Forces in Turbine Generator Stator Windingst

BY J. F. CALVERT*
Associate A. I. E. E.

Synopsis.-Experience has shown that the forces on turbine saturation is considered, and when neglected. Approximate
generator armature windings during short-circuit conditions are formulas are derived for the forces on coil sides within one layer of
large. The windings may be seriously damaged if adequate the conical end winding, and for the straight coil extensions from
consideration is not given to the design of the coil bracing and the slots. A numerical example is given in each of those cases.
support. The applicability of certain of these formulas to the phase con-
In this paper the forces are calculated for a particular case by necting rings is pointed out. The force on the entire conical
the use of images, by the change of flux interlinkages, and by the surface of the end winding is discussed qualitatively.
f H2 d s over some surface in air which surrounds the conduc- Certain desirable features of end winding bracing are mentioned.
tor under consideration. A new and relatively simple proof is In most of these cases, the instant considered is that following
given for the validity of the latter method. The limitations and de- a line-to-neutral short circuit when the maximum possible instan-
sirable applications of each method are considered. taneous current is flowing.
Forces on conductors in armature slots are computed when

I. General Discussion of Forces during can be determined from the works of various authors.1'2
Short Circuit In this paper, it will be assumed that the armature
A. General Problem. It is the purpose of this paper currents can be computed with reasonable accuracy,
to consider the maximum forces exerted in turbine (see Appendix A). The other currents, or perhaps it
generators on the various parts of diamond coil stator should be said magnetomotive forces, will be deter-
windings. Conditions at the instant when the maxi- mined through the very useful principle applied by the
mum possible instantaneous current flows in the arma- same authors, namely, that of the tendency for currents
ture winding on a line-to-neutral short circuit are given to flow in such directions as to maintain constant flux
the most attention. A great deal of work has been interlinkages in all the closed electric circuits in the
done by previous investigators to establish methods metallic parts of the machine.
for pre-determining the magnitudes of the currents, The flux distributions will be established by means of
which result from instantaneous short circuits. The the relation Jf H'd 1' = 0.4 w I (see Appendix B). The
various cases, such as line to line short circuits, line to usefulness and applicability of this principle has been
neutral short circuits, etc., have been considered in thoroughly demonstrated by the quite accurate solu-
great detail. Much work has been published on the tions of a number of the more important two dimen-
calculation of forces on bus structures and switches. sional flux fields in electrical machines.s,12
However, it is believed that very little, if any, work The forces will be calculated in this paper for two
has been published concerning the forces exerted on dimensional fields by one of three methods; each of
the windings of rotating machines during transient these will be demonstrated later in detail. The first,
conditions, is based on the use of images, and that the force exerted
A very brief consideration of the magnitudes of the between two parallel conductors is proportional to the
short circuit currents and the proximity of these in the product of the currents, to the length of the conductors,
coil sides will quickly convince one of the importance and inversely proportional to the distance between the
of these forces. Peak values of 50,000 amperes in conductors.4'5
parallel conductors with a spacing of about 134 in. The second method is one which is very commonly
from center line to center line of the coil sides is quite employed. It is based on the rate of change in inter-
a reasonable possibility in well designed machines of linkage (and, hence, of stored energy) with the dis-
only moderately large ratings. placement of a conductor while the currents are main-
Three things are necessary tained constant. The third method is based on the
1. Thredethingsereinecesa inothe
short-cicudy
r in the study:
cfact intheanforce
that on any current carrying conductor
1. The
2. determination of
The determination short-circuxfitds,
of the flux fields, andent
c existing electric field can be determined from
3. Methods of calculating 3. Methods
the forces exertedl
ofcluaigh oe f H2 d s, taken over some surface surrounding the
on the conductor.'
conductors after the currents and the flux fields are
kno *
In one instance the force will be computed by each of
these three methods to show that for the same boundary
The peak value of the armatture current per coil side cSonditions, all give the same results. In other cases,
*Eleetrical Engineer, Westinghouse Elec. & Mfg. Co., East it may then be assumed that the most convenient
Pittsburgh, Pa. method is the best. However, the third method
tTaken from work presented for a Master's thesis under the mentioned above will appear to be the most promising
University of Pittsburgh-Westinghouse Cooperative plan. whrsautinm tbecsdrd.Tsisvy
Presented at the Middle Eastern District Meeting No. 2, of the ______
A. I. E. E., Philadelphia, Pa., October 13-15, 1930. 1. For references see Bibliography.
178

30-158
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 179

commonly the case, because the enormous currents integral of H2 around each conductor. This will be
necessary to produce forces of real consequence are done for the case of embedded armature conductors
often so situated that they do saturate some of the when saturation is not considered.
adjacent iron parts. Figs. 1, 2, and 3 illustrate the flux distributions in
B. Division of Problem. The problem is arranged in rectangular slots for three different current values.
parts according to the most convenient divisions of the (These are taken from an article on graphical field
flux fields. These will be taken up as follows: mapping,8'11 where they were drawn in accordance with
1. Forces on conductors in slots. the principle outlined in Appendix B of this paper).
2. Forces within either layer of the conical surface Substantially the same results were obtained in a later
of the end winding exerted on the sides of the coils.
3. Forces on the straight coil extension from the
slot.
4. Forces on the phase connecting rings.
5. Forces on the whole conical surface of the end
winding.
jwlJ
II. Detailed Discussion of Forces during
Short Circuit
A. FORCES ON CONDUCTORS IN SLOTS
In the study of the forces acting on the embedded
portions of the armature conductors, there are three
main problems to be considered, i. e.:
1. Is the magnitude of the compressive force on
the coil sides great enough to damage or rupture the +

insulation?

FIG. 2-CURRENTS EQUAL IN VALUE BUT OPPOSITE

IN DIRECTION
Iron of infinite permeability and resistivity

paper by Messrs. Robertson and Terry, who made a

very careful mathematical analysis.7
Fig. 1 represents the current conditions on a single-
phase line-to-neutral short circuit in machines with
an armature throw greater than 23, or a single-phase
line-to-line short circuit with any commercially practi-
cable coil throw. Fig. 2 represents some instantaneous
condition during short circuit on a machine with a
chorded winding, while Fig. 3 corresponds to some other
instantaneous relation. These figures illustrate possi-
ble flux and current relations for a number of short-cir-
The
_, <4\
X effect of saturation will be considered neglected.
cuit conditions if saturation can be later.
FIG. 1CURRENTS EQUAL AND IN THE SAME DIRECTION Calculation by the Use of Images. By comparing
Iron of infinite permeability and resistivity Figs. 1 and 4, it will be seen that the influence of the iron
boundaries of the former are replaced by an infinite
2. Are there repulsive forces between coil sides series of conductors in the latter.8 (In Appendix C
which might send a coil into the air gap (carrying the are listed the more usual boundary conditions found in
stator wedge with it)? electrical apparatus which can be represented or re-
3. Are there forces which may rend apart the placed by appropriate series of images.) It will be
strands of one coil side? found that the corresponding conductors, shown in
As previously mentioned, it is desirable in some Figs. 1 and 4, not only carry exactly the same currents,
one instance to use all three of the methods for calcu- but exist in flux fields of identically the same form and
lating the forces, i. e., those involving-(1) images, densities. It follows that the forces exerted on the
(2) change of interlinkage with displacement, and (3) conductors must be the same in the two cases. With
180 CALVERT Transactions A. I. E. E.

