Professional Documents
Culture Documents
PROEFSCHRIFT
TER VERKRUGING VAN DE GRAAD VAN DOCTOR
IN DE TECHNISCHE WETENSCHAP AAN DE TECHNISCHE
HOGESCHOOL TE DELFT OP GEZAG VAN DE RECTOR
MAGNIFICUS DR. R. KRO NlG, HOOGLERAAR IN DE
AFDELING DER TECHNISCHE NATUURKUNDE, VOOR
EEN COMMISSIE UIT DE SENAAT TE VERDEDIGEN OP
WOENSDAG 14 JUNI 1961 DES NAMIDDAGS TE 4 UUR
DOOR
CONTENTS
INTRODUCTION
THEORY
CHAPTER I
1
2
3
4
5
11
30
39
44
50
ANALOG COMPUTATION
CHAPTER 11
6
7
8
58
62
68
11
12
13
CHAPTER IV
14
15
70
74
80
82
84
MEASUREMENTS
The Induction Motor Constants .
The Elec trica1 Transmission Shaft
REFERENCES
SAMENVATTING
87
93
INTRODUCTION
CHAPTER I
THEORY
RI
+ dd~.
(1.1 )
p = L (0) I,
(1.2)
UI
IRI+ ~~ I,
(1.3)
T = tIp = t PI;
(1.4)
J_ d tf
2
dt
+ .1 P dl
cl(
(1.5 )
dT
dt
+ .1 d P I
2
dt
1.
P dl
(1.6)
dt'
= dT
dt
+ J_ [dtf
f
L - l P_P d(L- l P )]
dt
+ J_ [dP (L- l P ) _
2
_ dT
dt
- Tt -
2"
dt
dO
dt
P dL- l P _ (P I - I) dP]
dt
dt
aL- l P
(1. 7)
----ae .
Hence, the electrica1 power flow into the machine, from (1.3, 1.7),
UI
IRI
+ dT
dt
J_ dO
2
dt
P aL- lp.
(1.8)
ao
UI
IRI
+ ddtT + M
dO
dt'
( 1.9)
where M stands for the torque exerted by the machine on the outside 10ad.
By substraction of eq. (1.8 ) from eq. (1.9), we obtain
M
12
)
d 2 0 dO
M ( dt 2 ' dt' 0, ... ; t
aL- l
-t P----ae p .
(1.10)
Upon substitution of eq. (1.2 ) into eq. (1.1 ) and rewriting eq. (1.10), we remou1d the descriptive equations of the general machine into the simp1e forms
RL- 1 P
+ dP
dt' o =
M+~
_ OL- 1
P
P,
ae
(1.11 )
where L has period 2n in 8. The eqs. (1.11 ) are app1icab1e to every rotating
eletrical machine, such as motors, generators and electrodynamical measuring
systems. Indeed, we have deduced eqs. (1.11 ) by similar methods as used by
WHITE and WOODSON [21], or OLLENDORFF [22].
In eqs. (1.11 ), L is a symmetrical matrix. Frequently, it is profitable to
submit such a matrix to a collineatory transformation [23] and particularly
to an orthogonal transformation [24]. By means of such a transformation, a
symmetrical matrix can always be reduced to a diagonal form, while energy
relations remain invariant. But, as we are free to carry on the reduction only
partially, it is also possible to create a certain freedom in the ultimate reduced
form. Let us take the case where L is transformed by the orthogonal matrix
S = 8- 1 to the form L. = S- lLS, and where L. may attain a nondiagonal
form simpIer than the original form of L. We add that, gene rally, S will depend
on 8. Now, by means of the transformation matrix S, we transform eqs. (1.11 ) to
0= M
+t
( PS- 1) (S- l
oSL~lS-l
S) (S- lP),
( 1.12)
( 1.13)
U. = S- l U, I. = S- l I, p. = S- l P, R. = S- l RS, L. = S- l LS,
(1.14)
and performing the differentia tions as to tand 8,
R L
- 1
+ .1 P
+ S- l as d8 P + dP.
08 d t '
(S- l
2.
as L -1
08'
(1.15)
dt '
OL. - 1 + L - 1 OS- l
08
08
s) P '.
( 1.16)
OS- l S
------ae
= 0 = S- l
as OS- l
a8
+ rra S,
(1.17)
o = M + J.2P. (S-100
as L + oL.
00
- 1
- 1 _
L.-l S-100
oS) P.
( 1.18)
In eq. (1.18) as weIl as in eq. (l.l5), we meet the product S-10S/0 0 = olnS/oO,
being a square matrix whose elements are, in general, functions of O. It will
make sense to define a matrix e in such a way, that
_
_.
S-
exp e,
as _ oe
- 1
- 1
In
S, S
e,
U. - R. L.
o=
- 1
p.
de
+cit
p.
- [(oe
+ t p.
a6 L.
00 -
a6'
as dO _ de
00 dt - cit
..e -
dP.
