1.

Inpure pumped storage installations, the category in the future, we would expect storage plants do act chiefly as spinning
upper and lower reservoirs are usually to see a significant increase in the number reserve in thermal networks, even at stand-
small and close together. Therefore, the of reversible units adopted. still and in the pumping cycle, if provisions
cost of the equipment becomes a greater We believe, also, that as the various for rapid starting and change-over are
proportion of the installation total cost operating authorities become more familiar made. From this point of view the
than for a conventional hydroelectric with the advantages which the Deriaz economy of additional equipment for quick
installation. The equipment savings real- turbine offers in flexibility of performance, change-over should be considered.
ized by the elimination of the separate there will be an increased interest shown in There is also, however, a factor of
pumps, separate valv-es, and coupling, semiaxial machines. In this respect, the nontechnical character which affects the
together with the savings resulting from authors' revision of their list of installations economy of pumped storage: the organiza-
the decreased station height, are of very to include the Valdecafias station in Spain tion and the commercial status of the
great significance. where construction is well advanced for the producers, distributors, and consumers of
2. The higher speeds possible with modern installation of three Deriaz reversible pump power and the purchase agreements be-
design offer substantial savings to the turbines is appreciated. These units will tween them. If peak load must be bought
generating equipment. A compromise operate over a head range of 250 to 165 feet from other networks, the price per kilowatt-
design of generator for the pump turbine generating 115,000 bhp (brake horsepower) hour can be extremely high; in such cases
is required, but we have known installations under 243 feet and pumping 3,180 cubic it may be good strategy to propose a
where the optimum speeds for turbine and seconds against 238 feet net head with a pumped storage plant as a wedge in contract
pump would differ by only 4%. Also, power required of 96,600 bhp. discussions !
as mentioned by the authors, the runaway Very thorough considerations on re-
characteristic for the single runner reversible Otto J. Hartmann: Mr. Joliffe presented versible pump turbines have been presented
pump turbine is less exacting on the genera- a very interesting review of the oppor- by Mr. MacKay. We definitely agree that
tor designer than would be the runaway tunities for application of pumped storage in the future, reversible units will be used
characteristic for the corresponding turbine plants in the Northwest region of the much more frequently everywhere, es-
at the same head. United States. It seems that the Pacific pecially for pure pumped storage installa-
region is in an ideal situation as regards tions. An increased adoption of Deriaz
It is agreed that each individual installa- electric energy and that, in the near machines may be expected for the lower
tion will still need to be considered on its future, it will not be necessary to apply range of head. Apart from technical
merits and that other considerations such auxiliary power sources such as pumped considerations, pumped storage with sepa-
as the possibility of underground installa- storage in that area. rate pumps and turbines has a long tradition
tion may affect the final analysis. It is Particular considerations with regard on the European Continent, which results
anticipated that a considerably number to economics have been outlined by Mr. in a certain inertia. This psychological
of potential sites between 300 and 1,300 Dale. It was very interesting to hear that factor does undoubtedly exist both with
feet in Great Britain will eventually be the Westinghouse Electric Corporation has the users and manufacturers of such
developed. It is interesting to note also found that great savings would be possible equipment.
that of the 45 installations listed in Table if the spinning reserve could be reduced. The addition to Table I, for Valdeca-las
I, at least 33 are for heads of 1,200 feet The authors would appreciate knowing power station in Spain, as given by Mr.
or less. Assuming, therefore, that ap- more about these investigations which MacKay, is gratefully acknowledged. The
proximately the same percentage of in- may provide a way to evaluate objectively authors appreciate the contributions by
stallations will fall into the same head the grade of availability of power. Pumped the discussers.

There have been several significant

Electromechanical Energy Conversion in contributions in the use of the discipline
of the continuum to describe machine be-
DouUle Cylindrical Structures havior quantitatively. To the extent
that one may generalize about the ap-
proach used, it might be said that some6"7
use line-integral, others use energy-
ROBERT M. SAUNDERS flow,8-12 while still others use energy-
FELLOW IEEE
storage"3 methods to relate internal
boundaries and conductors to terminal
Summary: Terminal characteristics of larger study of which the machine is but characteristics. Ollendorf's6 work (as in-
machines without saliency on either mem- a part, and those concerned with the in-
ber are derived from the solution of Max- terpreted by Mishkin7) is noteworthy
well's equations in the air gap, using ternal construction or design of the ma- since it was the earliest paper to use rigor-
conductor configurations which may be chine itself. The end view of the investi- ous Maxwellian field theory to describe
present in actual machines. First, the gator influences the choice of method used
harmonic current sheets are generated. to describe the machine. Those inter-
Then energy methods are employed to ested in the behavior of the machine in a Paper 62-135, recommended by the AIEE Basic
formulate the equations of motion for both Sciences Committee and approved by the AIEE
the electric and mechanical terminals. complex system tend to think of a Technical Operations Department for presentation
at the AIEE Winter General Meeting, New York,
The formulation is general, with respect to machine in terms of the lumped-param- N. Y., January 28-February 2, 1962. Manuscript
both the machine construction and the eter discipline of circuit theory; those in- submitted October 30, 1961; made available for
number of harmonics included. printing April 5, 1962.
terested in the design of a machine more ROBERT M. SAUNDERS is with the University of
often turn to the discipline of the con- California, Berkeley, Calif.
tinuum, as formalized by Maxwell, Gauss, This work was supported, in part, by the Simon
Fund of the University of Manchester, England,
*NVESTIGATORS of electric machinery and Laplace. The general techniques fQr while the author was a Simon Fellow in the Elec-
theory seem to have formed two dis- exploiting the discrete or lumped-param- trical Engineering Laboratories. This paper is a
considerable extension of an unpublished paper
tinct groups: those concerned with know- eter approach of generalized circuit written by the author and E. C. Guilford, of the
ing enough about a machine to be able to theory established by Kron,I Gibbs,2,3 and University of Washington. The author is indebted
to S. A. Stone of Oregon State University for his
incorporate its characteristics into a Ku4i5 serve the former group. critique of the manuscript.

