce a phase shift of the
proper sign to decrease the net angular displacement between the two
machines for a given value of transmitted power. Analyze the feasibility
of this proposal.
2. Find the steady-state power limit of two synchronous machines con-
nected through external series reactance of 1.0 unit. The internal reactance
of the generator is 0.6 unit and that of the motor, 0.7 unit. The terminal
voltages are held at 1.1 and 1.0 units, respectively. Plot curves of power
versus 0, the angle between internal voltages, (a) with constant field currents
that give the specified terminal voltages at particular values of power and
(b) with the terminal voltages held at the specified values.
3. Find the equation of curve E of Fig. 3, Example 1.
4. Solve Probe 2 by constructing Clarke diagrams.
6. Find the transfer reactance between E l and E2 in the circuit of Fig. 7
and show that the maximum power of that circuit is the same as the maxi-
mum power of the circuit of Fig. 8 with equivalent e.m.f.'s and reactances
as defined by eqs. 5 to 8.
6. Discuss the effect of parallel resonance between Xl and x, or between
x 2 and Xr in the circuits of Figs. 7 and 8.
7. Draw power-angle curves and discuss the stability of two machines
connected through resistance and enough series capacitive reactance to
neutralize the inductive reactances of the machines themselves.
8. Draw the power-angle curve of a salient-pole synchronous condenser
operating with reversed excitation.
9. Find the steady-state stability limit of two identical salient-pole syn-
chronous machines, one operating as a generator, the other as a motor.
Each machine has direct-axis synchronous reactance of 1.0 per unit and
quadrature-axis synchronous reactance of 0.5 per unit. The machines
are connected through an external reactance of 1.0 per unit. The field
currents are adjusted to maintain terminal voltages of 1.0 per unit at each
machine. Neglect resistance and saturation. What would be the error
due to neglecting saliency?
INDEX
Acceleration curves, 270 Constant flux linkage, theorem of, 13,