some modifications, the forces will be calculated from Then in Fig. 4, the total force on the upper con-
the conditions shown in Fig. 4. ductor is
Methods are available for considering the shape, n=+w 2 I2 In
dimensions, and separation of rectangular conductors F 44 (10)5 (2)

also a =Vy2 + x2
From symmetry, it can be seen that the forces in the
x direction cancel, because corresponding to every
pull to the right there is a duplicate one to the left.
Then the y component of force is
n=+ co
__2_I 2(3)
[ X Fv > 445 (1O)1
(1))6 VY(3) X.2 V Y2I+0
v~~~~~~~~~~~~~~~~~
+y2+ X.2
2
445 (1O)5
I2=ny
2±n
(4)

To avoid the labor of making this summation to

values where the contributions to F, are negligible,
a further simplification may be made. It will be ob-
served that I,, 12. If each value of I, is assumed to
be spread out evenly so that the group form a sheet
FIG. 3-CURRENT IN THE UPPER CONDUCTOR TWICE THAT OF
THE LOWER AND IN THE OPPOSITE DIRECTION fi
Iron of infinite permeability and resistivity 7y

in computing forces, if adjacent sides of the conductors h x x

are parallel.4'5 However, there do not seem to be ____ ,
any methods worked out for the case of conductors
placed in other positions relative to one another. .
It will be assumed for the present that the current in FIG. 5-REPLACEMENT OF CERTAIN SERIES OF CONDUCTORS
each conductor Fig.
in 4 is concentrated at the center. IN FIG. 4 TO AID IN CALCULATING THE APPROXIMATE FORCES
Then the force between two parallel line conductors through the centers of the conductors, instead of being
in Fig. 4 will be assumed concentrated at the centers, then
2 I2I, dx
F
445 X (10)5 a (1) In I2 SW =
(5)
and the total force in the y direction, from Fig. 5, is

,~ nlir.. n-' 1 15
-1i r1 r- 1 f 445 (1O)6 STO (y/2~~~~~~~~~~~~~~21I22
y +X2) (
(dx)=

r4iYWFt"L. ! q Li lL.XI IL..J1 =Yi

IJ L La Y
l l I!l U 1..)
2I2 y =y3 / X2_
INFINITE I._JSERIES 1' 1 1 1 1 r 1 e\x=-S
445 (1O)~S4
L__- .J L L_ L_ L
LJ L_J 4

i L4 L-L- J
-J - Li- LJL- l- F5 = 14.1 X 1O- S multiplied (8)

OF CONDUCTORS,
ANFIG. 4-REPLACEMENT OF THE SLOT SIDES THE
TO OBTAIN IDENTICAL
OF FIG. 1 BY by [Number of layers below the conductor
FLUX FIELD minus the number of layers above the
conductor]
where F = force in lb. per linear in. of conductor. Where the layers are both imaged and real, as shown
12 = current in one conductor. in Fig. 5.
Ina = current in the other conductor. In Fig. 1, where the currents are equal and in the
a = distance between the conductors in inches. same direction,
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 181

For the upper conductor, dL = d r N212 10-81 (19)

dL2 3.19
For the lower conductor, d = S ( ) (20)
122
F8=14.1x10-8x ,,. (10) dM d F NOm
,
1
(101)
319 (10-s)
(1
These forces are toward the base of the slot. dy dy L 2 sv (
Calculation of the Force from the Rate of Change of Substituting equations (18), (20), and (21) in (15)
Interlinkage. This method is very commonly em- I22
ployed. In the following case, it is based on Maxwell's 2F, = 8.85 [(3.19 + 2 X 3.19) X 10-1] 2 (23)
equations for the stored magnetic energy in ergs in
two electric circuits, i. e. 1I2
I1 ' 12 L2' (11) -F = 42.45 X 10 j (24)
100W' = ,J +2I_____M
29 2 + 1M)
By supposing that the lower conductor is moved
The single-phase short circuit will be considered. downward, and the upper held fast, it will be found that
Then
I1 = 12
If, for the instant when the current is at the peak
(12) 1Fv= 14.15 X 10-8
SW
(25)

value, a conductor is allowed to move by a small Equations (24) and (25) are in very close agreement
amount, then with equations (9) and (10), respectively.
d WI'F/ = ' (d y') (13) Calculation of the Forces from the Integral of H2
or d Wx' = Fx' (d x') (14) Around the Conductor. Any surface will be chosen
which surrounds a conductor. It will be imagined
But the work done must be at the expense of the that the chosen boundary is rigid, and that forces
stored energy, and the forces in lb. per in. of conductor exist between this surface and the conductor. The
are forces on the surface will act as follows: there will
I22 (dL d L2 dM ) be tension along the flux tubes producing a force in
F,=8.85 22 dy
+
dy
+2
dy
(15
dynes per unit of area.
122 /d Li dL d M \F' (H'1)2 (26)
Fx = 8.85 2 dx + dx + 2 d x (16) 8r
As before, the x components of the force will be and there will be pressure per unit of area exerted
considered first. The values of L1, L2 and M are single perpendicular to the flux surfaces (designated by a
valued and continuous functions of the displacement. g g
In Figs. 1, 2, and 3, the values of Li, L2 and M may F' = _ (H')2 (27)
either increase, decrease or remain constant, but it will 8wr
be the same for a motion + A x as for - A x. Then (H' lines per sq. cm.)
d Li d(L2 dM
=0, -xdx= 0, and dx = 0, individually, The force on the conductor will be equal and op-
dz dz posite to that acting on the surface. (What is be-
so that lieved to be a new proof of this principle is given in
Fx = 0 (17) Appendix D. It is a proof which depends only on
(as was found before from the solution by the aid of the more commonly used laws of electricity and magnet-
images). ism, and which requires no detailed mathematical
The forces in the y direction will now be considered. study).
Robertson and Terry7 mnade calculations which showed Converted into units of lb. per sq. In. and lines per
the usual assumption was a good one, namely, that sq. in., equations (26) and (27) become
the flux went straight across the slot. Proceeding H \2
on this basis, suppose that the upper conductor is F = + 0.0139 ( OOJ(28)
moved downward a distance d y.
Then N\
d L1 _ d N1cp1 (108 and°18 F = -0.0139 io000l (29)
dy dy L 1 08j0(8 respectively.
182 CALVERT Transactions A. I. E. E.
Applying equation (29) to the conditions indicated forces tending to separate the strands of any armature
by Fig. (1), it is found that between conductors conductor. It will be assumed, first, that a small
12 cylindrical hole is cut through a coil, as shown
HX = 3.19 S and above the upper conductor in Fig. 6. In this hole there is one small conductor,
which is electrically in parallel with the main con-
12 ductor, but the two are not in electrical contact at
H2 =2 X 3.19 S , and the same results are found for the section shown. The flux density at this point
will be just the same as if the insulating space sur-
2FY and 1F, as are given by equations (24) and (25), rounding this strand were of conducting material,
respectively. Since the pulls on the two sides of the
slot are equal and opposite, F, = 0 (as was found
before).
Forces Tending to Drive a Conductor out of the Slot.
In any case where the winding pitch is between 23
and full pitch, a three-phase short circuit may be ex-
pected to produce a flux distribution at one instant
somewhat like that shown in Fig. 2, where the currents
are equal but opposite in direction.
It can be shown by any of the three methods de-
veloped that for Fig. 2 (when saturation is neglected)
122
2FV =-14.15 X 10- S, (30)
FIG. 7-DIAGRAMMATIC SKETCH FOR THE INSTANT BEFORE
The minus sign indicates a force tending to drive SHORT CIRCUIT
the upper conductor out of the slot. Also
122 for this space may be assumed to be very small. By
= + 14.15 X 10-8 S (31) making the hole extremely small, the direction of the
force is more easily perceived. The flux density H'
and acts toward the bottom of the slot. (due to the main conductor) must be single valued,
It will be found that the force 2F, will be consider- so that as the isolated conductor is made smaller the
ably less than the maximum value existing on a single- main field on its two sides becomes more and more
phase short circuit. However, it is still important, nearly parallel. The force acting on this conductor is
because the wedge must carry the load. This force (A I)
should be given consideration, particularly in machines zA F' = H' 10 (32)
j_~~~~~~~~~~~~~~~1
r|1- $ r and is in the direction of the weaker resultant field.
/" t | This is along the ampere turn line toward the m. m. f.
center. Hence, all parts of the conductor are com-
N_r lt lpressed toward the m. m. f. center. The strands
li § > +further out are in stronger fields. They are acted on
\ / with greater force so that there are no effects due to
ei , ll/; V -the inertia of the strands which may tend to separate
them. From an examination of Figs. 1, 2, and 3, it
will be seen that there probably are no combinations
of current values possible in armature conductors,
which will result in electromagnetic forces, tending to
FIG. 6-SECTION OF FIG. 1 WITH A SMALL PART OF THE
CONDUCTOR ELECTRICALLY IN PARALLEL BJUT INSULATED FROM disrupt the coil sides by the separation of the strands.
THE REMAINDER Any forces which separate the strands must be brought
about by pulsating forces which press the whole con-
which can have high currents per slot, and which ductor against its supports (bottom of the slot and the
have wide slots. wedge) with such violence that the strands are shaken
If the coil were slightly loose so that the wedge apart.
could receive a series of impact blows (which would Reduction of Forces Due to Saturation. Saturation
arrive at twice the normal electrical frequency of the of the armature teeth will greatly reduce the forces.
machine), the possibility of failure would be con- In order to consider this feature, a brief survey of
siderably greater. ;the general flux distribution is necessary.
Forces on Individual Strands. It is intended, next, The particular condition which will be considered is
to determine whether there are any electromagnetic that existing at the instant one-half cycle after a single-.
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 183