OCTOBER 1963 Saunders-Energy Conversion in Double Cylindrical Structures 631

the machine and, also, since it was the first conductors embedded on either the stator
paper to make use of the magnetic vector or rotor or both:
potential, which is used in this paper also. 1. The Fourier series representation is
The energy approach employed in this obtained for a current sheet which represents
paper was used by White and Woodson'3 the conductors in rotor or stator slots.
although, with the exception of the d-c The series will relate the position and
machine, they did not include harmonic number of conductors at any given position
with the current flowing into the terminals.
effects. Additionally, they postulate si-
nusoidal current-sheet distributions, a con- 2. For any one current sheet on the stator
or the rotor, Laplace's equation is solved (A) (8)
dition not often enjoyed in electric ma- in the air gap to obtain the magnetic
chines. vector potential to any place within the Fig. 2. Replacement of embedded con-
In virtually all previous treatments, air gap, as a function of the spatial co- ductors with a surface current element
either the analysis has been confined to a ordinates and time.
consideration of the steady-state funda- 3. The electric equations relating the
mental space component of current, or currents and voltages are obtained, to-
gether with the forces or torques established establish a magnetic field may be thought
superposition has been used to make in the device, using energy methods. of as the fictitious current sheets which
the problem more tractable. In the latter lie on the stator and rotor surfaces. Cur-
method, the effects produced by each rent sheets may result from distributed
space and time harmonic component of Transformation of Co-ordinates
windings, concentric windings, or con-
the current have been found; then the centrated windings. The transition from
total effect has been determined by super- Although electric machines may be ana-
lyzed in cylindrical co-ordinates (Fig. 1), conductors embedded in slots to an in-
imposing the results of each individual finitely thin current sheet on the smooth
component. This approach is valid in as White and Woodson"3 do, there is some
convenience in employing rectangular surface of the rotor or stator has been
some cases only. In this paper super- demonstrated by Hague.14 He shows
position is used, but only up to the point co-ordinates, as Ollendorf6 does. Aside
from some reduction in symbols needed that, as far as the air-gap field is concerned,
where energy storage is involved, where the slot-embedded conductors may be re-
the validity of superposition ends. While to describe the field conditions, rectan-
gular co-ordinates do not result in a Bessel placed by a current sheet consisting of a
this purity of approach adds a certain very thin continuous conductor fastened
complexity to the analysis, the author function solution when skew is present.
The co-ordinate system is not a key issue to the surface of an otherwise uniform
believes the treatment here is correct steel surface, as shown in Fig. 2. Hague
within the assumptions made. Thus, the and the results will be the same in any
event when related to physical machines. argues that the field conditions in the air
objective of this paper is to develop a gap just over the slot of Fig. 2(A) and
rigorous, unified approach for describing The transformation which will change
cylindrical to rectangular co-ordinates is those just over the current element of
the general dynamics of a double- Fig. 2(B) must be the same, since the line
cylindrical magnetic device, using field 0 = ax and r = K1 exp (y/K2) where a,
K1, and K2 are arbitrary constants. As integral represented by the paths shown
theoryas the medium of analysis. The re- will be the same. Of course, this argu-
sulting equations include all the space and will be shown later, it is convenient to let
a=2er/rr, where Tr iS the length of the ment does not take into account the field
time harmonics due to the manner in associated with the slot itself, which in
which the windings are arranged, but repeatable section of the winding. It is
also convenient to let K1 = K2= Dr/2, and this paper is being considered as a non-
none due to irregularities in the magnetic interacting energy storage realm with
surfaces such as those caused by slot to express the radius r in terms of the
air-gap co-ordinate 8. Thus the trans- respect to the air-gap fields. Thus,
openings or saliency. using Hague's reasoning, the slot-em-
formations become
bedded conductors may be replaced by a
Method of Analysis x=-G0, y=-lnt 1+ -i1 current sheet of infinitesimal thickness
2r 2 ( Dr) having the same distribution of surface
Given below is the procedure to be em- It is interesting to note that the radial current as the slot-embedded conductor
ployed in analyzing a given electro- transformation also becomes a linear one if configurations. While it may be difficult
mechanical device where currents flow in 23<«D , for in such cases an expansion of to formulate the exact equivalent current
logarithmic function as a series yields sheet, such a current sheet is a helpful
adjunct to the analytic solution of the
Drr 28 125 2s 1 23 \3 1 fields within the air-gap machine.
r=[ 2 r(
2 2Dl)
)+( 3 Dr)r
. (2)
Dr Fig. 3 shows the co-ordinate system
If 23/DT<0.05, the series may be ter- employed in this analysis. On the rotor
mipated with the first term without undue surface, a 3-dimensional co-ordinate sys-
error; this results in a linear transforma- tem separate from that on the stator is
tion y=8 for the radial co-ordinate. It erected, the rotor co-ordinates being
should be emphasized that in the de- labeled xT, yT, and zr. On the stator sur-
velopment to follow, this condition is not face a si-milar co-ordinate system is
a necessary one. erected: x', yS, and z'. The distance be-
tween these two co-ordinate systems at
any instant is x. The effective air-gap
Current Sheets length is given by g. The rotor steel be-
low the rotor surface is considered to have
In electromechanical energy converter an infinite permeability, as is the steel in
Fig. 1. Cylindrical co-ordinates theory, the basic motivation tending to the stator above the stator surface. Fig.
632 Saunders-Energy Conversion in Double Cylindrical Structures OCTOBEOR 1963
 different amplitude. Under these coi where
tions the Fourier series representing 4N
pulses is given by b, = - Ks,(-1))KduKp, (llb)
Tr
co

Thus it is seen that the current sheet con-

|J| = iYE?(a, cos2rvTr x+b sin2r'rr tains the same pitch and distribution fac-
v =1
tors as do the magnetomotive force dia-
(6) grams implied in so many treatises on elec-
tric machinery. If a fractional-slot wind-
where ing is involved, it is better to apply the
Fig. 3. Rectangular co-ordinate system 7rvWs k
more general relationships, equations 6-8,
sin to determine the harmonic coefficients.