Air-gap line, 124 43
Amortisseur, see Damper windings Coupled circuits, inductances of, 8
Amplidyne, 142, 147; see also Rotat- time constants of, 22
ing amplifier Coupling, coefficient of, 10
Armature currents, 41 Cross-current compensation, 163
Armature leakage reactance, 30 Cylindrical- (round-) rotor machine, 1
Armature reactance, in exciter, 203 saturation in, 119
Armature resistance, 35
effect on damping, 237 Damper windings, 214
typical values of, 40 effect on machine constants, 32, 34,
Armature short-circuit time constant, 217
39, 43, 51 effect on stability, 219, 229, 240
Armature time constant, 39, 43, 51 purposes, 215
typical values of, 40 types, 214
Dampers, see Damper windings
Braking, d-c., 233 Damping, calculation of, 226, 234
negative-sequence, 219, 227 effect on power-angle curve, 243
Braking action of dampers, '216 effect on stability, 219, 229, 240
Build-down of exciter voltage, 181,188 explained by induction-motor
Build-up of exciter voltage, 171, 179, theory, 221
197, 199, 205 negative, 238
positive-sequence, 219, 234
Ceiling voltage, definition, 171 Damping power, 234
increase of, 207 flow of, in network, 239
required value of, 208 Decrement, 44; see also Time con-
typical values, 172 stants
Charts, Evans and Sels, 303 "Diactor," 153
synchronous-machine, 307 Differential field on exciter, 160, 165,
transmission-line, 303 173
Clarke diagram, 256, 258, 272 Direct axis, 2
Coefficient, leakage, 11, 37 Direct-axis sub transient reactance, 32,
of coupling, 10 66
of dispersion, 177 effect of damper windings, 217, 218
Collector-ring voltage, nominal, 170, typical values of, 40
171 Direct-axis subtransient time con-
Compensation, cross-current, 163 stant, 38, 42, 43, 49
Condenser, synchronous, typical con- effect of damper windings, 218-9
stants of, 40, 217 typical values of, 40
317
318 INDEX
Direct-axis synchronous inductance, Exciter, electronic, 141
25, 27, 40, 63 gear-driven, 148
Direct-axis synchronous reactance, 25, hydrogen-cooled, 149
63 loaded, 203
measurement of, 27 motor-driven, 148
typical values of, 40 pilot, 138, 140, 141
Direct-axis transient inductance, 31, saturation curves, 204
40, 44, 65 with series winding, 145
Direct-axis transient open-circuit time with two or more field windings, 202
constant, 36 Exciter response, 166
modification of, for point-by-point calculation of, 173
calculation, 93 by graphical integration, 178, 179,
typical values of, 40 199,205
Direct-axis transient reactance, 31, 65 loaded exciter, 203
measurement of, 44 nominal, 193
typical values of, 40 point-by-point, 178, 192, 201
Direct-axis transient short-circuit time rotating-amplifier pilot exciter,
constant, 37,41, 43, 49, 50 199
typical values of, 40 self-exci ted exciter, 187
D-c. components of short-circuit cur- two field windings, 202
rents, 42 unloaded exciter, 174, 179, 187, 192
Dispersion, coefficient of, 177 comparison of loaded and unloaded
Distortion curve, 204 exciters, 205
Drives for exciters, 147, 172 comparison of various systems, 190,
Dynamic stability, 168 200
definitions, 169
Eddy currents, effect on exciter re- effect of eddy currents, 194
sponse, 194 how to increase, 207
Electronic pilot exciter, 141 loaded exciter, 203
Electronic voltage regulator, 141, 142 nominal, 170
Equal-area criterion, applications of, calculation of, 193
87, 88, 89 definition, 170
Equivalent circuit, of induction motor, effect of ceiling voltage, 208
222 typical values, 172
of synchronous machine for cal- quick, 172
culation of positive-sequence Exciter voltage, referred to armature,
damping, 235 91, 172
of transformer, 11 Exponential decay, 18
Equivalent synchronous reactance,
279,286 Faraday's law, 4
Excitation, wide-range, 164 Field collar, 215
Excitation limits, 165 Field current, calculation of, during
Excitation systems, 137 build-up of exciter, 206
relation to stability, 165 during short circuit, 42
response of, see Exciter response Field flux linkage, calculation of
types of, 140 change of, 91, 113
Excitation voltage, 70 Field time constant, open-circuit, 36,
Exciter, 137, 138 40, 93
direct-connected, 148 short-circuit, 37, 40, 41, 43, 49, 50
drives for, 147, 172 Flux linkage, 3
INDEX 319
Flux linkage, in terms of Park's vari- Magnetic circuit, relation of induct-
ables, 58 ance to, 6
theorem of constant, 13, 43 Motor, synchronous, typical constants
Follow-up method, 20 of, 40
Formal integration, 178 M ultimachine system, conservative
Frohlich equation, 178 criterion for steady-state sta-
bility, 276
Governor action, effect of, on steady- damping in, 238
state stability, 268 Mutual inductance, 5, 6
Graphical integration, 178 versus rotor position, 55, 56
Harmonics, suppression of, by damper Negative damping, 238