phase line-to-neutral short circuit has occurred. Figs. 7 paths, and constant interlinkages were assumed, the
and 8 show the resultant flux immediately before the m. m. f. of each member would be practically equal.
short circuit, and one-half cycle later, respectively. Oscillographic records show that the currents of the
It is assumed in Fig. 7 that the armature, A, is open- field winding, f, will not produce an m. m. f. anywhere
circuited, but that the field winding, f, is excited so as to nearly comparable with that of the armature winding
maintain normal voltage at the generator terminals. during the first cycle after the short circuit occurs.
No currents of fundamental frequency are flowing in As the mutual reactance of the rotor and stator circuits
the damping circuits D. The armature is short-cir- is high, there must be a very large current flowing on
cuited at the instant (0 = 0) when the flux "trapped" the surface of the rotor. The leakage path for this
in the electric circuits, A, D and f are as shown. current has a greater reluctance than that of the arma-
ture. From Fig. 7 the initial flux linkages of the rotor
are greater than those of the stator. Therefore, in
order to maintain these very nearly the same in Fig. 8
as in Fig. 7, the total rotor m. m. f. must be greater
than that of the stator. This means that at a very
great distance from all circuits the total m. m. f. of the
-At | 1l 1:. 3q + A rotor will prevail. Hence, there is a very slight amount
l/1 l X/ of air gap flux threading both the armature and rotor
\¢( @ l l %> circuits in a direction opposite to that of the armature
leakage flux.
The total armature leakage flux probably will be
somewhat greater than the no load air gap flux, because
of the difference in distribution of the two fields. A less
FIG. 8-DIAGRAMMATIC SKETCH FOR THE INSTANT ONE-HALF percentage, of the leakage flux links all the winding.
CYCLE AFTER THE SHORT CIRCUIT OCCURRED In addition to this, there must be enough armature
flux to offset the linkage of the small amount of air gap
It is assumed that in each circuit current will flow flux which links both windings. The armature leakage
which tends to maintain constant flux linkages in its flux will require relatively few ampere turns for the core
own particular circuit. compared with the number required for the teeth and
Then when the rotor has turned 180 electrical deg., slot.
the resultant flux will be as shown diagrammatically in The retaining ring leakage flux will be prevented from
Fig. 8. The sum of the gap and field leakage flux increasing greatly during the first half cycle by eddy

_UUD :W
U- 11 h
FIG. 9-APPROXIMATE DIRECTIONS OF THE FLuX IN THE GAP TEETH AND CORE

(j b + zf4 ) must remain practically up to its initial

. currents and also by saturation. Eddy currents in the
value, because this constitutes the flux linkages of the sides of the rotor teeth will oppose increase in rotor sAlot
rotor winding. However, nearly all of the gap flux leakage flux. Consequently, the air gap flux can de-
must pass- inside the armature conductors, because the crease only slightly during the first half cycle. This
latter must have a net interlinkage opposite to that means that the total rotor magnetomotive force due
which the gap flux, alone, would produce if it returned to the winding currents and rotor surface currents will
through the stator core. The armature must have exceed the total m. ntr. f. of the stator by an amount
sufficient leakage flux to maintain its interlinkages approximately equal to the no load normal voltage
up to the initial value, and in the initial direction. m. m. f.
If the stator and rotor had precisely equal leakage If it were intended to map the complete flux field, cer-
184 CALVERT Transactions A. I. E. E.

tain steps would be necessary. Determine the magni- tion of the teeth is neglected, and when it is considered.
tude of the armature current as described in Appendix By equations (28) and (29) the forces are as follows:
A, or by any suitable method. Give to the rotor a
m. m. f. greater than that of the stator by an amount
approximately equal to the no load gap ampere turns. Saturation Saturation
Determine the rotor winding ampere turns relative to neglected considered
the stator from tests on the nearest similar machine.
Locate the
Lopposte remaining rotor
therematorwining current on the
ngs.orcurraw the rrotor surface
suxrfaeld
f Electromagnetic force on upper con-
ductor (toward base of slot) 1062 695
......
opposite the stator windings. Draw the flux field for Electromagnetic force on lower con-
the non-magnetic parts by assuming no saturation. ductor (toward base of slot) ......
354 210
Complete the field in the iron on the assumption of
slight saturation and uniform permeability. Calculate Sum = total force on lower conduc-
the ampere turns for the iron parts and attempt to tor (toward base of slot) ......... 1416 905
redistribute the ampere turn lines to take care of this. This shows that saturation is a most important factor
It is only intended here to consider the effects of to be considered in calculating the loading on the coil
saturation in a single slot pitch.
Fig. 9 shows the approximate direction of the flux
field in the gap teeth and core on a single-phase line-to- l so ^x2
neutral short circuit for the instant shown in Fig. 8.
The effect of saturation will be considered in the
zones Y-Y of Fig. 9. The flux in this zone will travel > m
nearly tangentially across the slot and teeth. Except
for a negligible number of ampere turns required for
the core, the ampere turn lines from the stator conduc-
I I taorZ7z exwz o FIG. 11-DIFFERENCE IN DENSITIES ON THE Two SIDES OF A
l7W l 320 TOOTH UNDER LOAD CONDITIONS

