Paper No.

1411
SUPPLY SECTION

621.313.32

OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

By J. H. WALKER, M.Sc, Ph.D., Member.
(The paper was first received 26th May, and in revised form 21st August, 1952.)
SUMMARY
By using the well-known 2-axis theory of Blondel the paper develops
on a simple geometrical basis the steady-state characteristics of synchronous salient-pole machines. From these it is shown that appreciable economies can be realized by relating the short-circuit ratio,
and thus the synchronous reactance, of an alternator to the stability
characteristics required by the load. It is also shown that further
economies can be obtained in both alternators and synchronous
condensers by designing the excitation system to supply positive and
negative field currents.
Power-station operation can be simplified and made more reliable
by the use of a combined watt- and VAr-meter with a dial based on
the power chart evolved in the paper.
The common basis underlying the operation of the salient-pole
alternator, synchronous motor and condenser, and synchronous
reluctance machine is shown by a simple generalized diagram, and
the operation of these types of motors is briefly discussed.
Equations are evolved for determining machine characteristics with
greater accuracy than is given by the geometrical method.
(1) INTRODUCTION
The fundamental 2-axis theory underlying the operation of
salient-pole machines was first enunciated by Blondel, and since
then many papers have been published1 that develop from this
theory the appropriate equations for design and operation. For
instance, the equations appropriate to transient conditions were
developed by Park in his classic contribution; again Gibbs, 2 in
a recently published monograph, comprehensively covered operation under both transient and steady-state conditions, but since
he expressed his results in the notation of matrix algebra they
are not in a form suitable for ready application to practical
problems. Although other papers, such as those by David 3 and
Angelini,4 have dealt with particular aspects of the application
of 2-axis theory to steady-state operation, there appears to be
no comprehensive analysis suitable for immediate practical
application to normal running conditions.
The present paper develops for the various types of synchronous
salient-pole machines the diagrams and general equations suitable
for practical use under steady-state conditions; it further shows
how the application of these results may in certain cases lead to
improvements in design and operation.
(2) THE VECTOR DIAGRAM
(2.1) Assumptions
(a) All load changes take place within a time interval longer
than the short-circuit transient time-constant of the machine.
The analysis in the paper is thus restricted to steady-state conditions as defined by this assumption.
(b) In all cases the machine is operating in parallel with a
large-capacity system, i.e. it is connected to a so-called "infinite"
busbar whose voltage is unaffected by load changes in the machine
in question.
(c) Magnetic saturation can be neglected. This is justified on
the grounds that saturation has an appreciable influence only in
Written contributions on papers published without being read at meetings are
invited for consideration with a view to publication.
Dr. Walker is with the British Thomson-Houston Co., Ltd.

the region of operation corresponding to rated load and power

factor, whilst the more important characteristics of synchronous
machines, such as stability, usually relate to the under-excited
(more-or-less unsaturated) condition where the error is negligible.
If in the first case it is desired to allow for the effects of saturation, the direct- and quadrature-axis synchronous reactances Xd
and Xq used in the paper can be replaced by the corresponding
equivalent reactances Xd(fiq) and Xq{eq) in the manner described
by Crary. 5 The use of these reactances will considerably
increase the labour of preparing the diagrams and charts described
in the paper.
(2.2) The Vector Diagrams of the Salient-Pole Machine
The conventional 2-axis vector diagram of a salient-pole
alternator supplying a lagging-power-factor load is shown in
Fig. 1. The line OA represents the terminal e.m.f. E, and /

Fig. 1.Two-axis voltage vector diagram.

Xa

Salient-pole alternator
1 1 per unit, Xq = 0-7 per unit.
Power factor == 0-9 lagging.

the armature current lagging behind the terminal voltage by the

angle <f>. The line AB is drawn at the angle <f> to the perpendicular
AY, a length AF proportional to IXq is marked off on it and the
imaginary pole axis is then drawn through F from O. The two
hypothetical components Id and Iq of the armature current can
then be obtained by drawing the vector representing ld perpendicular to the pole axis (imaginary), OF, and the vector
representing Iq parallel to the same axis. The mutually perpendicular vectors IdXd and IqXq can then be drawn, the extenr
sion of OF to OG then giving the internal voltage EQ. It should
be noted that throughout the paper the "per unit" system is
used so that rated r.m.s. current per phase is unit current, rated
r.m.s. voltage per phase is unit voltage, direct-axis reactance Xlf
is direct-axis reactance in ohms times rated current and divided
by rated voltage, and similarly for Xq. It thus follows that Eo
is the per-unit internal voltage. If the line AF is extended to B
such that AB represents IXd, OBA is the conventional vector
triangle of the cylindrical-rotor synchronous machine.
It can be seen that the internal voltages obtained by the two
methods scarcely differ, but the pole angle 8 obtained by the
13]

14

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

2-axis method is substantially less than given by round-rotor

theory.
The corresponding diagram for an alternator operating at a
leading power factor is shown in Fig. 2. It can be seen that

Fig. 4.Two-axis current vector diagram.

Salient-pole a l t e r n a t o r
Xd - 1-1 per unit, Xq -- 0 - 7 per unit.
Power factor - 0 - 9 lagging.

Fig. 2.Two-axis voltage vector diagram.

Xd

Salient-pole alternator
0 - 7 p e r unit.
1 1 per unit, Xv
Power factor - - 0 - 9 leading.

there is an appreciable difference between both the internal

voltages and pole angles given by the two different methods.
In both Figures the polygon OAHG represents a voltage
vector diagram. The terminal voltage is constant so that the
only operational variables are armature current, power factor
and field current. It is therefore necessary to transform the
voltage vector polygons of Figs. 1 and 2 into those appropriate
to currents; in addition, for simplicity of representation, it is
desirable that the vector polygon be replaced, in effect, by a
vector triangle corresponding to that normally associated with
the cylindrical-rotor theory. The latter change is shown in
Fig. 3, in which HG has been extended to B so that HB represents

representing the contribution from the salient-pole effect to the

internal voltage; in Fig. 4 it can be represented as the similar
contribution to the load field current.
By eliminating all but essentials, the simple vector diagram of
Fig. 5 can now be drawn, which shows quite clearly the relationships between short-circuit current, armature current and load
field current. This simplified current vector diagram now
affords a ready means of determining the relationships between
armature current, power factor and field current. If Xq Xd
the diameter of the semicircle becomes zero, and the vector
diagram reduces to the simple diagram of the round-rotor
machine as shown by the dotted line.
(3) POWER DIAGRAMS
(3.1) The Alternator Power Diagram

With a fixed terminal voltage it is obviously possible to

determine from Fig. 5 the active- and reactive-power loading of

Fig. 3.Two-axis voltage vector diagram.