2 T 2J7rv
a,= Tr irvWS
- Nj cos jT
-
Tr
For fractional-slot windings it is interest-
1'=0 ing to note that the fundamental term
4 shows one of the elemental current Tr
will have a lower amplitude than some of
sheets, discussed above, lying in the x-z Sr-1
the harmonics. (The author deprecates
plane on one of the stator or rotor sur- --KSU Nj cos-j (7) the use of the term "subharmonic," since
faces. Note that the elemental current Tr Sr
o=o such terms are not defined in the Fourier
sheet may be skewed by a distance at the ST-1 theorem and tend to trap the uninitiated.)
end of the conductor of Tek; the thickness 2 J' 2,rv. (8) Another special case is that of a squirrel-
of the elemental current sheet is con- bu=-Ksv
tr Njsin-j
~~~ST cage winding, where there may be as many
sidered to be negligible. The width of the different current sheets as there are bars.
elemental current sheet is w, in the x
direction. It is somewhat more convenient to rei re- In such a case, the analysis will show a
Such a conductor, skewed as it is, will sent this Fourier series as a single triigo- current sheet for the jth bar of
have a surface current density of metric function of the phase angle.
co J3=ij(1 - ±iz){k±_ z Ks, X
1 J1= w,Ni
cos
(3)
( IA= cicos (-
x2-u (9) u=l
(Tr
V, =1I
Components in x and z directions will be
Jx Ni tan X= Ni k, (4)
The resulting expression for the curr 'ent cos
2
x-Qv
T r -l)rs- l ls zl} (12)
sheet at any place on the surface tIhus
WS ws
becomes In this expression it appears that there
Ni F would be a current in the rotor whose
JZ-- (5) average value was other than zero if the
Ws J=i(lTx ;±sk Z ) CV X
total number of bars per repeatable
For a distributed winding, the slots section is odd. However, such is not the
would usually be equally spaced (although case, as an additional constraint is im-
(27ru 27rv
this is not absolutely essential), and the cos x-4u-- Trskc zj_ (1 .Oa) posed:
distribution of the amplitudes of the cur- Tr TrI n_-i
rent sheets might be as in Fig. 5. In this Ei,=O (13)
figure the current sheets indicate a re- If no skew is present, then all the curri j=o
peatable section in r,; this is where the is z-directed and the expression for Llie
Inasmuch as this constraint must be met
amplitude of the pulses shows a repeating current sheet becomes
instantaneously, there will be an addi-
pattern. In this repeatable section there
will be k pulses, each one conceivably of a tional pulsating current present to satisfy
J= iz cucos (-- ) (F Ob) this requirement. Such an additional
v =1 current will produce high-frequency
While it is always possible to determnine torques and currents which can be con-
a, and b, from the above, the case of c .lne
z
sidered not as contributing productive
tributed windings as usually found in
polyphase machines is a special one of
some interest. In windings of this t3
all the coils have the same number Of
Tpe z
h3
turns and the identical geometry. In X h2 h2
Dn.
Ws
other words, they are all identical so tI lat h0 _1 h 1
they may be treated in a special fashi
In such cases the expression for the cl -ir- - D |hk
rent sheet is
I8
I Fs. O
_
IL.
fz F
<ry>

z ']
Ayo

J i)sk2rv(
Fig. 4. Elemental current sheet on rotor or
stator surface
bL , si
' )
la) Fig. 5. Generalized elemental current sheet
distribution along rotor or stator surface
OCTOBER 1963 Saunders-Energy Conversion in Double Cylindrical Structures 633
 Y' One may substitute the defining relation- a2Ax+ 2Ax a2A a2Ax=

zV
d
-/,~~~~~~~~~~~~~~~~~~~~~~~~
a
c

b
' STATOR ship for the magnetic vector potential
in equation 16A, so that it reduces to
a2Ay a2A aa 2Ay a2A
(21a)

(21b)
aA Ox2 Oy2 az2 ata
VX E+- (16b)
- Ax - at =0
yr a2Az + a2Az +-a2Az -,e a2Az =0 (21c)
b a If the curl of a vector is zero, it is possible ,x2 0y2 rZ2 (t2
,- =-, xr 4.
L that the vector itself may be derived from
e c
."
d FROTOR some scalar: that is, As is customary in such partial differ-
I.v ential equations, a solution for the x com-
OA ponent is assumed of the form
E+- =-V 4 (17)
Fig. 6. Line-integral paths at
Ax Fx(x, z) Yx(y) Tx(t)
= (22)
Equation 17 does not uniquely define A,
since there are many vector functions such that F, is a function of x and z, Yf
torque, but rather as a possible aggrava- whose curls are the same. It is conven- is a function of y alone, and T, is a
tion of the noise problem. On the other ient to define the divergence of A as function of time alone. One set of func-
hand, if the number of bars per repeatable tions which will satisfy equation 21 is the
section is an even number, there is no V- A= -ttSE
Olt
(18) following:
pulsating current present and there is OD

maximum coupling between the funda- which will then define A uniquely. Sub- Fx = E [Kvix cos ( a,x - ,z) +
mental components of the squirrel cage c=1
stituting equations 14 and 17 into equa-
and the stator, provided that the repeat- KU2J sin (cacx-(3z)I (23a)
able section on the stator is equal to or
tion 15 and making use of equation 18
results in
greater than that for the rotor (T,8 > rrr).
