Professional Documents
Culture Documents
Synopsis.-There are three methods of obtaining the flux dis- and it can be used for design calculations. It is comparatively easy
tribution in a magnetic field. and it is the quickest method of the three. The graphical method of
First: By test. Templets or models can be made of the field plotting magnetic fields used in this paper is described in a compan-
to be explored and the flux distribution can be obtained by test as ion paper, "Graphical Determination of Magnetic Fields-
described in a companion paper "Graphical Determination of Theoretical Considerations" by Messrs. A. R. Stevenson, Jr. and
Magnetic Fields-Comparison of Calculations and Tests" by R. H. Park.
Messrs. E. E. Johnson and C. H. Green. This paper shows how the graphical flux plots can be used very
Second: By mathematical analysis. This method is accurate successfully in design calculations. In fact, the performance
and consistent results can be obtained. If the problem is very characteristic of a salient-pole synchronous machine can not be
complicated, however, the mathematical solution is very laborious predetermined accurately without the use of flux distribution
and sometimes impossible. coefficients.
Third: By the graphical method. This method is quite accurate * * * * *
.0~
SlotoLWiddh
Ai r-G&p ______________
FIG. 2-AIR-GAP COEFFICIENT OBTAINED BY PLOTTING FIG. 4-FLuX DISTRIBUTION AT No LOAD IN THE AIR-GAP
GRAPHICALLY THE FLUX DISTRIBUTION AROUND A TOOTH WITH OF A SALIENT POLE SYNCHRONOUS MACHINE FOR THE MAXIMUM
FINITEIWIDTH AND DEPTH AS USED IN PRESENT DAY MACHINES AND MINIMUM PERMEANCE POSITIONS. 10 PER CENT FLUX
PULSATION
Forty graphical plots were made similar to Fig. 1
with various ratios of slot width to air-gap, slot width If the number of teeth spanned by a pole varies when
to tooth width, and with a ratio of slot depth to slot the pole moves through a tooth pitch, the pole flux
width equal to four. The air-gap coefficients obtained will pulsate if the pole has no-low impedance damper
by these plots are shown in Fig. 2. These curves check winding, etc. Fig. 4 shows a 10 per cent flux pulsation
Carter's work very closely and thus there is practically from the maximum to the minimum permeance posi-
no difference between the finite and the infinite tooth tions in a machine which has a small number of wide
width and slot depth for open slots which are used stator slots and a small air-gap. This flux pulsation,
in present-day synchronous machines.
0.1 -id-
Slot W dth
v
t5 C
X
Inductor Alternator Pole Width Which Gives the Maxi- ture and the dotted flux wave was obtained by test
mum Effective Voltage. There seems to be an increasing with an exploring conductor placed- on the surface of
demand for high frequency generators. for supplying the armature. The flux wave fundamental is 1.11
power to induction furnaces, and high-speed tools, for times the maximum value of the flux wave and the flux
testing, and for experimental work. Usually if the wave third harmonic is 0.085 times the maximum value
frequency is above 2500 cycles per second, the induc- of the fundamental.
tor'type alternator is used. Fundamental and Third Harmonic in the Air-Gap
Flux Wave at No Load. Seventy-five hypothetical
IT A,,jGap flux plots at no load were made of poles used in present-
I >t > + rPoleWidt day machines whose pole faces were arcs of circles.
40 mo~ Pitch --=I TXT The pole shapes used are included in the limits of the
following three variables:
FIG. 6-INDUCTOR ALTERNATOR POLE WIDTÆ FOE MAXIMUM0P Minimum Gapfrom 1 to3
EFFECTIVE VOLTAGE AT NO LOAD
The flux waves were analyzed for their fundamentals
The voltage induced in an armature coil of an induc- and third harmonics, the values of which are shown in
tor alternator depends upon the flux pulsation, as shown Figs. 8 and 9. In Fig. 8, the fundamental A1 of the
in Fig. 5. If the pole width is 100 per cent of the pole flux wave is expressed as a decimal fraction of the maxi-
pitch, and if the air-gap is uniform, the flux wave will
be a rectangle, and, obviously, the flux pulsation and tthe
induced voltage will be zero. As the pole width is ll
decreased, Fig. 5, the flux pulsation increases. The fudmetl
effective value of the induced voltage, however, in- and valof
wA hich
creases until it reaches a maximum, and as the pole own i ta
width is further decreased, the effective voltage de- f7ise d as a fn
creases. Thus it is very desirable to shape the pole of I \
an inductor alternator so that the voltage induced inl
the armature coil is a maximum. The insert in Fig. 5 /
shows how the effective voltage varies with wI the poleis
width when the air-gap is five per cent of the pole pitch.l/ I
In this case the maximum effective voltage at no load
occurs when the pole width is 38 per cent of the pole
pitch. A number of these flux plots were made for|_
various Tios oftisa ir-gap to pole pitch and the or-l of
responding maximum effective voltages were obtained
andrplotted incFig. 6. It canxbe seen thatanpolerwidth ' J F x /
equal to half of the pole pitch should be used only when
the air-gap iS infinitely small. For a 10 per centXA 1
air-gap, pitch.
the pole~~width
~ should
~ be ~ of ~the pole
~ 0.35 ~ pitch.