Salient-pole alternator
Xd -- I 1 per unit, X,, =- 0 - 7 per unit.
P o w e r factor = 0 - 9 lagging.

lqXd and AB represents IXd. Then BC is drawn parallel to OG

and a semicircle is drawn on the diameter OC. By simple
geometry DB = OG == Eo and OC --= E(Xd/Xq - 1), so that
the triangle CBA together with the semicircle CDO give the
vector polygon of the alternator in a simpler form than that of
Fig. 1. In order to transform this into a current polygon it is
merely necessary to divide the four parametersrepresented by
the four sides of the polygon ODBAby Xd, thus obtaining
the current vector polygon OABD of Fig. 4. In this Figure, OA
represents E/Xd, the short-circuit current of the alternator with
no-load excitation, AB the armature current, and DB the internal
voltage divided by Xd, i.e. the short-circuit current with full-load
excitation. Since, with the assumption of no saturation, Eo also
represents the full-load field. current /y, E0/Xd can also be
written l//Xd. In Fig. 3 the vector OD may be considered as

Fig. 5.Simplified 2-axis vector diagram.

Salient-pole alternator
Xd 1 1 p e r unit, Xq -- 0 - 7 p e r unit.
Power factor = 0 - 9 lagging.

an alternator for a given armature current and power factor.

As an example Fig. 6 shows the variation of field current and
load angle for varying power factors, the armature current
being maintained constant. A constant armature current simply
requires the locus of the extremity of the current vector / to
move on a semicircle with its centre at O. Thefieldcurrent is
given as in Fig. 5 by the intercepts of the two semicircles on the
rays drawn from C to the extremities of the current vectors.
Since power-factor angles are measured from the ordinate Oy

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

!
Per-uriit kW
Power factor

0-6

0-25
0-50
0-75
Per-unit K.VA. lagging

Fig. 6. Circle diagram for constant armature current / with varying power factor and field current If.
Salient-pole alternator
= 1 1 per unit, Xq - 0-7 per unit.
Power factor = 0-9 lagging.

Fig. 7.Diagram for rated^constant field current If with varying power factor and armature current /.
Salient-pole alternator
Xd 1-1 per unit, Xq 0-7 per unit.
Power factor 0-9 lagging.

15

16

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

as zero, projections of the current vectors on the vertical axis maximum and then falls again. Similar considerations apply
represent active power and similar projections on the horizontal in Fig. 8 to operation either as a generator or a motor.
axis represent reactive power. A counter-clockwise rotation of
(3.3) Machine Stability
vectors is assumed in all these diagrams to be positive, and
negative values of <f> correspond to lagging power factors and It is now necessary to consider the question of stable operation.
positive values to leading ones. It then follows that in Fig. 6 In general when an alternator is operating on full load any small
projections of current vectors falling on the horizontal axis to fortuitous increase in the torque of the prime mover is followed
the right of Oy represent lagging reactive power, and to the by a corresponding increase in the pole angle 8; the latter
left leading reactive power. This Figure is thus an elementary increase leads to an increase in the active-power output of the
form of power chart.
alternator which restores the rotormore or lessto its original
An alternator connected to a large power system may be pole angle. However, if the alternator is initially operating at
subjected to substantial output variations, and under these its maximum active-power output point the small fortuitous
conditions it is useful to the power engineer to have a diagram increase in the torque of the prime mover is followed by a
showing, for a given initial steady-state condition (e.g. rated decrease in the output of the alternator. The prime mover will
load), the variation of armature current and power factor caused thus continue to increase the pole angle S until finally the speed
by output variations with constant field current. An elementary of the alternator rises above synchronism and the unit falls out
form of such a diagram is shown in Fig. 7, and in this the locus of step. The corresponding condition in a synchronous motor
of the armature vector is determined by drawing a number of may occur for example in a rolling-mill drive. If the motor is
rays through C and then marking off along each ray from the running on afixedfieldexcitation, the introduction of a piece of
periphery of the semicircle a length representing I//Xd; this material which is either too thick or not hot enough may lead
to an increase in the active-power output which exceeds the
construction is an extension of that given in Fig. 5.
The resulting smooth curve drawn through the extremities of maximum output for that particular field excitation and the
these rays is called the "limacon of Pascal," although in Fig. 7 motor will immediately drop out of synchronism.
It is thus important that a power diagram should indicate for
it appears to be a semicircle.
the various possible operating conditions the maximum output
points at various values of field current. As can be clearly
(3.2) The Generalized Power Diagram
seen from Fig. 8 these points are the maximum and minimum
In the diagrams so far considered, the power-factor angle <f> ordinates of the limacons, and the two smooth curves joining
has been restricted to the range 90', i.e. they have covered the these points indicate the limits of stable operation; the dotted
operation of synchronous salient-pole machines as generators portions of the limacons thus indicate the unstable region of
operating from zero lagging to zero leading power factor. It the diagram.
is obvious that by extending the diagram of Fig. 7 to cover the
These stability limits start from the intersection of the perfull range of <f>, i.e. O-36O;>, it will also give the performance of
pendicular at E/Xq with the horizontal axis and decrease
the machine as a synchronous motor operating from zero
asymptotically towards the perpendicular at E\Xd. This perleading to zero lagging power factor. This has been done in Fig. 8,
pendicular is, of course, the theoretical stability limit of the
cylindrical-rotor machine in which the loci of the extremities of
Generator y
the armature-current vectors are circles with centres at O. It
can thus be seen that when a salient-pole alternator is operating
2: -1-77
at about rated output and power factor, the maximum output
given by the 2-axis theory is slightly greater than that given by
cylindrical-rotor theory. However, if the machine is operating
at a low leading power factor, e.g. charging a long transmission
line during periods of light load, the 2-axis theory shows that
under these conditions the machine has a much higher maximum
output than is given by cylindrical-rotor theory. This
RVA lagging.,
increased output, however, can in general only be obtained
K.VA leadingby operating with reversed field current in the rotor winding, the
maximum value being obtained in the stable region of the
diagram at the abscissa corresponding to EfXq, i.e. at zero
leading power factor. The armature current and thus the output
at this point are represented by E/Xq, and the required negative
field current, E(llXg \IXd), corresponds to the diameter of
the circle between the ordinates at E/X and E/Xd.
(3.4) Properties of the Armature Current Loci (Limacons)
It is of interest to note in Fig. 8 that, for the given values of
Xd and Xq, the limacon for full-load field current is almost
Fig. 8.Armature-current loci for fixed field currents.
indistinguishable from a circle, for If = 1-2 amp it has a slight
Salient-pole machine
flattening
at one side, for If ~ 0-57 amp the limacon takes on
Xri 1 1 per unit, Xq -~ 0-7 per unit.
the special form known as a cardioid with the cusp at the abscissa
Rated power factor = 0-9 lagging.
E\Xq, and for lower values of If the limacon develops a
and in addition the limacons (armature-current loci) correspond- loop inside the circle as shown in the example for If 0 18.
ing to various fractions of full-load field current have been This loop can be considered either, in the normal case, as the
locus of armature current with negative field current, or as the
included.
In Figs. 6 and 7 it can be seen that as the power-factor angle cf) same locus with positive field current after the machine has
varies from 90 to +90, the active-power output rises to a traversed a pole angle S greater than 180 and resumed stable

