Symmetrical Co-Ordinates Networks: Trigonometrical Called Only (33) Portion Dealing Unsymmetrical

Presented at the 34th A nnual Convention of
the American Institute of Electrical Engineers,

Atlantic City, N. J., June 28, 1918.
Copyright 1918. By A. I. E. E.
METHOD OF SYMMETRICAL CO-ORDINATES APPLIED

TO THE SOLUTION OF POLYPHASE NETWORKS
BY C. L. FORTESCUE
ABSTRACT OF PAPER
In the introduction a general discussion of unsymmetrical
systems of co-planar vectors leads to the conclusion that they
may be represented by symmetrical systems of the same number
of vectors, the number of symmetrical systems required to define
the given system being equal to its degrees of freedom. A few
trigonometrical theorems which are to be used in the paper are
called to mind. The paper is subdivided into three parts, an
abstract of which follows. It is recommended that only that
part of Part I up to formula (33) and the portion dealing with
star-delta transformations be read before proceeding with Part II.
Part I deals with the resolution of unsymmetrical groups of
numbers into symmetrical groups. These numbers may repre-
sent rotating vectors of systems of operators. A new operator
termed the sequence operator is introduced which simplifies the
manipulation. Formulas are derived for three-phase circuits.
Star-delta transformations for symmetrical co-ordinates are given
and expressions for power deduced. A short discussion of har-
monics in three-phase systems is given.
Part II deals with the practical application of this method to
symmetrical rotating machines operating on unsymmetrical
circuits. General formulas are derived and such special cases,
as the single-phase induction motor, synchronous motor-genera-
tor, phase converters of various types, are discussed.
INTRODUCTION
IN THE latter part of 1913 the writer had occasion to investi-
gate mathematically the operation of induction motors under
unbalanced conditions. The work was first carried out, having
particularly in mind the determination of the operating char-
acteristics of phase converters which may be considered as a
particular case of unbalanced motor operation, but the scope
of the subject broadened out very quickly and the writer under-
took this paper in the belief that the subject would be of interest
to many.
The most striking thing about the results obtained was their
symmetry; the solution always reduced to the sum of two or
more symmetrical solutions. The writer was then led to in-
quire if there were no general principles by which the solution
of unbalanced polyphase systems could be reduced to the solu-
1027
1028 FORTESCUE: SYMMETRICAL CO-ORDINATES [June28
tion of two or more balanced cases. The present paper is an

endeavor to present a general method of solving polyphase
network which has peculiar advantages when applied to the
type of polyphase networks which include rotating machines.
In physical investigations success depends often on a happy
choice of co-ordinates. An electrical network being a dynamic
system should also be aided by the selection of a suitable system
of co-ordinates. The co-ordinates of a system are quantities
which when given, completely define the system. Thus a system
of three co-planar concurrent vectors are defined when their
magnitude and their angular position with respect to some fixed
direction are given. Such a system may be said to have six
degrees of freedom, for each vector may vary in magnitude and
phase position without regard to the others. If, however, we
impose the condition that the vector sum of these vectors shall
be zero, we find that with the direction of one vector given,
the other two vectors are completely defined when their magni-
tude alone is given, the system has therefore lost two degrees
of freedom by imposing the above condition which in dynamical
theory is termed a "constraint". If we impose a further con-
dition that the vectors be symmetrically disposed about their
common origin this system will now have but two degrees of
freedom.
It is evident from the above definition that a system of n
coplanar concurrent vectors may have 2 n degrees of freedom and
that a system of n symmetrically spaced vectors of equal mag-
nitude has but two degrees of freedom. It should be possible
then by a simple transformation to define the system of n
arbitrary congruent vectors by n other systems of concurrent
vectors which are symmetrical and have a common point. The
n symmetr cal systems so obtained are the symmetrical co-
ordinates of the given system of vectors and completely define
it.
This method of representing polyphase systems has been
employed in the past to a limited extent, but up to the present
time there has been as far as the author is aware no systematic
presentation of the method. The writer hopes by this paper to
interest others in the application of the method, which will be
found to be a valuable instrument for the solution of certain
classes of polyphase networks.
In dealing with alternating currents in this paper, use is
made of the complex variable which in its most general form
19181 FORTESCUE: SYMMETRICAL CO-ORDINA TES 1029
may be represented as a vector of variable length rotating about

a given point at variable angular velocity or better as the re-
sultant of a number of vectors each of constant length rotating
at different angular velocities in the same direction about a
given point. This vector is represented in the text by I, E,
etc., and the conjugate vector which rotates at the same speed
in the opposite direction is represented by f, E, etc. The effec-
tive value of the vector is represented by the symbol without
the distinguishing mark asI, E, etc. The impedances Za, Zb,
Zab, etc., are generalized expressions for the self and mutual
impedances. For a circuit A the self-impedance operator will
be denoted by Zaa or Za In the case of two circuits A and B
the self impedance operators would be Zaa Zbb and the mutual
impedance operator Zab. The subletters denote the circuits to
which the operators apply. These operators are generally
d
functions of the operator, D = dt and the characteristics of
the circuit; these characteristics are constants only when there
is no physical motion. It will therefore be necessary to care-
fully distinguish between Za 7a and Ia Za when Za has the form
of a differential operator. In the first case a differential opera-
tion is carried out on the time variable Ta in the second case the
differential operator is merely multiplied by Ia.
The most general expression for a simple harmonic quantity
e is
e = A cos pt - B sin pt
in exponential form this becomes
A +jBEJPI+ A-j B>j P
e- 2 t+ 2
(A + j B) E PI represents a vector of length V A2 + B2 rotating
in the positive direction with angular velocity p while (A - j B)
E-i P is the conjugate vector rotating at the same angular
velocity in the opposite direction. Since E1 PI is equal to
cos pt+j sin Pt, the positively rotating vector E = (A +j B) ei t
will be
A = A cos pt- B sin pt + j (A sin pt + B cos pt)
or the real part of E which is its projection on a given axis is
equal to e and therefore A may be taken to represent e in phase
and magnitude. It should be noted that the conjugate vector
.l is equally available, but it is not so convenient since the
1030 FOR TESCUE: SYMMETRICAL CO-ORDINA TES [June 28
operation dt e. Pt gives - j p e-i Pt and the imaginary part

of the impedance operator will have a negative sign.
The complex roots of unity will be referred to from time to
time in the paper. Thus the complete solution of the equation
x -1 = 0 requires n different values of x, only one of which
is real when n is an odd integer. To obtain the other roots we
have the relation
1 = cos 2 r r + j sin 2 ir r
- 2 -r r
Where r is any integer. We have therefore

1 .2irr
n = X
and by giving successive integral values to r from 1 to n, all

the n roots of xn- 1 = 0 are obtained namely,
n -. 2r 2
a, = = cos
n
s+sn2nn
a2 =
4irn = CoS
4T
4w .
+j sin
4w
n
a3=E = cos
6w +j sin 6 wr
n 1
= EC2J = 1
It will be observed that a2 a3.... a, are respectively equal to

a,(. -l)
a,2 a,3 ....
When there is relative motion between the different parts
of a circuit as for example in rotating machinery, the mutual
inductances enter into the equation as time variables and when
the motion is angular the quantities eJwt and e - jwt will appear
in the operators. In this case we do not reject the portion of
the operator having e-iwt as a factor, because the equations
require that each vector shall be operated on by the operator
as a whole which when it takes the form of a harmonic time
function will contain terms with Ejwt and e--jw in conjugate
relation. In some cases as a result of this, solutions will appear
with indices of e which are negative time variables; in such
cases in the final statement the vectors with negative index
should be replaced by their conjugates which rotate in the
positive direction.
1918] FORTESCUE: SYMMETRICAL CO-ORDINATES 1031
This paper is subdivided as follows:

Part I.-"The Method of Symmetrical Co-ordinates." Deals
with the theory of the method, and its application to simple
polyphase circuits.
Part II.-Application to Symmetrical Machines on Unbal-
anced Polyphase Circuits. Takes up Induction Motors, Gener-
ator and Synchronous Motor, Phase Balancers and Phase
Convertors.
Part III. Application to Machines having Unsymmetrical
Windings.
In the Appendix the mathematical representation of field
forms and the derivation of the constants of different forms of
networks is taken up.
The portions of Part I dealing with unsymmetrical windings
are not required for the applications taken up in Part II and
may be deferred to a later reading. The greater part of Part I
is taken up in deriving formulas for special cases from the
general formulas (30) and (33), and the reading of the text fol-
lowing these equations may be confined to the special cases of
immediate interest.
I wish to express my appreciation of the valuable help and
suggestions that have been given me in the preparation of this
paper by Prof. Karapetoff who suggested that the subject be
presented in a mathematical paper and by Dr. J. Slepian to
whom I am indebted for the idea of sequence operators and by
others who have been interested in the paper.
PART I
Method of Symmetrical Generalized Co-ordinates
RESOLUTION OF UNBALANCED SYSTEMS OF VECTORS AND
OPERATORS
The complex time function A may be used instead of the har-
monic time function e in any equation algebraic or differential
in which it appears linearly. The reason of this is because if
any linear operation is performed on A the same operation per-
formed on its conjugate A will give a result which is conjugate
to that obtained from A, and the sum of the two results obtained
is a solution of the same operation performed on A + X, or 2 e.
It is customary to interpret A and A: as coplanar vectors,
rotating about a common point and e as the projection of either
vector on a given line, A being a positively rotating vector and*
A being a negatively rotating vector, and their projection on

the given line being
+
E (1)
2
Obviously if this interpretation is accepted one of the two
vectors becomes superfluous and the positively rotating vector
E may be taken to represent the variable "e" and we may de-
fine "e" by saying that "e" is the projection of the vector E
on a given line or else by saying that "e" is the real part of the
complex variable .
If (1), a, a2 ... .al-1 are the n roots of the equation xn - 1 = 0
a symmetrical polyphase system of n phases may be represented
by
Eli = El
E21 = a
E31 = a2 E11
..............
.........
(2)
Aft1 = an-l A1
Another n phase system may be obtained by taking
E~12 E12
=
E22 = a2 E12
E32 = a4 E12
.............. (3)
(n-1)A12
En2 = = a2 (flE12
and this also is symmetrical, although it is entirely different
from (2).
Since 1 + a + a2 + an-l = 0, the sum of all the vectors
of a symmetrical polyphase system is zero.
If E1 A2 E3.... EA be a system of n vectors, the following
identities may be proved by inspection:
19181 FOR TESCUE: SYMMETRICAL CO-ORDINATES 1033
n
+ A1 + aE2 +E an-aE(n
a2 (-3 .
- 1 + a2+E3 + a..E3+.2(
n
El +a-IE-+ a4En +,, . En(-1

n
+ + A
El+ar-'E2+ a'(-l E3 + anl(-n)
E
+ E +a2
+2 aEn
a-2 3
n
E1 + aE2a a3
+ a2
+ a(r1)+ar+ a2 ++(ri)a3+a(n)E
n ~ ~ n 4
+ a-2 11 + a E2 + a' E3 + a... a--1En
n (4)
+ a-(r-l)-A1 + aA2 E2 a3 E+a
a-1 (l
+ a-(n-1) El + a-'2 + a-2E3 + ..a-(n-1) E

n
An_ 91 +a)2 + E3 + ....En

n
+ a1 -Al +a'E2+ a2E3 +..-.an-' E
+ a-2(n- 1) El +at E2x+paEs s+ fr A in t E abov

n
+ a-(n- 1) (r- 1) t r1i2+....a(-)(-)f

n
+ a-' El + a-1t2 + a-2 123 +...a-(n En

n
It will be noted that in the expression for Al in the above
formulas if the first term of each component is taken the result is
n - orE1. If the succeeding terms of each component involving
1034 FORTESCUE: SYMMETRICAL CO-ORDINA TES [June28
E2 E3 . ... En respectively, are taken separately they add up to ex-