co

In these cases equation 12 loses the term 892A

Yx =
E [K3
i,=1
cosh (aV2+f3v2-jufq2w2)12y+
1/rr and the number of independent (19a)
current sheets is one-half the number of K,,4Xsinh ( avl+&32-,.soi2W2)"2y] (23b)
bars. which is known as the wave equation. In 0 s
For any other irregular winding (such as the air gap no currents are flowing since Tx=i=. E 1,7x sin (r7wt+ f,) (23c)
concentric windings in fractional-horse- the current sheets are all concentrated ,q o
--

power motors) the method employed in on the surfaces or boundaries of ttit air Even in the largest known electric ma-
the earlier part of this section will gap. Thus ,iJ' is equal to zero and the chines and d are of the order of ten
a

apply, provided that the indicated Fourier wave equation to be solved is or less, pe = 1/c2 (c being the velocity of
analysis is carried out. In the end, one 02A light); thus (at±2+fl2-_IEf72C2)2)-4(a 2+±I32)
has an analytic expression for the current v 2A -,4E =O ( 19b) if 77co < <c, a condition met by all machines
at2
sheet as a function of the skew and the presently known to the author. If effect,
distance in the direction of the rotor or In an analogous fashion it can be shown then, the wave equation may be reduced
stator current sheet's motion. that to a Laplace equation since the term
C2A /Ct2 would be negligible with respect
jA0C12,>
2 p

v 2qx (20a) to the other terms. Thus, the magnetic

Vector and Scalar Potentials in the vector potential termss involve only the
Air Gap In the air gap there is no charge distribu- physical co-ordinates x, y, and z, and are
In working with magnetic field quanti- tion, so that the right-hand side of equa- independent of the supply frequency and
ties it is convenient and desirable tion 20a is zero: its time harmonics.
to make use of the magnetic vector po- If the wave equation reduces to
a242 (20'.)
tential, which is defined in the following V242-1sE--=0 Laplace's equation, it can be shown that
49t2 the magnetic vector potential must lie
fashion:
in the same direction as the current sheet
B=VXA (14) equations
The solution of the wave 19b

and 20b will yield an expression for the J. Consequently, there will be no com-
where B is the flux density and A the scalar and the magnetic vector potentials ponent of vector potential in the y
magnetic vector potential. In working in the air gap. From these quantities the direction of the machine, since all of the
with the magnetic vector potential in- energy conversion process may be com- currents are considered to lie on sheets
stead of with the flux densities and mag- pletely described mathematically. on the stator surface. This reduces the
netizing forces directly, the algebra is cut Solution of either equation 19b or 20b number of constants to be evaluated from
approximately in half, because the mag- would provide a unique solution for the nine to six, since equation 23b is now zero.
netic vector potential automatically in- magnetic vector potential because of the To evaluate the constants it is neces-
-cludes all the components of the flux interlocking relationships in the air gap. sary to examine the boundary conditions
density or the magnetizing force. It is somewhat more straightforward and in the machine. This is done by apply-
To solve the field problem in the air gap, the physical significance is retained, if ing Ampere's line-integral law at the
two of Maxwell's equations are employed: one works with the magnetic vector boundaries. To evaluate Ampere's law

VXH=J'+-
at
VxH=J'+~- EE
cE (15)
(15)
potential;

tion, the
thus

Because equation 19b is

three
solves

scalar
a equation l9b.
vector equa-
equations which
it is necessary to know the magnetizing
force H at the stator and the rotor sur-
faces. This can be computed by using
OBE= (16) need to be solved will then result as equation 14 and the magnetic vector po-
tential described by equations 23a and
634 Saunders-Energy Conversion in Double Cylindrical Structures OCTOBER 1 963
23c. The Ampere law evaluation in the Table 1. Independent Variables and External Forces
x-y plane is shown in Fig. 6,
Electrical
J'*ds= f H*dl=Hab Ax+HbcAy+
Mechanical Stator Rotor
HcdAX+HdaAY = JzAx
but Hb,= Hda and H5d=0; hence Hx= Displacement, q . ...............
. Q2S,... Q3S, 8 r Q,r Q3r....... Qnr
Velocity, q. .............xo . .. i . i2S. .......im3 ...... ilr, . i2 . i3r
3 . in
Jz. Since J. is known in terms of the External forces, F .......Fm. F ... mVC V3s
...3c. vmcs . vlr,.. v2cr,. . Vs3Cr.s.vncr
conductor configuration, two constants of
eqution 23c will be determined. Eval-
uating the same line integral at the rotor
surface will result in a third constant in
equation 23c. Applying this technique the energy storage in the air gap or the seven independent equations of motion.
in the y-z plane will evaluate the remain- energy flow in or out of one of the surfaces In the analysis of squirrel-cage machines,
ing constants. on the air gap. In this section the method if there are m stator windings and 2n
Finally, the expression for the magnetic of finding the equations of motion of the rotor bars (where n is an integer), then
vector potential due to the jth current machine or device using energy-storage there would be (n+m+ l) degrees of
sheet on the stator is given by methods is discussed. freedom and as many equations of mo-
Using energy rather than energy-flow tion. In working with the independent
methods, the equation of motion may be and dependent variables it is often desira-
Aj8=ij{1x +1 ' cos (a,8X, found through the use of the Lagrangian ble to establish a table showing the varia-
formulation. (A more complete treat- bles which may be present. This is done
ment is contained in reference 9, chap. 1.) in Table 1. For each independent cur-
O3 3vszs)(Kiv is cosh y Sys+ According to Lagrange, the equations of rent there will be a current sheet. Thus
motion for a rigid body are determined there will be m+n current sheets and m+
K2Uj5 sinh yv'ys)J (24) from the relationship n magnetic vector potentials in the mra-
chine. To find the total stored energy
d( W --2=
aq,
dt (agr
Fr (26)
(6 which would enable one to form the par-
and for the kth current sheet on the rotor tial derivatives is almost impossible, ex-
where qr is the displacement variable, (r cept in trivial cases. However, in equa-
is a velocity variable, £ is the Lagrangian tion 28 it is noted that there are two par-
Ak' = ik [lx +3Z)KlT (h
cos xr- function describing the co-energy stored tial derivatives. For electric variables,
,

within the system, and Fr is the exter- the energy stored in the electric field is so
Ovk r _ ,8vrrZ)( K,kr cosh 'Yuryr + nal force applied to the system. All of small it may be neglected., Hence, only
this is shown in Fig. 7, in which the the term Wgl/7aq must be considered.