windings, 216 Negative-sequence braking, 219, 227
Heating of rotor, 244 Negative-sequence currents, heating
Hunting, 216, 238 effect of, 244
Hydrogen cooling, 149 .Negative-sequenee impedance, 226
effect on stability, 220
Igni tron exciter, 141 Negative-sequence reactance, 34, 67
Impedance, reflected, 9 effect of damper windings, 217-9
short-circuit, 9, 11, 16 typical values of, 40
Impedance-type voltage regulator, 142 Negative-sequence resistance, 35, 226
Inductance, 3, 7 effect of damper windings, 217, 219
direct-axis subtransient, 32, 40, 66, typical values of, 40
217-8 Nominal collector-ring voltage, 170,
direct-axis transient, 31, 40, 44, 65 171
leakage, 13, 177 Nominal exciter response, 170
mutual, 5, 6 calculation of, 193
quadrature-axis transient, 33, 40, 66 Nonsalient-pole machine, 1
relation to magnetic circuit, 6 saturation in, 119
short-circuit, 9, 11, 16
synchronous machines, 24, 40 Offset current wave, 17. 42
versus rotor position, 53 Open-circuit transient time constant,
Induction machines, stability of, 302 36,40,93
Induction-motor theory, 221 Oscillogram, 41
Inductive reactance, sec Reactance analysis of, 47
Inertia, effect of, on steady-state sta-
bility, 267 Parallel operation, 163
Internal voltage, transient, 75 Park inductances, 63
Park's transformation, 57
Leakage coefficient, 11, 37 Park's variables, 57, 59, 60
Leakage flux, of exciter field, 176 Permeance, 6, 7
of transformer, 16 Phase balancing, 216
Leakage inductance, 13 Pilot exciter, 138, 140
of exciter field, 177 electronic, 141
Leakage reactance, armature, 30 Point-by-point calculation, of change
Linkage, see Flux linkage of field flux linkage, 93
Loads, characteristics of, 294 of exciter response, 178, 192, 201
effect on steady-state stability, 296 of field current of main generator,
206
Magnetic amplifier, 145 of simple R-L circuit, 21
320 INDEX
Point-by-point calculation, of swing Reactance, Potier, 40, 122
curves, of multimachine system, quadrature-axis subtransient, 33, 40,
114 217, 218
of salient-pole machine, 90, 99, quadrature-axis synchronous, 29, 40,
117 64
Positive-sequence damping, 219, 234 quadrature-axis transient, 33, 40, 66
Positive-sequence resistance, 35 saturated ~nchronous,279
typical values of, 40 short-circuit, 9
Potier reactance, 122 synchronous machines, 24-35, 40,
typical values of, 40 63-9, 122, 217-9
voltage behind, 122 zero-sequence, 35, 40, 64, 218
Potier triangle, 122 Reflected impedance, 9
Power-angle curves, damping in- Regulex, see Rotating amplifier
cluded, 243 Relay, 15
for constant terminal voltages, 254 Reluctance, 7
of saturated round-rotor machine, Reluctance power, 81
286 Resistances of synchronous machines,
of two-machine system with losses, 35
264,266 typical values of, 40
salient-pole machine, 80, 82, 83 Response of excitation system, see
steady-state, 80, 82 Exciter response
transient, 82, 83 Rotating amplifier, 142, 144, 146, 199
with and without voltage regulators, Rotor heating, 244
311, 312, 313 Rototrol, see Rotating amplifier
Round-rotor machine, 1
Quadrature axis, 2 saturation in, 119
Quadrature-axis sub transient react-
ance, 33, 66 Saliency, definition of, 2
effect of damper windings, 217, 218 effect on stability, steady-state.' 278
typical values of, 40 transien t, 82
Quadrature-axis synchronous react- Salient-pole machine, 1
ance, 29, 64 computation of swing curves, 90, 99,
typical values of, 40 117
Quadrature-axis time constants, 39 flux paths in, 26
effect of damper windings, 218-9 power-angle curves, 80, 82, 83
typical values of, 40 representation of, 89
Quadrature-axis transient reactance, saturation in, 126
33,66 Saturated synchronous reactance, 279
typical values of, 40 Saturation, 118
Quick-response excitation, 172 effect on steady-state stability, 279
effect on sub transient reactance, 33
R-L circuit, 19 in exciter, 176
Reactance, 2, 7 in induction motor, 119
armature leakage, 30 in nonsalient-pole machine, 119
direct-axis subtransient, 32, 40, 66, in salient-pole machine, 126
217, 218 in synchronous machine, 119, 126,
direct-axis synchronous, 25, 27, 40, 127
63 in transformer, 118
direct-axis transient, 31, 40, 44, 65 Saturation curves, of exciter, 204
equivalent synchronous, 279, 286 of synchronous machine, 121, 129
negative-sequence, 34, 40, 67, 217-9 Saturation factor, 280
INDEX 321
Second-harmonic components, 42, 44 Stability limit, steady-state equation
Self-excitation of main exciter, 140, for, 261
175, 187 of nonlinear system, by graphical
Self-inductance, see Inductance method, 297
Semilog plot of exponential, 19, 49, resume, 308
50, 51 Steady-state power-angle curve, 80, 82
Separate excitation of main exciter, Steady-state stability, see Stability,