insulation in the slot. In Fig. 9 it is seen that for cer-

e

lX Z2 - - 7L- Xtain other conductors the flux will travel nearly radially
in one tooth which bounds the slot, and it would be
expected that saturation would have an even greater
effect in reducing the force on these conductors. Due
to the dissymmetry of the flux distributions in the
adjacent teeth, there should be some tangential force
exerted on the conductors.
Normal Torque of a Machine. While the torque of a
rotating machine is not within the province of this
paper, some mention probably should be made of it.
It has been pointed out several times in the preceding
-X n-1- - - paragraphs that when there is no appreciable satura-
I tion, there is no tangential force exerted on the conduc-
FIG. 10-DENSITIES IN THE SLOT AT Y-Y OF FIG. 9 CURVES tors in the slot. In general, this approximates the
SHOW SLOT DENSITIES FOR 50,000 AMPERES PER CONDUCTOR normal running conditions of many machines. The
(1) when saturation is considered, (2) when saturation is neglected torque must be found near the teeth tip in the difference
and it is assumed that the flux goes straight across the slot in densities on the two sides (see Fig. 11). With
saturation, the flux will not enter circumferential sur-
tors in this slot will proceed almost directly toward the faces in a radial direction and some tangential pull will
rotor currents. These will spread enough to supply the result from the corresponding components, at these
requirements of half of each adjacent tooth as well as surfaces. In the old surface wound machines the torque
the slot itself. The values which will be used are ap- was carried almost entirely by the armature con-
proximately those which existed in an actual machine. ductors.* It should be noted that when the field map
There will be 50,000 amperes per coil side, or 100,000 has been drawn, equations (28) and (29) can be used to
ampere turns for the slot. The densities in the core are determine the forces exerted on the iron and copper
such as to require less than 1000 ampere turns in the
iro bac ofteso ic ,
an hec a eng *This is not a new idea, but has been studied in part both
theoretically and experimentally. See bibliography referencee 13.
Iected. Fig. 10 shows the densities In the slot with each Prof. A. D. Moore has a,lso used for some time in his lectures
conductor carrying 50,000 amperes, when the satura- these general facts concerning torques.
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 185

parts in any portion of the field. This applies not only from other metallic parts. By making just that
to rotating machines, but to contactors, and to any assumption, the force on any conductor may be com-
other apparatus in which the flux distribution can be puted by the successive use of equation (1). Thus,
determined. 2 12 1 \
Conclusions concerning the forces on conductors in F = 445 X 105 a l+M2+Y3+. + N- 1) (33)
armature slots.
1. The force pulsating at twice normal frequency Where:
and pressing the coils toward the bottom of the slot I = maximum armature current per coil side.
on a single-phase short circuit may have a value of 695 a = spacing center to center between coil sides in
lb. per in. of conductor (or greater). one layer of the conical surface of the end
2. On three-phase short circuits in machines of winding.
less than full pitch, there will be forces tending to N = number of coils per phase group.
drive a conductor out into the air gap, carrying the
wedge with it. These forces, while less than those just 4'7VDZ2VS
mentioned above, can be very serious if a coil is loose so
that the forces on the wedge arrive as impact blows. ,' /
3. The forces exerted toward the sides of the slot
are very much less, and only exist when the saturation , /
in two adjacent teeth is appreciably different.
4. There are no electromagnetic forces directly
acting to disrupt the conductors by separating the
strands one from another. However, the hammering
of the conductor on the, bottom of the slot, if the con-
ductor is loose, may cause some internal motion. \
This might be capable of wearing out the insulation M M 0 P
between the strands very quickly. FIG. 13-APPROXIMATE FORCE AGAINST THE SIDE OF THE COIL

B. FORCES WITHIN THE CONICAL SURFACE OF THE

END WINDING As before, the
conductors
correction factors for the shape of the
are neglected. These factors will reduce
Fig. 12 indicates the approximate paths of the the force to some extent. There will be a different cor-
flux linking the armature end winding on a single- rection factor for the force between each pair of condue-
phase line-to-neutral short circuit. All currents in the depending
parallel conductors are in the same direction, so that 3 a, or (Nu-p1) a, etc.
In order to calculate the force per inch of conductor
within one layer of the conical end winding of the ma-
chine for which the last numerical example was com-
puted, it is noted that
I = 50,000
a =1. 79
N= 7
Then from equation (33)
-' F =154 lb. per in. of conductor.
Fio. 12-APPROXIMATE PATHS OF THE FLUX LINKING THE Fig. 13 shows the probable sort of force distribu-
DIAMOND COIL ARMATURE END WINDING tion in the plane of the windings based on the conditions
indicated in Fig. 12.
these coil sides tend to pull together. The maximum Fig. 12 shows that greater densities exist at the inner
force will be exerted on the inside conductor at sections bend of the coil. The force must fall off toward the
X-X, because the greatest concentration of flux is in outer end of the coil for there are less and less coil
this elbow of the phase group. Toward the outer or sides outside of the inner one. Then as the outer bend
diamond ends of the coils there will be a force between of the coil is reached the repulsion from the coil sides in
the coil sides in the two layers tending to open out this the other layer of the phase group will tend to increase
angle. That is the coil ends have a tendency to bow the force.
out to a sort of semi-circular shape. The flux distribu- Conclusions concerning the forces on the conductors
tion around the semi-straight portion of the coil, be- witthin the conical surface of the end winding.
tween the inner elbow (X-X) and the diamond (outer) 1. The forces on a single-phase short circuit tend to
end, has much the same shape as if it were due to a pull all the conductors together within one layer of the
group of long straight conductors in air, far removed conical end winding.
186 CALVERT Transactions A. I. E. E.

2. The force is greatest on the inner coil sides of The currents indicated are as follows:
the group at the bend (or elbow) of the coil nearest the Ia = the armature current per coil side (as defined
frame. It decreases from there toward the outer bend in Appendix A).
of the diamond coil. If = the current in the field winding for the instant
3. In between these two bends in the coil the force shown in Fig. 8.
may be 150 lb. per in. of conductor, or greater, on the iso = the rotor surface current on the outside of the
assumption that the force may be computed as would be rotor. This is practically the continuation
done for long straight conductors in air. The fact that of the damping currents of the rotor body
the space within the phase group is limited indicates out into the retaining ring. The straight
that the force should be greater than calculated, but if
the correction for the coil shape is neglected, this is
compensated for to a certain degree.
4. This force is sufficient to distort the winding
unless bracing blocks (or the equivalent) are located
between coils at quite frequent intervals.
5. Conductors which are two strands wide are very
much weaker than those which are one strand wide in
the direction of this force. It is necessary that a manu-
facturer know the strength limitations of the coils and /
stay well within them. It should be noticed that while
the resistance to shear between strands is very low, KK
that the copper is held on the ends, so that tests on a
short bar are not satisfactory. Undoubtedly testing
coils to destruction on actual machines furnish the most /
reliable data on which to check calculations.
C. FORCES ON THE COIL EXTENSIONS FROM THE
FRAME
Tests on actual machines have shown that the
straight parts of the coils extending from the slots over
the retaining rings are almost as liable to failure as are FIG. 15-VIEW OF FIG. 14 IN THE RADIAL PLANE AT THE
the coil sides within either of the two layers of the CENTERLINE BETWEEN POLES
conical end winding.
The conditions illustrated in conjunction with Figs. 7 parts of armature coils extending directly
and 8 will be used for the basis of the following calcula- out from the slots furnish the m. m. f. to
tions. It was concluded that at the instant indicated carry thesge currents out into the retaining
Srwrowrf - ring to oppose the annature. If it were not
5>'ADv
-9""' for the arrangement of these armature con-
Mr g ;3 03 zV<>8z6 ductors, there are reasons to believe that
l5yqroq I,0/4 ,,,,n,zwith a tapered magnetic retaining ring this
V _fi2£ Eg [2 i g ) surface current would turn and travel
950 tangentially very soon after it left the rotor
*-I/=-;--4
gr ,zr<,9M,M < body. However, this seems impossible
when the armature winding is so near, and
/- [r [m [_] [1 _ l_l \ this belief has been very forcefully upheld
i L:5101 5i LI Li '/ in a test to destruction. It is assumed,
FIG. 14-DIAGRAMMATIC SKETCH OF THE CURRENTS FLOWING then, that this rotor surface current i6,
IN THE END ZONES JUST OUTSIDE THE FRAME travels axially across the outside surface
of the tapered section of the retaining ring.
by Fig. 8, the rotor m. m. f. was equal to that of the tr =the transient current flowing in the retaining
stator plus an amount approximately equal to the field ring to oppose any change in the retaining
m. m. f. at the no load normal voltage condition. ring leakage flux.
This applied to the main magnetic circuit. A detailed .jr = t;he retaining ring leakage flux.
study will be undertaken now in an attempt to picture isq = the retaining ring surface current on the in-
the current and flux distributions just outside the ends side surface which opposes any change in
of the "active" armature iron. the flux entering or leaving the ring from
Figs. 14 and 15 indicate, in a diagrammatic way, the the inside.
currents which must flow in this zone. With the currents defined, it will now be possible to
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 187