~~Inube thetheshrom fte flux plots wee1Seo
Alg
to give the maximum effective voltage at no load. FIG. 7-MAGNETIC FLUX DISTRIBUTION AROUND A POLE
Under load the maximum effective voltage should AT NO LOAD WHEN A SALIENT-POLE SYNCHRONOUS MACHINE
occur when the pole pitch is a little less than given by IS EXCITED BY ITS FIELD WINDING
the curve in Fig. 6.
- IN THE AIR-GAP ~~~~~~mum I i.9 value hof the
hr fluxamncA
wave which ftefu
is taken as unity.
II. MAGNETIC FLUX DISTRIBUTION aei
OF A SALIE:NT-POLE SYNCHRONOUS MACHINE AT expressed as a decimal fraction of the fundamental
NO LOAD WHEN EXCITED ONLY BY THE FIELD COILS which is taken as unity. The polarity of the third
Fig. 7 shows the flux distribution around the pole at harmonic is also given and it is considered minus
no load when a salient-pole synchronous machine is when it is as shown in the insert of Fig 9. The pole
excited by its field coil. The full line flux wave was shape, which will have a flux wave with a zero third
calculated from the flux plot at the surface of the arma- harmonic. can also be obtained from Fig. 9. For
144 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS Transactions A. I. E. E.
example, if the ratio of the minimum gap to the pole voltage can appear at the open delta. Since the multi-
pitch is 0.02, and if the ratio of the maximum to the ples of the third harmonic flux wave are usually small,
minimum gap is 1.5, then the ratio of the pole arc to and since both the armature coil pitch and distribution
the pole pitch should be 0.67 to obtain a flux wave at further decrease the voltage produced by these flux
no load which has no third harmonic. multiple third harmonics, the multiple third harmonic
Calculation of the Open Delta Voltage and the Delta voltages can be neglected. Thus, the calculation of the
Circulating Current of a Synchronous Machine at No third harmonic voltage at the open corner of a delta-
connected armature winding is very simple, if the
amplitude of the flux third harmonic is known.
Lt
t.3 ArcS
Pole Pitc
Pole oL_ IsE = Normal phase voltage of the armature winding
at no load
1 _000
-4--L ~A3s Amplitude of the flux wave third harmonic
A t 4 2Xg t42 t expressed as a decimal fraction of its funda-
mental Fig. 9
1.0
I_kp -Armature coil pitch coefficient for the funda-
~~~~~~~~~~mental
0e _ 4 ;0 kd = Armature coil distribution coefficient for the
fundamental
i_ _ _ _ _ t _ _ _ _ _ t! B --k3p = Armature coil pitch coefficient for the third
B
019
i _ < d _ _ _ __ 2.0 _ _
_harmonic
_03+ 2.25 _ k3d -Armature coil distribution coefficient for the
0@7
+4
_____
Gapl
xe_
%t ~o'SPthird
35l;harmonic
E3 = Open delta voltage (third harmonic)
_ _ Z _ _Minimum
0.7
I I 1-1 I
0 0.01 0.02
Minirnurn Gap
Pole Pitch
0.03 0.04 0.05
then
FIG. 8-FUNDAMENTAL OF THE No-LOAD FLUX WAVE IN k3p k3d
THE AIR-GAP OF A SALIENT-POLE SYNCHRONOUS MACHINE E3 = 3EA3 k k
Maximum value of actual flux wave equals unity
Fundamental A1 = A X B This method of calculating the third harmonic delta
2Z T T fFTT I I G#4 Ivoltage assumes that there is no saturation in the mag-
netic circuit and it ignores the effect of the stator
2.4 L X-1i I CA i aand
m W the rotor slots. These factors, however, are quite
small in most machines. The three-phase machine
A
O
Ple-
r
P Pitch Nuodwoe ,0 sae is shown inn Fig.
whose pole shape .6 7 had~ 18 slots
lt per
>_t
158 - LLpole, armature coil pitch 0.777, ratio of minimum gap
to pole pitch 0.037, ratio of pole arc to pole pitch
0.674, ratio of maximum to minimum air-gap 1.22,
and phase voltage of 3810 volts.
M~~uimum Gep Thus
0.5 H Minimum Ga
IT | | kd)1$0
tkp = 0.939 = 0.955
0.4 ;Wllool . Fromt-7_ k3p = 0.49 k3d = 0.646
YI Z ~oooo- 25'From
0.3 S _L 1I Fig. 9, A3 = 1.44 X 0.475 - 0.6 = 0.084
therefore
0.2
. Lt0.102f IIIIIIIIIIIIIIT1
0.01 0. 0.04 0.05
0.03 E3=
E3 3 X 0.084 X 3810
~~~~~0.49XX0.646
0.490x80x.6460=-339 Volts
Minimumn 0.955 0.939
DOIo Pitch
FIG. 9-THIRD HARMONIC OF THE NO-LOAD FLUX WAVE IN E3 by test = 300 Volts
THE AIR-GAP OF SALIENT-POLE SYNCHRONOUS MACHINE The delta circulating current at no load is
Maximum. value of fundamental equals unity .