boundary of this system is assumed For the mechanical equation of motion,
K2uJ. sinh -Y,Yj) (25) to be acted upon by n generalized only the term aWg/lqr must be con-
forces. These forces may be mechanical sidered, for there are no velocity-depend-
where or electric. If there are n degrees of ent terms in the energy relationship Wg.
freedom, then there must be n sets of Studying first the electric equations
independent variables and n independent and carrying out the partial differentia-
Klujs = K2,js coth -yvsg K2Uj =
UY
equations of motion. The Lagrangian tion indicated in equation 28 yields

Ki,k= -K2,,k coth -yjtg K2ukT= =

'Yr
r
function must be expressed in terms of
generalized displacements and velocities
which may be derived from the Hamilton-
ian function, which is expressed in
=_ a3ir
.I .+
W5 OW5 =-1I r a2AAz aAz
/lo JeL OisTO ay
Equations of Motion-Energy- ( 2Ax aAz a\(OAx- AAzV
Storage Method terms of generalized displacements,
velocities, and momenta. Since the air O1irtZ OsirOx az Ox
Having once obtained the description gap is a linear medium, the magnetic a2Ax .3Ax] (2a)
for the magnetic vector potential in the air energy stored therein is equal to the co- airay CY
gap at any position with respect to x, y, or energy stored in the air gap. This is to
z, it will be necessary to compute either say that the Hamiltonian and Lagrangian Now
functions are equal (3C = £). The Ham- m n
iltonian is given by Az= E
j=I
A s8+ EA
k=l
,r (30)
C= 3+V =f5(fH.dB)dv= W, (27)
Thus, for electromagnetic devices, the The rotor magnetic vector potential needs
equations of motion given by equation 26 to be referred to the stator. This may be
will become done by means of a Galilean transfonna-
tion (a special case of the Lorentz) when
ad, )-aWg
d (aWg Fr (28) the velocity is small compared with the
dt a0q velocity of light. The transformations
In the case of a wound-rotor induction are summarized in Table II.
"I
"I
"I motor there would be three stator and Substituting equation 30 in equation 29
three rotor currents. These, together and carrying out the indicated integra-
Fig. 7. Conservative system acted upon by with the one mechanical variable, would tions yields the following electric equation
external forces result in seven degrees of freedom and of motion for the ith circuit:
OCTOBE-R1963 Saunders-Energy Conversion in Double Cylindrical Structures 635
 conventional methods may be used to Table 11. Summary of Transformations
a wg sl v=l
s 4,,esvgctt
S.
j ' Cvj CVul develop equivalent circuits.
What now remains is to determine the Lorentz Galilean

coth (-y,/g) cos mechanical torque produced. This can

be computed in the same manner as xr .....
(XS -it) Xs xo,o
n
-j cvi3CvrKskviAr was done for the voltages, except that
2[(-yv )2_(eVI'ACos is
vi
the formulation is not quite so difficult, .
z-r£ (.....
.*
.......
o/
1 fxo 8\ ....
zs ....../. .. Zs
the derivatives being taken with respect
l
k=l ir ........

to the difference in the origin on the stator +/1(t 2/C2) C2 . f

acxr ) [(i+rj )X and the rotor x,:
i Yv 'Y U
'Yu Yv d Wv Wg=FM
5sinh y7,g5sinh y0Tg 2 dt &XO XOx

-ij'c,,jsc8c csch a2g X

sinmh g- sinjh y,g)] (31) The resulting force relationship is
j==1
FM= O_
--x0
dxo
a 3-rsI X COS (¢0Ui"'-0VjS+CeVXu)+
where n

dWy/cij'=stator flux linkages

co/

(y.,V)2(_Y,r)2K,,.,
I
Z
Cvk r Cv2r coth a2g X
/ {
2ax 3( S+ r)( 7 S _- r)
y
X k =l

T[i
STS) 231/2
+ (

'Yv {U 1+ cos (Oir-cPrj) (32a)

r [ --(Tsk 2 1/2
i =* C;1
Lik= u

sin
au( rsk -rskr)
(,,_r,l- Yv rLv2)sin (4,v,S,
Fm= i i. S -ik. r2 PX
_ ~~2 V=C j~=j L k=i
Ksk.=
av(rsk8 -Tsk)
2 tvkrluXo) ikl' .XQr) X CvjfCvk esch avg sin (0,js-
Similarly, for rotor flux linkages, 4)h -avXo) (33a)
co 2'31 ~ k=

ra ioTr i
l
Conclusions
}S)X J7C(irc8,sKsk rs) r sin (4uk +4aujCe,Xo)1t) (33)
|TU)'_ar2
-yv
Cos (0, i - As a result of this work, a rigorous
solution for the forces and for the coupling
coefficientsrelatingterminal characteristics
To determine the torque developed by of electromechanical energy converters
( yU, ^/~~r)eSev
the force, one must multiply by the has been formulated in terms of the
appropriate co-ordinate transform, which internal conductor configurations, the
(5inllhj2,rsilh<v)+ 2Xs is the radius, if a linear transformation magnetic boundaries, and the fields
(Aj (axo
GXCv t
from the cylindrical to the rectilinear produced. This paper should provide the
model obtains. foundation for future papers on the ap-
( t ) s If neither rotor nor- stator is skewed plication of these general results to specific
expressions for flux linkages and force are machines meeting the conditions estab-
lished herein.