consider the ampere turn surfaces. These emanate ring. The real current sheet, i,0, will travel nearly
from the rotor field winding and pass through the re- axially outward under the straight part of the armature
taining ring. Most of them terminate on the armature coil extensions, because it flows as a sort of image of the
conductors which carry current in the opposite direction armature current.
to the field windings. A few of these ampere turn sur- Oscillographic records of armature and field winding
faces go through the air gap and through part of the currents show that the armature m. m. f. is much the
stator iron to terminate in the field windings on the greater of the two. Yet the constant interlinkage
other side of the pole. theory demands that the total rotor m. m. f. be slightly
The effects of the rotor surface currents must be greater than the stator. This means that the m. m. f.
considered. The current on the inner surface of the of the rotor surface currents, i.0, is much greater than
retaining rings, i,i, will be neglected. This is probably that of the rotor winding If at the instant shown in
the wrong thing to do, but it is desired to be on the safe Figs. 8, 14, and 15. This real surface current, i0,s must
side in the force calculations. It is possible, and it be as much larger than the imaginary current sheet
even seems probable, that this current might nearly used to replace the saturation of the ring. The resul-
cancel the increase in m. m. f. in the end zone, due to tant current sheet of the real and imaginary one must
the increase in field current, but this is neglected. have almost axial directions of flow under the armature
The net effect of the current i, is practically zero. coils. The strength of the stator m. m. f. may actually
This component is shown by two current sheets. One be the greater at a short distance out in the tapered
is in the inner surface of the ring and one in the outer. section of the ring if the effect of the current sheet,
These flow in opposite directions. What the inner ii on the inside of the ring is considered.
sheet subtracts from the m. m. f. between a given pair The purpose of the foregoing discussion of current and
of ampere turn surfaces, the outer adds in again. flux in the end zone near the tapered section of the re-
SrAT.aw END PLA7rE

Ro To P P I A/G

FIG. 16-FLuX FIELD, IN THE END ZONE JUIST OUTSIDE THE END BELLS
Very small armature conductors are used for simplicity in mapping

There remains the rotor surface current, i,,, and a taining ring has been an attempt to get a graphic picture
certain contribution from the rotor windings. The of the actual conditions. Some approximations will
ampere turn surfaces arriving at the outer surface of the be made now to compute the forces on the armature
retaining ring can be replaced by an imaginary current windings.
sheet of just the right current density, which flows on a A resultant or equivalent current sheet will be
ring of infinite permeability. (The replacement of assumed to exist on the rotor retaining ring, and flowing
saturation by the suitable current sheet and unsaturated axially outward from the rotor slots. It will have a uni-
iron has been demonstrated by Th. Lehmann in some form current density and a total m. m. f. equal to that
of his excellent papers concerning graphical field map- of the armature plus the no load normal voltage rotor
ping.9,10'11) This imaginary current sheet may now m. m. f. The flux to be considered will be assumed to
be superposed on the real (transient) current sheet, remain within the zone of uniform width indicated in
i.0, flowing on the outside surface of the ring. Fig. 15. (The direction of the flux is approximately
The directions of these two component current sheets perpendicular to the plane of that drawing.) Satura-
are probably not the same. tion in the stator end plate is neglected, because it is a
The saturation in the ring is greatest in the tapered magnetic parallel on the stator core. The magnitude
section. Consequently, the leakage flux in the ring of the eddy currents in this plate cannot be estimated
will travel nearly axially across this section. A large satisfactorily. They will be neglected for the same
proportion of the ampere turn surfaces maintaining reason that the rotor m. m. f. was assumed at its maxi-
this leakage flux will lie in the ring in fairly near mum possible value, i. e., to be on the safe side. On
circumferential directions. The imaginary current this basis, Fig. 16 was drawn (using small armature
sheet used to replace this saturation must have the conductors), and Fig. 17 drawn from Fig. 16. The
same directions as these ampere turn surfaces inside the forces on the conductors may be computed from the
188 CALVERT Transactions A. I. E. E.
integral of H2 taken around each one. The wav the able magnitude. These act against the coil in the di-
flux encircles the group of coils indicates appreciable rection in which the individual strands are the weakest,
tangential forces on the outer ones of the group. How- but there are many of them in the coil side. Tension
ever, the coils may be well braced against this by roped- and compression along these strands operate to prevent
in blocking. The highest radial forces occur on the failure. The force falls off with distance from the slot
coil sides in the center of the group. A very easy means toward the first bend or elbow of the coil, and very
of approximating this force becomes evident from the rapidly beyond the bend.
work done concerning the conductors embedded in slots. 3. Experience indicates that this is a less likely place
It is as though the conductors lay in slots of width equal for failure to occur than within a layer of the conical
to the slot pitch. See section Y'-Y', Fig. 17. From surface of the end winding, but if the straight extension
I Y <NFrvo~~-"
rA '10
rtow EN PLA7ir | Y

I,' - 1 90 TOE f?rTA/NI

FIG. 17 FIG. 16 REDRAWN FOR ARMATURE CONDUCTORS OF NOIRMAL SIZE

equation (24), it seems the force on the conductor near- from the slots is long, special bracing should be supplied.
est the retaining ring will be about 4. A matter deserving real consideration is the
122 tendency to twist the conductor due to the radial and
2F,, = 42.45 (10-8) q.(34) q
tangential force exerted on the bend nearest the frame
the coil side at the inside of the phase group. Careful
~~~~~of
where q = the slot pitch in inches. bracing on each side of this bend (see section X-X,
lf q = 2. 185 and I2 = 50,000, then 2Fb, = 485 lb. per Fig. 12) m'st be used to so stiffen the mechanical re-
running in. of conductor. sistance of the coil and supports that it will remove this
Fig. 18 indicates the probable sort of variation load possibility of failure.
frowtheframe. kown. On the whle, they are eay to apply, bu
against the ede of the
coil.D. FORCES ON THE PHASE CONNECTING RINGS
AAw"A, Z*-o 7/-r* Rectangular copper straps are bent into semi-circular
/ SK§t2Z-Z/wM/29976WSarcs, and used to connect the terminals of the machine to
@ X ends of the various phase groups. These phase con-
~~~~~~~the
t\\ 1 l l'l ~~~~necting rings are considerably closer to each other than
inserted,FI 17d FIGs ~~ ~~~ 16n REDAW ARMATUReqatsp-
= FORe~~~toEanyCONDCTOS
heavy ironOF H NHOMLE
parts. SE GOPI
TheySIZare located H
in a relatively
T ~~~~weak part of the field of the armature end windings.
/ ~~~~~Theradius of curvature of the rings is rather large.
portaga 2Fs/rfnl=ue 42e.4 (1-)(4 agnilfreeete
Therefore, ntebn approximation
it seems a reasonable ers h rmto
wher q=Thefsotpitchtingrainches./ raroaprei ~~~~~~~assume that the forces
bAcn ountieachv sidetiofonthis connecting
phase(seen
thebndt rings
section X-Xt
> ~~~~~~~~~will
be about the same as if thev were straight conduc-
n , tors in air well removed from -other metallic parts.
O5ZT"IVIY The forces can be calculated by equation (1). How-
FIG.18-PPRXIMTE ORC AGANSTTHEDGEOF
HECTI,ever, the directions of the currents must be determined
from the conditions of the short circuit. The usual
Conclusions concerning the forces on the coil extensions forms of bracing, blocking, and clamping are well
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 189