Third harmonic A3 A X B -0.6h= ]
Load. A very interesting application of the flux wave Third harmonic synchronous impedance
third harmonic curves, Fig. 9, is the calculation of the Calculation of Air-Gap Ampere-Turns, UJsing the
voltage which appears at no load at the open corner of a Flux Distribution Coe;fficients Kf and K,. In order to
delta-connected armature winding whose coils do not predetermine the no-load air-gap ampere-turns accu-
have a two-thirds pitch. It is well known that only rately for a pole whose pole arc radius is less than the
the third harmonic or multiples of the third harmonic radius of the armature face, two flux distribution co-
Feb. 1927 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS 145
efficients must be obtained. It is well known that if The two flux distribution coefficients K, and,Kx,
the flux wave has a flat top, more flux (lines per pole) calculated from the predetermined flux wave, made
is required to give a certain effective voltage at the it possible to predetermine very closely the performance
terminals of the armature winding than for a peaked characteristics of a two-speed salient-pole synchronous
flux wave. In order to obtain the flux per pole ac- motor5. This two-speed motor had irregular shaped
curately, therefore, it is necessary to modify the flux poles arranged in pairs whose flux waves deviated ap-
equation which assumes a sinusoidal flux distribution preciably from a sine wave at either speed.
and to introduce the flux distribution coefficient4
K, which is the ratio of the area of the actual no load
flux wave to the area of its fundamental. 1_ X D01 __ Xr
Messrs. Doherty and Shirley introduced this flux o
t ovl [ | [ | 1 | | |MlnSmin t|
|D| 84i ~~~~~~~0.7 _ 0_1 _2 _0.03
0.0 0.04 0.05;
L O ffl
I.0 |
0.58
0
0.5.620.14 1 1 1 1 i i i Minimnum Gap
0w0C 0.04
LO 0 ~~~~~~~~~~~~~~~~~~~~~~~~Pole Pitch
;I olagPMcS I * rFIG. 11-POLE FACE FLUX COEFFICIENT KA
0.95109
I n I4-
I-I
UTTTI rl
11111
rr,I<wr F
t O l
aL94l L*
Ll 1 ll 0.92 F
m.0 Armature U.M.Fl
L1
Wa,~ ~~~~~~~~~~~~~~~~~~~~lxWv
This curve is plotted on the interpolar center line and it described. These armature flux waves were analyzed
neglects the small amount of leakage flux in the lower for their fundamentals and third harmonics, the values
corner of the -pole. The ratio of the area under the of which are shown in Figs. 13 and 14. In Fig. 13,
leakage flux distribution curve to the area under the the fundamental Adl is expressed as a decimal fraction
main flux distribution curve in Fig. 7 plus 1 gives the of the maximum value of the flux wave which is taken
leakage coefficient which is 1.16 for this pole. This
coefficient should be increased slightly to allow for the I I I ___ lu
c ~~~~~~~~~~~~~~~~~~~~~~~0.20
0.1 t002
205000
IoVyphoase Short Circuit. 'hen a polyphase salient- when a salient-pole machine is excited only by a
pole generator is short-circuited, the armature m. m. f. sine wavTe armature m. m. f. whose axis is in quadrature
is in opposition to the field m. m. f. If the armature with the pole center. The effect of saturation and of
resistance is neglected, the sustained air-gap short- the stator and rotor slots is neglected. The armature
circuit flux, which travels in synchronism with the pole, flux wave is made up principally of a fundamental
can le obtained readily by subtracting the armature and a large third harmonic. The flux wave funda-
flux wave in Fig. 12 from the field flux wave, Fig. 7. This mental is 0.54 times the maximum value of the armature
short-circuit wave is shown in Fig. 15. Fig. 16 shows m. mi. f., and the flux wave third harmonic is 0.43 times
the maximum value of the fundamental.
Seventy-five hypothetical flux plots similar to Fig. 17
were made of the air-gap flux which the armature
currents tend to produce when the armature m. m. f.
is in quadrature wvith the pole center. These flux
waves were analyzed for their fundamentals and third
harmonics, the values of which are shown in Figs. 18
and 19. In Fig. 18, the armature flux fundamental
I'.1
-%\OI.TAAIiE TNDUICED IN E(i'LORIN( C'oNDUCToR
ix-
P T'rim nt-GAP OF' A SALiIS.-I-PoLE M\ACHIiNE ATC
~I~T
'!