pressions of the form ' (1+a+a2 +... al-) which are all
n
equal to zero since (1 + a + a2. . . a - 1) is equal to zero. In like
manner in the expression for E2 E3. . . E,n respectively, all the terms
of the components involving each of the quantities E1 E2 E3.. . etc.
excepting the terms involving that one of which the components
are to be determined
Al add up to expressions
of the form Er
~~~~~~~~~~~
(1 + a + a2 + ....an-') all of which are equal to zero, the re-
maining terms add up to E2 E3.... E,, respectively. It will
now be apparent that (4), is true whatever may be the nature
of El E2 etc., and therefore it is true of all numbers, real complex
or imaginary, whatever they may represent and therefore
similar relations may be obtained for current vectors and they
may be extended to include not only vectors but also the oper-
ators.
In order to simplify the expressions which become unwieldy
when applied to the general n-phase system, let us consider a
three-phase system of vectors Ea Eb EC. Then we have the
following identities:
Ea_E + Esb+ Ec +E, + a Eb + a2 E,
Ea ~3 +3
Ea + a2 Eb + a E,
+ 3
Ea +Eb + Ec Ea + aEb + a2Ec
+a
Ea+a2Eb (5)
3 3
3
~E_ Ea+ Eo+ C + Ea + a Eb + a2 Ec
Ea + a2]Ab
3
+ a E1
(4) states the law that a system of n vectors or quantities

may be resolved when n is prime into n different symmetrical
groups or systems, one of which consists of n equal vectors and
the remaining (n - 1) systems consist of n equi-spaced vectors
which with the first mentioned groups of equal vectors forms
an equal number of symmetrical n-phase systems. When

n is not prime some of the n-phase systems degenerate into
repetitions of systems having numbers of phases corresponding
to the factors of n.
Equation (5) states that any three vectors Ea Eb Ec may be
resolved into a system of three equal vectors EaO .aO Eao and
two symmetrical three-phase systems Ea,, a2 Eal, a Eal, and Ea2,
Ea
-
Eb,i
/b
Ea=ao + al +Ea2~II
Ea=
EaO+ Eal/ Ea2 ~~~~ I,,~~~~~~~~~~~
Ec = EaO + a Eal + a2Ia2
Figs. (1) and (2) show a graphical method of resolving three

aectors a Eb and Ec into their symmetrical three-phase com-
ponents corresponding to equations (o). The construction is as
follows:-EL0 Ebo, E;oare= obtained by crawing a line from 0 to
1036 FOR TESCUE: SYMMETRICAL CO-ORDINATES [June 28
the centroid of the triangle Ea Eb E,. Ea1 Ebb, ,ci are obtained
by rotating EbEb positively through
through
an angle ~~~~2
7r and
3, E, nega-
3'
tively through the same angle giving the points a Eb and a2 E,
respectively. Eai is the vector obtained by a line drawn from 0 to
the centroid of the triangle Ea, a Eb, a2 E,; and Ebl and Ea lag
this vector by 3 and 3 respectively. To obtain Ea2 Eb2
Ec2, Eb is rotated negatively and E positively through the angle
23 giving the points a2 Eb and a E, respectively; the line
drawn from 0 to the centroid of the triangle -a, a2 Eb, a Ec is the
aCE
a~~~
4w~~~~~~~a
FIG. 2-GRAPHICAL REPRESENTATION OF EQUATION 5.
vector Fa2b2 an E2 lead this vector by the angles and

437 respectively.
The system of operators Zaa Zbb Zcc Zab Zbc Zc may be resolved
in a similar manner into symmetrical groups,
Zaa = ZaaO + Zaai + Zaa2
Zbb = ZaaO + a2 Zaai + a Zaa2 (8)
cc= ZaaO + a Zaai + a2 Zaa2J

Zab = ZabO Zabl + Zab2
Zbc = Zabo + a2 Zab1 + a Zab2 }(9)
Zca = Zabo + a Zabl + ab Za2

There are similar relations for n-phase srystems.
EXPLANATION OF THEORY AND USE OF SEQUENCE OPERATOR

Let us define the symmetrical sequences of nth roots of unity
in the following manner:
SO= 1,1,1....1.
SI= 1, a-', a-2.. .a--1)

S2 =1, a-2, a-4 .... a- 2(n-1)
Sr = 1, a r, a-2r .... a-(n-U)r (10)

S(r+1) = 1, a-(r+1), a-2(r+1). . . a-(n-1)(r+1)
S(n-1) = a-(n-1) a-2(n-) ... a-(n-1)2

Consider the sequence obtained by the products of similar
terms of Sr and S1. It will be
S(r+1) = 1, a-kr+1), a-2(r+l) .. a-(n-1)(1) (11)
Similarly
Sk 1, a-k, a-2k .. a-("-l)k (12)
and the sequence obtained by products of like terms of this
sequence and Sr is
S(r+k) = 1, a-(r+k), a-2(r+k) . a- (n-1) (r +k) (13)
We may therefore apply the law of indices to the products of
sequences to obtain the resulting sequence.
In the case of the three-phase system we shall have the fol-
lowing sequences only to consider, viz.:
S=1, 1, 1
SI= 1, a2, a (14)
82 = 1, a, a2
The complete system of currents Ia lb I, are defined by
S (Ia) SIO 0+ Sl al + S2a2 (16)
Similarly the impedances Zaa Zbb Zc, may be expressed in sym-
metrical form
S (Zaa) So ZaaO + S' Zaal S Zaa2 (16)
and the mutual impedances Zab, Zbc, Zca are expressed by
S (Zab) _S' Zabo + S' Zbl + 52 Zb2 (17)
Attention is called to the importance of preserving the cyclic

order of self and mutual impedances, otherwise the rule for the
sequenee operator will not hold. Thus, Zab, Zb6 and Zca are in
proper sequence as also are Zca, Zab, Zb,.
When it is desired to change the first term in the sequence of
polyphase vectors the resulting expression will be
S (Tb) So Iao + S' a2 Ial + S2 a 1a2
(18)
S (Ic) -SAIao
S1r+ a Ia 2l
+ Ia2 j
Similarly in the case of the operators S (Zab) we have
S (Zbc) = So ZabO + S' a2 Zabl + S2 a Zab2
(19)
S (Zca) = So ZabO + S' a Zabl + S2 a2 Zab2
Similar rules apply to the e.m.fs. Ea Eb E,
S (Ea) = SI EaO + S Eal + S2 Ea2
S (Eb) =Eao+ S a2 Ea + 2aa2
S (20)
S (Ec) = SO EaO + SI a Eal + S2 a2 Ea2
It should be kept in mind that any one of the several expres-
sions S (la) S (Ib) S (I), etc., completely specifies the system,
and each of the members of the groups of equations given above
is a complete statement of the system of vectors or operators
and their relation.
APPLICATION TO SELF AND MUTUAL IMPEDANCE OPERATIONS
We may now proceed with the current systems S (I), S (Ib),
S (IC) and the operating groups S (Zaa) S (Zbb) S (Zcc) etc. and
the electromotive forces in exactly the same manner as for
simple a-c. circuits. Thus,
S (Ea) = S (Zaa) S (,a) + S (Zab) S (Ib) + S (Zca) S (Ic) (21)
= (SO ZaaO + S5 Zaal + 52 Zaa2) (SO laO + S' Ial + S2 Ia2)
+ (SO ZabO + SO Zabl + SO Zab2)

(SO laO + S a2 Ia + S2 a Ia2)
+ (SO ZabO + S' a Zabl + S2 a2 Zab2)
(SO Iao + S a Iai + S2 a2 I.2)
i SI (Zqqo + 2 Z:bo) laO + S {Zaa2 + (1 + a2) Zab2I Ia,
1918] FORTESCUE: SYMMETRICAL CO-ORDINA TES 1039
+ So {Zaal + (1 + a) Zabl } 1a2

+ S1 {Zaal + (1 + a) Zabi} ILO
+ S1 {ZaaO + (a + a2) ZabO} Ial
+ S1 { Zaa2 + 2 Zab2} 1a2
a
*+ S2 {Zaa2 + (1 + a2) Zab2} Iao
{Zaal + 2 a2 Zab 1 } Ia1

+ S2
+ S2 {ZaaO + (a + a2) Zabo} 1a2 (22)
Or since 1 +a+a2 = 0, 1 +a -a2, 1 +a2-a and
a + a2 -1
S (Aa) = So (ZaaO + 2 ZabO) TaO + S (Zaa2- a Zab2) Ial
+ So (Zaai - a2 Zabl) 7a2 + S1 (Zaai- a Zabl) 0aO

+ S' (ZaaO- ZabO) Ial + S' (Zaa2 + 2 a Zab2) 1a2
+ S2 (Zaa2 - a Zab2) Iao + S2 (Zaal + 2 a2 Zabi) Ial
+ 52 (ZaaO ZabO) Ia2 (23)
Or since
S (Zbc) = So ZbcO + S' Zb,l + S2 Zbc2
SS0 ZabO + S' a2 Zabl + S2 a Zab2
we may write (23) in the form
S (Aa) = So (ZaaO + 2 ZbcO) 1a0 + So (Zaa2 - Zbc2) Ial
+ SO (Zaal - Zbcl) 1a2 + S1 (Zaal - Zb,l) laO
+ S' (ZaaO- ZbcO) Ial + S' (Zaa2+ 2 Zbc2) 1a2
+ 52 (Zaa2 - Zbc2) laO + 2 (Zaai + 2 Zbf,) Iia
+ S2 (ZaaO- ZbcO) 1a2 (24)
which is the more symmetrical form. We have therefore from
(24) by expressing S (Ea) in terms of symmetrical co-ordinates
the three symmetrical equations
S0 EaO = So { (ZaaO + 2 ZbcO) IaO + (Zaa2 - Zbc2) Ia
+ (Zaai - Zbcl) 1a2}
S1 Eal = S5 { (Zaal- Zbl) LaO + (ZaaO- Zbco) lal
+ (Zaa2 + 2 Zbc2) Ia2 (25)
$2 Ea2 = 82 {
(Zaa2 -
Zbc2) laO + (Zaal + 2 ZbC1) fal
+ (ZaaO- Zbco) 1a2}
1040 FORTESCUE: SYMMETRICAL CO-ORDINA TES [June 28
An important case to which we must next give consideration

is that of mutual inductance between a primary polyphase
circuit and a secondary polyphase circuit. The mutual im-
pedances may be arranged in three sets. Let the currents in
the secondary windings be I. L, and lw, we may then express
the generalized mutual impedances as follows:
(I) Zau Zbv Zcw
(II) Zbw Zcu Zav (26)
(III) Zcv Zaw Zbu
Each set may be resolved into three symmetrical groups, so
that
S (Zau) = S Zauo + S' Zaul + S2 Zau2
S (Zbw) = S ZbWO + S' Zbwl + S2 ZbW2 (27)
S (Zcv) = S Zc,o + Si Zcv1 + S2 Z,v2
and we have for S (Ea) the primary induced e.m.f. due to the
secondary currents S (IU)
S (Ea) = S (Zau) S (Iu) + S (Zav) S (Iv) + S (Zaw) S (Iw) (28)
Substituting for S (Ia), S (It) and S (Iw) and S (Zau), S (Zav)

S (Zaw) their symmetrical equivalents we have
S (Ea) = So (Zauo + ZbWO + ZcvO) I,o
+ SO (Zau2 + a Zbw2 + a2 Zcv2) Iui
+ SO (Zaul + a2 Zbwl + a Zcvi) Iu2
'uo
+ S1 (Zaul + a Zbwl + a2 Zclu)
+ SO (ZauO + a2 Zbwo + a Zcvo) 'ul
+ S (Zau2 + ZbW2 + Zcv2) 1u2
+ S2 (Zau2 + a2 Zbw2 + a Zv2) Iuo
+ 52 (Zaul + Zbwl + Zcvl) Iul
1u2
+ S2 (ZauO + a Zbwo + a2 Z uo) .(29)
1918] FOR TESC UE: SYMMETRICAL CO-ORDINA TES 1041
On expressing S (E.) in symmetrical form we have the following

three symmetrical equations
S0 taO = S' { (ZauO + Zbwo + ZcvO) Iuo
+ (Zau2 + a Zbw2 + a2 Zcv2) ful
+ (Zaul + a2 Zbwl + a Zcvl) Iu2}
SEal = S' I (Zaul + a Zbwl + a2Z1) luo
+ (ZauO + a2 ZbwO + a Z.v0) Iu, (30)
+ (Zau2 + Zbw2 + Zcv2) Iu2}
SZ E,E2 = S2 { (Zau2 + a2 Zbw2 + a Zcv2) Iu
+ (Zaul + Zbwl + Zcv1) Iu

+ (ZauO + a Zbwo + a2 Z 1u2)
1U2}
For the e.m.f. S (EAu) induced in the secondary by the primary
currents S (Ia) we have
S (Eu) = S (Zau) S (Ia) + S (Zbu) S (L) + S (Zcu) S (IC) (31)
Since S (Zbu) bears the same relation to S (Zcv) as S (Zav)
does to S (Zbw) and S (Zau) bears the same relation to S (Zbw)
as S (Z.w) does to S (Zc,) to obtain S (A.) all that will be neces-
sary will be to interchange Zbw and Z,v in (29) and change 1o Igl 1u2
to lao Ial and 7a2 respectively, this gives
S (Eu) = So (Zauo + ZbwO + ZcvO) laO
+ S (Zau2 + a2 Zbw2 + a Zcv2) 'al
+ SO (Zaul + a Zbwi + a2 Zcvl) 'a2
+ S1 (Zaui + a2 Zbwl + a Zcv1) 'aO
+ S1 (Zauo + a ZbWO + a Zcv ) Ial
+ S1 (Zau2 + Zbw2 + Z,v2) Ia2
+ S2 (Zau2 + a Zbw2 + a2 Zcv2) 'aO
+ S2 (Zaul + Zbwl + Zcvl) Ial
+ 52 (ZauO + ZbwO + ZcvO) 1a2 (32)
and the three symmetrical equations will be

So SuO = So I (ZauO + Zbwo + ZcvO) laO ]
+ (Zau2 + a Zbw2 + a Zw2 + a Zcv2) -al
+ (Zaui + a Zbwl + a Zcl) Ia21
S Aal= S1 I (Zaul + a2 Zbwl +Zcvl) 'aO
a
+ (Zauo + a ZbwO + a Z I,,) fal (33)

+ (Zau2 + ZbW2 + Zc,2) 1a2}
S2 Eu2 = S2 { (Zau2 + a Zbw2 + a2 Zc,2) 'aO
+ (Zaul + Zbwl + Zcv1) Ial

+ (ZauO + a2 ZbwO + a Zcvo) Ia21
The same methods may be applied to polyphase systems of any
number of phases. When the number of phases is not prime the
system may sometimes be dealt with as a number of polyphase
systems having mutual inductance between them:-For example,
a nine-phase system may be treated as three three-phase sys-
tems, a twelve-phase system as three four-phase or four three-
phase systems. In certain forms of dissymmetry this method is
of great practical value, and its application will be taken up later.
For the present part of the paper we shall confine ourselves
to the three-phase system, and dissymmetries of several dif-
ferent kinds.
The operators Zau Zaa, etc., must be interpreted in the broadest
sense. They may be simple complex quantities or they may
be functions of the differential operator dt For if
i = I (An cos n w t + B,, sin n w t)
it may be expressed in the form
i I (Airn 6jnwt + A
2j Bn,-j,wt
I 7 (34)
2 2
- real part of I
and any linear algebraic operation performed on 1/2 will give

a result which will be conjugate to that obtained by carrying
out the same operation on f/2 and since the true solution is
the sum of these results, it may also be obtained by taking the
real part of the result of performing the operation on i.
MODIFICATION OF THE GENERAL CASE MET WITH IN PRACTICAL
NETWORKS
Several symmetrical arrangements of the operator Zau etc.'
are frequently met with in practical networks which result in
a much simpler system of equations than those obtained for
the general case as in equations (29) to (33). Thus for example
if all the operators in (26) are equal, all the operators in (27),
except Se ZauO S Zbwo and S Z,,o are equal to zero, and these
three quantities are also equal to one another so that equation
(30) becomes
S Eao = S (Zauo + ZbwO + ZcvO) Iuo 1
Si Eal = 0 (35)
S2 Ea2 = 0
and equation (33)
SO Euo = So (Zauo + ZbWO + ZcvO) Iao
S' Eu, = 0 (36)
S2 Eu2 = 0
This is the statement in symmetrical co-ordinates that a sym-
metrically disposed polyphase transmission line will produce
no electromagnetic induction in a second similar polyphase
system so disposed with respect to the first that mutual induc-
tions between all phases of the two are equal except that due to
single-phase currents passing through the conductors.
If in (26) the quantities in each group only are equal, equations
(30) and (33) become
S EaO = SO (ZauO + ZbwO + ZcvO) 'uo
Si Eal = S' (Zauo + a2 Zbwo + a Zcvo) Iui (37)
S2 Aa2 = S2 (ZauO + a Zbwo + a2 Z0vo) 'u22
SO Euo = S (ZauO + ZbwO + ZtcvO) Ia 1

S' Eal = S1 (ZauO + a Zbwo + a2 ZcvO) Iai (38)
S2 Ea2 = Sl (Zauo+ a2 Zbwo + a Zcvo) Ia22
SYMMETRICAL FORMS OF COMMON OCCURRENCE
A symmetrical form which is of importance because it is of
frequent occurrence in practical polyphase networks has the
terms in group (I) equation (26) all equal and those in group
2
(II) cos times those in group (I) and those in group (III)
4
cos times those in group (I).
Snecs2w
Since cos 2 7
r a +a2 4wr we have on substituting
the values of the impedances in this case,
50 Eao SY {ZauO (1 +
= a + a2)} Iruo = 0
S' Eal = S' 12' ZauO0 IU (39)
S2 Ea2 = 212 Zauo Iu2
S Euo = S0 {ZauO (1 + a + a2)} Iao = 0
S' Eu1 = S' 11 Zauo lal (40)
S2 Eu2 =
S2 112 ZauO 1a2
The elements in group I may be unequal but groups II and
III may be obtained from group I by multiplying by cos
and cos 2w
3 respectively.
The members of the three groups will then be related as fol-
lows, the same sequence being used as before,
(I) zau Zbv zc w I
a+ a2 a + a2 a2
Zau +
a
2 ZW, 2 j
+2 (41)
a + a2 a + a2 a + a2
2II Zbv, 2 ZCWI 2 Zau
Consequently the following relations are true:

50 = a + a
2 Zau
S = a +2 a ZauO
2
a2 Zu
S2 Zbw2 = a S2 Zau2 (42)

2
52 ZbW2
v = 2 2S2 Zau2
aul2
Substituting these relations in (30) and (33) we have for this

system of mutual impedances
ZauO + ZbwO + ZcvO = 1
ZauO + a Zbwo + a2 Z",o = 11 Zc,o (43)
ZauO + a2 ZbwO + a Zcvo = 12 ZawO J
Zaul + Zbwl + Z,vl = 112 Zaul
Zaul + a Zbwl + a2 Zcv = 1 2 Zaul (44)
Zaul + a2 Zbwl + a Zcv1 = O
Zau2 + ZbW2 + Zcv2 = 12 Zau2
Zau2 + a Zbw2 + a2 Zcv2 = 0 (45)
Zau2 + a2ZbW2 a v2 = 1 2 Zau2
which on substitution in (30) and (33) gives
5 EaO = 0
S' Eal = S' {12 Zaul Iuo + 12 ZauO Iui + 1 Zau2

1 Iu2} (46)
SIEa2 Ill2
{1Zau2 IUO + 12Zaul Iul + 1 Zau0 Iu2}
Y Euo = 50 111 Zau2 Ial + 1 Zaul 1a2}
5' A51 8'l= {15 ZauO I + Ia2}47)
S2 Eu2 = S2 {1 Zazul Ial + 1 Zau o2} }J
The above symmetrical forms in which the factors cos 3

and cos occur apply particularly to electromagnetic induc-
tion between windings distributed over the surfaces of co-
axial cylinders; where if the plane of symmetry of one winding
be taken as the datum plane, the mutual impedance between
this winding and any other is a harmonic function of the angle
between its plane of symmetry and the datum plane. In other
words, the mutual impedances are functions of position on the
circumference of a circle and may therefore be expanded by
Fourier's theorem in a series of integral harmonics of the angle
made by the planes of symmetry with the datum plane. Since
the same procedure applies to all the terms of the expansion
it is necessary only to consider the simple harmonic case. In
the partially symmetrical cases of mutual induction, such as
that taken up in the preceding discussion, there will be a differ-
ence between two possible cases, viz:-Symmetrical primary,
unsymmetrical secondary, which is the case just considered, and
unsymmetrical primary and symmetrical secondary in which
the impedances of (26) will have the following values
(I) Zau, Zbc, Zcw
(II) a+2 a +a2 a + a'
) Zc,
(III) +a2~ a+a2 2 Zcw a + a2 Zau Z (48)
2 2 +a 2
The results may be immediately set down by symmetry from
equations (46) and (47), but the difference between the two
cases will be better appreciated by setting down the component
symmetrical impedances, thus we have
SI bWO 2 au 0
a
SOZcvo0
a
a'2 auO
S1 ZbWl 1 +a S' Zau (

2 2
S2 ZbW2 2
S2 Zau 2 (9
S V1= +2a2
2
22
Substituting these relations in the impedances used in (30)

and (33) they become
ZauO + Zbwo + ZcvO = 0 )
Zauo + a ZbWo + a Z0 = 1I ZauO (50)
Zauo + a2 Zbwo + a Zcvo = 1 ZauO J
Zaul + Zbwl + Zcvlj = 1 2 Zaul
Zaul + a Zbwl + a2 Zcv = 0 (51)
Zaul + a2 Zbwl + a Z,v1 = 11 Zaul J
Zau2 + Z6w2 + Zcv2 = 12 Zau2
Zau2 + a Zbw2 + a Zcv2 = 1 Zau2 (52)
Zau2 + a2 Zbw2 + a = 0 J
And we have from (30) and (33), or by symmetry
S Eao = So {1 Zau2 Iui + 12 Zaul 1u2}
S1 Eal = S' {12 Zauo fui + 12 Zau2 Iu2} (53)
S2 Ea2 S2 {12 Zaul Iul + 12 Zauo 1u2 J
So=Eo0
S' Aui = Sl {1l Zaui laO + 1 Zauo Iul + 1' Zau2 Iu2} (54)
S2 Eu2 = 52 {12 Zau2 laO + 11 Zaul Iul + 11 ZauO Iu2}
If the angle between the planes of symmetry of the coils and
2 7r
the datum plane are subject to changes, cos and cos
in the preceding discussion must be replaced by
27r+ a a2
cos (43!E + E)
= 2--e + a j (66)
where 0 is measured from the datum plane
In the strictly symmetrical case of co-axial cylindrical sur-
face windings in which the members of each group of mutual
impedances are equal, the result of substituting (55) in the

equations for induced e.m.f. will be
S Eao = 0
S Eal = S (12 ZauO Ej' Iul) (56)
S2 Ea2 = S2 (14 ZauO E-j' Iu2)
SoEuo 00
Eu0 =
82 Eui = S (14 ZauO I E aCa 1) (57)
S2 Eu2 = S2 (12 Zauo Ej' Ia2)2
In the case having symmetrical primary and unsymmetrical
secondary in which members of each group are different, but
in which there are harmonic relations between corresponding
members of the different groups, the impedances are
(I) Zau, Zbv, Zcw
(I)(2 f + 2 6-io )Zcw
a a2z
j +a2 ) au,(2 + a2 ) (58)
(III) (-v--2 ej0 + 2 e-j9 Zbv,

( a2 Ej0 + 2 cjO) ZcW, ( 2- El + 2 e-10)Zau
The symmetrical component mutual impedances will have the
following values in terms of ZauO Zaul Zau2
S zbWO = ( a < + a2 )
So ZbWo = ( E2
0
+ a2 e-J ) S ZauO
S1 Zbw, -( (22j + _
2 )8S Zaui (59)
S2 Zbw2 = ( + +a ) S Zau2
S' Zwl= ( 2 Ei + E2) S' Zaul

S2Zcv2 = ( + 2 )S2
Substituting these relations in the impedances of equations

(30) and (33) they besome
Zauo + Zbwo + ZcvO = 0
ZauO + a ZbWO + a' Zcvo = 142 ZauO E } (60)
ZauO + a Zbwo + a Zcvo = 12 Zau 0 j
Zaul + Zbwl + Zcvl = 112 Zau l 6 1

Zaui + a Zbwl + a2 Z,,,, = 14 Zaua 1 (61)
Zaul + a2 Zbwl + a Zcvl = 0
Zau2 + ZbW2 + Zcv2 = 12 Zau2 jo ]
Zau2 + a ZbW2 + a2 Zcv2 = 0 (62)
Zau2 + a2 Zbw2 + a Zcv2 = 14' Zau E i
which on substitution in (30) and (33) give
S EaO = 0
SEal = S {14 ZaulEjo IUO + l Zau 0E' Iui
+ 1 2 Zau2&6' 1u2} (63)
82 Ea2 S2 {12 Zau2 E-j0 Iuo + 12 Zaul E-j0 Iul
+ 1 2Zau0 Ei Iu2}
S0Eu0= SI {1 Zau2 E0 lIal+ 14 Zaui I-01a2} 1
Sl Eul = S' {12 ZauO IE-j Ial + 14 Zau2 Ejo Ia2} (64)
82 Eu2 = S2 {14 Zaul E0 Ial + 14 ZauO Cj Ia2}
0
In the case of unsymmetrical primary and symmetrical

secondary, we have for the value of the impedance in terms of
ZauO Zaui and Zau2
(I) Zau, Zbv, ZCW
(II) ( a2 ,ej + 2 -jo) Zbc,
a_ ___a a2
(2 eje + 2 Ce-j) ZcwP, (2 r;j + 2 2E_j@)) zau ~~(65)

(III) ( 02
( + 2 (-j) Zcw,
( 2 lE-0 + 2 C.-i) Zau, ( 2 6 + 2 3
The symmetrical component mutual impedances in terms of

ZauO, Zauli Zau2 are
(a f + a2{-jl Zu
ZbwoZbo= (-
Sl So 2 lEj0
E0 +2-
SOZaa2 2 fE-j0)
a e) SOSZau(
Zauo
S2 Zcv= ( 2 E0 + E-J) SoZau2
Sl Zbwl = ( + 2 E-jo S1 Zaul

2 2 ~~~~~~~~(66)
S2 Zbw2 = ( a2
2 + 22-) S22Zu
Zn_u2
(. 0 a2 Ej IZU
SI Zcvl = 2 + 2
S2Z a
ej 02 IE,jS2 Zau
S2cv2 = 2
i ;i)Zau2
2
And the impedances of equations (30) and (33) become

ZauO + Zbwo + ZcvO = 0
Zauo + a Zbwo + a2 Z,vo, = 142 ZauO Eji (67)
ZauO + a2 Zbwo + a Zcvo = 12 ZauO Ei0j
Zaul + Zbwl + Zcv1 = 14 Zaul1 &
Zaul + a Zbwl + a2 Zcv, = 0 (68)
Zaui + a2 Zbwl + a Zcv, = 14 Zaul E1i
Zau2 + Zbw2 + Zcv2 = 14 Zau2 E-j0

Zau2 + a Zb.2 + a2 Zcv2 = 12 Zau2 E0 (69)
Zau2 + a2 Zbw + a Zcv2 =0
And on substitution in (30) and (33), or by symmetry from
(63) and (64), we have
t
JalaO- So {1 Zau2 E
Iul + 12 Zaul E-j0 1u21 1
S1 Eal = S {11 ZauO Cj0 Iul + 12 Zau2 IE0 Iu2} (70)

82 Aa2 = S2 { 1 Zau l Ei Iul + 14 ZauO E-0
Iuj 2 J
SI Euo = 0
S1 Eu1 = S' {14 Zaui E-j IaO + 1 Zaui E-j IuI
+ 12 Zau2 6E-j 1a2} (71)
S2 Eu2 = S2 {111 Zau2 E"0 IaO + 12 Zail Ej0 lal
+ 14 Zau0 C I a2} J
A fuller discussion of self and mutual impedances of co-axial
cylindrical windings will be found in the Appendix. It will be
sufficient to note here that in the case of self inductance and
mutual inductance of stationary windings symmetrically dis-
posed if they are equal
Mab Mbc = Mca (A os
3
)
j (72)
Laa = Lbb = L -c= Maa = Mbb = Mcc_ = n
If the windings are symmetrically disposed but have different
number of turns
Laa = Maa = 2 A
Lbb = Mbb = 2 Bn (73)
Lcc = = 2 C.,
Mab = A Bn cos 3 )
Mbc= (V BCn Ccos)S

2 (74)
Mca =
( Cn An cos ) j
If the coils are alike but unsymmetrically spaced Laa Lbb LCC have
the same values, namely 2 An and
Mab = 1(AUcosnOfl)cos 2 7r
3
Mab = z 2 n
X (An cos n 0 1) COS + (An1r sin n 01) sin 2 n |
Mbe =2.1 (AUcosnO02) COS 2 n r
=
Mca2 (AUcosn03)cos 2ni
+ (An sin n 02) sin 2 n 7
If they are unequal as well as unsymmetrically disposed but are

otherwise similar Laa Lbb LC have values as in (64) and
Mab = (V/An Bn cos n 01) cos 2nr

3
+ (.\/An Bn sin n 01) sin -37
Mbc= I \(/Bn Cn cos n 02) COS 2

3 (76)
+ (V\Bn Cn sin n 02) sin 2 n3
r
Mca =j (V\ICn An Gi) cos nr

cos n
+ (.\CnAnsinn 03) sin2nj

r
3
Where the windings are dissimilar in every respect the expres-
sions become more complicated. A short outline of this subject
is given in the Appendix.
In the case of mutual inductance between two coaxial cylindri-
cal systems, one of which A, B, C is the primary and the other
U, V, W the secondary, the following A
conventions should be followed:
(a) All angles are measured, taking . u
the primary planes of symmetry as data
in a positive direction. Rotion
(b) The datum plane for all windings
is the plane of symmetry of the primary
A phase. v W
(c) All mechanical motions unless B C
otherwise stated shall be considered as FIG. 3-CONVENTIONAL
positive rotations of the secondary DISPOSITION OF PHASES
cylinder about its axis. AND DIRECTION OF Ro
(d) The conventional disposition of
the phases and the direction of rotation of the secondary wind-
ing are indicated in Fig. 3.
We shall consider five cases; Case 1 being the completely sym-
metrical case and the rest being symmetrical in one winding, the
other winding being unsymmetrical in magnitude and phase, or
both, but all windings having the same form and distribution of
coils.
Case I. A ll Windings Symmetrical.

Mau = Mbv = Mcw = f An cos n 0
Mbw = Mcu = May = 2 An cos n (-3 + f) (
Mlv = Maw = Mbu =- A,, cos n (4- + f)

Case II. Primary Windings equal and Symmetrical, Secondary
Windings unequal but otherwise Symmetrical.
Mau = 2 An cos n6, Mb, = I Bn cosn 0,
Mcw.= 2 Cu cosn 0
Mbw= 2 Cn cos n 3 +
Mcu= An cos n (27r +
Mav = 2 Bn cos n (2i + a) (78)

Mc, = Bn cos n + f
C cos n
Maw = Cn + a)
Mbu = An cos n (4i7 +)

Case III. Primary Windings Unequal but Otherwise Symmetri-
cal, Secondary Winding Equal and Symmetrical.
Mau =2Ancosn0, Mbv= 2Bncosn 0, Mcw= C,,cosn c
Mbw- Bn cos n (3W+ O)

=
Meu = Cn cos n (23 + )
Mav = 2 Ancos n (3 + 0) (79)
Mcv=2 Cn cos n (3 +)
Maw An cos n(3+3 )
Mbu = z B,, cos n (il + 0)
Case I V. Same as Case II except in addition to inequality

Secondary Windings are Displaced from Symmetry by angles a,
a2 and a3 whose sum is zero.
Ma = I (A,, cos a, cos n 0 + An sin a, sin n 0)
Mbv = z (Bn cos a2 cos n 6 + Bn sin a2 sin n 0)
Mc= (C. cos a3 cos n 0 + C, sin a3 sin n 0)
Mbw CnCos 3 COS n ( + 0)
+ Cn sin a3 sin n (23 + 0)}
Mcu - {An cos a, cos n (2L + 0)
+ An sin a, sin n(23 +o)}
Mav =2 B,, cos a2 cos n ( j + a)
+ Bn sin a2 sin n(23r + 0)} (80)
Mv Bncos a2 cosn (3 + 0)
+ Bn sin a2 sin n(43 + 0)}
Maw Cn cos 3 COS n(43 +)
+ Cn sin a3 sin n 4(3 +6)}
Mbu= 2 A cos a1 cos n + 0)
+ An sin a, sin n (4jr + )}
Case V. Same as Case III except that the Primary Windings

are Unsymmetrically disposed with respect to one anotheraswellas
being unequal.
Mau = z (A,cosca,cosn 6 + Ansina,sinn 0)

Mbv = z (Bn cos a2 COS n 6 + Bn sin a2sinn 6)
m = (Cn cos a3cos n 6 + Cn sin a3sinn 6)
Mbw 2 {Bn cos a2 cOsn( + )
+ Bn sin a2 sin n (23 +
Afcu = 4 {Ancos a,cosn( L + o)
+ An si ct si (2W1 + o) }
+ Cu sina(3 sin n (2W7 + o) }

Mc ({CnlCOSaC3 COSn ( 43W 0)
+ Cn sin caasirfn(43W +0) }
Maw= { Ancos acos n( 437+60)
4
+ An sin a, sin n (4379 + a) }
Mba = 2 {Bn coS a2 cos n (4W + a)
+.Bn sin a2 sin n (-. +6)}
The expressions for dissymmetry in both windings and for un-

symmetrically wound coils, etc., are more complicated and will be
dealt with in the Appendix.
The impedances Zaa Zbb, etc., Zau Zbv, etc., are functions of
Maa M^b, etc., Mau MbC,etc., and the resistances of the system.
The component of e. m. f. proportional to the current due to
mutual impedance is so small that it may generally be neglected

so that Zau becomes d dl~~~~d
Mau, Zbv = dt Mb, and so forth.
If the secondary winding is rotating at an angular velocity
a, 0 in equation (55) becomes a t and the operators Zaa, etc.
operate on such products as c-ixt lUl Ejct 1u2 where Iu and
1u2 are the variables.
The following relations will be found useful in the application
of the method in actual examples.
If D denotes the operator d and p (Z) is a rational algebraic
function of Z dx
yV (D) eax - e; (a) eax 1
y (D) {eax X} = eax y (D + a) X (82)
p (D) Y = eax y (D + a) Y e-ax
Where X and Y may be any function of x.
Star and Delta e.m.fs. and Currents in Terms of Symmetrical
Components
It has been shown in the preceding portion of this paper that
the e. m. fs. Ea Eb and Ec and the currents Ia Ib and 7c whatever
their distortion, may be represented by the sum of symmetrical
systems of e. m. fs. or currents so that the two expressions
S (Ea) = SO EaO + SI Eal + S2 Ea2 )
(83)
S (Ia) = S laO + Sl Iai + S2 1a2
completely define these two systems.
If we take the delta e. m. fs. and currents corresponding to
S5 EaO, Si Eal and S2 Ea2, Sl Ial, S2 Ia2, we have, since Eb,l leads Ea
by 2 and Eb,2 lags behind Ea2 by the same angle
S EbcO = 0
S' Ebcl =jA/3 S' EaIl
S2 Ebc2 = jV/3 S2 Ea2
S IbCO=-indeterminate from S (Ia) (84)

SIbc
Si j- ~~~1S1 I
=
.
52 Ibc2 = -i S2 1a2
And therefore if we take Eab as the principal vector

SO EabO = 0
S1 Eabl = j a sl'V3 Eal

S2 Eab2 = -j a2 S2 V3Ea2 (85)
S (Eab) = S Eabl + S2 Eab2
The last equation of group (85) when expanded gives
Eab = j '\/3 (a Eal - a2 Ea2)
Ebc = jV/3 (Eal Ea2) (86)
Eca = j \/3 (a2 Eal- a Ea2)
which may also be obtained direct from (83) by means of the
relations
Eab = Eb- Ea
Ebc = Ec - b
Eca = Ea Eo
Similarly
S2 lab = indeterminate from S (I)
SlIabl= j a ] a
'/3
82 Iab2 = j a2 [a2 (87)
S ( lab) = 82 labO + S' Iabl + S2 Iab2
with similar expression for lab 'bc and Iea which may be verified
by means of the relations
'a = Ica- lab + lao
lb = lab -Ibc + laO
Ic = bc- Ica + laO
Conversely to (84) we have the following relations
5 EaO = indeterminate from S (Jiab)

St Eal =S- Ebcl =-j SI Eaba
52Ea2- j S2EbC2= j S2Eab2 (88)

SO Iao = indeterminate from S (Jab)
S' fal = - j \/3- SI Ibel = j a2 -\,3 St lablI
S2 Ia2 = j V352 Ibc j aV/3S2Iab2
It will be sufficient in order to illustrate the application of

the principle of symmetrical coordinates to simple circuits to
apply it to a few simple cases of transformer connections before
proceeding to its application to rotating polyphase systems to
which it is particularly adapted.
UNSYMMETRICAL BANK OF DELTA-DELTA TRANSFORMERS
OPERATING ON A SYMMETRICAL CIRCUIT SUPPLYING A
BALANCED SYSTEM
Let the transformer effective impedances be ZAB ZBC ZCA and
let the secondary load currents be Iu Iv and 1w and let
the star load impedance be Z. One to one ratio of trans-
formation will be assumed, and the effect of the magnetizing
current will be neglected. The symmetrical equations are
o So (ZABO labO + ZAB2 Iabl + ZAB1 Iab2)

SI EUV1 - S' Eabl- S1 (ZA.B1 abO + ZABO Iabl + ZAB2 Iab2)
2 EUV2 O- S2 (ZAB2 labO + ZAB1 Iabl + ZABO lab2)
(89)
SIo l = 0
51 Z IU2 =EU2
Since the transformation ratio is unity and the effects of
magnetizing currents are negligible SI 'abl = S' 'uvi, S2 Iab2
S2 V2. And therefore by means of the relations (85), the last
two equations may be expressed
19181 FORTESCUE: SYMMETRICAL CO-ORDINATES 1059
Si EUV = SI 3 Z II
(90)
82 ,uv2 = S2 3 Z 1ab2
in other words, the symmetrical components appear in the
secondary as independent systems, 3 Z being the delta load im-
pedance equivalent to the star impedance Z.
Substituting from (90) in the second and third equation and
eliminating 7abO by means of the first equation, and we have
S' gab = S {(3 Z + ZABO ) ab I
+ (ZAB2
Z2
')iab2 }
52 0 S2 {( Z2AB32 )I
ZABO (91)
+ (3 Z + ZABO ZAB ZAB2 )

ZABO
b }
which, when S1 and S2 are removed, give two simultaneous equa-

tions in lab, and Iab2.
A modification of the problem may occur even when the load
impedances are symmetrical, as they may have symmetrical
but unequal impedances Z1 and Z2, to the two components
Iu1 and Iu2 respectively, as in the case of a load consisting of a
symmetrical rotating machine. The equations corresponding
to (89), (90) and (91) then become
o = So (ZA.0 'IabO + ZAB2 'abl + ZAB1 Iab2)
S1 Auv1 = S1 -ab1 - S1 (ZAB1 'abO + ZABO labl + ZAB2 Iab2)
S2Euv2 = Q -
S2 (ZAB2 IabO + ZABI labl + ZA.BO Tab2)
(92)
so Luo = 0
S1 Z1 Iu, = Su
S2 Z2 1u2 = u2
S1 tUv = SI 3 Z1 Iab I
S2 Euv2 = S23 Z2 lab2
1060 FORTESCUE: SYMMETRICAL CO-ORDINATES [June 28
S -gab Si ( 3,Zl + ZABO Z I)Iabi +
(ZABO - Z2I ) I
2 }
ZABO !. (94)
S2O = S2 {(ZABI- ABO) Iabi
(3Z2+ ZABO- ) Iab2 }
w
C
Effective Imp.= ZB
Effective IMP. - ZABI

B
V
FIG. 4-OPEN DELTA OR V CONNECTION.
In an open delta system ZAB1 = ZA2= ZABO - ZAB the trans-

formers in this case being both the same. Equation (91) becomes
in this particular case where ZABO is infinite
S' Aabi = S' {(3 Z + 2 ZAa) fab + ZAB Iab2}

} (95)
52 O = S2 IZAB Iab + (3 Z + 2 ZAB) Iab2J
and we have
labO = abl - Iab2 (96)
Similarly, instead of (94) we have
S1 Eabi = S {(3 Z1 + 2 ZAB) fabl + ZAB Iab2}

(97)
S2 O = S2 {ZAB fabi + (3 Z2 + 2ZB)
Z ab2 } (
The secondary voltages are obtained from (90) and (93) for
this latter case.
The solution of (95) gives

b 3 Z1 + 2 ZAB E
lab 1
(3Z1 + 3ZAB) (3Z1 + ZAB)
Iab2 =
ZAB Eab (98)
(3 Z1 + 3 ZAB) (3 Z1 + ZAB)
-TabO =
3Z1 + 3ZAB Eab
And we have
SlIai= Sl 3Z1 + 2ZAB

3 (Z1 + ZAB) (Z + 3 )I(9)
52 Ia2 = S2 ZAB Eb
3 (Z I + ZAB) (ZI + 3 )
And therefore
= Ea ZAB
3 )-g\ab
la = -+
3,ZAB (Z1 + ZAB) Z 1 + ZAB
Tb= zE ZAB (100)

Z1 + AB (Z 1 + ZAB) (Z + 3
)Eab
~~EC
C =ZAB
i
3
Three-Phase System with Symmetrical Waves Having Harmonics
We may express Ea in the following form:
Ea = 1 Ejwt + E2 Ej2wt + E3 ej3Wt
(101)
= 2En Ejiwt J
where En is in general a complex number.

If the system is symmetrical three-phase Ab is obtained by
2
displacing the complete wave by the angle - ~- ror
.2_.4 Xr 6
b 3 El-te3E2jt+ 3Ee3+
.2 t 4 X.6 ir
EC=e E Ejwt e 3E2 ej2wt E E3 Ejwt +
or since e
23
,X
2
a.,
.2w3
.3 1
= a etc.
=Elieiwt+E2iC2wt +EE3Cj3w + I
Eb = a2 El Ejwt + a E2 + E3 j3wi + (102)
a E2Ei2wt + E3Ej3wt +
EC = a El ejwt + 2
or
S(Ea) =-S {E39ii2wi + E6EjCwt + E9EjI9wt + 1
+ S' {E Iejwt + E4CEi4wt + E7 ej7wt + } (103)

+ S2 {E2 EJ2wt + E5 Ei5wt + E8 j8wt + }
S (Ea) -SI z (E3, Ej3nwt) + SI z (EU-_2 Ej(3n-2)wt ) +
S2 2; (E3n_l 8 (3-n - l) wt) (104)
This shows that a symmetrical three-phase system having
harmonics is made up of positive and negative phase sequence
harmonic systems and others of zero phase sequence, that is to
say of the same phase in all windings, which comprise the group
of third harmonics. These facts are not generally appreciated
though they are factors that may have an appreciable influence in
the performance of commercial machines. It should be particu-
larly noted that in three-phase generators provided with dampers
the fifth, eleventh, seventeenth, and twenty-third harmonics
produce currents in the damper windings.
In dealing with the complex variable it will be convenient to
use for the amplitude the root mean square value for each har-
monic. When instantaneous values are required, the real part
of the complex variable should be multiplied by \. In the
remainder of this paper this convention will be adopted.
Power Representation in Symmetrical Co-ordinates
Since the power in an alternating-current system is also a har-
monically varying scalar quantity, it may therefore be repre-
sented in the same manner as the current or electromotive force,
that is to say by a complex variable which we shall denote by

(P + j Q) + (PH + j QH), P + j Q being the mean value, is
the term of the complex variable of zero frequency, P represent-
ing the real power and Q the wattless power, /p2 + Q2 will be the
volt-amperes.
The value of the complex variable (P + j Q) + (PH + j Q )
may be taken as
(P+jQ) + (PH +j QH) = E I +E I (105)
with the provision that for all terms having negative indices the
conjugate terms must be substituted, these terms being present
in the product E I +E f, which is the conjugate of the product
(105). A similar rule holds good for the symmetrical vector
system
S (Ea) = aO + S' Eai + . . +S- (n-l)
(106)
S (Ia) = S' IaO + S' Iai + . . . +Sn1 I(a ) }J(10
The conjugate of S I is
S (Ia) =S Lao + S(n-l) lal + . +S' Ia(n-l) (107)
and the power is represented by
(P + Pb) + j (Q + Qb) = {S (a) S (a) + S (Ea)S(Ia) (108)
with the same provision for terms having negative indices. The
sign z signifies that all the products in each sequence are added
together.
z {S (fa) S (Ea)I = Z &+
SO {Iao0aO + f Ea
+Ia(n-1) Ea(n-l)}
+ z S' {ao Aal + lai Ea2 + Ia2Ea3+ * .
+Ia(n-1) EaO}
+ S2 {fa Ea2 + Ial Ea3 + fa2 E4 +
+Ia(n-1) Ea,}
+ 2 ( {!a Ea(n-1) Ial EaS + n1
+ Ia(n-1) Ea(n-2)}
The terms prefixed by SI, S2, S3 . .S(n-1) all become
zero and since Se becomes n
XS(Ia)S(Ea) =n{IaOEa0+IaiEa1+
+Ia(z-1) Ea(n-l)} (110)
In a similar manner it may be shown that
I S (Ia) S (Ea) = n { laO Eao + Ial Ea(n-1l) + 1a2 Ea(n-2) +
+Ia(n-l) Eal} (111)
and therefore
(P +jQ) + (PH +jQH) n{faOEaO + falEal +

+ITa(n-1) Ea(n-1)}
(112)
+ n {Ia0EaO + IalEa(n- 1)+ +Ia(n_l)Eal}
For a three-phase system the expression reduces to
(P + j Q) + (PH + j Qb) = 3 (fao Eao + lai Eai + fa2 Ea2)
+ 3 (lao EaO + Iai Ea2 + 1a2 .al)
In the above expression P + PH is the value of the instantan-
eous power on the system, P being the mean value and PH the
harmonic portion. When the currents are simple sine waves, Q
may be interpreted to be the mean wattless power of the circuit
or the sum of the wattless volt-amperes of each circuit. In
rotating machinery since the coefficients of mutual induction
may be complex harmonic functions of the angular velocity,
this is not strictly true for all cases; but if the effective impedances
to the various frequencies of the component currents be used, it
will be found to be equal to the mean wattless volt-amperes of
the system with each harmonic considered independent.
In a balanced polyphase system PH and QH both become zero.
The instantaneous power is a quantity of great importance in
polyphase systems because the instantaneous torque is propor-
tional to it and this quantity enters into the problem of vibra-
tions which is at times a matter of great importance, especially
when caused by unbalanced e. m. fs. A system of currents
and e. m. fs. may be transformed to balanced polyphase by
means of transformers alone, provided that the value of PH is
zero, while on the other hand polyphase power cannot be
supplied from a pulsating power system without means for
supplying the necessary storage to make a continuous flow of

energy.
PART II
Application of the Method to Rotating Polyphase Networks
The methods of determining the constants Za Zu, M, etc., of
co-axial cylindrical networks is taken up in Appendix I of this
paper. It will be assumed that the reader has familiarized him-
self with these quantities and understands their significance.
We shall first consider the case of symmetrically wound machines
taking up the simple cases first and proceeding to more complex
ones.
SYMMETRICALLY WOUND INDUCTION MOTOR OPERATING ON
UNSYMMETRICAL POLYPHASE CIRCUIT
Denoting the pole-pitch angle by r let the synchronous angular
velocity be wo and let the angular slip velocity be co,. And let
S1 Eai S2 Ea2 be the symmetrical components of impressed poly-
phase e. m. f. Let Ra be the primary resistance and R. the
secondary resistance. The primary self-inductance being Maa,
that of the secondary being Muu and corresponding symbols
being used to denote the mutual inductances between the dif-
ferent pairs of windings. Then by means of (39), (40), (56) and
(57)
S'1al Si {Ra ai+ 1 2 Maaa al
+ 12 Mau dt EJ(w0 i1)

S2 Ea2=S Ra 7a2 + 121 AMaa ddlt a2
+ +12111 Miau
M d C-_j(,@-wi)tS 2
S' Ei =0= S' {Ru ul + 1 Muu d Iis [ (114)
+ +11211 Mau
M dd- -j(wo -wl)t la
I
S2Eu2 = S2Ru Iu2 + 12 Muu d t 1u2
+ 12 Mau iEI -Ia2

denote 12 Maa by La and 14 Muu by LU, 14 Mau by M, the equa-
tions (1) become
S'Eal = S (Ra+ La )Ial
+ M dj(wow)l) IU
S2Ea2 = S2 (Ra + La d-t )Ia2

+ M d Ej(wo-WI)t Ju2 }
dt (115)
slEl = - {Ru +Ludt)8
+ M dtEdj(w -WI) tai }
S2A2 = =S {(Ru+Lu dt)Iu2
d t E(wo-) Ia2 }
+ M dt a
From the last two equations we have

dM
jjui - M dt E lj(wowl)'ai (116)
Ru + Lu d t
d
Ru + Lu dt
Substituting these in the first two equations of (115) we obtain
S E(Ra
alS1 + La d t)
Al2 d d _J(wow1
M2 d-J( -4 -t
WI
al (118)
Ru +Lu {dt -(wo- WI)t
S2 Ea2 =S2 (Ra + La d-)
M2 d d ~~ dt+(io
+(wo- wi4
) I (l)
RU + LU { d -j (wo-w1)
W -
If Eal Eal Eijwt and Ea2 = Ea2 ejwl the solution for IaI and 1.2
will be
Ial = a (120)
Ia2 = Z2(121)
z2
Where
Z1 Ra + jWo La+ R2 + w12 L 2 (Ru- j WI Lu) (122)
Z2 Ra + jwoLa +
RU2 + (2wo - w1) L2 {Ru- j (2 wo - w2) L"} (123)

The impedances Z1 and Z2 will be found more convenient to use
in the form
Z, = (Ra + K12 Ru) + j wo (La-K12 Lu) + Wo WIWi K12 RU
(124)
Z2 = (Ra + K22 Ru) + j wo (La- K22 Lu) - Wo -WI K22 Rs,
(125)
Where, as we will see later, K12 and K22 are the squares of the
transformation ratios between primary and secondary currents
of positive and negative phase sequence.
The last real term in each expression is the virtual resistance
due to mechanical rotation and when combined with the mean
square current represents mechanical work performed, the posi-
tive sign representing work performed and the negative sign
work required.
Thus, for example, to enable the currents S2 Ia2 to flow, the
wO wlwh K12 R, must be applied tQ
mechanical work 3 1a22 2wt-h
the shaft of the motor,
The phase angles of the symmetrical systems S' Iai S2 1a2

with respect to their impressed e. m. f., SI Ea, and S2 Ea2 are
given by these impedances so that the complete solution of the
primary circuit is thus obtained.
The secondary currents are given by equations (116) and (117)
and are
=l MRu Ial E =wLu K, Ialfjwlt (126)
u2 = R- j (2j2w-,)
wo w,) M
-
1a2 Cj(2uo-w-)t -12=

K 2 Ia Ej(2u'o-w1)t
RU + j (2 wowl)Lu (127)
In the results just given, M is not the maximum value of
mutual inductance between a pair of primary and secondary
windings but is equal to the total mutual inductance due to a
current passing through the two coils W and V through the coil
U
A
C B
wV
FIG. 5
U as shown in the sketch Fig 5 and the winding "A" when A

and U have their planes of symmetry coincident.
Where the windings are symmetrical the induced e. m. f. is
independent of the division of current between W and V, but
this quantity must not be used in unsymmetrical windings, or
with star windings having a neutral point connection so that
IaO is not zero.
The appearance of M in this equation follows from the equa-
tion
Iu + Iv + Iw = 0
so that
I - (Iv + IW)
The power delivered by the motor is
Po = 3 o w WK12Ki2Iai.u
al 2Ru - w K22 Ia22 Ru} (128)
- 2 wo-w,
The copper losses are given by

PL = 3 {IIai1 (RP + K12 RU) + 1a22 (Rp + K22 RU) (129)
The iron loss is independent of the copper loss and power out-
put. The iron loss and windage may be taken as
P, = Iron loss and windage (130)
The power input as
P =PO+ PL + PF (131)
The mechanical power output is P0 less friction and windage
losses.
Torque = 3 - K12 IJa2R-w-
2 K22Ia22 Ru
X 107 dyne-cm. (132)
The kv-a. at the terminals is
VPI2+ QI2 = The effective value of 3 (Eal Ial + Ea2 Ia2) (133)
This last result may be arrived at in the following way
S (Ea) = SI Eal + S2 Ea2)
(134)
S (Ia) = 52 fal + S71a2
Since 82f al is conjugate to SI Ial, etc.
The product of E8, and fLa is the power product of the two
vectors, S (Ea) and S (Ia) and omits the harmonic variation as a
double frequency quantity, the average wattless appears as an
imaginary non-harmonic quantity.
PI + i QI =: (SE Eal 1i,1 + SW Ea2 1a2 + S' Ea2 fal
+ 82 Eal 7a2) (135)
The SI and 82 products have zero values, since the sum of the
terms of each sequence is zero, hence-
P, + j QI = 3 (Eal lal + Ea2 PaL) (136)
\/p12 + QI2 = The effective value of 3 (Eal Pal + Ea2 fa2) (137)
The solution for the general case of symmetrical motor opera-
ting on an unsymmetrical circuit is nGt of as much interest as
certain special cases depending thereon. Some of the most im-
portant of these will be taken up in the following paragraphs.
Case I. Single-phase e. m. f. impressed across one phase of
three-phase motor.
Assuming the single-phase voltage to be Ebc impressed across
the terminals B C. The known data or constraints are
Ebc =j3 (al-Ea2) } (138)
Ia = 0, Ib =-Ic J
and therefore
la - 1a2 (139)
Ea, Ea2
Z1 Z2
Z2
Ea2 z1 -gal (140)
Substituting in (138)
Ebc * Z_
El=
Ea =- -'-z
N /5 1+z
1 + Z2 (141)
Ea2= j-*
and therefore
*Ebc
N/3. Z +Z2 (142)
Ebc 1
1a2 = J7/3 Zl + Z2
Since Ib = Ibl + Ib2 a2 Ila +a 'a2
E143
lb =-IC
zl + z (143)
Wo
=
( K22 R' - 2wo _wI K22 RXuIJ2 (144)
Pi + j Qi le (Z1 + Z2) + PF
= (145)
The power factor is obtained from (145) by the formula
cos ax = Pi+ Qi2 (146)
Substituting from (142) in equation (126) and (127) of the

general case we obtain for the secondary currents
ful =- K, Ebl+ 22sewlt
IZ1=-+K1 1 (147)
Its2 = j K2- + E j(2wo-wwi) {
Many unsymmetrical cases may be expressed in terms of the

operation of coupled symmetrical motors operating on symmetri-
cal systems. This is invariably the case with symmetrical poly-
phase motors operating on single-phase circuits. Since the
physical interpretations are useful in impressing the facts on
ones memory they will be given whenever they appear to be
useful.
Equations (141) and (142) show that single-phase operation is
exactly equivalent to operating two duplicate motors in series
with a symmetrical polyphase e. m. f. SI Eab impressed across one
motor, the other being connected in series with the first but with
phase sequeince reversed, the two motors being directly coupled.
Case II. B and C connected together e. m. f. impressed across
A B.
The data given by the conditions of constraint are
Eab = - Lca
(148)
EbC = O =j3 (Aal-a2)
We therefore have
Aal = Ea2 = - 3ab (149)
and
1la I =-
Eab
Eab }
(10
a2a2= 3 Z2
The remainder follows from the general solution and need not
be repeated here.
(150) shows that a motor operated in this manner is the exact
equivalent in all respects to two duplicate mechanically coupled
polyphase motors, one of which has sequence reversed, operating
in parallel on a balanced three-phase circuit of e. m. f. SI Ea
The secondary currents follow from substitution of (150)
in equations (126) and (127) of the general case.
Case III. B and C connected together by the terminals of a
balance coil, the impressed e. m. f. EAD applied between A and the
middle point of the balance coil. Resistance and reactance of
balance coil negligible.
The data furnished by the connection in this case is
lb = IC = _ (151)
2
and therefore
Ia aa la
_= ~~2
ia~~~~~'
2 la_
'a2 =al = 2
We therefore have
Ea, Zl la
Eal --2 (152)
Ea2 Z2 Ia
2
we have
Eab =j V-3 (a Eal- a2 a2)
=
jV3- (a Z1 - a2 Z2)
Ac = j V3 -Ia (Z1 - Z2)
ad =(ab + Ac)
Ia
-jv/3 2 {(a+ )Z,-(a2+ ) Z2( ~~~~~~~~~(153)
_ - 3 Ia(Z1 + Z2)
and therefore,
aa--14 z
la 12 (16Ead
~~~~~~~(154)
1918] FOR TESCUE: SYMMETRICAL CO-ORDINATES 1073
r ~0- WjWO- Wi
P W K2 WO- K22 5 Ia Rs (156)
3
Pi +j Qi =Ia1 (Z1 + Z2) + PF (156)
cos a = P Q (157)
Vp12 + Q12
0-
z8 --Po~eratr
w 60
0
0~
o z
zI
a-
10 100
2000 3000 4000
MOTOR TORQUE
5000 >
FIG. 6-CHARACTERISTICS OF THREE-PHASE INDUCTION MOTOR-