+ dl'
L. -1
-1 -
( 1.19)
(1.20)
oe)
a6
oL -1]
+ ae-
(1.21)
T = tIp = t PI = tILI =
=
t (SI.)
(SP.)
t l.p.
t P.I.
t I.SsP.
=
(1.22)
t I.L.I.;
UI = (Su. ) SI.
U.SSI.
U.I.
(1.23)
This method will clarify the description and the physical interpretation considerably, since we have created the possibility of splitting up such energy
functions into separate significant terms.
Now we will consider a specialization under rather broad assumptions. We
take a general three-phase machine, which is constructed so that, irrespective
of the angular rotor position, both the electric and the magnetic construction
of the machine will be invariant under the rotation over 2n /3 radians around
the coinciding stator and rotor axes. N ext, let the stator and the rotor variables be
denoted by the relevant symbols with subscripts : as, bs, cs and ar, br, cr respectively, implying that we are considering three stator coils Ca., Cb., Ces and three
rotor coils Ca T> Cb r> Cer Then, the appropriate L (0) matrix,
14
L (e) = L =
las . as
lbs. as
les. as
las. bs
lbs. bs
les. bs
lar. as
lbr. as
ler . as
lar. bs
lbr . bs
ler. bs
las. cs
lbs. cs
las. ar
lbs. ar
las. cr
lbs. cr
(1.24)
ler . cs
ler. ar
ler. br
ler. cr
- ms
-m s
ls
mo
m4
m2
m2
mo
m4
m4
m2
mo
- ms
ls
-mB
ls
- ms
- ms
ce)
m4
mo
m2
mo
m2
m4
... __ . .............
m2
m4
mo
-m r
lr
- mr
lr
-m r
- mr
. ...................
(1 .25)
-m r
-m r
lr
(e)
= lt
(e +- ~) >
(e + ~) >
mt
ce)
mt
mo = m
(e),
m2
= m (e+
mt
0, (t = s, r),
~),
0, (t
m4
(1 .26)
s, r),
= m (e+
4;),
apply. UsuaUy, one assumes lt and mt, (t = s, r), to be constants and m (e)
= m cos e [25, 26]. The idealization lt and mt to be constants, holds fairly
weU in practice [27, 28]. However, frequently the idealization m (e) = m cos e
hides the true perspective. To deduce a general odq-axes theory [29] of a threephase machine, we do not require any of the specializations mentioned. We
shall show, that even effe cts due to saliency, incorporated in the theory of the
generalized machine in quite aspecific way by WHITE and WOODSON [30], can
be handled without specific assumptions.
With this aim, we reduce the matrix L of eq. (l.25 ) by consecutive transformations to simpier forms, performing the procedure step by step. First, we
observe that the two main diagonal submatrices
(l.27)
occurring in eq. (1.25), can be diagonalized by simple methods, as performed
15
for a similar ma trix b y DUSCHEK und HOCHRAINER [31]. To be sure ofthe generality of our transformation, we could apply a general mathematical method
as outlined by FRAZER and DUNCAN [32] . However, with respect to our 3 X 3
L-matrix, the direct way of calculation is used because of simplicity. Thus, let
L t, (t = s, r), of eq. (1.27) be transformed by the orthogonal matrix St = S;- \
(t = s, r), to form the diagonal matrix L t. = St- 1 Lt St, then we have to
calculate the elements St, ik ' (t = s, r; i, k = 1,2,3), and t.i ' (t = s, r;
i = 1, 2, 3), of the square matrix St and the diagonal matrix A t respectively,
connected by
( 1.28)
H ence, temporarily dropping the indices t
(1.30)
where dots represent elements having the value zero. Eq. (1.29) is equivalent
to the system
=:k =~k
[ 1
-x
- x
=: 1[ ~:: 1
l -
Sak
['l
, (k
1, 2, 3).
(1.31 )
Hence, in order that at least one of the elements S i k out of each triple S ik'
(i = 1,2,3; k = fixed ), will be non-zero, the k> (k = 1,2,3), have to be
the roots of the characteristic lambda-equation to L ellt,
1-
det
-x
-X l =
- x 1- - x
-x
-x 1-
( 1.32)
16
X S 21 -
X Sa1
0,
+ 2x S21 - Sa1 =
S21 + 2x Sa1 =
X Su
X Su -
0,
0.
(1.33)
- - - - - --------------------------------------------------------------.
S2l
(1+x)s12+(1+x)s22+(1+x)s32
S122+S222+S322 = 1,
= 0, (three times),
(1.35 )
SnS12+S2lS22+S3lS32 = O.