for estimating these forces. Probably their direc- where a reasonably simple mathematical expression
tions are almost as important. The currents on can be obtained; but if the method involves the differ-
the retaining ring (the real transient current and ence of two large values of interlinkages by the use of
the imaginary current used to replace the satura- graphical field maps, it is not satisfactory. The
tion of the ring) will tend to produce an approximately method which uses the f HI d s over any surface
outward force on the phase group. The coils are rather enclosing a conductor is the most generally applicable.
centrally located between the iron parts of the frame It frequently involves the plotting of one rather
and end-bell, so that the effect of these latter parts difficult field map, but in such a case there is probably
should tend to cancel out. no way of avoiding this labor. (A simple proof is
It would be very desirable to have a definite, but given in Appendix D for the basic principle of this
not too great force, pressing the coils outward toward last method for the calculation of the forces.)
coil supports which are mounted on the frame. Such The forces exerted on the embedded armature con-
an arrangement would eliminate any need for bolts ductors during short circuit are computed and the
through the winding to hold it back toward the frame. great influence of saturation is illustrated. It is shown
There are two ways to accomplish this. Laminated that there are no electromagnetic forces tending to
iron near the stator end plate to furnish a strong at- disrupt the coil sides by rending apart the strands.
traction would achieve this result, but it is commercially A graphical picture is given for the normal torque of a
a rather impractical measure. A good conductor on machine. The forces between coil sides in one layer
the inner end-bell would carry eddy currents which of the conical end winding are computed. The forces
would offer a force of repulsion to the armature wind- on the straight coil extensions from the frame are
ing and tend to force it back toward the frame. This approximated. The forces on the whole conical sur-
is a practical expedient. An aluminum diffuser for a face are discussed, and methods of causing the force
double entrance fan will serve this purpose. However, to be definitely toward the frame are pointed out.
it is possible such a thing might be very badly done. The applicability of simple formula for approximating
If the entire inner end-bell were made of aluminum or the forces on the phase connecting rings is also pointed
copper, it would probably have high losses, and possibly out.
excessive heating under normal load conditions. The
reason would be that the path of the armature leakage ACKNOWLEDGMENTS
flux would be too greatly restricted. This damper The writer greatly appreciates the assistance of
would have to render a magnetomotive force almost Messrs. M. G. Leonard, A. M. Harrison, M. R. Lory,
equal to that of the armature. The smaller damper and R. D. Reed, for their assistance in making the field
will carry only a relatively small current and loss at maps, and also appreciates the helpful discussions
normal load. The tremendous increase in flux on with Messrs. L. A. Kilgore and C. M. Laffoon.*
short circuit will give sufficient current to produce a
definite force in the desired direction. Appendix A
In addition to the forces just discussed where the
resultant lies in a radial plane, there must be a tangen- ARMATURE SHORT-CIRCUIT CURRENT PER COIL SIDE
tial force acting on the phase group in the direction The maximum possible peak current per coil side
opposite to that of rotation. This torque will be due which exists at the end of the first half cycle after an
to the losses occurring in the end-bell, end plate, etc. instantaneous line-to-neutral short circuit is often
It is as though these iron parts were the secondary, computed by the approximate formula
and the stator winding the primary of a surface wound
induction motor, so that the burden of the torque falls I,,, X 1.5 X 2 -/ 2 X .99 (35)
on the armature conductors.
Some further considerations of the forces are given Xd
J -X ,
in Appendix E for certain limiting cases. I
I - 3.82 _rms (36)
SUMMARY X dt
Fairly detailed conclusions have been given at the
end of each section in the paper, so only a brief summary where,
will be made here. I - maximum short-circuit current per
Three methods of calculating the force exerted on coil side.
current carrying conductors have been demonstrated. Irm = r. m. s. value of normal load current
The method involving the use of images can only beused prci ie
when there is no appreciable degree of saturation. (The prci ie
most common configurations which can be imaged are XI' = subtransient reactance expressed as a
summarized in the Appendix C). The method in- _____decimal fraction.
volving the change of flux interlinkages is most useful *All of the Wsestinghouse Company.
190 CALVERT Transactions A. I. E. E.

Appendix B the boundaries may be replaced by the use of other

conductors are those of
BASIS OF GRAPHICAL FIELD MAPPING 1. Conductors between two planes which meet at
The basis for graphically mapping two dimensional a certain angle. The angle expressed in degrees when
flux fields, which is used in this article, has been pub- multiplied by some integer must give 180 deg.
lished several times,3,8,12 but is given here in an ab- 2. Conductors between parallel planes.
breviated form for the sake of completeness. 3. Conductors in rectangular slots.
Fig. 19 illustrates the flux field of a salient pole 4. Conductors in completely closed rectangles.
machine at the no-load normal voltage condition. In the first three cases the plane sides should have
either infinite permeability (and if an alternating
field is considered infinite resistivity, or unit permea-
bility and zero resistivity). In the fourth case if the
boundary is of infinite permeability there must be
as much current in one direction as the other so that
no ampere turn lines penetrate the boundary. If the
surrounding surface is of zero resistivity, it is not
necessary that the sum of the currents in the enclosed
.conductors be zero for the surface will carry currents
to receive any excess ampere turns from the conductors.
Appendix D
CALCULATION OF FORCE FROM f H2 d s
It is apparent that every element of current tends to
establish a field of force around it in such a manner that
FIG. 19-FLUX FIELD OF A SALIENT-POLE MACHINE AT THE N O- the flux lines form closed loops. Consequently at
LOAD N'ORMAL VOLTAGE CONDITION
certain conductors in any field the ampere turn lines,
or lines of no work, must converge at points of no flux
It will be assumed correct, and the relations necessary desit
that it be the true field will be illustrated (uniqueness y. These points will be called m. m. f. centers.
is assumed). If a unit pole is moved around the path However, not all current carrying conductors may
*
o p n o linking the current I2 and area A2 the work s l ~~~~~~~~~~susta1napzto
in a point of zeroeodnly h centers
density. The etr which hc
done is f. H' d 1' = 0.4 7 I2. This is all done along they might sustain if excited alone are submerged in
the flux line p - n of length 11'. If the elemental
tube of flux A 0 considered, has a mean width (A d2')
and mean density H2' in this length 12', then
H2' 12' = 0.4 7r.I2 (37) /
(Aq) 12 -= 0.4 7r I2 (38)

Similarly for the path

,Adl = 0.4 1r I1 (39)