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

operation on the loop. The curves showing the relationship

between torque and pole angle can be obtained from Fig. 8 by
plotting the projection of / on the vertical axis against the pole
angle 8. The resulting curves shown in Fig. 9 demonstrate more
Round-rotor theory

300

-1-6

360

Motor

Fig. 9.Variation of active-power output with pole angle d.

Salient-pole machine
Xa 1-1 per unit, Xq 0-7 per unit.
Rated power factor = 0-9 lagging.

clearly the effect of operating with a field current which produces

a loop in the limagon as in the case discussed above. If, for
instance, the machine is operating at a load of 0-1 kW per unit
and /^.= +0-18 per unit, then under carefully controlled
conditions the load could be correspondingly varied to follow
the harmonic curve corresponding to If 0 18 per unit until
the load finally resumed its steady value of 0 1 kW per unit.
During this procedure the rotor would have traversed a pole
angle S of approximately 190', and although the absolute
direction of the field current has not changed, nevertheless,
relative to the stator flux, it is 180 out of phase and is thus

17

negative. This final condition corresponds to operation on the

loop of Fig. 8. In normal operation such a procedure is impracticable, and operation on the loop is initiated with the
required negative field current. Fig. 9 brings out in a simple
manner the concept of saliency as a constant-amplitude secondharmonic torque superimposed on a fundamental sinusoidal
torque whose amplitude varies with the field excitation; and the
same concept, although not so obvious, underlies the geometrical
construction of Fig. 8.
(4) THE ALTERNATOR POWER CHART
(4.1) Constant Voltage
The form of power diagram shown in Fig. 8 is not suitable for
practical application by power-system engineers, since it does
not indicate, for example, the limitations imposed on the armature
field currents by the permissible temperature rises of the windings.
A practical power chart for cylindrical-rotor alternators has
already been proposed by Szwander,5 and the chart for salientpole machines in Fig. 10 follows similar lines.
Since it would be inadvisable to operate an alternator close to
the theoretical stability limit, the latter is replaced by a practical
stability limit. In the diagram, the curve is arbitrarily reduced
by a constant power value corresponding to (for example) 10 o
of the rating of the machine, so that at any load on the practical
stability limit curve such as 0 2 or 0 7 kW per unit the kW-load
can be increased to 0-2 + 0 1 = 0-3kWperunit or 0-7 + 0 1
= 0-8kW per unit before reaching the theoretical stability
limit.
Large alternators are usually supplied with main and pilot
exciters with no provision for supplying negative current to the
rotor field. In addition the main exciter is often, for simplicity,
not fitted with a bias field to overcome residual effects. Under
these conditions the excitation system is not capable of reducing
the alternator-field voltage (and current) to zero, and it is usual
here to limit the minimum field voltage to 5 % of that required
on rated load. This limit is shown as the segment on the lefthand side of the diagram.

Armature

Theoretical stability limit

(Round rotor theory)

heating limit

Power factor
Limit without KW restraint
kW limit

Field heating limit

(Round rotor theory)

10
5% excitation
margin
Fig. 10.Power chart for salient-pole alternator connected to constant-voltage system.
Xa = 1 1 per unit, Xg = 0-7 per unit.
Rated power factor = 0-9 lagging.
VOL. 100, PART II.

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

18

It is of interest to note that, if for simplicity the power chart

of this particular machine had been based on cylindrical-rotor
theory, the capacity of the machine in the region of leading power
would have been as shown by the dotted stability line and thus
considerably less than that given by the appropriate theory.
This form of chart can be of considerable assistance to the
power-system engineer in determining the capabilities of machines
under varying conditions of active-power loading and power
factors; it is, however, not in a form which can be readily
interpreted by the engineer responsible for the actual operation
of the alternators in a station. The diagram of Fig. 10 may be
modified into the form shown in Fig. 11 which is suitable for
Field current, amp per unit
KVAR
UAO OINO

LCADING

V j

J
/

0
kVAr

0-2

*= V k\
\

0-2

L\

Y\ V1 I y
\
0-G
0-4
Leading

0-8
-

7_\

1-0

J\

S
\

i
\
\

w1

1
1

0-4
0-6
Lagging
+*

Fig. 11. Simplified power chart.