.rN 'OLY'!I SE iOwRT ([RuC-I'r
measure of this distortional effect6f. The harmonics of can be readily obtained from Figs. 8, 9, 13, 14, 18, and
the armature flux have no distortional effect on the 19, the air-gap flux wave of a salient-pole synchronous
field flux fundamental, and since the harmonics of the machine can be approximated for any load condition.
field flux are usually small, the displacement angle can Additional practical applications of plotting magnetic
flux distribution curves will be given in a future
_.4_ _ r X .paper.
The author gratefully acknowledges the assistance of
Mr. L. P. Shildneck in plotting some of the flux dis-
tribution curves.
1.6
_I Discussion
PAPERS ON GRAPHICAL DETERMINATION OF
MAGNETIC FIELDS
0.40 _ _ __ _ X _ _ _ EE (STEVENSON AND PARK, JOHNSON AND GREEN, WIESEMAAN)
NEW YOREC, N. Y., FEBRUARY 8, 1927
1.5 ~~~~~C.H. Linder: Any simple experimental method of magnetic
field plotting is very useful in connection with a mathematieal
IB 0.1e_25
CIC _0_ i v 4 i D imun 6 1t
t .t_2.s
_ w
investigation of flux distribution. The accompanying Figs. 1 to 9
demonstrate the efficaey of the iron-filing method for determining
0.15 \ _ t > 0 003 ( 14 ! the general form of a magnetic field.
0
cl~~~G
0.1 .0 030 0.04 0.5
Pole Pitch
FIG. 18-FuNDAMENTAL OF THE AIR-GAP FLUX WAVE
WHEN A SALIENT-POLE SYNCHRONOUS MACHINE IS EXCITED
BY ONLY A SINE WAVE ARMATURE M. M. F. WHOSE AXIS IS
IN QUADRATURE WITH THE POLE CENTER
Maximum value of armature m. m. f. equals unity
Fundamental A,, = A X B
.0 00 0 .5 00 .0
t1[| |mS1MD11 1&pItFundatenuul ~~~~FIG. 1
_71111FFP 1:i_1111
.o[ppIpl,5O,ltl 1____
~~~~~~Thisis a four-pole
indaction motor with the rotor removed and the
e os| | | | | | N2}d | | | ;>vnstator windings excited with three-phase currents for a particular instant
IADC
I I 111 1 | 1311 ~~~ofbalanced operation. The iron filings are aligned to indicate the 4 poles
of the stator winding. The tuft of the filings around the circumference
71 ! | | | | | | | |- 1 t1 ~~of
I I I I I I I + I m I I I
the stator is due to the presence of the stator teeth.
FIG. 3
The strip of current carrying copper along the side of the slot in the right
hand figure represents, after a manner, the current in the field winding of a
salient pole machine. The flux indicated by the iron filings is equivalent
to the field pole leakage flux. Evidently, the flux density increases from FIG. 6
zero at the bottom of the slot to a maximum at the top of the current ele-
ment. rho center and left hand exposures have no practical application, This'circuit is identical with No. 5 except for the air-gaps inserted in the
bat merely indicate flux distributions with the current element In two horizontal arms.
positions.
FIG. 4
FIG. 5
A number of theoretical magnetic circuits have been studied, the magnetic FIG. 8
field being plotted mathematicaUy, graphicaly and experimentally. Such
a circuit of rectangular form is shown here. The coils on the perpendicular The air-gaps are situated under the excited coils. Before this 'field
legs are magnetized to send flux upward. Full potential, therefore, exists was photographed extreme difficulty was experienced in an attempt to
between the top and bottom horizontal arms. This accounts for the large construct the field graphically. The difficulty was one of magnetomotive
amount of leakage flux in the perpendicular gap between the coils. force distribution.
150 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS Transactions A. I. E. E.
C. M. Laffoon: The graphical method of mapping electric sufficient number of type cases can be solved then the solution of
and magnetic fields has been used by designing engineers for a any particular problem becomes quite easy. It is probable that
large number of years to determine the physical dimensions and twenty or thirty of these reference solutions would cover practi-
performance characteristics of electrical imachines. In most
instances, no particular effort was made to insure that the flux
distribution satisfied all of the theoretical conditions, yet very
accurate and reliable results were obtained. This was particu-
larly true in calculating the magnitude and shape of the e. m. f.
waves of rotating electrical machinery.
FIG. 9
The stator and a portion of two field poles or a synchronous machine with
a fifth harmonic cosine of magnetomotive force hmpressed along the stator
periphery. The flux distribution shown is the result. This particular
circuit was employed In certain studies made In conJunction with the prep-
aration of the paper "Synchronous Machines-Part I" by Mlessrs. Doherty
and Nickle.
It has only been during the last two or three years that special
consideration has been given to the problem of determining the
distribution of the magnetic field in the space occupied by and
adjacent to the electric conductors, with greater accuracy and
refinement, by means of both mathematical and graphical
methods of analysis. There can hardly be any question but that_______
the mathematical method of analysis is the most rigorous and
scientific, yet in most cases the graphical method is simpler FIG. 11-FLUx DISTRIBUTION IN THE CORE OF A TWO-POLE
and more convenient to use, and gives equally reliable results. TURBINE GENERATOR ON THE AssuMPTION OF UjNIFORM
It is for this reason that our own efforts have been confined to the PERMEABILITY
use of the graphical method.