In view of the result (1.34), the set (1.35) contains only two independent
equations; therefore, we can express S22 and S32 in terms of S12' yielding
(1.36)
where the upper and the 10wer signs belong together. Next, with respect to the
root 1. 3 = 1 +x, we arrive at
2S 23 = - S1 3-()" V2-3s 13 2, 2S33 = - S13+()" V2-3s l3 2, (1.37 )
where, too, the upper and the Iower signs have to be taken simultaneously
with but independently of the preceding pairs. To derive eq. (1.37), we did
not use the orthogonality condition
(1.38)
Substitution of eqs. (1.36, 1.37) into (1.38) and squaring the result will give
S12 2+ Sl3 2 = i(1.39)
It follows from (1.36, 1.37, 1.39), we may assign to one element out of th~
sextupie Sik' (i = 1,2,3; k = 1,2), any arbitrary vaIue; then the remaining
five elements are fixed. But, in order to emphasize the possibility of a clear
physical interpretation of the transformation, our elements S ik, (i, k = 1, 2, 3),
have to be real. To keep the arbitrariness in the choice of the values of the
sextupie as broad as possibIe, we are forced to put
- V2'3 cos
Su -
8 ,S13 -
( )'"
'
V2'3 SIn
8,
-VI cos
. /2
S32=vi[-tcosO( )'W3sinO]
S23 =
S33
Vl[-()'" sin 8
() "
W3 cos
+ ( )" W3 cos
( 1.40)
0, is incorporated in the arbi-
Vi cos [0 -
()'
~n] '
4nJ '
Vi- cos [ O-( )'3
(1.41 )
8] = Vfcos
0]
[()"'O -~ -(),,4;l,'
Vlcos
1,7'
W2
St
Vl
cos et
W2
[(=f)/"et+~-(=f)" ~]
, (1.42)
where the unprimed signs belong together and in their turn the one, two,
three primed terms belong together. Although the choice of the signs is free,
we choose the signs in such a way that
_ [ W2
W2
W2
Vi
( 1.43)
which agrees with the well-known PARK [12] or GOREV [13, 14] transformation
as specialized by LAIBLE [7] and broken up into two factors, as shown in eq.
(1.43), by AKHUNDAR [33]. For convenience of reference, let us define
At
Vl [
W2
W2
W2
+1
-t
-t
0 ]
, ... [ I
+W3. ,Ct ==
.
-W3
.
co~ et -si~ et
sin et
], (t = s,r),
(1.44)
cos et
S = [Ss 0 ] = [A s 0 ] [Cs 0 ] = AC
o Sr
0 Ar
0 Cr
(1.45 )
1,8
We caU the ot elements the zero elements, the dt elements the direct elements,
and the qt elements the quadrature elements. Further, to clarify the mathematical meaning of the odqA
axes transformation performed by St- I, its geometrical representation is shown
in fig. 1. There, the transformation At-I, being the
first transformation to be
performed in the sequence
of St- 1 = Ct- 1 At- I, is represented by the transition
from the original reference
frame a, b , e to the new
reference frame o, a', /.
The representation follows
from the requirement that
the vector {x a , X b, xc} spanned
on the orthogona1 reference
Fig. I. Geometrical rep resentation of the transforma tion
frame (a, b , e) has to be
St- l; AP = BP = CP = 3/2; lkl = I,
identica1 with the vector
k = a, b, c, 0, d, q, d', q'.
{x o, xa', xa'} spanned on the
reference frame (o, a' , Q'); then from the transformation law
Xo
[
:::
1
=
[ W2
Vl + ~
W2 W2
-i
-t
+W3 -W3
( 1.47)
I,
"
o = VI (iaV2+lbV2+ teV2) ,
(1 ) - )
e-d , = ,v/2
3" ea - 2 eb - 'tee ,
Q'~ VI (ibV3 - eV3) .
p.48)
Note that the reference frame (o, a', a' ) is orthogonal and that lol Jl d'l =
= lQ' 1= 1. Further, the transformation by Ct - 1 can be represented by aj finite
rbtation of, the frame of reference (o, a', Q') over the angle + ;e ~ [34, 35] :
yi~1dmg the firiab frame of l'eference (o a,
where the zero-axis acts as ,the
rotation axis.
Q ),
19
L'=A- ILA=
, (1.49)
m2 +m4 !
where empty spaces rep re sent elements having the value zero, and where the
quantities ls> ln m., mn mo, m 2, m4 are functions of e with the properties (1.26).
Next, inspecting the matrix of eq. (1.49), we observe that the nondiagonal
(3 X 3) submatrices occurring in eq. (1.49), constitute, up to a scalar factor,
matrices representing a fini te rotation ofa frame ofreference [34,35]. Indeed;
let us define the scalar quantities D and X by
1
D'/.
2
m2 +2
m4' - m -2 m4V 3]
mo
+m 2-m
--2--4V3 ,mo-
m2 + m4
cos X sin
.
X] ,
(1.51 )
-sm X cos X
1.-2ms
, ls+ ms
L'=
( 1.52)
20
Is- 2m s
mO+m2+m4'
,Dil,
,
, Dil,
, (1.5~)
L.=C-1L'C=
Dil.