Salient-pole alternator.

use by the station engineer. This gives a simple closed curve

within which the loading of the alternator must be confined
under all steady-state conditions of running. At present a
station engineer has in front of him a voltmeter (terminal
voltage), two ammeters (armature current and field current), a
wattmeter, a VAr-meter and a power-factor meter. These instruments may or may not be marked to indicate values corresponding to limitations relative to heating of the alternator windings,
etc., and in either case it is far from common practice, nor is
it easy, for the operator to relate the readings on these instruments
to the values given on a simple power chart such as that shown
in Fig. 11. This difficulty can be overcome by using a power
chart of the form shown in Fig. 11 as the dial of a combined
watt- and VAr-meter. The latter has already been described7
and this so-called "vectormeter" with the power chart of Fig. 11
as dial is shown in Fig. 12. From this the operator can see at a
glance both the active- and reactive-power loading on the machine
and the corresponding field and armature currents. A power
chart can be constructed only for one particular voltage, so that
in stations where the terminal voltage varies over a range a
modification to the above chart is necessary. Assuming the
normal operating voltage range of the station is 10%, five
charts are drawn on a film strip each corresponding to a 4%
change in voltage. This film strip passes across the illuminated
front of the meter, and the appropriate chart is automatically
moved in front of the screen by a mechanism in the meter
actuated from a contact-making voltmeter.
If the excitation and voltage-regulating systems are fitted with
some form of reactive-power limiter which follows very approximately the contour of the left-hand side of the chart, the instruments merely inform the operator of the loading of the machine,
and furthermore this limiting device does not permit transient
overloads. However, no such limiting device is required when
the vector meter and chart are provided since they enable the

Fig. 12.Combined watt- and VAr-meter with power-chart dial.

operator to see at a glance what action is required to avoid

instability. This meter arrangement thus eliminates the six
meters mentioned above and at the same time gives the operator
a clear and precise picture of the extent to which the set is
operating within its designed limits.
This can be illustrated by the operation of an alternator in a
station whose total output is a relatively small fraction of that
of the system to which it is connected. In such a station the
terminal voltage is normally maintained constant by the system,
and the alternator voltage regulator will only operate during
large load changes.
A machine in such a station and with a chart similar to that
of Fig. 11 may, during the night, be operating at 0 1 kW and
0-6kVAr (leading) with a field current of 0-2 amp. In the
morning there may be a request to increase the active power
quickly to 0-8 kW, and if the turbine governor is set up to give
this power the apparent-power change will proceed along the
0 2-ampfield-currentcurve leading to the alternator falling out of
step. However, if the operator has the vector meter and chart
in front of him, he can see at a glance that before such an activepower increase takes place he must increase the field current to,
say, 0-6 so that the subsequent active-power increase will still
be within the stable range of operation of the alternator.
(4.2) Effect of Variable Voltage
In most large power stations it is necessary to be able to vary
the continuous operating voltage of the alternators between
limits, such as 10%, of the normal operating voltage. The

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

effect of such a variation can be determined by redrawing the
charts of Figs. 10 and 11, in which = 1 0 per unit, for a
terminal voltage of, say, 0-9 per unit. The effect of such a
change on a machine designed for a range of 10 000 to 12 000 volts
is shown in the isometric solid of Fig. 13.

Per-unit kVA
Leading
Ij-0-09p.u.

Per-u,hit. kV/
Is Lagging
If*177p.vL.

Fig. 13.Diagram showing variations in alternator power charts with

variations in terminal voltage.
As in Fig. 10 the x-axis represents reactive power, but here
the z-axis represents active power and the y-axis terminal voltage. Plane E, which is parallel to the base plane F, represents
constant active power, the cylindrical surfaces G and K represent
per-unit values of armature current / of 1 -0 and 0-2 amp respectively, and plane H represents a constant limiting stability surface.
The concentric limacoid surfaces A, B, C and D represent
constant per-unit exciting currents of 1 -77 (full load), 1-5, 0-44
and 0 09 per unit respectively, the last corresponding to the
minimum field current permissible with the particular excitation
system employed in this case.
Fig. 13 shows clearly the effect of varying voltage on the
reactive power the alternator is capable of carrying. At
12 000 volts the lagging power at zero power factor is given by
pq, whilst at 10 000 volts the corresponding value is the much
greater value given by rs. Conversely at 10 000 volts the leading
power at zero power factor is given by ur, whilst at 12 000 volts
this figure is considerably greater as shown by tp. Similarly
at 12 000 volts the alternator can operate at leading power
factors with full-load armature current over the substantial
range represented by the back edge of surface G; the corresponding range of full-load armature current at 10 000 volts is represented by the very short front edge of surface G.
Although such a solid can be used for determining the performance of an alternator over the practical range of all its
parameters, its greatest usefulness is probably in the sphere of
academic exposition; in this latter case the two-dimensional
representation could profitably be replaced by a Perspex or
wire model.
(5) SYNCHRONOUS REACTANCE OF ALTERNATORS.
DETERMINATION OF THE APPROPRIATE VALUE
It has been common practice in the past for purchasers of
large generators to specify, inter alia, the value of short-circuit
ratio and thus the value of its approximate inverse (the synchronous reactance) that they consider appropriate. A common
value of the short-circuit ratio for hydro-electric generators
operating at 0-85-0-90 power factor has been 10, corresponding
to a value of synchronous reactance of about 1 1 per unit. It
would appear that the choice has been determined to a large