In Mr. Wieseman's paper, the graphical determination of
magnetic fields has been applied to salient pole synchronous
machines for the case in which the magnetization is produced'ta.'-r
by ,
either the stator or rotor windings alone. The flux distributionua
for these cases are also given in the paper on Additional Losses
in Synchronous Machines, by Mr. Calvert and myself. It is
interesting to note that the results are essentially the same in the
two papers. We have also applied the graphical method of ''-
determining magnetic fields to turbine-generators in connection the | <
with studies on additional losses and leakage reactances. In -'44~ k
this connection, the distribution of the magnetic field in the s o p b
following parts of the magnetic circuit of a turbine generator |
have been determined:
1. Air-gap space for the following load conditions:
a. No load with rotor excited to give normal voltage, ~i**
b. No load with the stator excited to give normal voltage,
C. Full load at power factors of zero, 80, and 100 per cent.
2. End-bell space under the same load conditions as for No. 1. .
3. Rotor and stator core at no load with the rotor excited
to give normal voltage. FIG. 12-FLUX DISTRIBUTION IN THE ROTOR OF A TWO-POLE
Part of these results are given in our paper on Additional TURBINE GENERATOR
Losses in Synchronous Machines. Some of the remaining cases
will be referred to in a discussion by Mr. J. F. Calvert. cally all of the special types of problems which could be found in
J. F. Calvert: In making flux plots, the greatest assistance making the two dimensional figures for rotating apparatus.
that one can have is another drawing of a similar field. If a In electrical machines the arrangement of the magnetomotive
Feb. 1927 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS 151
forces and the iron surfaces are usually such as to render a In Figs. 10 to 15 accompanying this discussion are shown graphi-
mathematical solution either very difficult or in many cases cal solutions to some other imagnetic problems of interest. In the
wholly impossible at the present time. Therefore, it would seem figures showing the flux distribution inside of the damper bars
that a good procedure would be to project the work by graphical under steady conditions, the exact location of the kernel or center
solutions which should later be verified or corrected by mathe- is rather difficult. The location of this point makes a consider-
matical solutions whenever possible. able difference in the appearance of the picture, but usually
Somewhat recently we have been doing work along similar makes very little difference in the total amount of flux and much
less difference in the flux turn interlinkages. Mathematical
solution, however, should locate these points beyond a doubt.
FIG. 13-FL1X DISTRIBITION ACROSS A ROIUND SLOT WHEN FIG. 15-FLuX DISTRIBUTION ACROSS A RECTANGUJLAR
WHEN THE CONDUCTOR IS NOT IN CONTACT WITH THE IRON
SLOT
F 1-F C
IS INTNCONTACT WYITH THtE IRON
DISTRIBUIO
THE CONDUJCTOR
J. S. Woodward (contributed after adjournment): In
discussion of the paper by Messrs. Stevenson and Park, Mr.
J. F. Calvert presented two figures showing the flux distribution
in a slot containing a conductor carrying current. The correct-
ness of these two flux plots presented was questioned, and Messrs.
Stevenson and Park asked me to calculate the distributions
by the mathematical theory, outlined in their paper. The
result of such calculation is shown in Figs. 16 and 17 herewith.
FIG. 16
in the bottom of the slot is shown in-the accompanying Figs. 18 Hence, by taking a r/xx = 0 for a value of (y) known to lie
and 19, which refer to Figs. 16 and 17, respectively. The sharp near the kernel, the (x) coordinate of the kernel will be obtained
curvature of the lines of flux and also of the lines of no work is to a high degree of approximation. Then, taking a Rla y = 0
of particular interest as it illustrates the general statement made along this (x) coordinate will be determined to good accuracy.
in Part II of the paper as to the action of these lines in the The accuracy can be made as great as desired by repeating this
vicinity of the kernel. process.
The two cases under discussion are of interest, as they bring up Th. Lehmann (communicated after adjournment): By their
the question of the location of the kernel. In free-hand plotting, ingenious extension of Professor Rogowski's method of calculation
the location of this point is important as it furnishes a basis for to interpolar fields, Messrs. Stevenson and Park have shown that
in certain cases the Fourier method has some advantages over the
method of conformal representation, even though the latter
method can be used to solve Poisson's equation, as has already
been shown by E. B. Christoffell.
But I maintain that Messrs. Stevenson and Park have fur-
nished by their -work a very welcome check on the graphical
method of determining lines of force. . The result of this com-
parison is satisfactory, and the same is true of the very cleverly
arranged experimental check by Messrs. Johnson and Green,
which corroborates perfectly the sketches of the lines of force
plotted by Messrs. Wieseman and Shildneek.