BALANCED THREE-PHASE
Evidently (155), (156) and (157) are identical to (144), (145)

and (146) if Ia is equal to lb 2 This will be the case if
the value of Ead = 2 times that of Eb,. The total heating of

the motors will be the same in each case but the heating in one
phase for Case III will be one-third greater than for Case I.
a.
ffi 1000 Eff ciency
p
etc are co d tPower Factor
z 80
Li60
_.) S 0 ., a.
Li.
40~ ~~__:Powe-r Factor --
a.
~:20 __A- "
- fnp-ut 200
z
Uov -K!W. output Z-
a. 1000 2000 3000 4000 5000 >6
o- M
MOTOR TORQUE Y
FIG. 7-CHARACTERISTICS OF THREE-PHASE INDUCTION MOTOR-SINGLE-
'PHASE OPERATION -ONE LEAD OPEN
This method of operation is therefore, as far as total losses,
I-
etc. are concerned, the exact counterpart of two polyphase
100
Li.
u,j60
0~~~~~~~~~
0
-Efficinecy
a_
20 20
z -
'K.W. Output a.
a. 0 4
1000 2000 3000 4000 5000 6000 >
MOTOR TORQUE
FIG. 8-CHARACTERISTICS OF THREE-PHASE INDUCTION MOTOR-
SINGLE-PHASE OPERATION IN MANNER INDICATED
motors connected in series with shafts mechanically connected,

one of which has itsphase sequencie reversed.
Figs. 6, 7 and 8 show characteristic curves of a three-phase
induction motor operating respectively on a symmetrical cir-

cuit, according to Case I and according to Case II.
Synchronous Machinery
THE SYMMETRICAL THREE-PHASE GENERATOR OPERATING ON
UNSYMMETRICALLY LOADED CIRCUIT
The polyphase salient pole generator is not strictly a symmetri-
cal maichine, the exciting winding is not a symmetrical polyphase
winding and it therefore sets up unsymmetrical trains of har-
monics in exactly the same way as they are set up in an induction
motor with unsymmetrical secondary winding. These cases will
therefore be taken up later on. A three-phase generator may
however be wound with a distributed polyphase winding to serve
both as exciting and damper winding and if properly connected
will be perfectly symmetrical. Such a machine will differ from
an induction motor only in respect to the fact that it operates
in synchronism and has internally generated symmetrical e. m. fs.
which we will denote by S1 Ea., S2 Ea2 the negative phase se-
quence component being zero; an e. m. f. SO EaO may exist but
since in all the connections that will be considered there will be
no neutral connection its value may be ignored. If the load
impedances be Za, Zb' and ZJ' they may be expressed by
Z,al = S Zao' + S' Zal' + S2 Za2'
and the equations of the generator will be
SE.a, St { (Ra + La dt Ial' + Za0' Ial'
+ Za2' 1a2 +M cd't E' 'Iu1 }
05= S2 {f(Ra + La t)/tIa2' + ZaO' Ia2* (158)