19

extent by the successful operation of machines similar in size

and operating conditions. It is common knowledge that in
Great Britain many large steam-turbine-driven sets operate with
short-circuit ratios as low as 0 5, admittedly under loading conditions which usually differ from those of a hydro-electric unit
and with a different type of alternator. Here again the choice of
ratio seems to have been determined by past successful operation.
It would thus appear that there is now a need to put the choice
of short-circuit ratio and thus of Xd on a firmer basis, and the
analysis given here enables this to be done. There are two
main considerations which determine the correct choice of the
value of Xd for a salient-pole alternator.
(5.1) Loading at Zero Leading Power Factor
Hydro-electric generators are frequently required to excite
long transmission lines during periods of light load, and this
condition sometimes involves, particularly during the early
stages of the development of a hydro-electric system, the
generators supplying the full rated power at zero leading
power-factor. As can be seen from Fig. 8, the correct value of
synchronous reactance for this condition depends on the system
of excitation used.
(5.1.1) Excitation System capable of supplying Positive Excitation Only.
In this case, the value of Xd must be slightly less than 1 -0 per
unit to enable the machine to carry the rated output at the rated
voltage E. This can be seen from Fig. 8, which shows that for
this condition E/Xd must be slightly greater than 1 -0.
(5.1.2) Excitation System capable of supplying Positive or Negative
Field Currents.
In this case, for rated output at zero power factor (leading),
the safe operating limit is determined by the condition that
E/Xq must "be slightly greater than 1 0 ; thus here it is Xq which
must be slightly less than 1 0 and not Xd as in the previous case.
Apart from the slight increase in cost of the excitation system,
this alternative could be provided at about 12% lower cost than
in the conditions given in Section 5.1.1, since a value of X of
1 0 would give a value of Xd of about 1-6, compared to
Xd = 1 0 for the conditions of Section 5.1.1 for the same
operating conditions.
A further point to be considered here is that a value of Xq,
which gives stable operation at rated output and zero leading
power factor with negative field current, may not give stable
operation at the same rating but at, say, 0-2 leading power
factor. This is due to the concavity of the stability limit curve
as shown for example in Fig. 8.
(5.1.3) Excitation System capable of supplying Positive or Negative
Field Currents and also of ensuring Stable Operation of the
Machine in the Unstable Region of the Power Chart.
In addition to the stability limit already discussed, there is also
the constant-flux stability limit that, as shown by David,3 is
produced by the use of a special excitation system as defined
above, and considerably extends the region of stable operation.
However, since the calculation of this limit involves, inter alia,
transient constants of the machine, it is not really a steady-state
condition and is thus outside the scope of the paper.
(5.2) Operation at or near Rated Power Factor
The location of a power station near to the main load centre
usually obviates the necessity of the generator supplying leading
power output to the system, and it is in these cases that there
has been the greatest lack of clarity and reliance on past experience concerning the correct choice of the value of Xd.
A consideration of Fig. 8 suggests two possible methods of
determining the most suitable value of Xd.

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

20

(5.2.1) Stability at Rated Load and Power Factor.

With this condition the alternator can be designed to be stable
(with a suitable margin) at rated load and power factor only;
the machine would, of course, operate stably from the rated
power factor to zero lagging power factor, though with reduced
output, in order not to exceed constant full-load field current.
Under these conditions the synchronous reactance could have the
values shown in Table 1.
Table 1

Stability criterion

Rated
power
factor
(lagging)

Xd
(per unit)

Shortcircuit
ratio

Output at rated
volts
Zero p.f. Zero p.f.
(lead)
(lag)
kVA
kVA
[per unit) (per unit)

Stable from full load

to zero lagging
power factor only

0-8
0-9
10

2-9
2-5
2-2

0-40
0-45
0-50

0-35
0-40
0-50

0-90
0-80
0-60

Stable from no load

to full active-power
output

0-8
0-9
10

11
0-95
0-85

100
110
1-25

0-80
0-90
100

0-75
0-65
0-30

(5.2.2) Stability at No Load.

Consider the condition in a large power station with an
alternator connected to the busbars but not supplying load. If
now conditions require the active-power load of the alternator
to be rapidly increased to full load, then if for any reason the
voltage regulator is under hand-control or, owing to an absence
of terminal-voltage change (see Section 4.1), does not operate,
and the field current is thus not increased, the machine may
fall out of step if the value of synchronous reactance is too high.
Suitable maximum values of Xd, together with the corresponding
short-circuit ratios to avoid this condition, are given in Table 1;
it may be a coincidence, but it is nevertheless interesting to note
that the values thus obtained are those which many customers
specify or manufacturers recommend for large alternators to
satisfy average operating conditions; i.e. those not requiring
large amounts of kVA at low leading power factors.
The values of Xd and short-circuit ratio given in Table 1 for
this condition and that of Section 5.2.1 are only to be regarded
as examples of the manner in which these machine quantities
may be specified. These quantities could if required be specified,
for example, to ensure that the machine operated stably on
rated output from rated power factor, say 0-8 lagging, to 0-5
leading power factor; this condition would require a value of
Xd differing from either of those given in Table 1.
The importance of selecting the correct value of Xd for a given
set of conditions must be emphasized. On the one hand, the
cost of an alternator will, in general, vary inversely with about
the fourth root of the synchronous reactance, so that a decrease
of Xd from 200% to 100% (i.e. doubling the short-circuit ratio)
will increase the cost by about 20%. On the other hand, for a
given rated load and power factor, the value ofXd exactly defines
the stability limit of the machine under steady-state conditions.
It therefore follows that it is preferable for a customer to
specify in detail only the operating conditions and not the
synchronous reactance or the short-circuit ratio of an alternator.
The choice of the best values for these constants should be left
to the design engineer.
(5.3) Parallel Operation of Alternators driven by Internal
Combustion Engines
It has been common practice in the past to select those combined values of the synchronous reactance of the alternator and