The fact that it was possible for Mr. Wieseman to obtain by the
FIG. 17 graphical method values wvhich check within 1 per cent with
Mr. Carter's equation for air-gap reluctance of infinite teeth,
drawing in the lines of no work. The location of the kernel is shows eloquently the great accuracy which can be obtained by
determined by the boundary conditions of the copper. the graphical method if it is used methodically and with
In Fig. 16, were the slot of infinite depth and not partly closed judgment.
at the top; the flux would go straight across the copper and the
kernel would become the line at the bottom of the slot. If the
slot were open at the top and not of infinite depth, the flux would
be concave downward and there would be one kernel located on
the vertical center line of the slot.
In Fig. 16, where the top of the slot is partially closed, the flux
is concave upward, and this condition, carried to the bottom of
the slot, causes the flux leaving the sides of the slot near the
bottom to enter the iron before reaching the vertical center line FIG. 19
and as a result there are two kernels, one in each lower corner of
the slot. Further, the charts and functional curves given in these
In Fig. 17, if the slot were not partially closed at the top there papers for air-gap reluctance, tooth barmonies, reactances, etc.,
would still be but one kernel, located on the vertical center line, will be of service to practical engineers, and will save them a
but this kernel would be farther from the bottom edge of the great deal of work, for which much credit should be given to the
copper than is the case in the figure. Partially closing the slot authors.
causes the flux near the top of the slot to have a curvature that One might perhaps think that the assumption of infinite per-
is concave upward, thus tending to depress the kernel. meability in the iron would reduce somewhat the practical use
In Fig. 17, it may be observed that the curvature of the flux of the sketches of lines of force given by the authors.
lines reverses approximately at the line marked 100. From I am of the opinion that, in spite of this, these sketches are
there down, the curvature is concave downward; hence the effect still interesting and valuable, even in the case where the poles
of the insulation predominates and we have but one kernel, are saturated, for the following two reasons. First, the satura-
located as shown. It is possible to obtain two kernels in the tion of the poles does not seem to have more than a slight in-
general case of a partially closed slot, when the conductor is fluence on the interpolar flux for a given useful armature flux,
insulated from the iron, by making the insulation thinner or by even though the body of the pole absorbs as much as 30 per cent
of the total ampere-turns. Further, even if one desires to take
account of the influence of saturation, the sketches of the lines of
force for A = cc can still be used as a basis.
J l I will later show how, with the aid of a sketch of lines of force
obtained on the assumption of infinite permeability, one can
deduce rigorously the actual distribution of the field when the
circuit is saturated.
FIG. 18 The sketches of lines of force given by the authors are, there-
fore, still of actual practical interest, even when there is consider-
making the depth of the slot smaller in comparision with the able saturation in the iron, and it has seemed worth while to me
width. to emphasize this fact.
In locating the kernel from a plot made by the mathematical J. F. Hl. Douglas: In one of the three papers a preference is
formula, where the flux lines close to the kernel are very flat, expressed for graphical methods and I wish to say that with this
two considerations may be followed. First, the general location newv development of a mneans of treating the interior of copper,
of the kernel will be clearly indicDated, a,s it is known to be in the locating the flux and lines of zero work in that way, we ha.ve per-
region of lowest flux density, the density at the kernel being haps the most valuable contribution that has been made to the
zero. Secondly, the flux lines surround the kernel, a,nd the subject for many years.
rate of change of flux, along any line passing through the kernel 1. E. B. Christoffel, Annali di Mathematica, 1867, ser. It, Vol. I, fasc. 10,
wrill change from positive to negative as the kernel is passed. p. 89.t
Feb. 1927 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS 153
The test methods which I have used-namely, the use of high- vector analysis enthusiasts call them), and the quantities con-
resistance templets-are inadequate for such cases . Neverthe- cerned are dealt with directly in their magnitude and direction in
less, I cannot quite pass the matter without protesting that the space, thus bringing out more clearly the physical relationships.
difficulty attached to making tests with templets is somewhat By the definition of vector potential
exaggerated. It is a very speedy method, both for determining H = -V X R (1)
total reactance and also for determining flux densities to any According to the magnetic-eircuitation law
point of the boundary. 47i=VXH (2)
I have used all three methods and templets, and I find the
labor with templets somewhat less than inintecaeo the case of usi( Substituting H from (1) in (2), gives R3
graphical distribution, and decidedly less than that using the
functions of complex variable. Since the vector potentiat is defined through its curl only,
an additional condition may be imposed; namely, that R is a
I do not think that the possibilities of the functions of complex
variable, however, have been fully appreciated. G
notthl
Generally ththe Hence thatsolenoidal
vector (V R = 0).
writers feel it incumbent to evaluate the whole field of force.