, Dil,
is effected if, and only if, the additional condition
0s- Or = X,
(1.54)
holds; for the rest, either Os or Or may be an arbitrary function. We can note
th at the quantity X = X (0, mod 2n) = X (0) (mod 2n), constitutes a nonlinear single valued measure ofthe true angle ofrevolution 0. We shaIl ex amine
this subject in more detail further on. Moreover, we see that if we take the
ideal case m (0) = m cos 0, then eq. (1.51) with eqs. (1.50, 1.26) yields X = 0,
and by using eq. (1.54) consequently yields 0s- Or = 0.
Now, having determined the transformation matrix Sin fuIl, viz. S according to
eqs. (1.44, 1.45) under the additional condition (1.54), we next carry out the
transformation
R = S- 1RS = [Ss
0
S r-
] [ rsE 0 ] [ Ss 0 ] = R.
0 rrE
0 S ,.
(1.55 )
(1.56)
+ Or
The quantities R. and oe /00 both occur in the transformed eqs. (1.20, 1.21 ) ;
and for the original definitions of R. and oe/oo we refer to eqs. (1.14) and
(1.19) respectively. Hence, we can now start the evaluation of the factor
1
[oe L
~ 00'
- 1 _
- 1
oe]
00
'
(1.57)
which occurs in the second term of the torque formula (1.21). Noting the
properties oe/oo = -oG/oO, L. = I., thefactor (1.57) becomes
Oe L.-1
t [a-o
- la-o
oG]
+ L.-
'
(1.58}
21
which is the symmetrica1 part of [oe joO]L. -1. Further, by mere inspection of
eq. (1.53), we write down immediately
lr-2m r
(Ls- 2ms) (lr-2mr)-(mO + m2 + m4)2'
Lr+mr
-(mO+m2+m4)
(ls- 2m s) (lr- 2m r) - (mO+m2+m4)2' 0,
-D'j,
0,
0,
0,
0,
0,
-D' j,
0,
L.-l=
- (mO + m2+m4)
(Ls- 2ms) (l,.-2mr)-(mO+m2+m4)2' 0,
- D' j,
0,
0,
0,
0,
0,
2"
(Oe
ar; L.
- 1
- 1
- L.
Oe) _
ar; -
2"
D'j,
(ls+ ms) (lr+mr) -D
ox
oOJ,
(1.60)
J=
(1.61)
u . RL - up + de lP
dt'
+ dIP.
dt '
L. ,
22
2"
D'j
ox oL.- 1
(ls+ ms) (lr+ mr) -D 00 ' -7ie'
(1.62)
t.,
m., Lr. mr to be
(1.64)
. (1.59)
are reduced to constants, the second factor of (1.64) to the value 3m / [(ls+
+ ms) (lr+mr) - (3m /2) 2]'/', the third factor to the value zero.
The first step in our sequence of transformations being performed, we stipulate
that our generalized transformation with S according to eqs. (1.44, 1.45, l.54)
includes effects due to saliency, and th at it stillieaves a certain freedom in the
choice of the condition of constraint to be put on the pair es, er. It will be
shown further on that this freedom offers the opportunity to transform the
nonautonomous differential equations regarding an induction motor and
regarding an electrical transmission shaft as weIl, into a set of autonomous
differential equations. In studying the stability of motion of such systems, the
benefits of the transposition of a nonautonomous into an autonomous set of
differential equations, are very great and important [36].
Before we bring about our second step in the sequence of transformations, we
shall derive, in addition to the properties (l.26), another set of restrictions to
be required of the quantities Is> m,,, Ir> mT) mo, m2, m4. With this aim we calculate
the magnetic energy function T = V.L.I. (compare eq. 1.22), where L. is
given by eq. (l.53 ). The result
It- 2mt > 0, It + mt > 0, (t = s, r), (ls- 2m s) (lr- 2m r)- (mO+m2 + m4F > 0, (ls+ ms) (lr+ mr) - D > 0,
to hold, irrespective of the value of
tities
e.
(l.66)
where all those scalar quantities have period 217, /3 in e. We call no, "0 and Go,
the transformation ratio from the stator to the rotor side, the coupling coefficient
and the (BLONDEL) stray coefficient respectively and with respect to the zero
components only; n, " and G have equivalent names and meanings, but refer
to both the direct and the quadrature components only. Taking a technical
point of view, the positive quantities just defined are important machine
23
quantities and are reduced to the well-known machine constants, ifwe take the
ideal case i., m., ir, mr to be constants and m (0 ) = m cos o. Therefore, we
factorize L. ofeq. (l.53) in terms ofno, ''0' n and" according to
L.