A~~ B
From the last two equations A B
FIG. 20-A. A CONDUCTOR WHICH MAINTAINS A M. M. F.
1j/A d1' R1 I, A1 CENTER
12'/A d2 R2 I2 A2 (40) B. A CONDUCTOR WHICH DOES NOT MAINTAIN A
M. M. F. CENTER
where R is the reluctance and uniform current density
is assumed so that I is proportional to A. The last the main field and only swirls or waves may be left
series of relations is sufficient for graphically mapping to indicate the tendency (See the upper conductor in
any two dimensional field in the non-magnetic parts, Fig. 1). The forces exerted will be considered for
if the currents are located and the saturation known these two cases, namely, (1) when the conductor con-
approximately. tains an m. m. f. center and, (2), when it does not.
Appendix C These two possibilities are illustrated in Figs. 20A
and 20B, respectively. Figs. 21A and 21B show exactly
IMAGES6'8 the same conductors existing in exactly similar fields,
The most common two dimensional field in which but with different boundary conditions. These are
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 191

strictly two dimensional problems. Unit depth per- The force on a small area of the material carrying
pendicular to the drawing will be assumed. In Fig. the current sheet (a b c Fig. 21A) will be considered
21A two current sheets, a b c and d e f, have replaced first. This element of the conducting sheet will be
the flux lines of Fig. 20A, which were in these positions. chosen so small that it can be considered as plane sur-
The current in these sheets travels parallel to that of face. Two views of it are shown in Figs. 22A and 22B.
the conductor, but in the opposite direction. The
current densities are such that the ampere turn lines c
leaving the conductor may terminate in these sheets at /,a/' W/r /
precisely the points where they formerly crossed a / 6 "f,o
flux line. />'f'c
In Fig. 21A the lines c d and f a form boundaries for
the iron of infinite permeability and infinite resistivity
and are so arranged that the flux lines link the rectangu-
lar conductor in the same direction as before. In I 9i 9\K
Fig. 20A, these were portions of ampere turn lines. / b
In both cases the flux crosses these lines perpendic-
ularly. Thus it is seen that the boundaries of flux
tubes, which appear in these two dimensional fields, /1
may be replaced by sheets of current which travel
perpendicular to the plane of the drawing, and which / X
have such current density and sign that the ampere 41r'
turn lines arriving there may terminate in the current
sheet. Ampere turn lines, which are lines perpendicular
C /

~ d
-i c
FIG. 22A-A PART OF THE CURRENT SHEET a-b-c OF FIG. 21A

4
1 F- 1 -t
1I In Fig. 22B the ampere turn lines are shown arriving
u
from the rectangular conductor and terminating in
the sheet. When the small conductor is moved
A B (Fig. 22A) it will be supposed that there is a rectangular
FIG. 21-A. FIG. 20A WITH SPECIAL BOUNDARY CONDITIONS iron block behind it whose surfaces coincide with the
TO ISOLATE THE CONDUCTOR
B.-FIG. 20B WITH SPECIAL CONDITIONS TO ISOLATE C
THE CONDUCTOR /

to flux may be replaced by surfaces of imaginary iron

bodies that have infinite permeability (and for a-c.
fields, infinite resistivity).
Fig. 20B represents a current carrying conductor p
without an m. m. f. center. It has been arranged in /
Fig. 21B with a current sheet of like sign at a - d, and /n//
one of opposite sign at b - c. The latter is numerically '7 /
equal to the sum of the other two currents, but of o
opposite sign. The ampere turn lines d - c and a - b
of Fig. 20B are iron surfaces in Fig. 21B. Again the
magnetic circuit in the iron is so arranged that the flux
lines link the original conductor in the same manner
as before. /
The problem is to determine the forces on theR
rectangular conductors. These are, obviously, theFI.2-A THRVEOFIG2A
same in Fig. 21A as in Fig. 20A, and the same in Fig.
21B as in Fig. 20B. In Figs. 21A and 21B the forces on ampere turn lines. Work is done on the small con-
the rectangular conductor must be equal and opposite ductor, but not on the iron block, because the latter
to the integral of the forces taken over the boundary moves perpendicular to the flux and has the same
materials in each case. It will be found easier to intensities at one surface as on the opposite. The
make the computations on the boundary. change in stored energy in the whole field is
192 CALVERT Transactions A. I. E. E.

l A ~ (A The same arguments as given above for the con-

,A W 2 (41) ductor which contained an m. m. f. center (Fig. 21A)
may be applied to the conductor which had no m. m. f.
Where A i is the current in the small element and center (Fig. 21B).
A 4 the flux which passes through the iron block. It must be remembered that the force on the rec-
This flux has the same magnitude as that which occu- tangular conductors which is really sought is equal and
pied this space before the movement of the small opposite to the integral of the force taken over the
conductor and iron block.
but
A 4) = H' (A 131) (A 12') (42)
and
Aw' A F' (A 13') (43) X)°C
H' (Al12') (A i)
= -
AF' 20
where
I= energy in ergs f2
A 4) = number flux lines cut through (measured in
maxwells.)
A 11' = length of conducting sheet in the direction of
the flux lines. FIG. 23A-SIMILAR TO FIG. 21B BUT WITH ONE BOUNDARY NOT
A 12' = length of conductor in the direction of current ALONG EITHER PART OF THE ORTHOGONAL SYSTEM
flow.
A 13' = distance the conductor A 11' X A 12' iS moved imaginary bounding materials introduced in Figs.
inward perpendicular to the ampere turn 21A and 21B.
lines. It is worth noticing that any section of a magnetic
H' = field intensity near the conductor. field may be given new boundary conditions without
A i = current flowing in the element of the sheet. changing the enclosed flux field. It is not necessary
A F' = force acting on the small conductor (measured to restrict the chosen boundary to either one or the
in dynes). other part of the orthogonal system as was done above.
If a unit pole is moved around the path o p q n in For instance, suppose that Fig. 20B were changed as
Fig. 22B linking the current A i, the work done is all shown in Fig. 23A. The new current sheet d - a'
along the line q r or A 11' because in this figure H' is
zero outside the current sheet.
H' (Al1') = .4 w (Ai) (45) d ac
Substituting the value for A i in that obtained for
A F' and rearranging
AF' (H' )2 (46)
( 11') (A 12') X_
8
which is the force per sq. cm. Converting this to
the force in lb. per sq. in. and lines per sq. in.

F =-0.0139 ( ) (47)
The above force is one of repulsion tending to force the
current sheet outward. The weaker field is on the
outside (see the complete Fig. 21A). FIG. 23B-THE NEW BOUNDARY d -a' OF FIG. 23A RESOLVED
The~
foc ~
on th iro sufcsc-dadf-ams INTO THE CUSTOMARY ORTHOGONAL COMPONENTS
be ones of attraction inward and the force per unit
area is has just the right density and the iron body is brought
(H \2
F = ± 0.0139 (100) (48)
Up behind it.
It might be imagined that this new boundary is
°°° ~~~~composed of a large number of steps (Fig. 23B), the
(Because this is commonly proved and given in surfaces of which lie either parallel to the flux or to
handbooks for use in magnet calculations no proof is the ampere turn surfaces. To obtain the resultant
given here). force on a small section, it is only necessary to compute
March 1931 FORCES IN TURBINE GENERATOR STATOR WINDINGS 193

the two components of force indicated in the latter LIST OF SYMBOLS

figure. FLUX, FIELD INTENSITY, AND RELUCTANCE
Appendix E = flux in maxwells
ad) = armature leakage flux per pole
LIMITING CONDITIONS
A serious attempt was made to gain a view of the a4sg
LfN
I = field leakage flux per pole
= air gap flux per pole
forces exerted on the entire phase group by mapping r = retaining ring leakage flux
fields for certain limiting conditions. The distribution Osi = leakage flux entering the retaining ring from the
of the ampere turn surfaces on the retaining ring were under side
=
H' field intensity (lines per square cm.)
H = field intensity (lines per square inch)
\R = reluctance
CURRENTS (all in practical amperes)
Ia = r. m. s. value of armature current per conductor
at normal load
I2 = maximum possible armature current in upper
conductor in a slot on short circuit
I = current in an imaged conductor
If= field current per slot at some particular instant
following short circuit
a
\ = current in some current sheet
iso = current on the rotor surface and retaining ring
surface near the air gap