+ Zall lalt + M dE -jwot IU2 }
o = (Ru + Lu d u)
IUl + M d -Jwot Iala
O (Ru + Lu t) I2' + Mdt EtIa2'

The last two equations give

d
I1' - d t
C-jwot la 1
Ru + Lu dt dt
d (159)
dt
Iu2 - d eiwot 'a2'
RU + Lu dt
which on substitution in the first two equations of (158) give
the equations
Ra+ La d d
M d t (dd-
t jj o)oj
al1I
{ RU LdUy
+ (L d t o) J I
+ Zaol Ial' + Za2' Ia2' Eal
(160)
Zal Iali + Ra +La
dt dt
Al2 d d jWA
_ dt ( dGt +Jw) a2 ZaO' 1a20= 0
or if
Eal = Eal Eju"'o (161)
the impedances ZaO, Zal, Za2 become ordinary impedance for an
electrical angular velocity wo and equations (160) become
(Ra + j wLa + ZaO) Ial' + Za2' Ia'2 = EaI
Zal' fal'+ {ZaO'+ (Ra+ K22 RU) +j 2 Wo (La - K22 Lu)- (162)
K22 Ru} Ia2
K 0 J
It is apparent that in the generator the impedances

Ra + j wo La = Z1'
and { (Ra + K22 Ru) +j 2 wo (La- K22 Lu) - K2RuI = Z2'
take the place of Z1 and Z2 in the symmetrical induction motor

operating on an unsymmetrical circuit, and we may express
equation (162)
(ZaO' + Z1') 'al' + Za2' 'a2 = Eal
(163)
Za' 'al' + (ZaO' + Z2') 'a2 0
which gives
Zal' I
a2 = -
Ial
ZaO + Z2
Eal
'al'
(ZaO' + Z1') z
Za' +Za2'
Z21
Or in more symmetrical form
(Za0'
lal (ZaO' + Z1') (ZaO' + Z2')
Z2') Zal' Za2 (164)
'a2
Zal'+
(ZaO' + Z1') (ZaO' Z2')- Zal Za2' al
From (159) we have for the damper currents

1 = OifR > 0
'u = K2 1a2 Ej2uot (165)
2 M
where K2=j . +wo
Ru= j 2 wo Lu
A particular case of interest is when the load is a Synchronous
Motor or Induction Motor with unsymmetrical line impedances in
series-Equation (163) becomes
(ZaO' + Z1' + Z1) 'al' + Za2' Ia2 = Ea2
Zal Ial' + (ZaO' + Z2' + Z2) 1a2' = 0
Za0' + Z2' + Z2
a1 (ZaO'+Zl'+Zl) (ZaO'+Z2'+Z2)- Zal Za2 Ela (166)
Zal
1a2'= (ZaO'+Z1'+Z1) (ZaO'+Z2' Z2) - Zal Za2 a
An important case is that of a generator feeding into a symmetrical
motor and an unsymmetrical load. Let the motor currents be
1078 FOR TESCUE: SYMMETRICAL CO-ORDINATES [June 28
Ia, Ib, fc, those of the load la,, Ib,, 1,' and the load impedances
Za', Zb', Zr'. The equations of this system will be
1 Eal = S' {Z1' (fal + lal') + ZaO' 'al' + Za2' 1a2'
Si Eal = S' {Zi1 (Iai + fal) + Zi fI
al
S2 0 = S2 {Z2' (Ia2 + fa2l) + ZaO' 1a2 + Zal' Iail (

S2 0 = S2 {Z2' (Ia2 + Ia2') + Z2 Ia2} )
Or, omitting the sequence symbols and re-arranging-
Eal = Zi' lal + (Z1' + ZaO') fal' + Za2' Ia2
Eal = (Z1' + Z1) fal + Zl' Iall
(168)
O = Z2' fa2 + Zal' fall + (Z2' + ZaO') fa2'
O = (Z2' + Z2)fa2 + Z2'fa2'
These equations can be further simplified as follows:

o = (Z2' + Z2) a2 + Z2'Ia2'
O = - Z2 Ia2 + Zal' Ial' + ZaO' 'a2
(169)
O =- Z fal + ZaO' IaO' + Za2' Ia2'
Eal = (Zi'+ Zi) fal + Zl'fal'
A set of simultaneous equations which may be easily solved.
THE SINGLE-PHASE GENERATOR IS AN IMPORTANT CASE OF THE

THREE-PHASE GENERATOR OPERATED ON AN UNBALANCED LOAD
Let the impedance of the single-phase load be Z and let us
suppose it to be made up of three star connected impedances
Za'-= 3 Z,c + -
z
Zb' =
2
z
2c
=
the value of ZX in the limit being infinity. Then we have

Zao' = Zx + Z2 (170)
Zal' = ZC
Z a2' = Z_x
Equation (164) in the limit when Zx becomes infinite reduces
to
1al Z + Zl' + Z2'
I'a2 =
Ia2 Eai + Z2'
Z + Zl' (171)
The single-phase load being across the phase B C, the single-
phase current I will therefore be equal to IC or
I j -\/3 E-al
1
Z + fZl' + Z2' 1 ~~~~~~~~~~(172)
(12
Z +Zl' +Z2'J
IU1 = 0 if RU > 0
1u2 =-j K2 I Eju'ot
jK2 (173)
= - _jK 2 I Cj2u-ot
1u2IU=
A/3 J
Iu2 is double normal frequency
PI + j QI = 312 Z 1
PL + j QL = 3 I2 (Z1I + Z2) (174)
(P + j Q) + (PH + J QH) 3 Eb (f ++I)
In the case of the generally unbalanced three-phase load
P1 + j Q1 = 3 {(Iai2 + Ia22) ZaO'
+ lal fa2 Za2' + fal Ia2 Zal'}
PL + J QL = 3 { Iai2 Zl1 + Ia22 Z2'} (175)
(P +J Q) + (PH + j QH) 3 Eal (1a2 + fa2)
1080 FOR TESCUE: SYMMETRICAL CO-ORDINATES [June28
When the generator has harmonics in its wave form equations

(162) must be written
(Ra+ jwLa+Zai') Ia1'+Za2' Ia2' = Eai
Zal'Ial'+ {ZaO'+ (Ra + K2j2Ru) (176)
+ j 2 w (La -K22 LU)-2la2 R,, I Ia2' =
a2J
Where Eal is finite, Ea2 is zero and vice versa, the frequencies
being different in each case, we have therefore a solution for each
frequency depending on the phase and amplitude and phase se-
quence of the e. m. f. of this frequency generated. Of course
the values of Z1' and Z2' change with each frequency on account
of the change in the reactance with frequency, and a value must
be taken for w conforming with the frequency of the harmonic
under consideration.
Symmetrical Synchronous Motor, Synchronous Condenser, Etc.
As in the case of the generator, the synchronous motor has two
impedances, one to the positive phase sequence current of a
given frequency and the other to the negative phase sequence
current of the same frequency. But, since there is no quantity
in the positive phase sequence impedance corresponding to the
virtual resistance which indicates mechanical work in an induc-
tion motor, its equivalent is furnished by the excitation of the
field. Let us denote the e. m. f. due to the field excitation by
S Eall assuming it to be for the present a simple harmonic three-
phase system. Let PO be the output of the motor which will
include the windage and iron losses assumed to be constant.
Then for the synchronous motor on a balanced circuit of e. m. f.
S' Pal we have
S' Eal = S' {Ialr(Rat + j w La') + Eal)} (177)
S Sal faI = So { Ia12 (Ra' + j it La') + Q0 (178)

Where Qo is the imaginary part of the product, Pall fal. (178)
reduces to
Eal Ial cos a Iai2 Ra' + P0
=
3 (179)
Where cos a is the required operating power factor. Solving
for lai
IT I al=Eai,Q 2 cos a 1 4 1- 4 Ra' Po } (10

(180)
CS
Ra' cos2 at}
i\ 1- 33 Eal'2CS
= cos a 4 Rai' P0
Ia = Ea' 2 Rai 1 R 1- 3Ea12 cos2 a }
(cos a - j sin a) (181)
The apparent impedance of the motor is
2 R, sec 4o (cos a + j sin a) (182)
1 4- 1- 4 Po
3 EaV2 cos2 a
and
Eali Eal [1-22_ {i 1 1V 3 Eai2 C052

4Ra'P3Ea1os a
}
(cos a - j sin a) (Ra' + j w La')] (183)
The same equations apply to the case of the synchronous

condenser with the difference that the mechanical work is that
required to overcome the iron and windage losses only.
If we take
E,1 = Eal (cos a + j sin a) EjWoI = (A 1 + j Bi) EjWtt
(184)
Ea01 = (A1' + j B1') 'jwor }
we have
la 1 =
-2 Ra I
41 1- 3 3A 12 etn (186)
A1l = A21 (14'V'i 4RA Po) Ejwot (186)
Bit = {Bi- j La/R AI (1 4- 1- 34 Po

,) cjwo
(187)
Since a may be a positive or negative angle, the sine may be
positive or negative for a positive cosine, and therefore the power
factor will be leading or lagging accordingly as B1 is negative or
positive respectively. The double signs throughout are due to
the fact that for any given load and power factor there are always
two theoretically possible running conditions. However, since
we are concerned only with that one which will give the max.
operating efficiency, that is the condition that gives Ial the lesser
value, for a given value of P0 the equations may be written
2 (1(i - ju0ot
4 Ra Po
'al 2 Ral V 3 A i2 6~"
A,'l = 1 (1 + 2Vii- 41 ) EjiOt (188)
jwoL_ ('Al 4 Ra'P
And corresponding values for (180), (181), (182) and (183) may
be obtained by omitting the positive sign in these equations.
Another condition of operation is obtained by inspection of
(180), due to the fact that Ial must be a real quantity
34 Ra' PO
3 Ea12 cos2 a
must be > 1 (189)
this is the condition of stability. In terms of (184) it becomes

433RA'
A~
Po must be > 1 (190)
The same conditions apply to the synchronous condenser, the

total mechanical load in this case being the iron loss and windage
and friction losses.
Proceeding now to operation with unbalanced circuits having
sine waves the motor also having a sine wave. In addition to
equation (177) we shall have
52 Ea2 = S2 Z2' la2 (191)
The mechanical power delivered through the operation of this
negative phase sequence e. m. f. is given by PN where
PN = 3 1a22 R2 (192)
this quantity must therefore be subtracted from the value of P0

in all the equations in which Po appears when unbalanced cir-
cuits are used in connection with equations (177) to (190) inclu-
sive. These equations, however, give the conditions for main-
taining a given mechanical load and a given power factor in the
positive phase sequence component, but in practise what is re-
quired is the combined power factor of the whole system, or the
conditions to give a certain combined factor while delivering a
given mechanical load; this may be obtained as follows:
The negative phase sequence component is a perfectly definite
impedance and is independent of the load, and therefore the zero
frequency part of the product Ea2 1a2 may be set down as
Ea2 Ia2= P2 __
3 +J Q3 (193)
we have also for the positive phase sequence power delivered
(Al + j Bi) Iai = Ia12 Ra' + -P

3
_Pt;3
+ j (w Ial La' + Bll) Ial (194)
And the power factor is given by cos a, where
Ia Bi+ Q2
3
tan a = (195)
IaiAj + 3
2
From (194) we have

P
Al Ia, =Iai2 Ral + 3 - 3 (196)
B1 = w Ial La' + Bl' (197)
A12 + B12 = Ea12 (198)
The simplest method of solving these equations is by means
of curves. Taking arbitrary values of lal, B1 and A1 are chosen
Po
consistent with (198) so as to satisfy (195),
3 Al,
and B1' are
then obtained from (196) and (197). If there are harmonics in
the impressed e. m. f. but there are none in the wave form of the
machine, the machine will have a definite impedance to the
positive and negative phase sequence components of each har-
monic, so that there will be a definite amount of mechanical
work contributed by each harmonic which must be subtracted
from the total work to be done to give the amount of work con-
tributed by the positive phase sequence fundamental component,
the equations will be identical to (193), (194), (195), (196),
(197) and (198), if we take PN to mean the total mechanical

work done by the harmonics both positive and negative phase
sequence and P2 and Q2 to represent the products
T, (nEal nfal + nEa2 nfa2)
the zero frequency part only being taken into account.
When harmonics are present both in the impressed wave and
in the generated wave, the problem becomes too complicated to
treat generally, but specific cases can be worked out without
much difficulty.
Phase Converters and Balancers
The phase converter is a machine to transform energy from
single-phase or pulsating form to polyphase or non-pulsating
form or vice versa to transform energy from polyphase to single-
phase. The transformation may not be complete, that is to say,
the polyphase system may not be perfectly balanced when sup-
plied from a single-phase source through the medium of a phase
converter. Phase converters may be roughly divided into two
classes, namely-shunt type and series type.
INDUCTION MOTOR OR SYNCHRONOUS CONDENSER OPERATING
AS A PHASE CONVERTER OF THE SHUNT TYPE TO SUPPLY A
SYMMETRICAL INDUCTION MOTOR OR SYNCHRONOUS
MOTOR
Let Z, and Z2 be the positive and negative phase sequence
impedances of the motor, Z1', Z2' those of the phase converter.
Let SI ELa and S2 Ea2 be the positive and negative phase sequence
components of the star e. m. f. impressed on the motor as a result
of the operation. The single-phase supply will be one side of the
delta e. m. f. S Eib, which has positive and negative phase se-
quence components SI Eb,l and S Eb,2 the single-phase supply
being Eb= Ebcl + Ebc2.
The value of Z2' may be considered fixed for all practical pur-
poses and since in the induction motor phase converter the speed
is practically no-load speed, Z' is practically the no-load imped-
ance plus a real part obtained by increasing the real part of the
no-load impedance by the ratio of the normal no-load losses to
these same losses plus a the secondary losses due to the phase
converter currents. The latter may be calculated roughly as
even a large error in its value will have an inappreciable effect
on the actual results. We have therefore
S' Eal = S'j 1
52 bc2= Sk2i
Lla2 ( (199)
Si' Iai' = - S1 Ebc 1 (200)
S2 Ia2= S2 Lb)
In the common lead of motor and converter we have

ESi 2 E1c 1
Iai'lat
+ 1a2'S'+ Iai Ia2 = 0
V- Z, (202)
or, substituting from (200) and (201)
Sb2 = Lgb 1( z +
zt+ -i-) PU2 Z ) (203)
(S2l 2)
2
1 + 1
EbC, Z + 2 (204)
1 1
b4c1 = Zb
+ Z2 (201)
1 + 1
i;bc2 + Eb\ (206)
A
= 1 1
which give the complete solution for all the quantities required
with the aid of equations (200) and (201). For the supply
current 1
I = rbcl + Ibc2 + Ibcl + Ibc2 1
S Ibc = S b1c1 + S2 Ibc2 (207)

S Eb, = S1 Eb,l + S2Ebc2 )
P1 + i Ql = Ebc f (208)
In order to obtain a perfect balance we may consider the addi-
tion of an e. m. f. S2 j E,2.
-E= in series with the phase converter
N/3
whose value must be a function of the load and the phase con-
verter impedances, and therefore equation (201) will be replaced
by
(=S i
S2 Ia2=S2 + VZ) (209)
S2 la2 = S2 j Eb,
and since the balance is perfect Ebc2 is zero, and therefore
2 3j - S2 Z21Ia2l (210)
An e. m. f. equal and of opposite phase to the negative phase
sequence drop through the phase converter is required to pro-
duce a perfect balance.
Carrying out the solution in the same manner as in the imper-
fect converter, we obtain
1 1 1
7__
Ebc2 = 1
77
1 Eb. 1 (211)
17 _7
Z2 Z2 Z2
and since Eb.2 is zero and Ebcl = Ebc the single-phase impressed
e. m. f., we obtain
Ex2 Z2 ( zi += 1z1) Ebc (212)
and therefore from (210)
52 Ia2 = S2 j + Z1') 1
L (213)
S' Ial = - S'j (214)

S2 la2 = 0 (215)
SlIal = - S'j V/3Z (216)
Figs. 9, 10, 11 and 12 are vector diagrams of some of the princi-
pal compensated shunt-type phase converters. There will be no
difficulty in following out these diagrams if the principles of this
paper have been grasped.
Vp
Eai
2
A
/c Eca \ai bi
FIG. 9-VECTOR DIAGRAM OF SHUNT-TYPE PHASE CONVERTER OPERATED

FROM TRANSFORMER So AS To DELIVER BALANCED CURRENTS
Terminal voltages of phase converter SE'a
Terminal voltages of motor SlEal
Negative phase sequence e.m.fs. in phase converter S2 (OA2)
The Phase Balancer is a device to maintain symmetry of

e. m. fs. at a given point in a polyphase system. It may consist
of an induction motor or synchronous condenser with an auxiliary
machine connected in series to supply an e. m. f. always pro-
portional to the product of the negative phase sequence current
passing through the machine and the negative phase sequence
impedance of the balancer. It therefore has the effect of an-
nulling the impedance of the machine to the flow of negative
phase sequence current. Thus, in a symmetrical polyphase
network, where we have an unbalanced system of currents due to
certain conditions
S Ia - S' Ial + S2 Ia2 (217)
If a balancer be placed at the proper point the component S2 1a2

will circulate between the loads and the phase balancer, the other
component SI Ial being furnished from the power house. On the
other hand, if there be a dissymmetry in the impedance of the
system up to the phase balancer, the latter will draw a negative
phase sequence current sufficient to counteract the unbalance
due to any symmetrical load by causing the proper amount of
negative phase sequence current to flow to produce a balance.
The balancer may be made inherently self-balancing by insert-
ing in series with it a machine which is self-exciting and is able
Ea
ft,a a
<~~~~~~~- it
E'c~~~~~~~~~~~E
FIG. 10-VECTOR DIAGRAM SHOWING RELATIONS BETWEEN MOTOR
TERMINAL E.M.F'S., CONVERTER TERMINAL E.M.FS., AND SYMMETRICAL
GENERATED E.M.F'S., SAME CONNECTION AS FOR FIG. 9.
Negative phase sequence drops in phase converter S2Z2' 1
Conjugate positive phase sequence e.m.fs. S1(ABC)
to furnish an e. m. f. equal to the negative phase sequence imped-

ance drop. The combination thus has zero impedance to nega-
tive phase sequence currents. If in the neighborhood of a phase
balancer the loads have impedances
S Za = So ZaO + Sl Zal + S2 Za2
The equations of the system are
S1 E., = S1 ZaO lal + S1 Za2 1a2
i (218)
S2E2= 0 =S2ZaOIa2 + S2Zali Ia
The currents in the phase balancer are

SI2 Ia2 and SI Eai
z,
The solution of (218) gives S2 1a2 and S la, the former of
which are the phase balancer currents. The solution is
-la I =
ZaO
aal
Zao 2 - Zal Za2 (219)
-la2 --2 ZaO2 - Zal

=
Zal Za2
E
Eall J
E'a ai
o'S t~C
OA,
Aic
CB6C Dr
2'2
D B b
FIG. 11-VECTOR DIAGRAM OF SHUNT TYPE PHASE CONVERTER SCOTT

CONNECTED WITH COMPENSATION"BY TRANSFORMER TAPS
Terminal voltages of converter D'A and B'C'
Terminal voltages of motor S'Eal
The phase balancer is a voltage balancer and will maintain

balanced e. m. f. for any condition of impedance, and if the im-
pedance of the mains is unsymmetrical it will draw a sufficient
amount of wattless negative phase sequence current through
these mains to produce an e. m. f. balance at its terminals.
Hence the complete solution requires consideration of all the
connections in the network between the supply point and the
balancer. Two equations for each mesh and connection are
required, one of the positive phase sequence e. m. fs. and the
other of the negative phase sequence e. m. f., and these equations
may be solved in the usual way.
Series Phase Converter. In discussing the various reactions in
rotating machines we have made use of the terms "positive phase
sequence impedance" and "negative phase sequence imped-

ance." These terms are definite enough when dealing with rela-
tions between machines whose generated e. m. fs. all have the
same phase sequence, but require further definition when we
are dealing with relations between machines whose e. m. fs. have
different phase sequence. We shall retain the symbols Z1 and
Z2 for the values of the positive and negative phase sequence
impedances, depending upon the sequence symbo' S to define
whether these impedances apply to a negative or positive phase
sequence current. Thus, the phase sequence of the currents and
E'a
Ea2
E' Eb1
<
FIG. 12-VECTOR DIAGRAM OF SHUNT-TYPE PHASE CONVERTER WITH

AUXILIARY ROTATING COMPENSATOR TO EFFECT A PERFECT BALANCE
Terminal voltages of phase converter S Eat
Terminal voltages of motor S'Eai
Terminal voltages of compensator S2Ea2
e. m. f. will be defined by the apparatus supplying and receiving

power and the impedances of the transmitting apparatus will be
defined in relation to these currents. As an example a motor
series connected in counter phase sequence relation in a circuit
and driven in a positive direction will have impedances
positive phase sequence Z2
(220)
negative -phase sequence Z1
Where an auxiliary machine is defined as being of negative
phase sequence relation to other machines, it will have imped-
ances as given above to the positive and negative phase sequence

currents passing through the other machines.
A single-phase transformer winding tapped at the middle point
may be regarded as an unbalanced three-phase system where
Ea = 0, E = E,s Ec =-E
2 A. being the single-phase e. m. f The system may be repre-
sented by the equation
S Ea = SI Eal S2 Aa2
Eal
a2
FIG. 13-VECTOR DIAGRAM OF SERIES-TYPE CONVERTER.