N - IKrN-\
(l.68)
1
N= ................................ ,K=
: no n
n
, L=
"
"
is- 2m s
is+ ms
is+ms
. (l.69)
i. - 2m.
is+ ms
is+ ms
= N (r )-l K - l N,
(l.70)
an expression needed in the development of eqs. (l.62, l.63 ). Now, eqs. (l.69,
l. 70) at hand, we are in a position to discuss the terms (l.64) occurring in
eqs. (l.62, l.63 ) from a practical point of view. In this respect we remember,
that we started our theory by assuming the ferromagnetics to possess certain
acceptable idealized properties, able to explain nearly all phenomena occurring in actual rotating machines. H ence, it will have little sense at this
point of our theory to develop the mathematical treatment into details, whose
corresponding phenomena in nonideal machines will be masked by other
peculiarities evoked by properties not covered by our idealizations, notwithstanding such details may correspond to actual detail phenomena. In this
respect we raise the point that the quantities i., m., ir and mr are fairly constant in
practice [25, 26], and so do the coupling coefficients ''0' " as well. With respect
to ''0' ", we note that even the total range of values of coupling coefficients of
totally different but normal machines will fall on the interval (0.90; 0.98)
roughly. We shall translate our argumentation by treating N, rand K as
weIl as constant matrix quantities. However, the quantities a o = 1- "0 2 ,
a = 1- ,,2 may deviate appreciably from their average values. Therefore, we
shall retain the 0 dependency of a o, a. In this way, we know, we neglect in
advance terms arising from aK-l fao, apparently related to the differentiation
of simple functions of a o and a as to O. But, in view of the meaning of the
torque expressions (1.11, 1.63), the appearance of such terms are caused by
the dependency on 0 of the magnetic energy content of the stray fields and, as
the stray fields are tightly connected with the main fields and, moreover, the
energy content of the stray fields are certainly small as to the energy content
24
of the main fields, the influence of the terms arising from oK- I /oO must be
negligibly small. Thus, our argumentation states that the set
U.
RL - up
+ de P + dP.
dt'
(1.71)
dt '
(1.72)
will describe, up to an acceptably high degree of accuracy, the electrodynamical behaviour of the three-phase machine. At this point of our theory, it is
convenient to introduce the second step in our sequence of transformations,
a transformation suggested by eq. (1.70), i.e. "the reduction of all quantities
involved to the stator side". But, as N =t=- Ji-l, the relevant transformation will
not be orthogonal. As a matter of course, it is possible to take as a second step
in the sequence of transformations, the orthogonal transformation reducing
L. of eq. (1.53) to the diagonal form. However, in that case, the technically
important machine quantities (1.67) are mixed up in our equations in an unnatural way, and as the final gain - even from a mathematical point of view is doubtful, we should miss the mark.
To perform the transformation, let us define, in accordance with the notation
used in eq. (1.70), the reduced quantities
U' = NU" r = N - IJ., p' = NP" R' = NRN,
(1. 73 )
2:
using
(1.74)
3:
using
de
de
dt = dt
(1.75)
N,
4:
DIl,
i_
l_~
ls+ ms a'
(1.77)
25
Then follows easi1y the performance of the reduction to the stator side of
eqs. (l.71, l.72 ), yie1ding
U*= R* (L*)-lK - llp*
+ de P* +
dt
dP*
dt '
(l. 78)
M +! _ 1_ ~ OX P*JP*.
ls+ ms a oe
(l. 79)
Now we consider the product R* (L*)-l occurring in the 1eading term of the
right-hand side of the voltage equation (1 .78); we note
Is- 2m s
, (l.80)
(l.81 )
-"e'
e'
R* (L* )_lK_l =
= T - IK- l=
a(ls+ ms) ,
- "e"
elf
- "e"
elf
( l.82)
we meet another set of technically useful quantities, introduced by RDEN[37]. With respect to the constants e' and e", - constants in the ideal case
only - , we point at the natural way in which the related dimensionless quantities e' jwo and e" jw o enter the calculations of RDENBERG. We, too, prefer to
define the equivalent dimensionless quantities Y by
BERG
Yos
-XYs
Ys
Ys
( 1.83)
Tt
-XoYo,'
Yt
Yor
-
Yr
XY r
-
aWo(lt + mt)'
(t = S,T),
Yr
XY r
r = u juo, X =
woP' juo,
= wot,
ft =
~ ~8
x UX
(1.84)
dX
+ de
d .. X + di'
(1.85)
0 = f.l+!X]X,
where we effected the third and final step in the sequence of transformations.
The introduction of dimensionless quantities will be convenient for analog
computation [38] and offers valuable means for comparison of different
machines. For future purposes we derived the rather obvious relations
r = S- l NU, X = S- l NwoP, rx
Uo
Uo
S- l NRI.
Uo
( 1.86)
-.l(m 2 - m )V3
2 1
4
.
mO- 2 (m 2 + m4 )
( 1.87)
27
(1.88)
m (0) = mEflnCosnO.
n=O
-!(m2-m 4 )V3
3
-)
.*0
00
= 2 m :=~
(1.89)
fln sm nO,
mO-!(m2+m4)
=
(4n)
(2n)] . n
3 -cos n 0+ 3 sm 3"
00
[ cos n 0+
m ~!o'n
. + 1
P,
4;)]
3
00
-2 mE fln cos nO,
( 1.90)
n= O
.*0
D =
(~mr
00
n= O
00
o.