0 F m
i is 2TS,rriol ijCvj8C i8X
Nomenclature
coth ae g cos (4,v i - o vi")- a= Fourier coefficient, meters-'
A = magnetic vector potential, webers/
These stator flux linkage expressions will n meter
result in voltages due to coupling be- Z
iCrcu2rcus csch a2g X B=Fourier coefficient, meters-1
tween one coil on the stator and other coils B= flux density, webers/meter2
k=1 c=velocity of light, meters/second
on the stator, and between that coil and
c = Fourier coefficient, meters-1
coils on the rotor. Similarly, the rotor Y=rotor diameter, meters
flux linkage expressions will result in volt- Cos ('Mi8-,uk,-2 avXo) (31a) E=electric field intensity, volts/meter
ages due to coupling between rotor and g= air-gap length, meters
stator coils, and between one rotor coil h=current density per slot, ampere turns/
and other coils on the rotor. Since these O2 meter
are equations for flux linkages, removing
9Wg a 8Trs I.ol H=magnetizing force, amperes/meter
aix 2av
x
C= total energy stored in conservative
the currents will yield inductances. Thus, v=l systenm, newton-meters
636 Saunders-Energy Conversion in Double Cylindrical Structures OCTOBER 1963
i=-4listantaiteous current, amperes _0 MACHINES, Y. H. Ku. Proceedings of the Institu- theorem of electromagnetic field theory,1
tion of Electrical Engineers, London, England,
J: linear current distribution amperes/ vol. 99, pt. 4, no. 4, Dec. 1952, pp. 410-28. the air-gap fields due to conductors in
meter slots caii be produced by equivalent current
6. EINHEITLICHE THEORIE DER DREHFELD-
J= surface current distribution, amperes/ MASCHINEN AN HAND EINES MODELLES, F. Ollen- sheets on the stator surface. Presumably
meter2 dorf. Archiv fuer Elektrotechnik, Berlin, Germany, the current-sheet distribution will be some
Kd = winding distribution factor vol. 24, pt. 2, Aug. 23, 1930, pp. 129-50; Dec.
5, 1930, pp. 612-34; Dec. 30, 1930, pp. 715-30.
function of the width of the slot and length
Kp = winding pitch factor of the air gap.
Ks slot constant 7. THEORY OF THE SQUIRREL-CAGE INDUCTION
MACNINE DERIVED DIRECTLY FROM MAXWELL'S
K1, K2, K3, K4 = air-gap coefficients, webers/ FIELD EQUATIONS, E. Mishkin. Quarterly Journal REFERENCE
ampere-meter of Mechanics and Applied Mathematics, London, 1. ELECTROMAGNETIC WAVES AND RADIATING
I= length of air gap in the z direction, meters England, vol. 7, pt. 4, 1954, pp. 472-87. SYSTEMS (book), E. C. Jordan. Prentice-Hall
S=Lagrangian function, newton-meters 8. ENERGY FLOW IN ELECTRIC SYSTEMS-THE Inc., Englewood Cliffs, N. J., 1950.
mn=number of stator phases or independent VI ENERGY FLOW POSTULATE, J. Slepian. AIEE
currents Transactions, vol. 61, Dec. 1942, pp. 835-41.
n=number of rotor phases or independent 9. SOME REMARKS ON TEE ENERGY FLOW IN W. Shepherd (University of Manitoba,
currents ROTATING ELECTRIC MACHINES, F. Dahlgren. Winnipeg, Man., Canada): There are
Transactions of the Royal Institute of Technology,
N=number of conductors per slot asso- Stockholm, Sweden, no. 38, 1950. now several distinct approaches used in
ciated with one independent current 10. FLOW OF ENERGY IN D- C MACHINES, E. I.
teaching the theory of rotating electric
Q = charge, coulombs Hawthorne. AIEE Transactions, pt. I (Com- machines. It is possible for a student to
r = radial displacement, .radians munication and Electronics), vol. 72, Sept. 1953, take down from a library shelf half a dozen,
Sp= slots spanned by a single coil (coil pp. 438-45. different books called (approximately);
pitch) 11. FLOW OF ENERGY IN SYNCHRONOUS MA- "Electric Machines." and after an hour
S=total number of slots CHINES, E. I. Hawthorne. Ibid., vol. 73, Mar. of perusal come to the conclusion that
Sr =slots per repeatable section 1954, pp. 1-10. these books have little or nothing in common
t= time, seconds 12. TORQUE-ENERGY RELATIONS IN INDUCTION except their titles. Although such a situa-
3=kinetic energy, newton-meters MACHINES, P. L. Alger, W. R. Oney. Ibid., pt. tion is fraught with danger to the unini-
111-A (Power Apparatus and Systems), vol. 73,
v=volume, meters3 Apr. 1954, pp. 259-64. tiated, it is inevitable and, on the whole,
v = potential difference, volts 13. ELECTROMECHIANICAL ENERGY CONVERSION welcome.
'U= potential energy, newton-meters (book), D. C. White, H. H. Woodson. John The classical method1I with its massive
ws=width of slot opening at air gap, meters Wiley and Sons, Inc., New York, N. Y., 1959. detail, its emphasis on the steady state, and
Wg= stored energy in magnetic field, 14. ELECTROMAGNETIC PROBLEMS IN ELECTRICAL the compartmentalization of ideas (and
newton-meters ENGINEERING (book), B. Hague. Oxford Uni- machines) has gradually given way to
x distance measured in the direction of versity Press, New York, N. Y., 1929, pp. 183-85, various forms of unified machine theory
297.
motion, meters These older texts still have great value for
xo displacement of origin of rotor the more specialized student but are out
reference frame with respect to the of favor as introductory treatments.
stntor reference frame, meters As the author points out, the proponents
V =curl operator, meters-1 of unified machine theory may be roughly
a coefficient = 21rv/rr, meters-1 divided into two camps: those favoring
6 radial air-gap co-ordinate, meters the distributed-constant (continuum) ap-
e = exponential base
e-permittivity, farads/meter Discussion proach and those preferring the lumped-
constant (system) approach. Table III
= coefficient = (2-7ruV/r)(rsk/l), meters' indicates the relative positions of a repre-
y = coefficient = (a2 +,32)1/2 = a [1 + Richard T. Smith (TRACOR, Incorporated, sentative set of current English language
(Tsk/l)2J 1/2, meters-' Austin, Tex.): Prof. Saunders has made a texts and articles on machine theory.
= order of harmonic (time) valuable contribution in extending the Many of the texts cited cover two or more
0= angular displacement (space), radians dynamic analysis of the double cylindrical of the classifications, which should not be
Au-permeability, webers/ampere-meter structure to include time and space har- regarded as rigid or exclusive.