the inertia of the rotor and flywheel that avoid the possibility of
resonance between the engine impulses and the natural electromechanical period of oscillation of the rotor and flywheel.
This condition has to-day lost its importance, since it is now
possible to design the damper winding on the rotor in such a
manner that it reduces to negligible proportions any such
resonance effects.
(5.4) Inherent Voltage Regulation
All industrial and central-station alternators are now supplied
with automatic voltage regulators as standard equipment. Under
these circumstances the value of the inherent voltage regulation
is practically worthless as a criterion of satisfactory performance;
it should therefore neither be specified by the customer nor
should the manufacturer be required to state an approximate
value since even the latter condition may cramp the design of
the machine.
(5.5) Effect of Specifying Transient Reactance X'd
To a certain extent all the above considerations are overridden
by the choice of the value* of the transient reactance X'd of the
machine.
A normal large low-speed water-wheel alternator has a natural
transient reactance X'd of 30-35 %, and the values of the second
set of synchronous reactances given in Table 1 correspond to
this value of Xd. However, if a system analysis shows that the
machine must have a transient reactance less than the natural
value, the purchaser should not assume that the machine will
have a correspondingly reduced value of Xd. In the absence
of any special requirements relative to operation at leading
power factors, the extent to which Xd is reduced with X'd is
again a matter for the machine designer alone.
(6) POWER CHARTS FOR SYNCHRONOUS MOTORS
The characteristics of synchronous motors are given, as in the
case of alternators, by the fundamental diagram of Fig. 8; the
curves applicable to motors are below the horizontal axis and
are mirror images of those above the axis.
The procedure for developing power charts for synchronous
motors is thus exactly the same as that for alternators, although
the manner of arranging motor charts will differ from that for
generators.
In the first place the power loading on the motor is determined
entirely by the requirements of the apparatus being driven, and
in the second place the majority of motors operate with a fixed
field current, so that the amount of leading or lagging power
supplied to or drawn from the line is determined, for a given field
current, entirely by the power load.
Under these conditions a normal motor runs more or less
unattended, and there is little point in providing a combined
watt- and VAr-meter with a power chart as previously described.
Nevertheless a chart of the form shown in Fig. 14 can be of
considerable value in determining, in conjunction with the
characteristics of other parallel electrical loads, such as induction
motors, the optimum rating of the motor, the active-power
rating being determined, of course, by the shaft load on the
synchronous motor.
(7) OPERATION OF A SYNCHRONOUS CONDENSER
The main purpose of synchronous condensers on large power
systems is to maintain constant voltage by supplying varying
amounts of lagging or leading power to the line in accordance
with the conditions ruling at the moment.
The only active-power loading on the machine is represented
Here again the value of this parameter should not be settled capriciously, but
should be determined by a proper analysis of the system. As shown by Ross, 8 an
unduly low value of X'$ requires a relatively expensive machine.

WALKER: OPERATING CHARACTERISTICS O F SALIENT-POLE MACHINES

21

in the negative direction, it is possible for the machine to

supply 90% of its rated output at zero lagging power factor
without any change in the dimensions or characteristics of the
condenser. It is not possible to exceed this figure of 90% since
otherwise the losses in the machine itself will lead to unstable
operation as shown by the armature-current locus corresponding
to IfIXd~ 0-33. The vector diagram for this condition of
operation at maximum stable output is shown in Fig. 15(6).
Conditions are not usually as clear cut as the above discussion
would imply since in many cases the transient stability of the
system to which the condenser is connected may set a maximum
limit to the transient reactance of the machine. It is impossible
to lay down a hard and fast rule for the relationship of transient
and synchronous reactances in a given machine; for this reason
it is essential, in order to obtain the most economical arrangement, that the purchaser of a condenser should specify only the
transient reactance (preferably obtained from a network analysis
of the system) of the machine and the fraction of the rating at
zero leading power factor that it is required to supply at zero
Fig. 14.Power chart for salient-pole synchronous motor.
Xd = 1 -28 per unit, Xq =- 1 08 per unit.
lagging power factor.
Rated power factor -- 1 -0.
It is undesirable for either the purchaser or his consultant to
by its losses which are supplied from the line. The condenser specify the value of the synchronous reactance (or the shortis thus a special case of the synchronous motor, but, since the circuit ratio) since the values of these parameters which give the
losses are practically constant and represent only 2-3% of the most economic condenser for the given operating conditions can
rating of the machine, the power charts already described have usually be best determined by the machine designer.
little practical application. Nevertheless, a power chart in a
restricted form is of interest in investigating the performance of
(8) OPERATION AS A SYNCHRONOUS RELUCTANCE
synchronous condensers when operating at zero lagging power
MACHINE
factor, since on zero leading power factor the internal voltage
It can be seen from Fig. 8 that one locus for the armature
and thus thefieldcurrent are given by the arithmetic sum of the current vector is the circle drawn with centre on the .v-axis and
short-circuit current E\Xd per unit and the per-unit armature touching the two ordinates at E/Xd and E/X^ that corresponds
current. The synchronous-condenser operating characteristics to zero field current. It thus follows that a salient-pole machine
on zero lagging power factor are given by the curves in the left- will operate as a motor or generator with no field excitation
hand quadrant of Fig. 8 below the horizontal axis, and this although with a substantial reduction in power output. Advanarea has been considerably magnified in Fig. 15(o), which has tage has been taken of this phenomenon in the development of
the well-known synchronous reluctance motor; this machine is
Constant power line , kW (machine loss)
usually manufactured by simply milling a number of longitudinal
slots in the rotor of a standard squirrel-cage induction motor.
The number of these slots corresponds to the number of poles
of the stator winding, and the width of the slots is arranged to
give the maximum ratio of XJXq consistent with an acceptable
value of magnetizing current. The combined power chart and
vector diagram of such a motor is shown below the horizontal
008
axis in Fig. 16, and it can be seen that it operates inherently at
v-O-33
/

1-2

1-0

\
+0-48'

Lagging
0-8 0-6 0-4

0-2

kVAr
Leading
0
0-2 . 0-4 0-6

Theoretical
stability limit

Fig. 15. -Characteristics for synchronous condenser operating

on zero lagging power factor.

Generator

Per-unit KVAr
(lead) R
Per-unit k W

(a) Condenser supplying lagging output.

Xd ~ 1-7 per unit, Xa = 1 -0 per unit.
Machine loss -=2% of rated output.
(b) Current vector diagram at maximum stable power.
95% full load power.

been drawn for values of Xd = 1-7 per unit and Xq = 1-0 per unit
appropriate to a typical synchronous condenser. It can be seen
that in order to operate with a small positive field current it is
necessary to restrict the lagging power to 50% of the rating at
zero leading power factor. However, if the excitation system is
designed to reduce the field current to zero and then increase

Motor

Fig. 16.Combined power chart and vector diagram for a

synchronous reluctance motor.
Xd

2-22 per unit, Xq -= 0-78 per unit.