(1.91 )
.*0 .'*0
COS X, sin X]
x [ -sin X, cos X
00
(3 / 2) m
,\l
[COS 0, sin 0]
[(ls+ms) (lr+mr)]' /. i...J fln -sin 0, cos 0
n
1 92
,( . )
(_) v + l
n=O
.*0
m2 = m (0 + ~), m4 = m( 0+
2
4
- 1[(mO-m2)2
X -_ tg -1 -!(m
_I( -m )V3 D -"2
mo
oX
00
"2
m2 + m4)'
+ (m2-
m4 )2 + (m4- mO)2] ,
( 1.93)
(1.94 )
(V3) [
0(mO - m2 )
0(m 2- m4 )
0(m -m )]
2D (mO+m2)
00
+(m2+m 4 )
00
+ (m4+ mO) 400 o ' (1.95)
( 1.96)
2n/3 O.
28
contribution to the torque. Hence, by our direct way, we elude the prob1em of
the harmonical ana1ysis, at the expense of ang1e dependency of ft, Y s and Y r
The method developed for a three-phase machine may be app1ied, mutatis
mutandum, equaUy weU to synchronous generators having the stator construction invariant under revolution over 2n /3 radians around some stator axis and
having two mutua1 perpendicular planes of image symmetry with respect to
the rotor. Then, in applying the method, one is directed to define instead of
the positive quantities (l.67), the positive quantities no, nd , nq , ''0' "d, "q and
ao, ad' a q , complicating the derivation only slightly.
Ahead of applications to the theory, we list the principal equations, spread
over the three subjects:
1. onginal equations
U = RL- uP
+ ~~, o =
_ L- l
M +!P
P, P
ae
LI,
( 1)
where U, I, Pare columnmatrices whose elements are consecutively distinguished by the subscripts as, hs, cs, ar, hr, cr, attached to the symbols u, i, 'Ijl,
respectively; L is a symmetrical square e dependent matrix whose elements
lo t,g't', (g = a, h, c; t = 5, r) are generally nonzero and have periode 2n in e;
R has the diagonal form.
2
final equations
dX
rx + de
dr X + dr' 0 = ft + !X]X
Z = rx, x are columnmatrices whose
T =
-7
ft
XdrXq s - XdsXqn
(2 )
where T,
elements are consecutively
distinguished by the subscripts os, ds, qs, or, dr, qr, attached to the symbols y,
.c, x, respectively; more over,
Y os
Ys
- "Y s
- "Yr
Yr
Yr
(3)
+1
-1 .
,]=
29
Yot
x=
(4)
3. transformation equations
r=
ft
=
S- lNU/uo, X = S- lNwo'P/uo,
Z = rx =
S- lNRI/uo, i = wot,
(5)
(6)
- 1
V2
S = [S
~J, St =
Vi"
:2
- t
j2 -t
(7)
COS et
sin et
(8)
Z has the
30
miss a derivation of some salient points to found a sound base for the investigati on on the stability of motion in the three cases to be considered.
Let the stator and the rotor supply voltage systems,phases at, bt, ct, (t =
respectively), have negligible internal impedances, then the coil voltages
(g = a, b, c; t = s, r), are submitted to the constraints
+ 1 . -1] [
[ -1 +1.
.
-1
+1
uat ] _ [uct, at ]
Ubt Uat ,bt
Uct
ubt , ct
, (t = s, r),
S, T
Ug t>
(2.1 )
where the voltages U g t , g' t> (g :j::. g' = a, b, c; t = s, r), clearly represent the
coupled phase voltages, normally acting as the driving voltages. We note the
condition
uct' at ]
[1,1, IJ [Uat,bt
= 0, (t = s, r).
(2.2 )
Ubt , ct
0, (t
s, r).
(2.3 )
To draw conclusions from the current constraints, we use eq. (5), viz.
z = rX= S- lNRljuo,
(2 .5}
Bymeans of eq. (2.5) we express the voltage constraints (2.1 ) in terms of a
nonsingular matrix according to
~ 2 + 1 +1
+2 . ]
+1
+2
[ Uuat ]
bt
U ct
uct' at ]
[Uat ,bt , (t
Ubt , ct
s, 'r).
(2.6)
31:
~:: ] = ~ [!~
+! +~] [~::',:: ] ,
- 2 + 1 +4
Uct
(2.7)
(t = s, r).
Ubt , ct
1 [ + 1 -1
.]
. +1 - 1
- 1 . +1
u at ]
Ub t
=3
Uct
[uct,at ]
Uat,bt
(2.8)
,(t=s,r),
Ub t, ct
where we expressed the driving phase voltage matrix Ut, (t = s, r), in terms of
the coupled phase voltage matrix. The inversion (2.8) of (2.1 ) was possible by
a set of symmetry properties and is independent of the functional character of
the voltages U. t, gt, (g =I=- g' = a, b, c; t = s, r). Next, using the transformation
eqs. (5.8), we determine
as ]
u. o...