140, 174, 179 steady-state
Ship propulsion, 169 Steady-state stability limit, see Sta-
Short circuit, of synchronous genera- bility limit, steady-state
tor, 40, 92 Subtransient component of short-
Short-circuit currents, 41 circuit currents, 41, 42, 49, 50
Short-circuit impedance, 9 Subtransient reactance, 32, 33, 66
Short-circuit inductance, 9, 11, 16 effect of damper windings, 217, 218
Short-circuit ratio, 128, 169 typical values of, 40
in approximation for equivalent Subtransient time constants, 38,39,42,
reactance, 290 43,49
Short-circuit reactance, 9 effect of damper windings, 218-9
Short-circuit transient time constant, typical values of, 40
37, 40, 41, 43, 49, 50 Superposition, principle of, applied to
"Silverstat," 154, 156 measurement of reactance of
Sli p test, 29 synchronous machine, 27, 45
Stability, dynamic, 168 Sustained component of short-circuit
effect of damper windings, 219, 229, curren ts, 41, 42
240 Swing curves, calculation of, taking
effect of excitation system, 165 field. decrement into account,
steady-state, 247 99, 105, 114
conservative criterion for, 271, 276 effect of damping upon, 243
effect of governor action, 268 point-by-point calculation, round-
effect of inertia, 267 rotor machine, 117, 118
effect of loads, 296 salient-pole machine, 90, 99, 117
effect of saliency, 278 Synchronizing-power coefficients, 272
effect of saturation, 279 Synchronous condensers, typical con-
effect of voltage regulators, 309 stants of, 40, 217
importance, 247 Synchronous machines, 1
two-machine system, with losses: constants of, 24-52, 63-9, 122, 128,
264, 267, 268, 272 217-8
with negligible losses, 251, 256, effect of damper windings, 217,
258 218
transient, 78; see also Vols. I and II mathematical theory of, 52
assumptions regarding machines, revised equations, 73
79 saturated, nonsalient-pole, 119, 279,
effect of damper windings, 219, 307
243 salient-pole, 126, 307
effect of excitation system, 165 reactances of, 24-35, 40, 63-9, 122,
effect of field decrement, 98 217-9
eft'ect of saliency, 82 resistances of, 35, 40
effect of saturation, 127 types of, 1
round-rotor machine, 111 typical constants of, 40
Stability limit, steady-state, 249 vector diagrams, 67-79, 115, 120,
by Clarke diagram, 256) 258, 272 125-6
322 INDEX
Synchronous motors, typical constants Vector diagrams of synchronous ma-
of, 40 chines, round-rotor, 77, 78
Synchronous reactance, direct-axis, 25, salient-pole, 72, 77
63 subtransient state, 78
equivalent, 279, 286 steady state, 69, 72
quadrature-axis, 29, 64 transient state, 74, 76, 77, 78, 115
saturated, 279 with saturation, 120, 124, 125, 126,
280
Theorem of constant flux linkage, 13, Voltage, behind Potier reactance,
43 122
Tilne constant, armature, 39, 43, 51 behind synchronous impedance, 70
direct-axis transient, open-circuit, behind transient impedance or re-
36,40 actance,75
short-circuit, 37, 40, 41, 43, 49, 50 calculation of change of, 91
fundamental concepts of, 18 ceiling, 171, 172, 207, 208
of coupled circuits, 22 excitation, 70
of exciter, 179, 181, 182, 188, 190, nominal collector-ring, 170, 171
202, 208 steady-state internal, 70
of R-L circuit, 19 Voltage regulator, "Diactor," 153
quadrature-axis, 39, 40, 218-9 direct-acting rheostatic, 140, 153
subtransient, 38, 39,40, 218-9 effect on stability, 249, 309
synchronou&-machine,36-52,218-9 electronic, 141, 142
Tirrill voltage regulator, 149 impedance type, 142
Torque of induction motor, 223 indirect-acting rheostatic, 140, 158
Transformer, inductances of, 8 polyphase voltage, 158, 173
saturation in, 118 rheostatic, 152
short-circuited, 15 rocking contact, 157
Transient component of short-circuit "Silverstat," 154, 156
currents, 41, 42, 49, 50 static, 142, 144, 145
Transient inductance, see Transient Tirrill, 149
reactance Type BJ, 159
Transient internal voltage, 75 Type GF A-4, 161
Transient power-angle curve, 82, 83 Type J, 162
Transient reactance, direct-axis, 31, vibrating, 149
40, 44, 65 where used, 139
quadrature-axis, 33, 40, 66 Voltage-time curve, see Exciter re-
Transient stability, see Stability, sponse
transient
Transient time constant, direct axis, Water-wheel generators, typical con-
36, 37, 40, 41, 43, 49, 50, 93 stante of, 40
open-circuit, 36, 40, 93 Wheatstone bridge field rheostat, 164
quadrature-axis, 39, 40 Wide-range excitation, 164
short-circuit, 37, 40, 41, 43, 49, 50
Trapped flux, 17, 43 Zero-power-factor saturation curve,
Turbogenerators, typical constants of, 121
40 Zero-sequence reactance, 35, 64
effect of damper windings, 218
Vector diagrams of synchronous ma- typical values of, 40
chines, 69, 120 Zero-sequence resistance, 36