Where problems of boundary densities are concerned, which alone V X (v X R) = V (V . R) -R (V . V) = -V2 R (4)
would be of interest in loss calculations, and where problems of the Consequently, Eq. (3) becomes
total reactance or permeance of the field is in question, then 4 7ri = V2 R (5)
there is a very neat way of handling Schwartz's and Christophel's which is identical with Eq. (30) in the paper, V2 being the
theorem, which I don't recall having seen in print. That is a Laplacian operator. If, in Eq. (2), H were expressed in
method of graphical integration around the boundary. rational units, the factor 4 X would be entirely absent from the
Schwartz's and Christophel's theorem is a very simple matter of equations thus still further simplifying the result.
formulating, provided the boundaries are straight line boundaries, Inua two-dimensiol field, according to req ()
but extremely difficult to integrate, except for, the simplest c nentwofHimnaiona diectondis
z oR must*be nta
cases. Nevertheless, on the boundary, the functions are wholly in the
real~~~~
aj. a grpia itgainseslypsbe.
inr
t perpendicular directionTis folows d ct
cu rdirection. e omsthe
This follows directly from the
definition of the curl as a line integral. Consequently, the
I ask Mr. Wieseman one or two specific questions. I should equation of a line of force is R = const.
like, either now or in the written closure, information as to At a point in the two-dimensional field, consider the direction
whether Fig. 2 checks my 1915 work2 and whether Fig. 3 checks2 n in which R varies most rapidly. We then have from Eq. (1),
my 1924 work.6 for the absolute value of H:
I should like also to make one or two detailed comments. H = a n (6)
In Mr. Wieseman's paper is the statement: "If the number of so that
teeth spanned by a pole varies, a certain amount of flux pulsa- dR Hdn (7)
tion is necessary."
There is one shape of pole which will avoid all pulsations when This result indicates that an increment of R is equal to the flux
used with armatures of any amount of slotting and every amount between two lines of force to which the two values of R refer;
of slotting; namely that pole shape which gives a perfect sine Eq. (33) in the paper.
wave on the equivalent armature circuit. Such a punching could R. W. Wieseman: Perhaps the most interesting, as well as
be used with any amount of slotting opposite, without pulsation. the most important, application of graphical flux plotting at the
The proof of this proposition was given in the appendix of the present time is the calculation of the quadrature synchronous
paper read by Douglas and Kane in Chicago two years this com- reactance of a salient-pole machine. As a matter of fact, the
ing June. value of the quadrature synchronous reactance can be pre-
A rough calculation of Mr. Wieseman's coefficient as A q 1 determined only by a field plot. This quadrature synchronous
and as A D 1 for complete range of pole pitches, air-gaps, and reactance is one of the several coefficients which appear in the
pole-face curvatures used seems to lie within the range of 40 to paper Synchronous Machines, Part II, 1926, by Messrs. Doherty
60 per cent. It seems that the problem of obtaining extremely and Nickle.
high pull-out torque (I mean pull-out torques greatly in excess In a polyphase machine, the armature currents produce a
of present ratings) is not yet possible without a radical change of sine wave of flux which travels in synchronism with the poles.
pole design. When the armature m. m. f. axis coincides with the pole axis
It may be that there are some particular combinations of pole (for example, at zero power factor), the flux which the armature
pitch, air-gap, and pole curvature which will give smaller ratios of currents tend to produce is much more than the armature flux
the transverse coefficient, but it should now be clear that in order when the armature m. m. f. axis is in quadrature with the pole.
to develop extremely sturdy synchronous motors, it is needful Thus, the direct synchronous reactance is more than quadrature
that the ratios of A ql to A dl should be brought very much less. synchronous reactance.
Vladimir Karapetoff: (communicated after adjournment) Let the normal fundamental flux per pole = 0 and let the nor-
The mathematical treatment in Appendix C may he given also in mal armature currents produce a flux ¾ in the quadrature axis.
the language of vector analysis. While the authors are justified Let the fundamental of this armature flux = Og
in using ordinary partial derivatives, so as- to make the theory Let XIq equal the armature leakage reactance (expressed as a
comprehensible to a larger circle of engineers, yet, with an ever decimal fraction) in the quadrature axis.
increasing interest in vector analysis among younger engineers q
and physicists, the alternative abbreviated treatment, added Then the quadrature synchronous reactance XQ - -- + XiQ.
below, may not, for the sake of completeness, be out of place.
Several elementary works on vector analysis are now a,va,ilable, The values of ff1 as well as many other coeffiients, are given in
so that the exposition is given without proofs or definitions. The this paper.
great advantage of this new "'short-hand" language is that no With reference to the Laffoon and Calvert discussion: I note
axes of coordinates are used (crutches or scaffolding, as some that they also prefer the graphical method of obtaining the flux
distribution and the flux distribution coefficients for design
2. Potential Gradient and Flux Density, by J. F. H. Douglas and E. W. calculations.
Kane, TRANS. A. I. E. E., 1924, P. 982.