[
Ubs , cs
!) ,
[COS (T+<p)
(2.10)
where <p represents an arbitrary phase angle. Then, it foIlows from (2.8, 2.9),
rs
=~S =~[~
U
-lUs
- 1
COS
OS
- SIn
Os
+1
(2.11 )
32
o = u m /V2,
(2 .12)
s=
[cos (
.+91~~- es)]
sin (.+91+~-
(2.13)
es)
Before we enter into more detailed investigations on the three cases, we lay
down the rule, that we put
hence forward" the ideal case", implying X =
In the first case, the case of the normal induction motor, the final eqs. (2) yield
upon substitution of Y s according to (2. 13) and of Y r = Sr- 1N rUr/uo = 0
according to (2.9) taken at U,.
d
Ys + d.'
de s
-XYs
d. '
d
+ de s
d. ' Ys+ d.'
- cos( . + 91 +~ - es)
n)
sin (.+91+6- es
- XY r
X ds
, - XY s
X qs
(2.14 )
Xa,.
, -XYr,
Xqr
(2.15 )
Together with the condition e s- er = e, the specification of the function fl
and the initial conditions of the dependent variables, the set (2.14, 2.15)
describes the electrical-mechanical behaviour of the induction motor. If we
constrain the angle
by
es
(2.16)
and we denote the derivation with respect to by a dot over the relevant
symbol, then consequently
(2 .17)
where s represents the slip of the induction motor. Further, with respect to the
specialization of fl, we assume here, as weIl as in the cases below, the mechanical
33
(2.18)
.----~--------~r--
.----r-~------~r-~
IS
+ 1 - 1
U r- - l.3
[ - 1
'] [ucr, ar ]
[ + 1
uar ,br = -l
.
+1 - 1
+ 1
ubr,cr
_ 1
+
1
- 1
(2.20)
Then, substitution ofeqs. (2 .1 3,2.20) into the fin al eqs. (2) yields
34
Ys+ d.'
des
-crr '
+ des
Xlis
-XYs
Xqs
, (2 .2 1)
Xdr
(.+<p-~+ er)
Xqr
_ + sin
(2.22)
Compatible with es- er = e, we take the constraints
n
n
(2.23)
-XYs
,,
.
Yr 1. '
.
- XYs
+ (1-
- ( 1 - 'ia),
Yr
t)
lis
XqS ]
X
Xdr +
X qr
(2.25)
(2.27)
35-
In the foUowing section, we shall use the set (2.27) as the descriptive equations
appropriate to the double supplied slip ring motor.
To prepare the third case,
the case of an electrical
transmission shaft, let us
consider the appropriate
circuit diagram shown
by fig. 3. In case of both
machines we have to determine the fuU 1" of eq.
(2.9) and we already calculated the Ys matrix by
means of (2.13 ), fixing
uo=u m /V2 . But, aswe deal
with two (equivalent)
machines, having generFig. 3. Circuit Diagram ofthe Electrical Transmission Shaft.
aUy different circumstances regarding the rotor systems, we certainly must distinguish the Y r matrices
of the rotors involved. Moreover, as we have in Y s according to (2.13) the
transformation angle esstill at our disposal, we must distinguish also the relevant
Y s matrices. Let us use the obvious notation Yt! and Y t 2 , (t = s, r). Now,
inspection of the circuit diagram shows, that the original driving phase voltage
matrix at the rotors 1 and 2, both equa1 U r = -ro(1rl + l r2)' Upon substitution
of this expression into a pair of equations, each similar to (2.9), the first equation
with the index 1 and the second equation with the index 2 attached to the symbols
ofthe variables, we can determine the Y ri matrices, (i= 1,2). Then, by straightforward calculation we shaU arrive at a result, equivalent to the result th at
we want to derive by another argumentation. We take the other way, only to
avoid the introduction of a new constant Yr' Thus: we can look at the interconnected rotor system, as ifthe rotor phase resistances we re not equal to r" but
equal to ro +r" and, in addition, as if the voltage drop on a resistance ro due to
the phase current i gr2 should act as a fictitious driving phase voltage on the
rotor phase gr!, and vice versa, (g = a, b, c). Hence, changing everywhere rr
into ro+rr and taking
(2 .28)
=
36
where we used the equiva1ence of the machines, reason why we did not
distinguish R i and R;, nor N i and N;; but ,as generaHy e ti #- e t; and X i #- X ;,
the difference in notation as to S and Z app1ies. Here, comparing eq. (2.4),
Zo ri = ZOSi = 0 and tims, instead of eq. (2.29),
(2.30)
Consequently, the fictitious rotor phase driving voltage matrix (2.28) becomes
U ri = -
(2.31 )
where we restored the notation Z "j into (rX)rj. H ence, from eq. (2.9),
T ri = -
Uo n
-~ S ri-1S,.;(rX)r;, (2.32 )
ro + rs