No LOAD E.M.F'S. ACROSS MOTOR TERMINALS SlEal
No LOAD E.M.F'S. ACROSS CONVERTER TERMINALS S2Ea2
SINGLE-PHASE E.M.F'S. 2E
E.M.F.ACRoSS TERMINAL OF MOTOR UNDER LOAD EaEbEc
E.M.F. ACROSS TERMINAL OF CONVERTER UNDER LOAD E'aE'bE'c
where Eal = (
4 ~~~~~~(223)
Ea2 i\3
-.
= -j
If, therefore between the single-phase source of power and

the load we interpose a polyphase machine with e. m. f. - S2
Aa2, we shall have at the load terminals the e. m. f. SI Eal.
If we use an induction-type phase converter it will have imped-
ances to motor currents as follows
To positive phase sequence Z2'
(222)
To negative phase sequence Z1'
we therefore have the relations

St Eal = S' Ial (Z1 + Z21) (223)
S2 Ea2 S2 1a2 (Z2 + Zl') (224)
If the converter is doing no mechanical work, Zl' is large com-
pared with Z2' or Z2, and therefore the component of negative
phase sequence is small in the motor. The value of Zl' depends
upon the slip of the phase converter which will depend on the
mechanical load it carries as well as on the load carried by the
motors. Approximately the load currents due to the motors
produce the equivalent at the phase converter of a mechanical
load equal to one-half the rotor loss of the phase converter due
to these load currents. Substituting the values given in (221)
for SI Eai and S2 Ea2; we obtain
R
S'Ji V3 = S lIai (Zl + Z21)
(225)
-S2 j = S2 la2 (Z2 + Z1')
S'lal = Si'J
SV/3 (Z1 + Z21)
S2 1a2 = - S2 j _\3_(2E5
If instead of an induction-type phase converter a sy ichronous

phase converter is used an e. m. f. of negative phase sequence S2 La2
the generated e. m. f. of the phase converter must be introduced
in equations (224) and (225) and the value and phase of these
e. m. fs. will depend upon the load on the phase converter shaft as
well as the load carried by the motors. The equations will be
S1 Eal = S' Ial (Z1 + Z2') (227)
S2E2 = S2 Ia2 (Z2 + Z1') + S2 Ea2r (228)
or
51]i vz5 =S-' ial (Zl + Z2)
S2j 3X = S2 Ia2 (Z2 + Zl) + S(229)
-
-\2] 3
Sa2 (Z2 + Zl') + S2 A2'
19181
Let
then (230) becomes

E
~~~~~3
FORTESCUE: SYMMETRICAL CO-ORDINA TES
The last member of equations (229) is the equation of a syn-

chronous condenser. Assuming its windage, iron loss and in-
creased losses due to secondary reactions to be P0, we have by
equation (160) of the Section on Synchronous Motors
E Ia2 cOs a
1a2
= 1a22 (R2 + R1') +
= a2 + j b2
a2 = (a22 + b22) (R2 + R11) +

-P
PO
1093
(230)
(231)
(232)
Of the two quantities a2 and b2, b2 alone is arbitrary and depends

upon the excitation, a2 will depend upon the value of b2 and also
upon the losses. Solving therefore for a2 in terms of b2, we have
a2=
a=/3
2 V (R2 + R1')
'Vii - 4 (R2 + R11) {3 b22 (R2 + R1') + P (233)

E82
Since b2 is arbitrary we may now determine cos a2 =
a2 and the value of Ia2 in terms of the impressed e. m. f.
V/a22 + b22
will be by (181) of Section on Synchronous Motors
82la S2[.
S27a2-52
2
Es
L\/3 cosC2
2 (R2 + Rl') {
- 4
EF82(R2cos2
+ Rl') Po } Eja]
a12 J (234)
The effective value of 1a2 in terms of the effective value of A. will

then be
Ia2 E= cs
2 a 4R1V12F+ 1')Po
(R2 + Rl'){
1a32
E.2 cos2 a23}
(235)
and since the component of the e.m.f. generated in phase with the
current is determined only by the magnitude of 1a2 and the
motor losses, if we define its value by A2' the quadrature
component being B2' we shall have
A
E. COS a2 (1 V1-4 (R2~+ I?l') P" 26
Ea 2 E.2 COS2 0X2'Po
and
B2'12 - E.
-. sin a2 w (L2+ L21)
~~~~A21 (237)
E. sin a2 +
3 w (L2+ Li') cos a2 (

PO 2 .(R2 + Rl') 1
\ 4 (R2 + Rl')Po) (238)

and therefore we have
EJ rCOS 44(R2 +Rl) PoX

[cos2OZ2 (i+V 1
-
E2 I -- A3 E2COS2 a,)
j L si 2 E 3 a(2+ +R,) 2(]_

3 w (L2, + Li') cos a,2
sin a, + 2 Po (R2 + Ri')\
-4 (R2F.2+cos2
Rl') PO'lw)(29
a (ejw) (239)
The impedance of the phase converter to the flow of negative

phase sequence current is
2 (R2 + Rl') sec a
_ 1 _4 (R2 + Rl') Po
1- (240)
The balance will be at its best when Ia2 is a minimum with

cos a2 as the independent variable. This will be the case when
cos a2 is unity; that is to say when b2 is zero.
Recapitulating the results given above, we have for the general

case taking the single-phase e. m. f. E. as reference
Ilal S'j
S g3 (Z1 + Z2')
(24A,
(241)
S Ia2 -j (a2 +; b2) (242)
where b2 is arbitrary and
Es
a2 2V32(R2+ R1') {
/iV 4 (R2+ R 1') b22(R2+ R) + Po)
~E's (243)
Since b2 is arbitrary cos a2 is determined by
cos a2 a22a2+ b22 (244)
we may express la2 in terms of A. by

Ia2 - -S2iE 2 (R2+ R1) { 1-
_ 4 (R2 +R i) Po } ea2 (245)

E.2 COS2 a2
The effective value of 1a2 will be
G2 =VNa? + b22- Es cos a2

V/3 2 (R2+ R1')
1 _ 4 (R2+ R') Po }
E.2 cos2 a2 (246)
If A2' and B2' are components of Ea2' these being the generated
e. m. f. in phase and in quadrature with the current 1a2 we shall
have
Ea2= -j (A2' + j B2') (247)
and A 2' and B2' will have the following values
A2' = E. cosa2 /A4 4(R2+Ri') P\

2E)9 (248)
B2- E- {sinCa2+
3 w (L2+ Li') cosa2 /
Po 2(R2+RI')
V R2 os2 Po) } (249)

and Ea2' expressed in terms of EJ becomes
F2~~EN/-'/E
E2/~ [ coo
2
1r + o\1
(1a'i E2 COS2
4 (R2+ Ri')Po_
E. csa2
a
.s .
7151na2-[+
3 w (L2 + L') cosa2/2
2 P0 (R2 + R1')
V\/l E,2 + Rif)P }] (250)

The effective impedance of the phase converter to the flow of
negative phase sequence currents is
2 (R2 + R1') sec a2 . .
/ (R2~4 E82+R1')Po
COS2 a
(cos a2- J sln2) (251)
or
E.2 cosa2 (1 + V 4 (R2 + R1') Po -a2 (252)
Po 2 \ E82 cos2 a2J
In the above equations cos a2 is arbitrary or b2 may be con-
sidered arbitrary and cos a2 will then be determined.
Minimum Unbalance is obtained when cos a2 is made unity or
when b2 is made zero in equations (241) and (252).
Perfect Balance is obtained by driving the phase converter
mechanically so as to supply the mechanical power Po from a
separate or symmetrical source. Under this condition a2 and b2
both become zero when cos a2 iS unity. The only equation of
the system is then (241).
Currents and Power Factor in the Single-Phase Supply Circuit
of Series Phase Converter.
The e. m. f. is 2 F8 and the current supplied is
lbI
_b IbI2
lb
____ c_ I lb 2 -
1'c2 (253)
2 2
If we take
SIla, =Slj (al-j b1) (254)
Ib, - IC , - (a, - j b1) (255)
2 2
Similarly, since under the same conditions
S2 1a2 =-S2j (a2 + j b2) (256)
1b2 -t = 2/3 (a2 + j b2) (257)
22
and therefore
N/3_ {(al + a2)-j (bi-b2)}
IJ =2 (258)
where al, b1, a2, b2 are to be obtained by means of equations
(243) to (254). The single-phase power factor is given by
tan 0 = (259)
a, + a2
of these quantities a2 is usually the smallest and its value may be
obtained approximately by assigning to b2 a value which will
make the ratio b -b2 equal to tan 0, and obtaining the
a,
corresponding value of a2 by (242), the value of b2 may then be
recalculated from (259) by substituting the tentative value ob-
tained for a2. This procedure may be repeated until sufficient
accuracy has been obtained.
Single-Phase Power Factor in Shunt-Type Phase Converter.
The simplest procedure is to obtain a curve of admittances
for varying excitation of the converter and plot the power factor
obtained by varying the admittance with a fixed load. The true
and wattless power is obtained easily by means of (208) whether

the system is balanced or unbalanced.
Figs. 14, 15, 16 and 17 are vector diagrams of several
methods of using phase converters to supply a balanced 3-phase
e. m. f. to a symmetrical load such as an induction motor. The
diagrams are all based on a main machine having the same nega-
tive phase sequence impedance and the system in each case is
A
Ea
%I
a2 c
B'B
Converter Motor A
C' C_
Bt B~~~~~~
FIG. 14
SINGLE -PHASE IMPRESSED E.M.F. = B'C'
MOTOR E.M.F. = BC
NEGATIVE PHASE SEQUENCE E.M.FS. Ea2Eb2EC2
CONJUGATE POSITIVE PHASE SEQUENCE E.M.FS. AaAEAo
PHASE CONVERTER TERMINAL E.M.F. AB'C'
delivering the same amount of power at the same voltage and 3-

phase power factor without supplying any wattless power. It
will be noted that the scheme Fig. 14 has the lowest single-
phase power factor, Fig. 16 the highest and the rest arcing alike.
It may be remarked, however, that with the shunt-type schemes
adjustments can be made for power factor correction which will
result also in better regulation.
1918]
.~ ~
FOR TESC UE: S YMMETRICAL CO-ORDINA TES
APPENDIX I
Cylindrical Fields in Fourier Harmonics
When we have a diametrical coil around a cylinder concentric
1099
with another cylinder which forms the return magnetic path, and
the length of the gap is uniform and the coil dimensions are very
small, the field across the gap takes the form of a square topped
t Phase
A
Single
Phase
Motor A!
FIG. 15
SINGLE-PHASE IMPRESSED E.M.F. = B'C'
MOTOR E.M.F. = BC
PHASE CONVERTER E.M.F. = C
NEGATIVE PHASE SEQUENCE E.M.F Ea2Eb2Ec2
CONIUGATE POSITIVE PHASE SE(UENCE E.M.F. Ea1Ebl:cl
PHASE CONVERTER TERMINAL E.M.F. AB C
wave, which may be expressed in the form of a Fourier series

with the plane of symmetry of the coil as reference plane, and its
Fourier expansion is
(S
4B
(CO 8 3COS 3O@+
Scos SG -. . + ' . ) (1)
where .] is the avverage induction in the avir gap.
1100 FORTESCUE: SYMMETRICAL CO-ORDINA TES (June 28
al
-A
E b2
B'
%8 SPhase Auxiliary Motor B

nMettrBarnce
A A
5C''jc
FIG. 16-PHASE CONVERTER WITH AUXILIARY BALANCER.
/"
0 %> cOPhase Motor

0 A
y B' B
FIG. 17
SINGLE-PHASE IMPRESSED E.M.F. = XY
MOTOR E.M.F. = ABC
THERE IS A 2 TO 1 TRANSFORMATION OF E.M.F. FROM SINGLE-PHASI
TO THREE-PHASE IN THIS CONNECTION
With pitch less than ir the curve will have a different form, the
amplitude being greater on one side of the plane of the coils than
on the other, the areas of each wave will remain the same and
second harmonic terms will appear. Let 2 mo r be the new pitch
then the average amplitude of the induction will be the same as
before, namely B, and the value on one side of the coil will be
2 (1 - mo)B and on the other side 2 mo B so that the total flux
will be the same on either side. To obtain the values of the
coefficients we have
mor 2w
2 (1- moBfcosn Ed 0 + 2 moB cosn Od 0 = 2 An

0 mowr
moT 2w
2 (1- mo) B [ LsinnO - 2moB[ n sinno] =-2 A,

0 molw
4B { (1-mo) + mO sin n MO 7r }
4 B /1.
An = 4 n- sin n mo0r1J (2)
Let 2 mo0 = 22r, then (1 - mo) 7r i2r and

(os 0 +
CB cos 2 - cos 4 0 - 1cos 50
1
+ cos7 0 + 8cos8- Cos 10O. . .) (3)
A general expression for 63 where B is the average of the posi-
tive and negative, maximum value for any pitch coil would be
4B i i .
3=
r (-sin
n
nmO r cos n (4)
and includes all possible coil pitches. If the number of teeth in
a pole pitch be n,; in addition to the average induction as in-
dicated by (4), there will also be a tooth ripple of flux, the maxi-
mum value of which will depend upon the average value of the
induction at each point. The value of mo must be a fraction
having n, as denominator and an integral numerator. The
1102 FORTESCUE: SYMMETRICAL CO-ORDINA TES [Tune 28
value of the integral numerator is therefore always mo n. The
correct value for the max. induction will therefore be
(PBM= fJ4B I) .
(-sin n MO 7r COS n O (1-

1
(_ 1)monT KT cos nTO) (5)

where KT is the ratio of the average to the min. air gap. "mo"
must always be chosen so that mo n, is an integer.
If the length of the average effective air gap in centimeters
be d the value of B is given by
47r I N
0_ 2d gauss
where I is the maximum value of the current in the coil and N
is the number of turns. If d is given in inches we may write
B = 2d X 2.54 maxwells per square inch.
If we integrate (5) between the limits (0 - mo r) and
(0 + mo r) we shall have the total flux Sp through the coil
0+mo r
z -B sinnmo7rcosn0 d9
uSi
0-mo wr
7r 0+mo wrni
--(-l) 1~nr (_:( sinn mO cos n )Krcsn dO

0-m u
= 4 Bre K- sin n mo 7r sin n0

L Ij0-mow
0+mo w
-4Br I ( 'IT KT 2
1 sin
mo n r
sin (n-n7) 0
0-mo wr
sin (n + nT (6)
2 (n + n,)
The second expression is zero for all values of 0 which are
integral multiples of the tooth pitch angle, so long as mO n is
also an integer and therefore it is zero for all mutual inductive
relations of similar coils on a symmetrical toothed core we there-
fore have:
The induction through a coil displaced an angle 0 from the axis

of a similar coil carrying a current giving a mean induction B both
coils being wound on the same symmetrical toothed core is
_ 8Brl 1(12sin2 n mo 7r cos n 0) (7)
The second term in equation (6) also becomes zero when n,
becomes infinite independent of the value of 0. We may there-
fore safely make use of an imaginary uniformly distributed wind-
ing when considering self and mutual impedances. It will also
be shown later on, that with certain groupings of windings the
second term may be reduced to zero for every value of 0.
If N1 be the total number of complete loops in one complete
pole pitch, we may take 2N,ras the density of winding per unit
angle of the complete pole pitch. The mutual induction per
turn in a coil angularly displaced an angle 0 from another coil
N1
N
of winding density 2 with an effective total air gap 2 d and
with windings subtending an angle 2 ml r is given by
+mi 7r
ml 8N rd ( 1N 2 sin2n m 7r cos n(60-+j0') d O' henrys

-m r (8)
lO9drd 3 sin2 n mO r [sin n (+60)']__ henrys

16 N1 r I~ 16h
M1= 109 ird (3 sin2nmo rsin nm17rcosn 0 henrys (9)
Next, if the loop of which M1 is the mutual inductance is part of a
winding having distribution density of winding 2 2 and sub-
tending an angle 2 m2 r its mutual inductance with the other
winding will be
m2
M12 12 8N1N2r1
101 w2d J I n31 sin n mO r sin n mi 7r
-m2 7 cos n (6 + 6') d O' henry (10)
8NlN2r1 1
= O19ir2d - sin2 n mO 7r sin n mi wr
O' =m2 7
[sin n (6 + 6')J henrys
O' =-mt r
M12 = 16 N101 N21 1 sin2 n mO 7r sin n ml w

2snsnnm
sin n m2 7r cos n ) henrys (11)
This is the general expression for the mutual inductance be-
tween two groups of connected coils of like form on the same
cylindrical core. It should be noted how much the harmonics
have been reduced due to grouping.
When the coils are not of like design as in the case of a rotor
and stator and the pitch of the coils is different in one from the
other, sin n mo ir will not appear twice in the equation but one
of its values must be replaced by sin n m. 7r where 2 m. ir is the
pitch of the new coil. Equation (11) then becomes
1
MlaM- 1O- N. d
6NiNare 4 sin n mO ir sin n mx r
sin n:m1 7r sin n m2 r cos n 0) henrys (12)

This formula is strictly correct when m, is an integer and when
6 is an integral multiple of the tooth pitch. It is true for all
values of 0 if either mo or m, or both are unity.
By considering the axes of two similar groups of coils as coin-
cident we obtain the value of A1 L1 which is part of the self in-
ductance of the group, thus
106Nr2 d sin2 n mo7r sin2n m 7r (13)

The other factor that enters into the self inductance is the slot
leakage inductance which depends upon the number of turns in a
coil, the number of coils in a group and the width and depth of
the slot and the length of the air gap. Since with the value of
Ai L1 all the field which links the secondary winding has been
included, only the portion of the slot leakage which does not link
all the turns in the opposed secondary coil should be considered.
No hard and fast rule can be made for determining this quantity
since it depends upon the shape of the slots, there should be little
trouble in making the calculation when the data are given. De-
noting this quantity by A2 L1 we have
L1 = A1L1 +A2L, (14)
Symmetrically Grouped Windings. The above formulas give

the mutual impedance between groups of coils, each group of
which may be unsymmetrical. Generally machines are designed
so that, although the individual groups of coils due to fractional
pitch may be unsymmetrical, the complete winding is symmetri-
cal. When two coils are together in a slot this may be done by
connecting one group of coils opposite the north pole in series
with the corresponding group opposite the south pole; that is to
say, the group displaced electrically by the angle 7r. If therefore
we take equation (11) and consider the mutual induction as due
to a group having axis at 0 = zero and another having its axis
at 0 = 7r with a similarly arranged group of coils having its
axis at 0, we find that (11) becomes
16 N,N2 rl 1
M1 = 1 2d sin22 n mo7r sin nm 7r
sin n M2 7r (1- cos n )2 cos n 0 }henrys (15)

Similarly
Ala M-= lNi
10 Nadr II { 1 sinn nml 7r sin n m. 7r
sin n ml 7r sin n ma 7r (1-cos n 1)2 COS n } henrys (16)

Since 1- cos nir is zero for all even values of n it is evident that
(15) and (16) contain no even harmonics, moreover the above
formulas give the mutual induction between two similarly
connected groups of windings, but if (1- cos n r) is used only with
the first power these formulas give the mutual impedance be-
tween one pair of such symmetrically grouped windings and
another single group with axis inclined at an angle 0.
The value of self induction is
Al Li = 16N2l ( 1 sin2n mO sin2 n mi 7r

loll i2d
(1 - cos n 1r)2 (17)
A2 L1 is found in the same manner as before
L1 = A1 Li + A2 Li (18)
It is obvious from (15) and (16) that the effect of dissymmetry
is to introduce more or less double frequency into the wave form
of generated e. m. f.
It will be seen from an examination of (15) and (17) that, for
7
example, a winding of pitch 3 and subtending an angle
when connected in a symmetrical group of two has the same field
form and characteristics as a full pitch winding of the same
number of turns subtending an angle 27r.
There are many symmetrical forms of winding but all will be
found to be covered by the formulas (15) and (16).
Unsymmetrical Windings. These may take many forms which
may be classified:
(1) Dissymmetry of flux form due to even harmonics.
(2) Dissymmetry in axial position of polyphase groups.
(3) Dissymmetry in windings due to incorrect grouping of
coils.
(4) Dissymmetry due to unsymmetrical magnetic character-
istics of the iron.
Of these various forms of dissymmetry the most common is a
combination of (1), (2) and (3). These forms of unsymmetrical
windings may all be calculated by the formulas (11) to (16).
It is to be noted that the mutual inductance between a sym-
metrical and an unsymmetrical winding is harmonically sym-
metrical. Hence, if the field of a machine is harmonically
symmetrical, the e. m. f. generated will be also harmonically
symmetrical whatever may be the form of the windings.
The reciprocal nature of M is fully established by its form, for
it is immaterial in obtaining (16) whether we start out with the
winding whose pitch is m: or with that whose pitch is mo, the
result will be the same. The effect of saturation will be to tend
to alter the values of the coefficients of M but the general form
will not vary appreciably. We shall now consider some standard
windings of generators and motors.
Three-Phase Symmetrical Full Pitch. Here mo, mi and M2 are
0.5, 0.1666 and 0.1666 respectively. Using formula (15) all
the even harmonics disappear and (1 - cos n r)2 is equal to 4 or
zero.
16 N,N2 r1 / os
M12 O+1 d
0 + 4 cos 3
8 d 81
+ Cos6
1
625
+401Cos 7 0 + 651 Cos 9 ) (19)

1918] FOR TESCUE: SYMMETRICAL CO-ORDINATES 1107
Theoretical Symmetrical Three-Phase Winding. Here mo

- 0.5, ml - m2 = 0.333. Using formula (11)
M12 = _ 1609N,N2rl
3, 16 / 0 + 625cos
2r Icos 1 50
4 109w-72d62
+ 2401 cos 7 0 + 14641 cos 11 0 + . (20)