Power System Stability Vol III Kimbark

Uploaded by

Copyright:

Available Formats

Power System Stability Vol III Kimbark

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Power System Stability Vol III Kimbark

Uploaded by

Copyright:

Available Formats

POWER SYSTEM STABILITY

R. S. Blicq J. D. Irwin J. M. F. Moura

Dudley R. Kay, Director ofBook Publishing

IEEE Press Power Systems Engineering Series

Dr. Atif S. Debs Dr. Donald W. Novotny

Dr. M. El-Hawary Dr. A. G. Phadke

Mr. Richard G. Farmer Dr. Chanan Singh

Dr. Charles A. Gross Dr. E. Keith Stanek

Dr. G. T. Heydt Dr. 1. E. Van Ness

Edward Wilson Kimbark

IEEE Press Power Systems Engineering Series

© 1956 by Edward Wilson Kimbark

Printed in the United States of America

Library of Congress Cataloging-In-Publication Data

Kimbark, Edward Wilson

This is the third volume of a three-volume work on power-system

XIII Excitation Systems 137

Since the problem of power-system stability is to determine whether

1. Cross sections of synchronous machines, (a) round-rotor type, (b) salient-

Synchronous machines are classified into two principal types-

proceeding to define these reactances we may profitably review the

Fundamental concepts of inductance. The concepts of electric cur-

FIG. 2. Illustrating flux linkages. FIG. 3. A line of magnetic flux con-

*The flux linkage of a part of a circuit, such as a winding of a machine, which is

According to Faraday's law of electromagnetic induction, any change

It is always a positive quantity, because, according to the right-hand

e= - ~ (Li) volts [5]

holds, however, the self-inductance can be defined by

This relation is ordinarily used in defining the unit of inductance, the

Mutual inductance. There are some groups of inductively coupled

,pI + L I 2i2 + Lisis · · .

Here L 11, L 22 , L 33 are the self-inductances of circuits 1, 2, 3; L 12 is

any two circuits is the same in both directions; that is, L 12 = L 2 1,

and so on. The definition of self-inductance, eq. 11a, is more general

Relation of inductance to the magnetic circuit. The magnetomotive

The flux linking a second electric circuit of n2 turns due to current it

where JJ. is the permeability, A is the cross-sectional area, and l is the

Inductive reactance in ohms of a circuit of inductance L is given by

Summary of concepts of inductance and inductive reactance. The

3. Derivative of flux linkage with respect to current.

Inductances of two coupled circuits (the transformer). Consider

FIG. 5. Two inductively coupled circuits.

nal voltages, as marked in Fig. 5. The instantaneous terminal volt-

In the steady state, with sinusoidal alternating currents and voltages,

Short-circuit impedance and inductance. If the secondary circuit is

VI = ( ZIt - -Z122) It [27]

The first term of this expression, ZII, is the self-impedance of the

The driving-point inductance is

The last two equations are similar in form to eq. 28.

Coefficient of coupling and leakage coefficient. Consider two induc-

and the mutual inductance of coils 1 and 2 would be

Similarly, with current i 2 in coil 2 and no current in coill, if the flux

and the mutual inductance would be

coil 1 is reduced more than that of coil 2, and

where k2 < 1. Elimination of nt/n2 between eqs. 40 and 41 gives

L 12 = Vk 1k2L 11I.J22 = k vII..JI1L2; [42]

'The short-circuit inductance given by eq. 31 may then be written

FIG. 6. Equivalent T circuit of the inductively coupled circuits of Fig. 5.

provided that the secondary current be divided by a and the voltage

k2 nl = L 12 < a < L 11 = ! nl [46]