22

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

a low lagging power factor. A high power factor could be

obtained only by an impermissibly small radial air-gap to give a
low value of magnetizing current.
For this reason the reluctance motor is used only in those
applications requiring less than about 10 h.p.
These machines are occasionally used as reluctance generators,
and Fig. 16 shows that in this case the machine can generate only
at a low leading power factor. This may introduce difficulties
in practical operation because in industrial installations none of
the loads connected to the infinite busbar may have the characteristics required by the reluctance generator.
(9) ALGEBRAIC ANALYSIS
(9.1) The Excited-Rotor Machine
From the geometry of Fig. 1 it can be seen that the pole angle 8
is given by

S-arctan

(1)*

,E//^-sin

and from similar considerations the internal voltage Eo, and

thus t h e field current If, for t h e given values of E, / , a n d <f> is
obtained from
Eo = IXdsin

(8 - <f>) + E c o s 8

(2)

T h e value of <f> for a n y required value of 8 is easily derived

from eqn. (2) a n d Fig. 1 a n d is given by

f;

0 8 arc tan \

E sin 8

' VI . .

tan8/J

(3)

Correspondingly, by derivation from eqn. (1), the value of /

for the same value of 8 is given by
1

XAcos </tan 8 + sin <f>.

)

(4)

and the values of all three variables on the right-hand side of

this equation are now known.
If it is required to calculate only the curves of Fig. 8 but
not those of Fig. 9, then eqns. (1) to (4) can be simplified to a
single equation, which involves much less labour, as follows:

~ ~\ cos 8

(6)

in which r is the length of a ray, drawn in Fig. 8, from the

value of E\Xq on the horizontal axis, to the required locus.
The curve traced out by this equation for varying values of 8
is, of course, the limacon of Pascal.
The theoretical stability-limit curve is plotted from the following equation:
E

^=1--^-

1- \
x.

. (7)

xj '

in which d and y are co-ordinates of the stability curve. The

derivation and method of applying this equation are given in
Appendix 13.1.
It is interesting to note the types of limacons which are obtained
for various values of E^Xd. If EJXd > (E/Xg - E/Xd) then,
for example, the smooth outer curve of Fig. 8 is obtained; as
EfJXd -> (E/Xg EIXd) the curve exhibits an increasingly
marked depression at the left side on the horizontal axis. When
E0/Xd - m\IXq - UXd), eqn. (6) becomes
.

. (8)

which is the equation to the cardioid, and the depression now

becomes a cusp with its point at the origin of r.
With Eo/Xd < E{\IXq l/Xd) the limacon develops a loop,
as can be seen in Fig. 8, this loop being the armature-current
locus for negative values of field current.
When EJXd = 0, i.e. zero field current, eqn. (6) simplifies to

These are, in essence, for steady-state operation, the four

fundamental equations of the salient-pole machine from which
COS
all steady-state operating characteristics can be derived.
(9)
These equations can be applied to the calculation of, for
example, the curves of Fig. 8. The values of Xd and Xq being which is, of course, the equation to the circle in Fig. 8. It is
known, a required locus, say that for the field current corre- also the equation for the current locus of the reluctance machine
sponding to rated terminal voltage load and power factor, is (see Fig. 16).
calculated as follows.
It can be seen from eqn. (6) and Fig. 7 that the equation for
The value of 8 at rated load is calculated from eqn. (1) and per-unit active power is
from this the value of internal voltage and thus the field current
P = rE sin 8
on rated load are obtained from eqn. (2). To facilitate further
calculation a number of suitable values are assumed for 8 and
28 .
(10)
the corresponding values of <f> and / are calculated from
= HT S i n
eqns. (3) and (4), the previously calculated value of EJXd at
To find the value of 8 at this maximum power and thus the
rated load being treated as a constant. The outer curve of
Fig. 8 can then be plotted from these results. A similar procedure maximum power, eqn. (10) is differentiated with respect to 8 and
is adopted for the remaining loci in Fig. 8, in each case the equated to zero, thus obtaining
starting point being the calculation of E0/Xd for a given value
of / at the rated power factor.
If, as previously suggested, E and / are expressed as per-unit
values, then, at rated voltage and load, E and / are unity. To
By substituting these values of 8 in eqn. (10) the value of maxicalculate the above set of curves for a different value of E, say mum power is readily obtained.
at 90% of terminal voltage, it is only necessary to re-evaluate
Reference was made in Section 3.4, in connection with
eqns. (l)-(3) using E- 0-9 per unit instead of E = 1 per unit. Figs. 8 and 9, to the fact that the power curves represent a
The curves of Fig. 9 can now easily be obtained from the fundamental torque with a second-harmonic torque superimposed
previous calculations, since
on it. This is also shown, in algebraic form, by eqn. (10).
Per-unit active power = El cos <f>
(5)
(9.2) The Reluctance Machine
* A counter-clockwise rotation is assumed to be positive. The zero position (see
Fig. 8) for the armature current vector is along the active-power axis above the apparentFor the excited salient-pole machine the armature current
power axis. The zero position for the internal voltage vector is along the apparentand power factor are specified at the outset, but with the
power axis and to the right of the ordinate at EIXq.

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

reluctance machine it is necessary to determine these quantities
for a specified active-power loading. In order to retain the
simplicity of the per-unit system it is necessary to assume
an approximate full-load current in amperes and use this in
calculating the per-unit values of Xd and Xq. It is important
to note that the extent to which the assumed value differs from
the value obtained from the following equation has no effect
whatever on the accuracy of the per-unit value of current
calculated from, for example, eqn. (14). The assumed value of
per-unit current should, of course, lie within the theoretical
current range of the machine, i.e. E\Xd to E/Xq.
The power output is given by eqn. (10) after omitting the lefthand term since the machine has no rotor excitation and
is zero. Thus

By differentiating with respect to 8 and equating to zero, the

value of 8 for maximum power is found to be 45, which is
obvious from Fig. 16.
The equation for maximum power thus becomes

E 1

(l-XA

*'"*
2\Xq
X d)

The current and power factor at any given active-power loading
are given by
E [~1 +

x
<f> arc cos f -

vci . . . .
(15)*
in which p is the per-unit rated load referred to the maximum
output [eqn. (13)].
Since the losses in this type of machine are high they must
be added to the active-power output to give the per-unit rated
load.
At maximum output, i.e. p = 1, eqn. (14) reduces to

(11) ACKNOWLEDGMENTS
The author is indebted to the Directors of the British ThomsonHouston Co., Ltd., for permission to publish this paper. He also
wishes to thank Mr. L. D. Anscombe, M.A., Member, of the
same Company, and a number of colleagues for advice and
criticism, and to acknowledge his debt to another colleague,
Mr. N. Kerruish, M.A., for assistance in the mathematical
analysis. Messrs. Nalder Bros, and Thompson Ltd., London,
kindly supplied the meter details of Fig. 12.