Here the third group of harmonics is entirely eliminated.
2
Three-Phase Symmetrical 3 Pitch Winding. Here mo =
0.333, ml = m2 = 0.166. Using formula (15)
N,N2 r
M =33 161092 =
/
(cos 1
+ 625cos 5
21 cos 7
os71+14641col+
+ cos 11 0 + (21)
...
2
which gives the same result as (20).
FORMULAS FOR SALIENT POLE MACHINES
The formulas given in the. preceding discussion are appropriate
for distributed winding and non-salient poles. Where salient
poles are used the field form due to the poles with a given wind-
ing will be arbitrary so that with the polar axis as reference we
shall have
2wT Na Ia 2 (An cos n ) (22)
=d
Where 63 is the induction through the armature or stator. When

the poles are symmetrical An cos n a might be chosen at once for
this condition and in this case we do not require coefficients of
mutual induction between pole windings, since the value of (B
is obtained by considering the mutual reaction between pole
windings to be such as will produce symmetry. We may how-
ever assume 63 to be perfectly general in form in which case the
flux through a coil of pitch 2 mo 7r is
Od4
Na la r l
7
2 An sin n mo 1r cos n) (23)
We have therefore for the mutual induction between one pole
and a group of coils at an angle a and subtending an angle
2 mi wr
4 Na,N rl / A,snnmwin1cs
Mai 1019d 2sin
n mo7rsin n ml7rcos
n/ (24)
and where there is symmetry due to grouping of windings, we
have
Ml ON,2
4 NaNd rmArsinnm
(1- cos n 7r)2 cosnO} (25)
where Na is the number of turns for one pole and (25) applies to
one pair of poles and the corresponding group of coils. When there
are more than one pair of poles in series and the corresponding
groups of winding are also in series, if it is desired to consider
the mutual inductance of the complete winding, the result given
above must be multiplied by the number of pairs of poles.
If in equation (16) we take
Nal.
2--sin n ma -=Na
1 (26)
and ~ ~
7rn
= B,
it becomes
Mla = 32N,Nd I { Bn2 sin n mt 7r sin n mo 7r
sin n ml r (1-cos n )2 cosn } (27)

which is the expression corresponding to (25) starting with the
winding flux form. (25) and (27) must therefore be identical
and we have
32NiNare . 4NaNire
109 d B10s9nnm 7- lO9d A,
or
3n - 8sin n m (28)
and
7iO d 2 (B. sin n mo r cos n) (29)
and is the induction wave form for a single turn of the winding.
The expression for the mutual inductance between windings
of the same core for salient poles is obtained in terms of the pole
flux wave form by substituting in the formulas 88 s1n nn 7r
m.
for n 7r . We have therefore the following formulas for salient
poles.
General expression considering only one pole and one group of
coils.
2WrNaIa
CB3a =(10 d (n cosn) (a)
20d A sinnm cos nO) (b)
Mat = 4 NaNirl
rI
An2I sinn mO 7r sin n ml ir cos n (c)
109 d
2N1N2r l An sinnmm
nm
1097rd n3 sin nm wslnrnm7rsinnmlr
sin n m2 7r cos n0) (d)
A, La = 4irNgarl / An mz r
sin nnzr (e)
109d n
A Ll =- 12101Ni2rl ( An si n mo r sin nm1 ))

r
(f)
ird n sin n M..r
General expressions considering only poles to be symmetrical.
Considered on the basis of two poles, Na being turns on one pole.
3a 2B NaIa {An
r
= 2
lOd A((1-cos n r) cos n (a')
= {A sin nmo 7r
I,
A n sin
d
=2020 m r (-cos
i nmxw csl)on n 7r) cos n } (b')
4NaN, rl An 07snn1
Mal = _ N109 d sin 1 mo r sin n mi r
(1-cosn r)cosnO} (c')

2NiN2rl f An sinnmO7r s
109 ir d na sin n m. snn m0rsinn m1r
sin n m2 ir cos n 0 } henrys (d')
IiLa 4irNa2rl fz{0n sinnmz7r (1-cosn7r) } (e')
Al Li
2 N12 ri 2;
lO9 r d
Ann sin2 nmiO r sin2nmi 7r }
sin n m, r (
General expression with both polar and winding symmetry.
s=a 2 7i Na Ia
10 d
An (1-cosnrr) cosn0) (a")
inI, sin n MO r
20 d
2
An sin n m (1-cos n r) cos n0 (b")
Mal 4NaN,
109d
rl z An sin n mo(r
sin n ml 7r cos n7r)2
cos n 0 } (c")
2N1N2 rl { An sin nmowr .

10'r d n3 sinnm'mr s n m0rsinnm1ir
Sin nIm2 7r (1- COS n r)2 COS n 0 } (d")
Al La =a 4 7 Na2 r l A sin n M r (1cos n r)2 (e!')

109 d n)
2 N12 r 1 f An sin2nmo0isin2n Ml7
A1L1 =- 10 rd z n sinnm.7n
(1cos n r)2} (f")
In using any of the formulas given above for machines having

more than two poles, it must be divided by the number of pairs
of poles and likewise the expression for M or A1 L must be multi-
plied by the number of pairs of poles, which leaves the formula
for these quantities unchanged.
Let us next consider the actual induction in the air gap with
a distributed winding operating with three-phase currents. Let
im1 be the magnetizing current of the first phase i42 and im3
those of the other phases. The induction due to one group of
coils of phase 1 is
(5= 810N1irdd, { I sinnmoirsinnimi7r cosnO}
- (30)
and if the phase displacement of 2 and 3 from 1 be 'P12 and 'P13

(B2 = 8 Nr d 2{ sinn mO ir sin n M2 7r cos (nO - 'P12)
(31)
8 N3 dd { 2 sinnmo0rsinnm2 rcos (n 0-Ps3) }
(32)
For symmetrically grouped coils the formulas become
(B =8_Nl ijmi2 (1
lOir d~n
sinn mo7r sin n ml r (1-cos n 7r)
cos n } (33)
N=
8 n2d { '2 Sinnmoirsinnm2ir(l- cosnir)
cos m (O-4 12) (34)
310d 2 sin n mo r sinnm37r(1- cosn r)cosm
(0- (P13)} (35)

For a symmetrical three-phase motor with full pitch coils
MO = 0.5, ml = M2 = M3 = 0.166 (33), (39) and (35) become
of the four
8Nim f 2 11
(Bl 8 10 ird
=
^ cos O--cos
9 -9cos 7O
3O + T5 cos 5O +
81 121 169os3+()
which is the field due to one group of coils alone. The wave is
flattened by the third group of harmonics but all the other
harmonics are peaking values. There is therefore a decided
gain in such a wave form of flux since it permits of high funda-
mental flux density.
The maximum value of flux is approximately
Bmaz = 0.823 80 f d gaus (37)
where d is given in centimeters.

1.677 N,
d im
Bmax= 7d maxwells per square inch,
with d given in inches.
For the total winding the resultant induction will be the sum
of B1, B2, and B. If we take the symmetrical winding with
angles between planes of symmetry
(P122-3 and (p13- 3', we have
=
cos n = ein@
2 +
IE-jn 0
2
cos n 0- a-
3-2 + an 2
(38)
4 r e_)
0
cos ( an + a-n 2
If we multiply these three quantities successively by Imi,