FiG. 24

approximated at the center line between poles. The

two layers of armature conductors in this radial plane ( -vlx
were plotted as one layer since only the force on the |
entire group was desired. The fields were plotted as
though they were two dimensional, because this had
given fairly good results when compared with compass
readings on stationary flux fields in this end zone. ,
It was recognized that the stator and end-bell were /
neither perfect magnetic nor perfect damper surfaces; '
but the field was plotted (Fig. 24) as though it were the /
first, and then (Fig. 25) as though it were the second. /
These both indicate that the winding at its outer edge
would have a force component toward the frame. / ( V
Aside from approximations necessary to plot these FIG. 25
fields, it was felt that the omission entirely of the
power component of the damping currents on the bell i,. = component of retaining ring current flowing to
and frame might be a fallacy which would completely oppose change in d)
nullify the value of any quantitative results. How- is = retaining ring current on the inside surfaces
ever, this method will probably have merits in parts flowing to oppose change in q$,I
of machines where the currents travel either only
axially or only tangentially, and these two figures are INDUCTANCES AND REACTANCES
submitted for this reason, and because they do give (All inductances in practical henrys per in. of length
a partial picture of conditions in the end zones. along the conductor)
194 CALVERT Transactions A. I. E. E.

L1 & L2 = self inductances of circuits 1 and 2, respec- 9. "The Graphical Determination of Two-Dimensional
tively. Magnetic Fields in Regions where La Place's Equation is Satis-
fied also in Regions where the Curl of the Field is not Zero,"
= circuits. . I and II, by Th. Lehmann, La Revue Generale de l'Electricite,
M mutual inductance of the same
Xd = subtransient reactance expressed as a deci- Vol. 14, 1923.
mal. "Sketches of Magnetic Fields in Iron," I and II, by
10.
FORCE, ENERGY, AND POWER Th. Lehmann, La Revue Generale de l'Electricite, Vol. 17, 1926.
F force in dynes per cm. of length and also per Fields 11. "Determination of the Partial and Resultant Magnetic
in Saturated Dynamos," by Th. Lehmann, La Revue
sq. cm. Generale de l'Electricite, Vol. 18, 1927.
F = force in lb. per in. of length and also per sq. in. 12. Additional Losses of Synchronous Machines, by C. M.
w = energy ergs. Laffoon and J. F. Calvert, A. I. E. E. TRANS., Vol. 46, 1927.
AXES AND DIMENSIONS 13. Forces on Magnetically Shielded Conductor, by J. H.
Morecroft and Alva Turner, A. I. E. E. JOURNAL, Jan. 1929.
x =distance measured across a slot, or across a .p 25.
conductor in the end winding in the same
direction with respect to the strands.
= distance measured radially inward in a
= Discussion
slot, or in the same direction with respect J. A. Terry: Of the three methods employed to determine
the magnitude of the forces, the second one. based upon the rate
to=distanconductoreatanydpoits,
a = distance between conductors.
of change in interlinkages, is the one most commonly used. The
reason for this is the simplicity with which the desired result can
s = a surface in air surrounding a conductor. be obtained from the inductance of a svstem, which of course
A1 & A2 = cross-sectional areas taken in current carry- must be quite accurately known in order that the operating
ing conductors. characteristics of a machine may be understood. Thus the
SW = slot width. forces on the end projections of alternator windings neglecting
the effect of the induced rotor currents may be obtained by use
tpitch.
q = slotq pltch. of the reactance formulas presented by Mr. P. L. Alger in an
1 = length as defined at the place used. Institute paper (1928) The Calculation of the Armature Reactance
A d = width of an incremental flux tube. of Synchronous Machines.
N = number of coil sides per phase The inclusion of the effect of saturation on the forces between
group.
N1 & N2 = number of turns per coil side-in coil 1 the coils is a good contribuition to the knowledge of the subject
and coil 2, respectively, since, in general, saturation materially complicates analyses, and
nearly always requires the free use of approximations which can
GENERAL only be justified after the solution is obtained by comparing the
M resulting flux densities with those which would result from the
Where primes are used
Whererimess w', H',
y', w,
as F',, y, H', L', and M',
L', ad magnetomotive force obtained in the solution. There is one
dynes, centimeters, ergs, lines per square cm. and important point which Mr. Calvert might have mentioned in
henrys per cm. of conductor are intended. connection with the comiparison between the forces with and
Where they are not used as F, y, H, then lb., in., without saturation. The comparison has shown that, on the
and lines per square in. are intended. basis of the same slot current in both cases there is a total of 1416
Where subscripts as F,, 2F,, are used, forces in the lb. per in. saturation.
including length, neglecting satturation, and 905 lb. per in. length
As a matter of fact saturation materially
y direction on coil 1 and coil 2 are meant. reduces the reactance of the machine and thus increases the
Subscript m, as in dn. indicates mutual flux. current flowing during a short circuit so that the comparison on
the basis of the same current in both cases does not present the
Bibliography true effect of saturation. Tests have shown that the commercial
1. Short Circuits of Alternating Current Generators, I and II turbine alternators saturation may decrease the reactance which
by C. M. Laffoon, A. I. E. E. TRANS., Vol. 43, 1924. limits the short-circuit currents to 0.6 of the unsaturated value.
2. Reactances of Synchronous Machines, by R. H. Park and If this new current were then substituted in the force expressions
B. L. Robertson, A. I. E. E. TRANS., Vol. 47, No. 2, 1928. the force would be (1.6)2 X 905 or 2300 lb. However, the
3. "Graphical Flux Mapping," I, II, III, IV, V and VI, saturation for such a current would be greater than for the
by J. F. Calvert and A. M. Harrison, Electric Journal, Vol. assumed current so this force is fortunately not correct. It
26, 1928. appears offhand that, depending upon the location and degree of
4. "Static Inductive Apparatus," by J. F. Peters (a mime- saturation, the forces in case of a short eircuit might be either
ographed pamphlet for the Westinghouse Design School) greater or less than in the unsaturated case.
published at East Pittsburgh, Pa. In Appendix A a more accurate formula for the line-to-neutral
5. "Repulsion Between Strap Conductors," by H. B. short-circuit current which consists of replacing xd" by
Dwight, Electric World, September 15, 1917. Xd" ± x2 ± x0
6. "Electricity and Magnetism," (Fifth Edition), by °might just as well have been given; X2 and x0
Sir Ja.mes H. Jeans, Cambridge University Press, Londonl,
England, 1925. are the negative- and zero- sequence reactances respectively.
7. Analytical Determination of Magnetic Fields-Simple This wsill give a slightly hbigher short-circuit current in most
Case of Conductors in Slots, by B. L. Robertson and I. A. Terry, commercial machines.
A. I. E. B. TRANS., Vol. 49, 1929. Since the repulsive forces between the coil sides of a slot is of
S. Graphical Determination of Magnetic Fields, by A. R. interest in the design of slot wedges it is of interest to study these
Stevenson, Jr. and R. H. Parks, A. I. E. E. TRANS., Vol. 46. for some typical cases. Consider that the currents in the upper
1927. and lower coil sides are numercially equal but out of phase by an