(12) REFERENCES
(1) MORTLOCK, J. R., and DAVIES, M. W. H.: "Power System

Analysis" (Chapman and Hall, 1952), p. 179.

(2) GIBBS, W. J.: "Algebra of Electric Machine Analysis"
(Technical MonographsThe British Thomson-Houston
Co., Ltd.).
(3) DAVID, H., and FAVERAN, J.: "Stability of Alternators with
Series Excitation connected by a Long Line to a High
Power System," C.I.G.R.6., 12th Session 1948, Vol. 3,
Paper No. 305.
(4) ANGELINI, A. M.: "General Topogram for Synchronous
Machines," ibid., Vol. 2, Paper No. 114.
(5) CRARY, S. B.: "Power System Stability, Vol. 1" (Chapman
and Hall, 1945), pp. 114-150.
(6) SZWANDER, W.: "Fundamental Characteristics of Synchronous Turbo-Generators," Journal I.E.E., 1944, 91,
Part II, p. 185.
(7) British Patent No. 568166.
British Patent Application No. 11295, 1952.
British Patent Application No. 13137, 1952.
(8) Ross, P. G.: "Some Notes on the Transient Reactance of
Water-wheel Alternators" (ProceedingsB.T.H. Summer
School, 1952).
(13) APPENDICES
(13.1) Derivation of the Equation for the Pull-Out Point of
a Salient-Pole Machine
For simplicity assume terminal voltage E = 1.
Then

and eqn. (15) to

where

At zero output, i.e. p 0, eqn. (14) reduces to

7=--or
A

and eqn. (15) to

23

sin 8 = IqXq and Eo cos 8 - IdXd

Id / sin (8 </>) and Iq --- I cos (8 <f>)

Let

/ cos <f> -- y and / sin (f> - - x

therefore

sin 8 = Xq(y cos 8 + y sin 8 tan <f>)

(18)

. (20)

Eo cos 8 = Xd(y sin 8 y cos 8 tan <j>) .

. (21)

<^ arccosO = 90

(19)

(10) CONCLUSIONS
The paper has shown that a simple vector diagram, based on
the 2-axis theory, can be developed for a salient-pole machine.
This diagram can form the basis of a power chart that, used in
conjunction with a combined watt- and VAr-meter, gives an
operator a ready means of maintaining the load on an alternator
within the machine operating limits, particularly those relating
to stable working. It is shown that in order to obtain the
cheapest and most economical machine it is preferable that the
purchaser should specify the operating conditions only, and that
the choice of appropriate machine constants such as synchronous
reactance should be left to the machine designer.
* For the derivation of these equations see Appendix 13.2.

The pull-out point is where /cos (f>, i.e. y, is a maximum and it

is required to find the maximum value of y for a given 0 .
For maximum y, dy\d<^> = 0.
Differentiating eqns. (20) and (21) with respect to <f> and
putting dy/d<f> 0
d8
/
cos -jy = Xqy(~

. r,d8
,ao
,dS
. ~ i i\
e
s
sm odcf>
,, + cos o t a n <pJ -\- sin o sec 2 p J
(22)

and
sin 8 -j-. --= XHy(cos
dcp
\

2
8 -,-. + sin 8 tan d>
n cos 8 sec <f>) . (23)
T
d<p
d(p
J

Eliminating dS/dcf) from eqns. (22) and (23) and simplifying:

Xd cos2 8 + Xq sin2 8 = XdXqy tan

(24)

WALKER: OPERATING CHARACTERISTICS OF SALIENT-POLE MACHINES

24

For per-unit values of E other than 1 0 this equation is

written

By using the relationship of eqn. (20) it follows that

* s
yxQ
tan o = - j r

y i

1 xXq

Divide eqn. (24) throughout by cos2 8 so that

v ~v

Since for any values of E, Xd and Xg, d always lies between

1 and 0, the values of y' from the equation
(28)

This reduces to
.

(25)

The co-ordinates of any point on the theoretical stability

curve of, for example, Fig. 8 are thus given by x and y in
eqn. (25), since this stability curve, in effect, passes through
all the points of maximum active-power output for the various
limacons shown in this Figure.
A more convenient form for calculation is obtained by
replacing the x-co-ordinate by d:

can be calculated, once and for all, as in Table 2, so that for a

given machine the values of y in eqn. (27) can be obtained by
multiplying the values of y' by (E/Xq EIXd).
(13.2) Derivation of Current and Power-Factor Equations of
Reluctance Machine
In Fig. 16 the diameter of the circle is given by E(l/Xq 1/Xd).
Let this be replaced by 2k. The vertical NQ is a measure of
the power output; let this output as a fraction of the maximum
output [eqn. (13)] be represented by p. From eqn. (6), RN in
Fig. 16 equals 2k cos 8. Let OM in Fig. 16 equal i.
Then from the triangle NQO
(29)
and from triangle RNQ
(2k

Eqn. (25) then reduces to

>

S^ 8)2 -

4fc2 COS2 8

. .

(30)

cos2 8 =

This gives

Substituting this in eqn. (29) gives

Table 2
d

005
010

0-20
0-30
0-40
0-50
0-60
0-70
0-80
0-90
100

Replacing / and k by their actual values and simplifying, gives

4 4-140
2-700
1-600
1070
0-735
0-500
0-327
0196
0100
0033
0

f1

LFig. 16
Again from
pk
I

vo . . . .

(32)