a2i, aIml and add, we have
Imi { - (1 + a- + a((+')) + 2
X (1 + an 2 + a-(n-1)) ) } (39)
and giving n successive odd values from 1 up, we find for (39)
the following values
n = 1 (39) becomes 2 I e-ii
n- 3 " a 0
19181 FORTFESCUE: SYMMETRICAL CO-ORDINATES 1113
3 --
n= 7 0'
2 E
n = 11
3'-fT C6j7o
~~m1
~~~2 2n
flfl 21m15jfl2 2nw \/~~~~3
3
We may therefore express 63 by

(13 real part of
16 NiIml 1-sin n MO 7sin n Ml7r-1 cos n ir)
X ~l sin2
2w- (40)
It will be obvious that if we proceed around the cylinder in the
negative direction of rotation at an angular speed w and I.,
is equal to Iml EjwI, for n = 1 the value of B1 will remain
constant and real, hence B1 must be a constant field rotating at
angular velocity w in the negative direction. The value of B may
be expressed in harmonic form, but in this form it does not illus-
trate the rotating field theory so aptly. The harmonic form is
given below and is simpler in appearance than (40).
_16 N1 imlIl
d~~n2 sin n MOw7sin n ml r(1 -cosnwr)
lOw
sin2 2nrCos n) (41)

For a symmetrical three-phase motor with full pitch coil
(ino 0.5 mi1= 0.166) 63 becomes
1NlOird ~Cos 0+1cos 50+ 1Cos 70

+
1 1oICs1
coslO+ cos3O (42)
This gives for the maximum induction approximately
1.075 X 12 N1 iml 1.29 N1 im
(1max 10= d rd gauss (43)
where d is measured in centimeters.
ma _=3.28 X N1 m1 maxwell per square inch (44)

where d is measured in inches and N is the total number of
turns per pair of poles.
APPENDIX II
Graphical Construction for Obtaining Symmetrical Components
The graphica method for finding the symmetrical components
of S (Ea) given in the text7serves as a geometrical interpretation
EC~~~ bs
FIG. 18--GRAPHICAL METHOD FOR OBTAINING THE SYMMETRICAL COM-

PONENTS OF THE THREE-PHASE VECTORS Ea ,Eb, EC
of equation (5), but the graphical method shown in Fig. 18 is

much simpler and more convenient, the construction is as
follows. Find E and F the centroids of the two equilateral
triangles with B C as base: With 0 the centroid of triangle
A B C as centre describe the two circles passing through E and F
then; E 0 extended till it touches the circle through E at the
opposite end of the diameter gives Eal; Ebl and ELi are obtained
by laying of points on the circle 60 degrees from E. Similarly
if F 0 is extended to meet the circle through F at the opposite
end of the diameter we obtain Ea2; and Eb2 and EC2 are obtained
by the same construction as before.
The proof of this construction is as follows: If G and H are
1918] FORTESC UE: SYMMETRICAL CO-ORDINA TES 1115
the apexes (not shown in figure) of the equilateral triangles

having B C as base
OB+OC+OG OA +OG -_a+OG
3 3 3
OB+OC+OH -OA+OH _-Ea+OH
3 3 3
- a2. 0 C = - a2 (OD + D C)
-a .OB = -a (O D + D B) --a (OD-D C)
- a2. 0 C- a. OB = - (a + a2) 0D + (a- a2) .D C
=OD + j \'-D C
= OG
That is
OG =-(aEb+ a2E)
Similarly
OH = - (a21b
aEb )
and therefor
a,
QE-_ Ea + aEb+a2]c -
3
and OF- +a2
The construction when Ea + Eb + I is not zero is so obvious

,
that it is not necessary to show it here.

If lines be drawn from Ea to Ea2, Eb to Eb2, Ec to P.2 these
lines will be parallel to Ea' Eb' and E,' respectively and will meet
at a point O'. The vectors O' Ea O0 Eb and O' Ec give the
values of the e. m. f. across each member of a star delta bank of
transformers when operated on the three-phase circuit S (9a)
with ratios changed so as to give a balanced secondary triangle
of e. m. f's.
1116 SYMMETRICAL CO-ORDINATES [June 28
DISCUSSION ON "METHOD OF SYMMETRICAL CO-ORDINATES

APPLIED TO THE SOLUTION OF POLYPHASE NETWORKS"
(FORTESCUE), ATLANTIC CITY, N. J., JUNE 28, 1918.
J. Slepian (by letter): During the past eighteen months
it has been my very good fortune to have been in close contact
with Mr. Fortescue and to have had many interesting discussions
on the ideas embodied in this paper. Since I have had so long
to think oveT and digest these ideas, I think I may be pardoned
for going at great length here into the viewpoint I have reached.
I feel all the greater need for a long discussion, because on reading
the paper I see that the great wealth of material to be presented
in a limited space has crowded out much detailed explanation
and appeal to analogy, which in discussions with Mr. Fortescue
have contribtuted so much to my clearer conception of the ideas
presented here.
The method given here had its origin in considering the oper-
ation of balanced induction machines under unbalanced condi-
tions. Using the "coordinates" proposed here, the theory of
these machines may be given with beautiful simplicity. I think
I am right in saying that the utility of the method is practically
entirely limited to the case of rotating induction machines.
Purely static apparatus tieated in this way does not show any
simplification. When one considers, however, that almost
every practical alternating-current circuit contains at least
one rotating machine, namely the generator, the broad field
of application of the method becomes apparent.
The root of the ideas given here is old and was given early
in treatments of single-phase motors. The simple constant
rotating nature of the flux in a balanced polyphase motor was
well known. It was discovered that the more complicated flux
in a single-phase motor could be resolved into the sum of two
such constant rotating fluxes of opposite rotations, with the
magnitude of these fluxes not necessarily equal. It was ob-
served also, that if the balanced polyphase states, which would
give independently each of these rotating fields, were super-
imposed, the resultant state would give correctly the currents.
voltages, torques, etc., of the single-phase motor. For a recent
discussion of this, see Mr. Lamme's paper, page 627, Volume
I, TRANSACTIONS 1918.
Mr. Fortescue has generalized this method of resolution of
the flux in a polyphase machine under unbalanced conditions
into the sum of fluxes, each ccrresponding to a balanced condi-
tion, to a similar resolution of an unbalanced system of any
polyphase quantit es whatever.
In my discussion here I shall, for simplicity, confine myself
to the three-phase case. To illustrate the resolution of a set
of unbalanced three-phase currents into symmetrical or balanced
components, consider three arbitrary currents Ia lb I, flowing
into the terminals of a three-phase, star-connected apparatus.
(Fig. 1). Mr. Fortescue shows that this set of currents may be
had by adding together the following three sets of currents.
1918] DISC USSION A T A TLA NTIC CIT Y 1117
Method of Symmetrical Coordinates. First a current lao in each

phase. These three currents then are all equal and all in
phase. They combine to give a current in the neutral. Second:
a current faG in phase a, a current a2Iai, in phase b, and a cur-
rent aLl in phase c, where
a =- 2 +j2 /3
Ia
FIG. 1 FIG. 2
is a cube root of unity. This set of currents is clearly a balanced

polyphase set of what is called positive sequence. (Fig. 2). Third
a current Ia2 in phase a, a current a Ia2 in phase B and a cur-
rent a2 1a2 in phase c. (Fig. 3.). This set of currents is clearly
a balanced polyphase set of what is called negative phase se-
quence. If these three sets of cuirents be made to flow simul-
taneously in the three-phase apparatus the resultant current in
aIa2 Ia2
FIG. 3
each phase will be respectively Ia, lb, I,. The values of 'aO, Ial,
Ia2 are obtained from equations (5). They are:
lao = 1/3 (Ia + lb + Ia).
Ial = 1/3 (Ia + alb + a2 I,).
1a2 = 1/3 (Ia + a2Ib + aI).
In a similar way, any set of three voltages to neutral acting
on the three-phase apparatus can be resolved into the sum of
three sets of balanced voltages of zero phase sequence, positive

phase sequence, and negative phase sequence, respectively.
The sequence operators, SO, SI and S2, which Mr. Fortescue
introduces, have a close analogy, it seems to me, with the j used
in the usual treatment of alternating currents. In the elemen-
tary theory, it is found useful to resolve an alternating quantity
into two components, one in phase with some reference alter-
nating quantity, and the other component in phase quadrature.
If ii and i2 are the magnitudes respectively, of the in-phase and
the quadrature components, their resultant I, is denoted by
i = il + ji2. Here j may be looked upon as a unit alternating
quantity in phase quadrature with a reference alternating
quantity. Thus, the equation i = lil + j i2 states that the
quantity i may be obtained by adding together il times a unit
in phase quantity, and i2 times a unit quadrature component.
In the same way, the symbols SO, S1, S2 may be regarded as
unit pol1phase vectors* of zero, positive and negative phase
sequence. Thus S= (1, 1, 1), if referring to currents, rep-
resents a unit current in each phase of a three-phase apparatus
the three currents being in phase with some reference alternat-
ing quantity. Similarly, SI = (1, a2, a) represents a unit
balanced, three-phase current, of positive phase sequence, in
the three-phase apparatus, the current in the first phase being
in phase with the reference alternating quantity. Likewise
S2 = (1, a, a2) represents a unit balanced, three-phase current of
negative phase sequence. Thus the equation (15) S (Ia) =
S0 1 + SI Ia + S2 la2 states that a system of three currents
S (I.) is equal to the sum of three sets of currents, the first set
being of the type So, that is of zero phase sequence, the second
set of the type S1, that is of positive phase sequence, and the third
set of type S2, that is of negative phase sequence. The three
currents of any set are obtained by multiplying the three cur-
rents of the corresponding unit polyphase vector by the complex
quantity indicated. Thus the currents in the second set are
obtained by multiplying the three currents, 1, a2, a, respectively,
by Ia1.
The utility of resolving three-phase currents and voltages in
this way, when applied to rotating balanced machines, lies in
this fact, that a symmetrical set of voltages of any phase sequence
applie,d to the machine will produce a symmetrical set of curients
of the same phase sequence, and that a symmetrical set of currents
of any phase sequence flowing into the machine will produce a
symmetrical set of terminal voltages of the same phase sequence.
This fact is well known in the theory usually given of balanced
polyphase apparatus, although it is seldom explicitly stated.
Other methods of dividing a set of unbalanced three-phase
quantities acting on a balanced machine into components would
*This very fortunate term is due to Mr. C. T. Allcutt and expresses the
idea of the paragraph most succinctly.
19181 DISCUSSION AT ATLANTIC CITY 1119
not have this simplicity. For example: suppose the three actual
currents in the three phases, and the three actual voltages of the
terminals relative to neutral were taken as the components or
coordinate of the currents and voltages. Then one component
of current alone, that is a current in one nhase only, would
not produce only the corresponding component of voltage, but
would produce voltages in all three phases.
Neglecting saturation, for a given rotor speed, the symmetri-
cal currents produced in a balanced machine by symmetrical
applied voltage, are proportional and in a definite phase relation
to the symmetrical applied voltage. Thus the currents may
be obtained by multiplying the individual members of the sym -
metrical set of applied voltages by some complex number. Sim-
ilarly, a symmetrical set of currents flowing into a balanced
machine produces a symmetrical set of terminal voltages which
may be obtained by multiplying the individual members of the
symmetrical set of currents by some complex number. The
first complex number could be called the symmetrical admit-
tance of the machine, and the second the symmetrical impedance.
This symmetrical impedance and admittance will be different,
of course, for symmetrical components of different phase sequence.
Thus for a star-connected, ungrounded neutral machine, the
admittance for symmetrical voltage of zero phase sequence will
be zero; if the neutral is grounded, the impedance to zero phase
sequence current will be principally the leakage reactance be-
tween phases; if the rotor is running near synchronism in the
sense of positive phase sequence, the impedance Z1 to positive
phase sequence current will be large, while the impedance Z2 to
negative phase sequence will be small. The complete expres-
sions for these impedances Z1, Z2, are given in equations (122),
(123). The relations between current and voltage components
are given in equations (120) and (121). In these four equations
is concentrated the whole theory of symmetrical machines
operating under unbalanced conditions. But so simple are these
equations and their physical meaning so clear, that once under-
stood, they enable us to predict qualitatively, without com-
putation, the behavior of rotating balanced machines under any
unbalanced condition whatever.
Consider, for example, the simple picture of the action of a
phase balancer which the above treatment gives. The balancer
is merely a balanced machine across the polyphase line running
near synchronism. It offers high impedance for the symmet-
rical voltage of normal or positive phase sequence, but offers a
very low impedance for any negative phase sequence component
of voltage. Thus the negative phase sequence voltage is
"short-circuited out," and balance on the line is preserved.
The harmful effects of slight unbalance in the terminal voltage
of a polyphase machine upon the machine's rating is also clearly
shown. Since the negative sequence impedance is very low, a
small negative phase sequence voltage will produce large negative
phase sequence currents with their attendant heating. This

suggests a better quantity to denote the degree of unbalance of
a line than the one commonly used. It is the ratio of the nega-
tive phase sequence component of the line voltage to the positive
phase sequence component.
Balanced stationary apparatus also enjoys the simplicity
pointed out in balanced rotating machines, namely that sym-
metrical voltages of any phase sequence produce symmetrical
currents of the same phase sequence, but here the impedances
for positive and negative phase sequence are always the same.
The zero phase sequence impedance is generally different.
When we pass to unbalanced apparatus; things become more
complicated. Symmetrical currents of one phase sequence no
longer produce e. m. fs. of that phase sequence alone, but the
total e. m. fs. are unsymmetrical and contain components of
other phase sequences. Let us consider in detail an unbalanced
star impedance, with grounded neutral. Let the phase impe-
dances be Z., Zb, Zc.
Let us first study the e. m. fs. to neutral produced by a
zero phase sequence current. If laO, Iao, IaO, are the three cur-
rents, the three e. m. fs. are Zalao, ZbIao, Zalao. Resolving
these three e. m. fs. into symmetrical components, we find by
equations (5) that the zeroth order component is:
3 (ZaIao + ZbIaO+ ZcIao) = ' (Za + Zb + Zc) Iao
The positive phase sequence components will be:
I (ZAao + aZbIao + ac2ZcIao) = " (Za + aZb + a2Zc) lao
The negative phase sequence component will be:
3 (ZaIao + a2Zblao + aZcI0ao) = " (Za + a2Zb + aZ,) I..
Thus the total e. m. f. using the symbols S'S' S2 would be
written:
So [3 (Za + Zb + Zc) lao] + S' [13(Za + aZb + a2Zc)IaoI +
S2 [I(Za + a2Zb + aZc] lao = So (ZaoIao) + S' (ZalIao)
+ S2 (Zaao)
Where Zao = 31 (Za + Zb + Zc)
Za. = 3 (Za + aZb + a2Zc)
Za2 =
3 (Za + a2Zb + aZ,)
Now consider the e. m. f's. given by the positive phase se-
quence currents, la, a2Iaa, aIal. They will be Zalal, Zba2Ia
Zcalai.
Resolve the system of em. f's. into symmetrical components.
We find for the zero phase sequence component:
3 (ZaIa1 + Zba2Ial + Zc,al.) = (Za + a2Zb + aZr) I.,1
Za21a
1918] DISCUSSION AT ATLANTIC CITY 1121
For the positive phase sequence component we find

4 (Z0101 + aZba2I0l + a2ZcaI01) - 4 (Za + Zb + Zc) 101 =
ZaoIal
For the negative phase sequence component we find
3 (ZaIa1 + a2Zba2Iai + aZcalaj) = 1 (Za + aZb + a2Zc) Ila =
Za lIa 1
Lastly, consider the e. m. f.'s produced by negative phase
sequence currents, a2, a 1a2, a21a2. The zero phase sequence
component will be
4 (ZaIa 2 + Zbal. + Zca2Ia2) = 4 (Za + aZb + a2Z") 1a2 -
ZalIa2.
The positive phase sequence e. m. f. component will be
3 (Zala2 + aZbaTa2 + a2Zca2Ia2) = 4 (Za + a2Zb + aZc) Ia2=
Za2Ia2.
The negative phase sequence e. m. f. component will be
4 (ZaIa2 + a2Z-aIl2 + aZca2Ia2) = 4 (Za + ZI + Zc) 1.2 =
ZaOIa2
The components of voltage are then expressed easily in terms
of the quantities Zao, Zal, Za2, defined in equations (8).
Let us bring these results together in tabular form where the
relations between them can be observed.
Voltage Components
Currents So or zero SI or positive S2 or r egative
phase sequence phase sequence phase sequence
Zero phase sequence,

So (Iao) = (Iao Iao lao) ZaoIao ZalIao Za21ao
Positive phase sequence,
SI (1.0) = (lal, a2Ia alal) Za2lal ZaoIal Zallal
Negative phase sequence
S2 (Ia2) = (Ia2, aIa2, a2Ia2) Zal1a2 Za2Ia2 Zaolao
Studying the table above, we see that each symmetrical

component of current gives rise to symmetrical voltages of all
three phase sequences. We notice first the e. m. f. component
which is of the same phase sequence as the current by which it
is produced, is obtained by multiplying the producing currents
by Zao. That is the unbalanced apparatus may be said to have
one component of impedance, Zao, which gives e. m. f. 's of the
same phase sequence as the currents. This component of im-
pedance does not change the exponent of the S symbol defining
the phase sequence of the currents. Thus Zao alone correspond-
ing to the currents S0 (Ila) gives the e. m. fs. S0 (Z.1100); corres-
ponding to the
1122 SYMMETRICAL CO-ORDINA TES [June 28
Currents SI (Iai) gives the e. m. fs. SI (Zao7al); and corres-

ponding to the
Currents S2 (1a2) gives the e. m. fs. S2 (Z2 '.s2).
Now let us examine how Zal enters in the above table. Cor-
responding to the zero phase sequence.
Currents SI (lao) we find the e. m. fs. SI (ZaiIao) of positive
phase sequence. Corresponding to the positive phase sequence.
Currents S (Iai) we find the e. m. fs. S2 (ZaiIai) of negative
phase sequence. Corresponding to the negative phase sequence.
Currents S2 (1a2) we find the e. m. fs. So (ZalIa2) of zero phase
sequence.
If we agree now to the following definitions of the symbols S
when affected with higher exponents:
53 =50.; S4 = S'; SI = S2; S6I S S7 = 5'; etc
We see that the way in which Zal terms enter can be summarized
in this way:
The unbalanced apparatus has a component of impedance,
Za., which increases the exponent of the phase sequence symbol
by unity.
Lastly, let us see how Za2 enters the above table. We find
that:
Currents S0 (lao) give e. m. fs. S2 (Za21ao).
Currents S' (Iai) give e. m. fs. So (Za2Iai).
Currents S2 (Ia2) give e. m. fs. S' (Za2Ia2)-
This may be summarized by saying that the unbalanced appara-
tus has a component of impedance Za2 which increases the
exponent of the sequence symbol by two.
All ihese results will be obtained automatically if we suppose
that symbols S, SI, S2, axe attached respectively to Zao, Zal, Za2,
and when multiplying the current components, S (lao), S (Ial),
S2 (Ia2), the exponential laws hold. Thus:
S (Zao) S5 (lao) = S (Zao lao); Sl (Zal) S- (Iao) = S (Zallao)
S0 (Zao) S' (Ial) = S' (Zao Ial); S' (Zal) S (Ia ) = 52 (Za al).
So (Zao) S2 (Ia2) = S2 (Zao a2); S' (Zai) S2 (lao) = S3 (Zal Ia2).
=
So (Za 1 a2)
S2 (Za2) S(Iao) = S2 (Za2Iao)
S2 (Za2) *S'(Iai) = S3 (Za2Iai) = S0(Za2Ial)
S2 (ZQ2) S2 (1,2) = S4 (Za2Ia2) = S (Za2Ia2)
The whole result can be expressed by writing the total imped-
ance, S(Za) = So Zao + S8 Zal + S2 Za2-
To get the e. m. fs. corresponding to any currents, we multiply
together the total impedance by the total current, following
merely the rules of algebra, and interpreting higher powers of S
as described above. Thus we get:
S(Ea) = S(Za) S(Ia)
= (S0Zao + S'Zai + S2Za2) (SOIao + SlIal + S2Ia2)
= S(Zaolao + ZalIa2 + Za2Ial) + Sl(Zao Ial
+ Zallao + Za2Ia2) + S (ZaoIa2 + ZalIal + Za2Iao)
We get here the foundation of a complex three-phase algebra,

which performs the same functions for three-phase systems as
the usual complex quantity does for simple alternating current.
This leads to another interpretation or mode of viewing the
sequence symbols, S, SI, S2, and here again we may profit by
analogy with the j in the complex algebra of simple alternating
quantities. Suppose i is a current in phase with some reference al-
ternating quantity. If this curient flows through a resistance r.
the resulting e. m. f., ri is also in phase with reference alternating
quantity. If the resistance is of unit value the resulting e. m. f.
is i. Suppose, however, that the current flows through a
reactance whose impedance is x ohms. The resulting e. m. f.
will be x I in magnitude, but it will be in phase quadrature with
the reference alternating quantity. This is taken care of in the
usual theory by affecting the expression for the reactance with a
j, thus xj, and making j have the property that when multiplying
a vector or alternating quantity it does not change the value of
the vector but merely advances its phase by ninety degrees
This is a different meaning fiom what was given before. j here
no longer represents a unit vector or alternating quantity. It
now represents an operator, or a symbol for advancing the phase
of any vector which it multiplies. Where j by itself is referred
tc as a vector, it should be understood that ji is meant; where
i is a unit vector in phase with the reference vector.
It is clear that operating twice successively by j upon a vector
merely reverses the phase of that vector. Thus j (j.i) = - 1,
which leads to the rule of multiplication J2 = -1. Lastly, we
see that the e. m. f. induced by a current flowing through an
impedance may be obtained coirectly both in phase and magni-
tude by multiplying the current by a number of the form r + jx,
which represents the complete impedance. Thus the complex
algebra of simple alternating quantities is born.
We may attach significance as operators in a similar way to
the symbols S, SI, S2. Now they shall no longer represent poly-
phase vectors, but shall merely be operators which when written
next to a polyphase vector, change it into a polyphase vector
of another phase sequence. The quantities which the S's affect
shall be considered as polyphase vectors. Thus the symbol I
shall stand for three currents, each equal to I. Operating on I
with S means multiplying each of these curients by unity.
Thus S is a unity operator and does not change the polyphase
vector upon which it operates. A separate symbol for it might
have been omitted.
S' I or S operating on I, means that the three currents
I, I, I, are to be multiplied by 1, a2, a, respectively; or that the
first current is to be unchanged, the second to be advanced in
phase by 120 deg. and the third by 240 deg. S' by itself shall
not mean anything unless it is understood to be followed by 1
in which case it stands for the unit polyphase vector of positive,
rotation obtained by operating with S on the three currents
(1, 1, 1).
Similarly, S21 oI S2 operating on I, means that the three cur-
rents I, I, I, are to be multiplied by I, a, a2, respectively; or
that the first current is to be unchanged, the second advanced
in phase by 240 deg. and the third by 120 deg. S2 by itself
shall not mean anything unless it is understood to be I, in
which case it stands for the unit polyphase vector of negative
phase rotation obtained by operating upon the polyphase quan-
tity (I, I, 1).
With the meaning now given to S, SI, S2, it at once follows
that they satisfy the law of exponents. For example, S (S'I)
means leave the first cui rent I unchanged; advance the second
cui rent I, in phase by 120 deg., and then again by 120 deg;
advance the third current, I, by 240 deg. and than again by
240 deg. It is clear that the final results are exactly the same
as the results of operating with S2 on I. Also it is clear that
multiplication by an operators and a constant Z is commuta-
tive. Thus ZS'I = S'Z1I. This serves as the foundation of
the complex thiee-phase algebra.
a a
50 l Ohm
I Ohm
Sh lOhnt1Oh
~ ~~ ~ ~~~~~yOh
1Oh
-'1/2 OhM
,V2 V
Ohn
1/2 3 j Oh Ohms
C b c IN~bb
FIG. 4 FIG. 5
In the complex algebra of simple alternating quantities, it is

easy to give an illustration of the operator J as in impedance.
In fact, a reactive impedance of one ohm gives for any current
an e. m. f. which may be obtained from the current by multi-
plying by J. Can we similarly illustrate the three operators
SI, Si, S2.
SO, of course, is easy. Take a grounded star of resistances,
each of one ohm. Then we find for the currents So (I) = (I, I, I)
the voltages, (I, I, I) = S I; for the currents S2 (I)
(I, al, a21) the voltages (I, al, a21) = S2I.
Now for SI. Take a grounded star in which the impedance
of phase a is 1 ohm, of phase b, (- aVWj)
\ ohms = a2 ohms,
and of phase c, ( + 2 \/3j) ohms = a ohms. Now we
find for the currents S (I) = (I, I, I), the e. m. fs. (I, a21, al)
S'I = S (SoI) for the currents SI (I) = (I, a2I, al), the e. m. fs.
(I, al, a21) = S21 = S' (S'I); for the currents S2 (I) =
(I, aI, a21), the e. m. fs. (I, I, I) = SI = SI (S21). Fig.5.
Lastly, S2. Take a grounded star in which the impedance of
phase a is 1 ohm, of phase b, (- + 2 \ITJ) ohms, and of
phase c, (-4- \Iv3j) ohms. Then we find for the currents

So (I) = (I, I, I), the e. m. fs. (I, al, a21) - S21 = S2S01;
for the currents S' (I) = (I, a2l, al) the e. m. fs. (I, I, I) =
S = S2S'I; for the currents S2 (I) = (I, al, a21) the e. m. fs.
(I, a2I, aI) = S11 = S2S21. Fig. 6.
In these examples, negative resistances appear. This is not
surplising, as it is clear that unbalanced apparatus, when cur-
rent of one phase sequence flows, may feed energy into impressed
I Ohm
Ohn& -1/2 Ohm

-lk
_V-3-j Ohm. +Y203i Ohms
FIG. 6
e. m. fs. of another phase sequence. The negative resistance

may be represented physically by a series commutator a-c.
motor, driven at constant speed in the opposite of its motoring
direction of rotation.
The resolution of an arbitrary three-phase star impedance
into its three-phase components is merely the problem of finding
three three-phase impedances of the S, S' and SI type, respec-
tively, which put in series as in Fig. 7 will reproduce the given
three-phase star.
Zao aZ a2Za
SI Zao S' Zal s2 7
FIG. 7
It is clear that admittances may be treated in the same way

as impedances above. Thus the general three-phase admittance
may be written in terms of its components, thus:
S (Ya) = SO Yao + Sl Ya, + S2 Ya2
The curients produced by a general three-phase voltage S (Ea)
= SEao + S'EaI + S2Ea2 will be:
S (Ia) = (S Yao + S'Yal + S2 Ya2) (SEao + S'Eal + S2Ea2).

= So (Y EaO + YalEa2 + Ya2Eal)
+ S' (YaoEa, + YaiEao + Ya2Ea2)
+ S2(Yao Ea2 + YalEal + Ya2Eao).
In the complex algebra of simple alternating currents we pass
formally from an impedance R + jx to the corresponding
I
admittance by taking the reciprocal, - and multiplying
numerator and denominator by r - jx to reduce it to
the standard form.
Thu
r -jx r x
r + Jx r-jx r2 + x2 r2 + x2
Similarly in this three-phase complex algebra we obtain the

admittance corresponding to an impedance S0ZaO + S'Zal
+ S Za2 by taking the reciprocal 0 + S!Za, + SZa2 But
now to reduce to standard form we must multiply numerator
and denominator by
So (Zao + aSlZal +a2S2Za2) (S0Zao + a2S'Zal + aS2Z.2).
Thus:
1 S0Zao + aS'Zal + a2S2Za2
S0Zao + S'Zal + S2Za2 S0Zao 4 aSlZal + a2S2Za2
50 Zao + a2SlZal + aS2Za2
S0 Zao + a2SlZal + aS2Za2
S0 (Zao2+ G2ZaiZa2 + aZalZa2) + S1(a2ZaoZa1 + aZaoZal + Z2a2)
+ S2 (a2ZaoZa2 + Z2a, + aZaoZa2)
D
SO (Z2ao Zal Za2) + SI (Z2a2 Zao Zal) + S2 (Z2 - Zao Za2)
- -
a
S0(Z3ao + Z'al + Z'a2 - 3 Zao Zal Za2)

Remembering that So is an operator which does not change
the quantity upon which it acts, and therefore that a symbol for
it might have been omitted, we may leave off the S in the
denominator and get for the final admittance:
S(Y) Z2 ao - Zal Za2
S Z3ao + Z3a + Z3a2 3 Zao Zal Za2
+ S Z2a2 -
Zao Zal
Z ao + Z3a + Z3a2 - 3 Zao Zal Za2
+S2
Z2al -Zao Za2
Z+ ao
+ Z3 + Z3a2 3 Zao Z4l Za2
The principles just enunciated enable us to put together three-

phase apparatus and calculate the resulting three-phase com-
ponents just as is done in single-phase apparatus. For apparatus
in parallel, we add admittances; for apparatus in series we add
impedances.
So far, in the static apparatus considered, the mutual induc-
tance between "legs" of the three-phase apparatus has been
assumed zero. When these mutual inductances are not zero,
things become more complicated. A three-phase apparatus
with grounded neutral is, of course, a four terminal network, and
as is well known, in the most general case would require
22
=
6 complex constants to characterize it under steady
state conditions. It is clear that a complex three-phase expres-
sion So Zao + SI Zai + 2 Za2 depends on only three complex
constants, and hence cannot represent the mest general three-
phase apparatus.
The relations between the symmetrical components of voltage
and current on the general static three-phase network are given
in equations (25). One way of summarizing these equations
would be to say that the three-phase apparatus has three dif-
ferent impedances for the three phase-sequence currents. Thus
the impedance to zero phase sequence current as given by
equations (25) is:
So (Zaao + 2 Zbco) + S' (Zaal - Zbcl) + S2 (Zaa2 - Zbc2);
the impedance to positive phase sequence currents is
So (Zaao - Zbco) + Sl (Zaal + 2 Zb,l) + S2 (Zaa2 - Zbc2);
and the impedance to negative phase sequence currents is
So (ZaaO - Zb,o) + S' (Zaal - Zbcl) + S2 (Zaa2 + Zbc2),
Another way of summarizing the equations (25) would be to
say that the actual three-phase apparatus is equivalent to a
three-phase network having the three-phase impedance
S0(Zaao - Zbco) + Sl (Zal- Zbcl) + S2 (Zaa2 - Zbc2),
in series with a network whose impedance to zero, positive and
negative phase sequence current are respectively, S3 Zbco;
SI 3 Zb,l; S2 3Zbc2.
With the ideas developed in this paper, solutions of problems
on symmetrical rotating machines with unbalanced static appa-
ratus may be worked with comparative ease. As an example
I shall consider the case of a generator, with ungrounded neutral,
acting on a given star load, of leg impedances, Za, Zb, Z,. SUp-
pose the machine is symmetrical, i. e., non-salient poles, and a
damping squirrel cage on the field. In this case the generator
has an impedance Z, to positive phase sequence current, and a
small impedance Z2 to negative phase sequence currents. The

positive phase sequence impedance Z, is what has been called
the synchronous impedance of the machine such that if E is the
no load voltage, E - Z11 is the voltage on a positive sequence
balanced load I. The problem is to find the currents and volt-
ages on the load, and the voltage between load neutral and
generator neutral.
We have given that lao = 0.
We calculate the impedance drops of positive phase sequence
and put their sum equal to the generated positive sequence
voltage. Thus:
Zl Ial + Zao Ial + Za2 1a2 = E.
Similarly, since the generated negative phase sequence voltage
is zero,
Z2 Ia2 + Zal Ial + Zao 1a2 - 0.
Solving for Ial and 1a2
+ ZaO
la,i= E Zl Z2 + (Z1 + Z2
Z2) Zao + Z2ao - Zal Za2
'Ta2 =E -Zal
IZ Z2 + (Z1 + Z2) Zao + Z2ao -Zal Za2
The terminal voltages will be:
Eai -E Zi 'al Zaolai1 + Za2 'a2 E F

Z2ao - Zal Za2
0
+ Z2 Zao
Ea2 = - Z2 1a2 = Za1 'al + Zao Ia2 = E Z2Z

0
al
The vJoltage between load neutral and generator neutral will

be:
Eao = Zal Ia2 + Za2 lal = E *Z2al Zao Za2 + Z2 Za2

A special case of interest is where Zb - Z, are zero, that is, a
single-phase short circuit. We then have:
Zao = 3 Za Za 3 Zal Za2 = 3 Za
la,i= E 3 Zl Z2(Za++(Z13 Z2)
+ Z2) Za
I -Za
E3 Zl Z2 + (Z1 + Z2) Z,
19181 DISCUSSION AT ATLANTIC CITY 1129
Z2 Za
3 Zl Z2 + (Zl + Z2) Za
Ea, =E. Z2Z
Ea2 = E .3 Z1 Z2 +Z2(Zl
Za
+ Z2) Za
Z2 Za
ZZ
Eao =E
3 Z Z2 + (Zl + Z2) Za
If just previous to the single-phase short circuit the generator
was unloaded, Za becomes infinite and our final results are:
I1a =- Ia2 E
Eal -Ea2 = Eao = E z2

ZI + Z2
If the generator above did not have a damper winding on the
field, it would be an unsymmetrical machine having a single-
phase rotor. As we know, the theory of unsymmetrical rotating
machines is of considerable complication. Mr. Fortescue has
promised a treatment by these methods of the general unsym-
metrical rotating machine in a future paper, which I look
forward to with great interest.
C. P. Steinmetz: In dealing with calculations or investi-
gations of polyphase systems, or, as usually the case, three-phase
systems, the difficulties which we meet are not so much mathe-
matical difficulties, but are what I may call mechanical difficul-
ties. The equations, while mathematically not complicated,
lead to expressions which are so complicated and extensive in