X =skew angle, radians monics. It is hoped that the author may The lumped-parameter methods usually
u=order of harmonic (space) be encouraged to extend his analy sis to involve the resolution of the machine into
o-=number of repeatable sections in machine salient pole machines. I have used an lumped windings which are then treated
Tp span of a single coil (pitch), meters analysis somewhat similar to the steps by topographical methods,1'6 by the use
Tr =length of a repeatable section of the given in the section "Method of Analysis," of transformations of variables,7' 4 or by
Fourier period, meters but not including harmonics, to calculate the application of basic physical and
=s distance between center lines of adja- certain machine self- and mutual reactances. mathematical laws to idealized physical
cent slots, meters My approach was (1) to obtain current- models.15'20 Generalized machine theory
TS, distance of slot skew over length of sheet representations of the actual currents; using the 2-reaction approach is now widely
core, meters (2) to solve for the field scalar potential, taught in Britain where some schools also
0= phase angle (space), radians satisfying postulated boundary conditions; introduce tensor analysis towards the end
4 =phase angle (time), radians and (3) to use the actual winding configura- of the undergraduate course, usually to
1 = unit vector tions and the field of step (2) to compute power option students. In the United
flux linkage, from which self- and mutual States and Canada there is great diversity
reactances follow directly. in the approach and emphasis between
References different universities; reference 16 repre-
sents the most popular approach.
Cameron W. Blachford (University of The various forms of the continuum
1. GENERALIZED THEORY OF ELECTRICAL MA-
CHINERY, G. Kron. AIEE Transactions, vol. 49, Illinois, Urbana, Ill.): The author suggests approach, notably the field approach21-23
Apr. 1930, pp. 666-83. replacing the current-carrying conductors and the Lagrangian approach,24.25 are
2. THE MODERN APPROACH TO ELECTRICAL in slots by infinitely thin current sheets still unpopular as teaching methods. It
MACHINE ANALYSIS, W. J. Gibbs. Engineer, located on the stator surface, having a can be argued that both of these methods
London, England, vol. 192, 1951, pp. 467-9, width equal to the width of the slot. This tend to cloak the physical reality of the
485-8, 517-20, 546-8, 578-80. is certainly a helpful technique to use machine in sophisticated mathematical
3. TENSORS IN ELECTRICAL MACHINE THEORY when determining the air-gap fields. equations. Moreover, at the stage of
(book), W. J. Gibbs. Chapman and Hall, Ltd., I would like to know how accurate this specialization of undergraduate electrical
London, England, 1952.
approach might be when applied to a engineering courses, machines and field
4. TRANSIENT ANALYSIS OF ROTATING MACHINES
AND STATIONARY NETWORKS H1Y MEANS OF ROTAT-
machine. Hague's work, referred to in theory are taught to different "options"
ING REFERENCE FRAMES, Y. H. Ku. AIEE reference 14 of the paper, is entirely theo- and, regrettably, the mathematics involved
Transactions, vol. 70, 1951, pp. 943-57. retical and says that only the fields imme- in continuum theory is often outside the
5. ROTATING FIELD THEORY AND GENERAL diately over the slot and current sheet will scope of many engineering courses. This
ANALYSIS OF SYNCHRONOUS AND INDUCTION be the same. According to the equivalence latter criticism can also be applied to the
OCTOBER 1963 Saunders-Energy Conversion in Double Cylindrical Structures 637
 Table 111. Methods of Rotating Machine 10. TRANSIENT ANALYSIS OF ALTERNATING CUR-
RENT MACHINERY (book), W. V. Lyon. John
literature on the subject is contained in
Analysis Wiley and Sons, Inc., 1954. a dissertation by Guilford.2
11. SYMMETRICAL COMPONENTS (book), C. F.
Prof. Blachford questions the validity
Continuum Methods Wagner, R. D. Evans. McGraw-Hill Book Com- of the assumption that the slot-embedded
(Distributed System Methods pany, Inc., 1933. conductors may be replaced by surface
Constants) (Lumped Constants) 12. SYNCeRONOUS MACHINES (book), C. Con- currents. While it is true that the field
cordia. Johni Wiley and Sons, Inc., 1951. conditions immediately over the slot will
Line-Integral2l -21 Topographical5'6 13. THE GENERAL THEORY OF ELECTRICAL be different from those immediately over
Energy-Flow27-30 Transformations of Variables MACHINES (book), B. Adkins. Chapman and Hall, the surface conductor, all indications are
Energy-Storage24,25 Tensors7 -9 Ltd., London, England, 1959. that this is a secondary effect in terms of the
Symmetrical Com- 14. ELECTRIC ENERGY CONVERSION (book), total energy storage in the air gap of the
ponents"11
Two-Reaction Theory'2l13
Y. H. Ku. Ronald Press Company, 1959. machine. In validating the above as-
*PO Components14 15. ELECTRO-MAGNBTIC MACHINES (book), R. sumption, one strong indication is obtained
Idealized Physical Models -20 Langlois-Berthelot. Macdonald and Company, by comparing the calculated mutual in-
Classical1I4 Ltd., London, England, 1953. ductance with test results; preliminary
16. ELECTRIC MACHINERY (book), A. E. Fitz- results on six machines indicate excellent
gerald, C. Kingsley. McGraw-Hill Book Company,
Inc., second edition, 1961. correlation, providing the Carter coeffi-
cient is used to obtain the effective air gap.