3. The Reluct ance of Some Irregular Manet ic Fields, by J. F. H. Douglas, Mr. Douglas stated that the templet method is very speedy,
TRANS. A. I. E. E., 19i5, P. 1067. and that the 'labor involved with templets is somewrhat less
154 WIESEMAN: GRAPHICAL DETERMINATION OF MAGNETIC FIELDS Transactions A. I. E. E.
than in the case of using the graphical method. There are a of functions of a complex variable in connection with these
few simple cases of magnetic-flux distribution where the templet problems. The valuable work of Dr. Carter in this connection
method can be used to advantage, especially in illustrating flux is well known, and is especially interesting because of the recent
fringing to students. When, however, the flux distribution is to publication of another paper in which he applies the method
be obtained in the many different magnetic parts of dynamo to many new distributions which had not been attempted before.4
machinery, I prefer the graphical method, and I have found that The application of the Schwarz and Christoffel transformation is
the results obtained by this method are quite accurate. very difficult when it is necessary to integrate around more than
Mr. Douglas requested information as to whether Figs. 2 four angles, and the suggestion by Professor Douglas of a graphi-
and 3 checked his work. In Mr. Douglas' 1915 paper, the ratio cal method of integration should be of great assistance in the
of the slot width to the tooth width is very much larger than is application of the theory of functions of a complex variable
used in Fig. 2 of my paper. Consequently, no check can be to these problems.
made. Fig. 6 in the article by Messrs. Douglas and Kane, in The authors are very grateful to Dr. Lehmann for emphasizing
1924, practically checks my Fig. 3. that their sketches of magnetic lines of force are of actual practi-
Mr. Douglas stated that the flux pulsations can be avoided cal value, in spite of the fact that saturation was neglected.
with armatures having any number of teeth, if the pole gives a The article by E. B. Christoffel, which he mentions, we think will
perfect sine wave of flux on the equivalent armature circuit. be a valuable addition to .the bibliography, although his discus-
SSalient-pole machines practically never have perfect sine fluxsion came in so late that we have not had a chance yet to look
waves and, therefore, this method of eliminating flux pulsations it up. The use of conformal representation for the solution of
cannot be used. Poisson's equation has been done by St. Venat in the solution of
A. R. Stevenson: Messrs. Laffoon and Calvert are right in the torsion of rods; see, for example, Love's "Theory of Elas-
saying that the graphical method is the most convenient one for ticity." Herr M. Strutt, in the Archiv fur Elektrotechnik,
use in the everyday designing of electrical machinery, but the April 7, 1927, has applied this method to the approximate solution
mathematical method is of importance in the preparation of of the case of a current-carrying rectangular iron conductor
the 20 or 30 typical flux plots which Mr. Calvert mentioned as of constant (high) permeability.
being of great assistance in sketching similar fields. Although Mr. E. E. Johnson: While the iron-filing method gives
these typical flux plots can be determined to any degree of ac- excellent indications of the form of the magnetic field, the field
curacy by the graphical method, in some cases it takes a great plots so determined must not be interpreted too strictly as re-
deal of experimental sketching before the general outline of the gards field intensities. When making the iron-filing dis-
distribution can be determined; whereas, in such cases, the tributions, the filings, which are preferably of cast iron, are first
mathematical method will sometimes give a more accurate answer ~~distributed on some suitable plain surface, such as white paper.
with less work, in less
with less less
work,in time.
The m. m. fs. are then applied and the whole structure is gently
The best articles on graphical plotting are contained in the
tapped to allow the filings to take their proper conformations.
long series of articles by Lehmann, In the process of tapping, the iron filings in the near vicinity
of our paper. He carried the graphical .method of plotting of highly magnetized iron surfaces have a tendency to skip
much farther than we did, including in the problem the saturation along the paper and gather in clusters on those surfaces. Also,
in the iron. If anyone desires to study the graphical method even in uniform magnetic fields of high intensity the filings cluster
further, he could not do better than to refer to Lehmann's work. together in strings. This clustering leaves free open spaces,
Mr. Calvert, in his discussion, has submitted six pictures of from whih it might be inferred, on superficial examination,
flux distributions sketched by the graphical method. The two that the flux density is low at these points.
showing the flux distribution in a square bar in a square slot
did not look quite right to us. They are sufficiently accurate for Mr. Green and I are very grateful for the comments of Mr.
all practical purposes; but we do not agree with the location of Laffoon and Mr. Calvert. The experimental method which we
Woodward has submitting adihave used for determining the field distribution inside of current-
the kernels, andMr
which hershows the flu disribuio int square br carrying regions
whinh
sqe slots, asefluxdesterminedi thesemsathematicale
aperm b
a
. cases where is not
it might always
with profitconvenient although
be employed. Thethere matywas
method
used in the particular case of the alternator field poles as a check
be
descre in*ou
described our paper. which Mr. Stevenson and Mr. Park did in their
Professor Douglas' discussion is of special interest because of upon the work
the articles he has already published on this subject, especially paper.
with regard to the templet method of determining flux distribu- 4. "The Magnetic Field of the Dynamo-Electric Machine," The
tion and also because of his remarks about the use of the theory Institution of Electrical Engineers, Vol. 64, No. 359, November, 1926.