form as to make any such calculations extremely difficult.
In dealing with the balanced polyphase system, this difficulty
has been overcome by the introduction of the equivalent single-
phase system, by considering the polyphase system as resolved
into a number of single-phase systems, each comprising the
circuit from one of the phase wires to the neutral point of the
system. This method however, fails in the unbalanced poly-
phase system-and naturally practically all existing commercial
polyphase systems are more or less unbalanced-and the theory
of the vector method given to us today in a more extensive
description by Mr. Fortescue, gives the solution by showing us
in the case of the general three-phase system that it can be
resolved into two balanced three-phase systems of opposite phase
rotation. We can apply the same plan to other polyphase
systems.
V. Karapetoff: Mr. Fortescue deserves the gratitude of the
profession for bringing out a new method for numerical compu-
tations in unsymmetrical polyphase systems, and also for

applying the method to a number of practical cases.
Expressions (4) represent the result of solution of certain
equations, which equations unfortunately are not given in the
paper. The expressions quoted are unnecessarily involved, and
it is difficult to see their method of derivation. Moreover,
it may be shown that they do not represent the most general
case of resolution of an unsymmetrical system of vectors. The
following procedure would seem to be preferable.
Definition. A multiple-angle symmetrical polyphase system of
vectors is defined as one in which the vectors are numbered not
consecutively, but skipping a certain
number of vectors. Thus, Fig. 8 1
represents a triple-angle seven-phase
clockwise system, because three angles 3
are comprised between phase 1 and 6
phase 2, and also between phase 2
and 3, etc. Such a generalization of
the concept of polyphase systems is
useful when the number of phases 4
and the number of angles skipped are
prime numbers, so that all the phases
may be numbered consecutively, 2
without omitting or repeating any. F
If the total number of phases n is a FIC,. 8-A TRIPLE-ANGLE
prime number, then the total num- SYSTEM
ber of possible combinations or
multiple systems is (n - 1), the angles between the consecu-
tive phases being
2
n
2 X
n
. . (n-1) 2.
n
It is
assumed that the numbering in all the systems is either clock-
wise or counter-clockwise, so that there is no confusion between
a positive and a negative phase sequence.
Theorem 1. An arbitrary system of n unequal and unsym-
metrical vectors ,without residue, may be represented by (n-i) sym-
metrical multiple n-phase systems. A system without residue
is defined as one in which the sum of the vectors is equal to zero.
Let the given vectors be E1, E2 .... . En, and let the (n - 1)
unknown systems be denoted by A, B, . M, where A is a
single-angle system, B is a double-angle system, etc. It is
required to prove that the following equations are consistent
and may be solved for A, B .... . M:
E1= A 1+B+. + M1
E2= A2+-B2+. + M2 (1)
. ..... . ... ... . .. . .. . . ... .
En = A. + B. + . . . + M.
These equations correspond to eqs. (4) in Mr. Fortescue's paper
when 2 E = 0. In order to preserve the same conventions as in
the original paper, the component systems are assumed to be

numbered clockwise (though the vectors are rotating counter-
clockwise). Then if a is an operator which rotates a vector by
an angle 2 7r/n counter-clockwise, the preceding equations
become
El= AI + B,+ . . . 1[i
E2 a-'Al + a-2B1 + . . . + a-(n-1) M1 (2)
. . . . . . . . . . . . . . . . ..................................
En= a-(n-l) A1 + a-2 (n-1) B1 + . a-(n-1) (n-l)Ml J
These n equations contain (n - 1) unknown vectors A 1, B1,

M1, but only (n - 1) equations are independent of each other,
because of the condition z E = 0. To solve for Al multiply
the second equation by a, the third by a2, etc. and add them
together. The result is
nA,= El+aE2. .+an- E, (3)
This checks with the second line of Mr. Fortescue's equations (4)
By analogy we obtain
nBv= Ei+a2E2+ . . . +a2(n-1) E (4)
and
n M1 = E1 + an-l E2 .... a(n-l) (n1-) E (5)
The theorem is thus proved because equation (2) can be solved
for the (n - 1) unknown vectors in terms of the given E's.
Theorem 2. Referring to theorem 1, if the given system of
vectors has a residue, say, equal to a vector R, this residue may
be split into n arbitrary vectors, pi, P2, pn, in phase or out
of phase with one another, and equations (1) become
Ei = pi + Al + B . . . ......
. + Ml
E2P2 A2 B2. .. ..+ M2 (6)
En= Pn+An+Bn + .... . + M2 f
This proposition follows at once from the fact that the vectors-
(E1 - pi), (E2 - p2), etc. form 'a system without residue, so that
theorem 1 applies to them. Mr. Fortescue considers only a
particular case when
Pi1 P2= ....= Pn= R/n
= (E1 + E2+. . En) /n (7)
and writes this expression as the first term on the right-hand
side in all the formulas of his equations (4). This is perfectly
justifiable in so far as he is after a symmetrical solution, but
there may be practical cases in which a more general representa-
tion according to equations (6) is preferable. In this case the
expressions (E1 - pi), (E2 - P2), etc. must be used in equations
(3) to (5), instead of E1, E2, etc.
The net result is that only a system without residue may be
resolved in a perfectly definite manner, while a system with a
residue first must be converted into one without residue by
subtracting arbitrary parts of the whole residue from each

vector.
With a three-phase system the double-angle component simply
becomes a three-phase system of an opposite phase rotation,.so
that it is not necessary to introduce the question of multiple-
angle systems. From a practical point of view the paper would
gain much if the author would begin directly with the three-
phase system on the basis of his equations (6), and leave the
general consideration of n-phase systems for the end of the paper.
For reference purposes one would not have to study the general
theory and perhaps get discouraged before obtaining the required
bit of specific information.
The introduction of the sequence operator S does not seem to
be necessary, at least for the three-phase system. A proper
cyclic notation should accomplish the same purpose with much
less theory and much less writing. As a matter of fact, the
index of S may be obtained correctly from the subscripts in the
expression to which it refers. For example, in Mr. Fortescue's
formula (22) on the first line the sum of the subscripts of Z I is
1 + 2 = 3, and the order of S is 3 or, which is the same, 0. On
the second line the sum of the subscripts of Z I is 1 + 0 = 1,
and the order of S is 1. The same is true for all the fundamental
formulason pp. 1038 to 1042 aswell as in applications. Therefore,
it would seem that the operator S ought to be dropped, and the
formulas so rewritten that they would apply to any phase of the
system in cyclic rotation.
The fundamental formula (21) will then simply become
E= ZI+ W+I++ W_I_ (8)
Here E, Z, and I apply to any of the three phases, I+ is the cur-
rent in the next consecutive phase, and I_ is the current in the
preceding phase. For example, if E, Z, and I refer to phase
b, 1+ refers to phase c and I_ refers to phase a. At the same
time the equation is so written that any of the three phases may
be taken as the phase under consideration. W+ and W. are the
corresponding mutual reactances.
When the mutual inductance between a primary and a second-
ary polyphase circuit must be considered, Fortescue's equation
(28) applies, and in the simplified cyclic notation it becomes
E= X J+X+J++X_J_ (9)
where J, J+ and J_ are the secondary currents in the three phases.
The letter X is used instead of W to indicate that the mutual
inductance is between the phases of the primary and the second-
ary circuit. X without subscript covers the combinations of
the "corresponding" phases a - u, b - v, and c - w, while X+
refers to the combinations a - v, b - w, c - u of the phases "tied
one forward", and X_ refers to the remaining combination of
the phases "tied two forward", or "tied one backward".
Each of the quantities in equations (8) and (9) may be rep-
resented by means of symmetrical coordinates in a general

manner, without specifying phases a, b and c. Thus,
I = Io+Ip+ In
I+= Io+aIp+a-l In (10)
I-= 1o +a-l Ip+ aIn)
Similarly,
W+ WO + a Wp + a-'W ()
The first equation (10) refers to any phase, and states that the
current in that phase is a sum of the residue current Io, plus the
vectors Ip and In of the symmetrical coordinates of the same
phase. The subscripts p and n refer to the systems of positive
and negative phase rotation respectively. In an n-phase system
subscripts 1, 2, 3, etc. should be used in place of p and n, to indi-
cate single-angle, double-angle, triple-angle, etc. coordinates.
Substituting expressions such as (10) and (11) into equations
(8) and (9) one obtains comparatively simple polynomials with
three groups of terms: those without a, those multiplied by a,
and those multiplied by a-'. These final expressions give the
coordinates of E, and are the fundamental equations upon
which all applications should be based, without any need for the
sequence operator S.
Incidentally, I believe that the expression "symmetrical
components" is a more correct and descriptive term of the
method than the term "symmetrical coordinates", and I should
like to see it so changed while it is not too late.
I also suggest that the term "mutual reactance" be used
throughout the paper in place of the "mutual impedance". The
latter term implies a combined resistance and inductance action,
while the formulas in the paper seem to refer to the magnetic
inductive action exclusively. If the resistance component is
intended to be taken into account, a mathematical definition of
the mutual impedance becomes necessary, because two coils are
presupposed to have different resistances, and it is not clear how
these resistances enter into the expression for mutual impedance.
It may also be mentioned here that there are cases of unsym-
metrical polyphase connections which can be solved more
readily without resolving the given electrical quantities into
their symmetrical components. Several such cases are con-
sidered in the present writer's work entitled "Ueber Mehr-
phasige Stromsysteme bei Ungleichmaessiger Belastung (Enke,
Stuttgart, 1900). For example, in the solution of any arbitrary
n-phase star-connected system with given voltages E and phase
admittances Y, it is convenient to take the voltage e between
the two neutral points as the independent variable. Then, for
each phase the current I = Y (E - e), and according to the
first Kirchhoff law
1 = ZY(E -e) = 0,
I
so that
e ( YE)/ Y.
Knowing e, the currents in the individual phases are computed

from the first equation. In the most general case of a compli-
cated network with a-c. voltages inserted at different places, the
Kirchhoff equations of two kinds, written for the vectors of
currents and voltages, furnish a complete solution, and one has
to consider in individual applications whether or not the resolu-
tion into symmetrical components is desirable or not.
A. M. Dudley: Attention has been called to the practical
applications of this solution, and one or two may be mentioned
here. It has been long known that in the ordinary induction
motor in addition to the main rotating working field there may
exist several other fields of different amplitude and frequency
and phase rotation and that the existence of such fields is prob-
ably responsible for the operating freaks sometimes noticed in
motors. For example, a motor is sometimes found which will
not reverse its mechanical direction of rotation when running
light even though the leads be reversed so as to give the opposite
direction of phase rotation; or, a two-phase motor will be
found which when running light takes power from the line on
one phase and returns power to the line on the other, or a,
squirrel-cage motor will start from rest with considerably more
torque than it is capable of accelerating up to full speed so that
it apparently has a so-called sub-synchronous speed. These are
actual practical operating conditions often mentioned but never
determined quantitatively. Mr. Fortescue's analysis offers the
means of such quantitative study.
As a further practical instance may be mentioned the two
cases of single-phase connections referred to in the analysis of
equations (141), (142) and (150), one of which is equivalent to
two coupled polyphase motors connected in parallel and the
other to two coupled polyphase motors connected in series.
This was fully discussed from another view point in Mr. Lamme's
April paper before the Institute and the identity of the two
conclusions is a check on Mr. Fortescue's method.
To those who have expressed the idea that the paper is hard
to read I would commend Dr. Slepian's discussion also printed
herewith as illuminating both to the method and the results.
He makes the suggestion that the method as outlined may con-
stitute the basis of a new complex polyphase algebra which will
greatly simplify the study of polyphase network problems of all
kinds.
Dr. Steinmetz and Prof. Karapetoff have told us that as a
mathematical achievement and as a demonstration of theory
this paper is a masterpiece. I should like to venture the predic-
tion that as a practical working tool it will eventually come into
the greatest usefulness.
Let us then pay our respects to Mr. Fortescue both as a great
mathematician and as a scientist and also as the developer of a
practical tool which shall make easier the daily task of the engi-
neer in the ranks.
Charles F. Scott: Mathematics and engineering have their

varying conditions, and they have their supporters and their
critics. To a certain class of engineers the mathematician is an
abstract man who starts somewhere and finds something that
can lead to consecutive conclusions. On the other hand, a
practical man, using that term in the extreme, is apt to rely
wholly on experience and to have no use for the mathematics
he does not understand. We all know that the really valuable
position is the mean between these two, in which the theoretical
and the practical do go together.
The reference to papers of a mathematical and explanatory
character, and their relations to engineering reminded me some-
how of the first Institute paper I came in contact with. It was
through the pages of the Electrical World. I had been experi-
menting with alternating currents and was associated with men
who knew only slightly more about it than I did. Many things
seemed mysterious. Then the paper by Mr. William Stanley
on "Alternating-Current Phenomena," appeared, and that
mysterious thing which was the stumbling block of about
everybody who came up against alternating-current, the simple
single-phase phenomenon of current and voltage out of phase,
was cleared up. Mr. Stanley drew some triangles and explained
a few things. That put me on a new basis. I could see how the
mathematical explanation, the physical phenomena, were
related, and it gave me a real understanding of the alternating
current. A little later came the polyphase, and it was a step
from that to the work on the unbalanced polyphase system.
As we advance then, getting into more and more complicated
problems, these papers of solution and explanation come to
guide the engineer in his work.
What is this paper? When you look at a paper of this sort it
looks as if a mathematician had produced it, leading on and on
to interesting formulas, one after another. Now, to bring out
my point, I want to emphasize and say I do not think this is a
product of the mathematician, but of the engineer. Mr.
Fortescue did not work these things out from the theoretical,
mathematical standpoint, but as an engineer who was confronted
with problems in which unbalanced polyphase systems were
concerned. It may have been transmission or some kind of
particular machinery which involved problems which he did
not have the proper tools to solve.
Therefore, in studying the problems and seeing the inade-
quacy of the mathematical tools at hand, he went to work and
constructed tools to perform that work. Mr. Fortescue has
done just as some other engineers do, who work first with experi-
ments, then go to mathematics, and then back to the experiments.
That combination of first theory and then practise, back and
forth, constitutes the work of the engineer who can be con-
structive and do pioneer work. I think I am right in saying
that this is not a paper of a theoretical mathematician, but a
paper of a practical engineer, who is developing a new tool for

himself and offering it to others.
C. 0. Mailloux: This is the second great paper which Mr.
Fortescue has presented before this Institute. Much that I had
to say in the way of friendly criticism has already been antici-
pated by Prof. Karapetoff. I sympathize with Prof. Karapetoff's
views, and I share them, in regard to the possibilities of simpli-
fication of this paper. I thank Mr. Fortescue and many other
members of the Institute will also thank him, for having given
an introduction to the paper, because that introduction is really
the mathematical specifications in accordance with which the
paper has been prepared; and to me it was more valuable, as a
criterion of the value of the paper, than the paper itself, because
the paper itself cannot be read offhand. The bewildering exhibit
of subscripts to be found in it is something that will, well, make
one pause; but the introduction, to any one who is at all familiar
with mathematics, tells exactly the basis on which the paper is
written, what it aims to do, and substantially the method by
which it accomplished its aim. I am very thankful, indeed,
personally, that that was done.
Part I is also very useful, but if it had followed the method of
presentation suggested by Prof. Karapetoff, it would have been
much simplified. The paper is so valuable, that it would be
well worth while to rewrite it for the purpose of making it more
easily digestible and more easily accessible to the great body of
practical men. It appears to have been written too much from
the point of view of the professed mathematician, who seeks
first to make a generalization, and then proceeds to experimenta-
tion to find practical and special cases. It would have been a
better arrangement if that generalization had been put in an
appendix, and if the statement had been made that from the
general case is derived the particular case which interests us,
namely, the case of the three-phase current. I think that to
discuss n-phases is of academic interest only at present, even
though it is possible that there may be some day when it will be
of practical importance. Until that seemingly remote day comes,
a generalization which includes n-phases might as well lie in an
appendix, out of harm's way, and especially where it would not
encumber the rest of the discussion. If Mr. Fortescue had
written the paper by putting into the appendix, the generaliza-
tion, the thing which makes it complete and comprehensive, he
would have simplified it and made it more useful.
It is because I think so highly of the paper, I would like to
see it made clearer. It is not a difficult paper, the mathematics
are simple, but they look complex, because one is bewildered by
a maze of subscripts and by many "operators." The fundamen-
tal idea of the operator is a splendid one, but it is just as well to
put that explanation in an appendix, in all its generality, and
then use it in the simple form that Prof. Karapetoff suggests.
Those who deal with the matter practically, would not then
have so many equations to analyze and not so many symbols to

deal with.
The analysis of the actions and reactions of the various forces
present in electrical circuits, and the interpretation of the
equations devised for expressing these actions and reactions,
under different conditions, whether they occur in balanced or in
unbalanced systems, may be facilitated by remembering that
all physical forces and all forms of energy are subject to the same
general principles of conservation and equilibrium, and, espe-
cially, that Newton's principle of dynamic equilibrium,-namely,
that under all conditions and at all times, action and reaction
are opposed and equal to each other, exactly-, applies to
electrical, as well as to mechanical, forces, powers, and energies.
Analogy requires that, in any electrical circuit, at any instant
of time whatever, the forces of action and those of reaction should
always be opposing and balancing each other exactly, just as
they do in mechanical systems; and it is known that they always
do balance exactly. Analogy also requires that, for electrical
circuits, when dealing with instantaneous values, the equations
of electrodynamic equilibrium should have the same general
characteristics as the general equation of dynamic equilibrium
or balance for mechanical forces. The latter, it is well known,
reduces to an algebraical sum of four terms, representing four
distinct kinds or amounts of force, or power, or energy, according
to the case, whose resultant, when a state of equilibrium occurs,
is equal to zero.
The four kinds or amounts of force, power, or energy, in terms
of which every dynamic reaction whatever, in mechanics, can
be completely expressed, find exact parallels in electrodynamics.
Here, also, we find four kinds of force, power, or energy; we have
reversible forces of two kinds, corresponding to those which, in
mechanics, produce or result from changes in kinetic energy, or
changes in potential energy; we have the dissipated force, which
is expended in overcoming ohmic resistance, corresponding to
the force lost in overcoming friction in mechanics, and we have,
for the fourth kind, the force, power, or energy representing the
balancing actions or reactions which are necessary to maintain
the equilibrium that must exist at every instant of time between
action and reaction. The first two kinds represent energy which
is not immediately dissipated but is displaced and stored in the
system; the third kind represents energy which is immediately
consumed in the system; and the fourth kind represents energy
which is put into the system from an outside source, or else
which is taken out of the system by overflow into another system
or circuit, as may be required to maintain equilibrium.
These considerations show that even in an unbalanced elec-
trical system, so-called, there must still be dynamic balance or
equilibrium. The balancing force, power, or energy, according to
the case, appears in the fourth term, where it represents the
compensating force, power, or energy by which equilibrium is
maintained.
Charles L. Fortescue: I maintain that the paper is quite

simple. The mathematical portion is not difficult to under-
stand, anybody can follow it if he takes the pains. I admit that
the appearance of the equations is cumbersome, but that is almost
impossible to overcome. The nature of the subject makes the
equations cumbersome.
Prof. Karapetoff in his discussion used the word "stimulation".
I wish to say that necessarily in a paper of this kind there are
many sources of stimulation, as I pointed out in my introduc-
tion, many of the ideas are not new. The idea of symmetrical
component three-phase systems is being used more or less by
others, but the theory has never been presented systematically,
and I think the idea of symmetrical operators is new.
Dr. Steinmetz's and Prof. Karapetoff's discussions are what
I may term conjugate discussions. Dr. Steinmetz discusses the
paper purely from the practical point of view. He thinks that
the system is capable of practical application, and will be of
great use for that purpose. Dr. Karapetoff points out the pos-
sibility, from a theoretical point of view, of the n-phase system.
I wish to say nothing would have given me greater pleasure than
to go into that purely theoretical matter much more extensively.
It is very fascinating and has great promise, but I felt that there
was needed, in this paper, some practical justification for present-
ing it. The theoretical part was so long drawn out, that I felt
it was necessary to show by practical illustrations why I presented
it.
I felt that the presentation of mathematical solutions did not
alone afford a justification for the paper, but that there must
also be a good practical reason for it, and I felt if I went into
all the theoretical ramifications of this very interesting subject
people would ask: "What is he about? What does he mean by
driving us through all this painful stuff without giving us a
good reason?" So I thought that the presentation of the theory
should be as concise and short as possible, and for that reason
I left out a good deal of explanation that some people think
ought to be there.
Prof. Karapetoff has apparently lost sight of the fact that
unless his arbitrary vectors are all equal to E1 + 12 + *
. .
n
they in turn may be resolved into (n - 1) symmetrical systems
and the "residue" will eventually reduce to the same value as
that given by me, namely n equal vectors of value
E i + E2 + .E. .E
n
The form in which the theorem is presented does not preclude
the adoption of any artifice such as the subtraction of n arbitrary
vectors to make the vector sum equal to zero, when any material
advantage is gained thereby. Such an artifice, in fact, is gener-
1918-] DISCUSSION AT ATLANTIC CITY 1139
ally adopted in the case of Y-connected circuits where the vector

sum is rarely zero on account of the presence of the third har-
monic. Furthermore, in any symmetrical system there is a
definite impedance to each component symmetrical current
including the "residue" as Prof. Karapetoff has termed the
constant vector, this is not true if an arbitrary "residue" is
adopted.
Prof. Karapetoff brings up another point, the use of sequence
symbols. In three-phase systems, once a person has become
familiar with the use of this method, it becomes quite a natural
system. He thinks quite naturally in terms of symmetrical
components, and he usually omits the use of the symbols. In
other words, the symbols are not absolutely necessary, but they
simply afford a sure method of keeping track of the mechanical
part of the mathematical work. If one makes a mistake in
carrying out the mathematical work, one can always go back to
the sequence symbols and check it up. If a quantity looks
unsymmetrical, and a mechanical error is suspected, one can
always follow it out with the sequence symbols and be sure of
getting the correct result. Furthermore, in dealing with a
system of more than three phases, the sequence operator be-
comes important, because there are so many cross mutual
inductances that the formulas become very much involved, and
it is necessary to have some reliable mechanical device that
will keep one in the right path; the sequence operators are such
guides. They are almost indispensable when dealing with
power both in three-phase and n-phase systems. Those are the
principal reasons for their use.
Prof. Karapetoff brings up the question-What is the "mutual
impedance?" I have not used the term perhaps quite correctly
as I have applied it to cases in which the dissipative forces are
not strictly proportional to the velocities or currents. Mutual
impedance may be defined in this manner: If we have two ter-
minals of a circuit carrying a given current, and two other
terminals of another circuit, the electromotive force produced
across the second pair of terminals, given a sine-wave current of a
given frequency is, the product of the mutual impedance between
the two circuits and the current in the first pair of terminals.
In some cases the current follows paths common to both circuits,
so that the mutual impedance may have a real component; in
other cases the energy component may be due to eddy currents
or losses in subsidiary circuits common to both.
Mr. Dudley has pointed out that the practical engineer must
always keep track of the theoretical side of his work. The
object of all mathematical investigation and theory should be
to carry our knowledge of operating conditions further, and to
investigate such obscure phenomena as arise from time to time.
Very often we find that a new tool has to be devised in order to
enable us to carry out our theoretical investigations without
becoming involved in too great complication.
I feel that a paper presented before a scientific body that

cannot be read by every member is hardly justified. I believe
that all can follow this paper if they sincerely try to do so, it
may take a little time, but the mathematics is not difficultt.
V. Karapetoff (by letter): In my Theorem 2, P2, P2 etc. are
arbitrary vectors such that ; p = R, where R is the total residue
of the given system of vectors. Each p maybe resolved geometri-
cally into a vector R/n, in phase' with total R, and another vector,
say p, so that p = p + (R/n). Hence ; p + (R/n) = R, or
z p' = 0. Thus, the system of vectors is without residue and
consequently may be resolved into (n - 1) symmetrical systems.
The system of n vectors each equal to R/n, in phase with total R,
is an irreducible residue.
When it is desired to sift out all the symmetrical components
from the given vectors, the form R/n for the residue should be
used. However, there may be conceivable cases in which there
might be some advantage in splitting the residue into certain
unequal or unsymmetrical parts, without attempting to extract
all of its symmetrical components. For this reason, Theorem 2
is stated in the most general form. I am glad to have my
attention called to the fact that only in the case where each p is
equal to R/n, we have the true or the irreducible residue.

Symmetrical Co-Ordinates Networks: Trigonometrical Called Only (33) Portion Dealing Unsymmetrical

Uploaded by

Copyright:

Available Formats

Symmetrical Co-Ordinates Networks: Trigonometrical Called Only (33) Portion Dealing Unsymmetrical

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Symmetrical Co-Ordinates Networks: Trigonometrical Called Only (33) Portion Dealing Unsymmetrical

Uploaded by

Copyright:

Available Formats

Presented at the 34th A nnual Convention of

the American Institute of Electrical Engineers,

METHOD OF SYMMETRICAL CO-ORDINATES APPLIED

tion of two or more balanced cases. The present paper is an

may be represented as a vector of variable length rotating about

operation dt e. Pt gives - j p e-i Pt and the imaginary part

Where r is any integer. We have therefore

and by giving successive integral values to r from 1 to n, all

It will be observed that a2 a3.... a, are respectively equal to

This paper is subdivided as follows:

A being a negatively rotating vector, and their projection on

El +a-IE-+ a4En +,, . En(-1

+ a-(n-1) El + a-'2 + a-2E3 + ..a-(n-1) E

An_ 91 +a)2 + E3 + ....En

+ a-2(n- 1) El +at E2x+paEs s+ fr A in t E abov

+ a-(n- 1) (r- 1) t r1i2+....a(-)(-)f

+ a-' El + a-1t2 + a-2 123 +...a-(n En

E2 E3 . ... En respectively, are taken separately they add up to ex-

(4) states the law that a system of n vectors or quantities

an equal number of symmetrical n-phase systems. When

Ec = EaO + a Eal + a2Ia2

Figs. (1) and (2) show a graphical method of resolving three

FIG. 2-GRAPHICAL REPRESENTATION OF EQUATION 5.

vector Fa2b2 an E2 lead this vector by the angles and

Zbb = ZaaO + a2 Zaai + a Zaa2 (8)

cc= ZaaO + a Zaai + a2 Zaa2J

Zbc = Zabo + a2 Zab1 + a Zab2 }(9)

Zca = Zabo + a Zabl + ab Za2

EXPLANATION OF THEORY AND USE OF SEQUENCE OPERATOR

SI= 1, a-', a-2.. .a--1)

Sr = 1, a r, a-2r .... a-(n-U)r (10)

S(n-1) = a-(n-1) a-2(n-) ... a-(n-1)2

Attention is called to the importance of preserving the cyclic

+ (SO ZabO + SO Zabl + SO Zab2)

+ So {Zaal + (1 + a) Zabl } 1a2

{Zaal + 2 a2 Zab 1 } Ia1

+ So (Zaai - a2 Zabl) 7a2 + S1 (Zaai- a Zabl) 0aO

An important case to which we must next give consideration

Substituting for S (Ia), S (It) and S (Iw) and S (Zau), S (Zav)

On expressing S (E.) in symmetrical form we have the following

SZ E,E2 = S2 { (Zau2 + a2 Zbw2 + a Zcv2) Iu

+ (Zaul + Zbwl + Zcv1) Iu

and the three symmetrical equations will be

+ (Zauo + a ZbwO + a Z I,,) fal (33)

+ (Zaul + Zbwl + Zcv1) Ial

and any linear algebraic operation performed on 1/2 will give

SO Euo = S (ZauO + ZbwO + ZtcvO) Ia 1

Consequently the following relations are true:

S2 Zbw2 = a S2 Zau2 (42)

Substituting these relations in (30) and (33) we have for this

S' Eal = S' {12 Zaul Iuo + 12 ZauO Iui + 1 Zau2

The above symmetrical forms in which the factors cos 3

S1 ZbWl 1 +a S' Zau (

Substituting these relations in the impedances used in (30)

impedances are equal, the result of substituting (55) in the

j +a2 ) au,(2 + a2 ) (58)

(III) (-v--2 ej0 + 2 e-j9 Zbv,

S' Zwl= ( 2 Ei + E2) S' Zaul

Substituting these relations in the impedances of equations