17. THE NATURE OF POLYPHASE INDUCTION
tensor approach which is probably the most MACNINES (book), P. L. Alger. John Wiley and Electrolytic tank studies are currently in
fundamental and comprehensive method Sons, Inc., 1951. progress to ascertain over what ranges of
used for analyzing electric circuits and 18. ELECTEOMECHANICS (book), H. H. Skilling. slot opening to slot pitch (w/'r) and air gap
machines. It seems likely that continuum John Wiley and Sons, Inc., 1962. to slot pitch (gQ/) this assumption is valid.
methods will continue to be largely re- 19. INTRODUCTION TO THE DYNAMICS OF AUTO- The equivalence theorem seems to be
stricted to the postgraduate level, since the MATIC REGULATING OF ELECTRICAL MACHINES another manifestation of the substitution
systems approach is now well established. (book), M. V. Meerov. Thornton Butterworth, principle used so often in engineering
The equations of a particular device or Ltd., London, England, 1961. problems to make them tractable. The
system derived by different methods are 20. ELECTRIC MACHINERY (book), C. C. Carr. Hague method, Schelkunoff induction and
John Wiley and Sons, Inc., 1958. equivalence theorems, and the method of
equivalent, 9 25.26 although the equivalence
is often obscure. Dr. Saunders' paper is 21. See reference 6 of the paper. images are specific examples of the sub-
a particularly good example of how two 22. See reference 7 of the paper. stitution method, wherein certain field
different techniques, in this case field 23. A SIMPLIFIED ELECTROMAGNETIC THEORY OF conditions are matched by fictitious sources
theory and the Lagrangian method, can be THE INDUCTION MOTOR USING THE CONCEPT OF in order to make feasible the solution of a
combined to advantage. The study of WAVE IMPEDANCE, A. L. Cullen,, T. H. Barton. field problem.
Proceedings of the Istitution of Electrical Engineers, Prof. Shepherd has contributed an
machines would greatly benefit from more London, England, pt. C, Sept. 1958, pp. 331-36.
of this type of work. extremely valuable extension to the paper;
The paper is also valuable in that it 24. See reference 13 of the paper. in fact he has almost written a second
makes provision for computing the results 25. ELECTROMECHANICAL ENERGY CONVERSION paper. I am certainly indebted to him,
of magnetomotive force harmonics, which (book), S. Seely. McGraw-Hill Book Company, for he adds a great deal to the paper. In
Inc., 1962.
are often ignored in the various lumped- general, I agree with Prof. Shepherd's
26. CURRENT EQUATIONS OF THE CAPACITOR Table III but I also would argue that it is
parameter methods. A considerable dis- MOTOR BY THE MATRIX AND LAGRANGIAN
advantage of the present method is that it METHODS, R. P. Comsa. Transactions of the not laid out the way I would do it. My
cannot be applied to salient pole structures. Engineering Institute of Canada, Montreal, Canada, contention is that the continuum methods
The author's views onl the possible extension vol. 2, no. 2, 1958, pp. 70-4. will develop only the differential equations
of the method to nonuniform air-gap ma- 27. See reference 8 of the paper. describing the distributed nature of the
chines would be appreciated. 28. See reference 9 of the paper. fields and conductors in the machine.
29. See reference 10 of the paper. Once these equations and motion have
been determined, then those methods of
REFERENCES 30. See reference 12 of the paper. rotating machine analysis which Prof.
1. ALTERNATING CURRENT MACHINES (book), Shepherd calls "system methods" will
A. F. Puchstein, T. C. Lloyd, A. G. Conrad. John become applicable. Thus, the methods
Wiley and Sons, Inc., New York, N. Y., third Robert M. Saunders. Turning first to the described on the- right side of his table
edition, 1954. question raised by Dr. Smith regarding the
2. PRINCIPLES OF ALTERNATING CURRENT MA- extension of this method to salient pole really represent the second and third
CHINERY (book), R. R. Lawrence, H. E. Richards. machines, it can be said that much of the parts of a process in which the continuum
McGraw-Hill Book Company, Inc., New York, reasoning herein is applicable to many method, deriving as it does the equations
N. Y., fourth edition, 1953.
other forms of electromechanical energy in motions, is the first part. It is my
3. THEORY OF ALTERNATING CURRENT MA- conversion devices. The most difficult contention that one without the other is
CHINERY (book), A. S. Langsdorf. McGraw-Hill
aspect in extending the method to devices sterile and has led us to many erroneous
Book Company, Inc., second edition, 1955. concepts of machine analysis, particularly
other than those with double cylindrical
4. THE PERFORMANCE AND DESIGN OF ALTERNAT-
structures lies in the boundary value under transient conditions. Thus, there
ING CURRENT MACHINES (book), M. G. Say. is no basic conflict between these methods;
Sir Isaac Pitman and Sons Ltd., London, England, problem posed by irregular surfaces. Sali- rather there is an interplay between them,
second edition, 1955. ent pole machines fall into this category;
5. ELECTROMECHANICAL SYSTEM THEORY (book), slot openings are another aspect of this and they are mutually dependent, one
H. E. Koenig, W. A. Blackwell. McGraw-Hill same problem. However, once the field upon the other.
Book Company, Inc., 1961. distributions have been determined, the
6. EQUIVALENT CIRCUITS FOR MACHINES AND methods in this paper are applicable.
TRANSFORMERS INCLUDING NONLINEAR EFFECTS, REFERENCES
G. R. Stemon. Proceedings of the Institution of Dr. Smith points out that the author is 1. END COMPONENT OP ARMATURE LEAKAGE
Electrical Engineers, London, England, pt. C, not unique in the application of similar REACTANCE OF ROTOR-GENERATORS, R. T. Smith.
monograph no. 68, 1953, pp. 129-43. methods to the analysis of machines, but AIEE Transactions, pt. III (Power Apparatus
7. See reference 1 of the paper. that the extension to space harmonics is and Systems), vol. 77, Aug. 1958, pp. 636-47.
8. See reference 3 of the paper. unique. My bibliography is not nearly 2. A FIELD THEORY ANALYSIS OF ELECTRO-
9. TENSOR ANALYSIS OF ELECTRIC CIRCUITS as complete as that contained in Dr. MECHANICAL ENERGY CONVERSION IN THE INDUC-
TION MACHINE WITH SLEEVE ROTOR, E. C. Guilford.
AND MACHINES (book), L. V. Bewley. Ronald Smith's paper,' which contains an extensive Doctoral Dissertation, University of California,
Press Company, New York, N. Y., 1961. bibliography on the subject; additional Berkeley, Calif., 1959.

638 Saunders-Energy Conversion in Double Cylindrical Structures OOCTOBIER 1963