FinaHy, T si given by eq. (2.13 ) under transposition ofe s into e Si ' T ri given by
eq. (2.33 ), w~ know the fuH Ti' (i = l , 2). Then, substitution ofthis Ti into the
final eqs. (2) yields,
cos( . +q;+~ - eSi)
Ys+ di'
d
de
de Si
di' - xYs
d
- xy ,.,
, - xYs
de"i
, Yr + d.' - di
, - ;cYr ,
de r i
+ di' Yr+ di
Xqsi
+
Xdr i
X q ri
(2.34)
(2.35)
where, as before, we omitted the zero components. We add th at here
37
Yr
(2.36)
due to the presence of the "slip resistances" ro' Now, to form a convenient set
of autonomous differential equations, we choose as a first condition of constraint,
(2.37)
a condition compatible with (2.35). Further, we introduce the new variables
(2.38)
and we suhmit the time dependence of ()r to the second condition of constraint
(2.39)
ti
The quantity s refers to the average slip of the two slip ring motors, whereas a
represents the difference in rotor position of two corresponding rotor phase
coils of the rotors. Substitution of eqs. (2.40) into (2 .34) leads to
cos la
-
'sln 21 a
)"'
d
Ys+ d?:
+ 1+( -) il,
d
ys+ d?:
,
d
, Yr + d?:'
-xYr
- xYr
X dsi
- xYs
xqsi
- s
X dri
d
+ s , Yr + d?:
xqri
(2 .41 )
38
t(xutt-Xut2),
(u = d, q; t
Xutv ,
s, r),
the new
(2 .42 )
(2.43)
Y.
(2.44)
Ys
-1 -XYs
+ 1
Ys
+i
- xYs -21a
- xYe
Ye - s
-xY. + s Ye
+i
Ys + 1 - XY s
1 .
-XYs
- 2a
+ 1 Ys
- s
- XYo
Yo
-XYo +s Yo
sin ia
Xds.
Xqs
Xdre
xqre
Xdso +
XqSO
Xdro
xqro
(2.46)
e- e
(2.47)
1,
ids,
i qs
i dre
i qre
, (2.45)
i dso
iaso
i dro
i qro
3. Particular Solutions
-II
l
= sc, ft
ft c, Xut
= Xut C, (u
= d, q; t = s, r),
l-l[l
(3.1)
XdSc
xas" =
x dr c
x qrC
Ys
+1
- 1 - XY s
Ys
. - XY s
- XY r .
Yr - sc
. -XYr + sc Yr
1
.
c- aYSYr) -SC(SCYs+Yr) J
_! [Yr(s
sC(sC- aYsYr) + Yr(scYs+Yr)
Ll
xYr(sc-aYsYr)
xY r(s Cy s+ Yr)
(3.2)
,
39
the well-known formula ofKLOsz [46]; the quantities e and Sk, representing the
machine constant of the KLosz-formula and the slip at breakdown torque
respectively, and, further, the quantity ft\ are given by
e=
~2ys
V(I+Ys2) (I + a2Ys2) ,
Sk =
Yr
VI I+Y
+a2ys2
s2,ft
(35)
.
_~c
I.t
(1 +e) are found in most text books on asynchronous machines [47]. In this
respect we add the remark that only few authors, although for instance LIWSCHITZ [48] does, establish exact expressions for the constants (3 .5 ). As in the
text books on induction motors, we, too, point to the slight influence of the
normal values of e as to the quantitative value of the stationary torque. But,
as we shall show further on, the stability of the stationary solution does depend
essentially on the parameter Ys'
40
With respect to the double supplied slip ring motor, we construct a particular
solution ofthe set (2.27) by the assumption
a
a C, fl = fl c, x"v
x C" v, (u
d, q; v
e, 0),
(3.6)
where the superscripts refer to the constancy of the relevant variables. Taking
a = a C, it follows by (2.24) th at () = 2, and consequently a = a C implies that
the actual angular speed of the rotor equals 2w o radians per second, where Wo
stands for the power supply angular frequency . Now, from (2 .27, 3.6),
(3.8)
Substitution of eq . (3.7 ) into the torque equation of the set (2.27 ) yields, after
some elementary calculations,
fl c
(3.9 )
Eq. (3.9) gives the (reduced and normalized) torqu e of the double supplied
slip ring motor running in synchronism. Such a motor appears to have particularities considerably resembling those of the norm al synchronous motor [41].
We believe the exact result (3.9) - exact with respect to the theoretical assumptions - to be new. We add the remark that, at first sight, one could imagine
the stability of motion of the particular solution to depend on a differential
equation obtained from (3.9) by giving a C a slowly varying time dependence
and - in accordance with eqs. (2 .23) - by substituting a second order differential equation in a C for fl c. Calling such a method a "quasi stationary
method", we shall show further on th at this quasi stationary investigations are
not very reliable. In particular the method will fail to predict the stability
problem with respect to the induction motor and will approximate the true
perspectives with respect to very slowly varying states of motion.
To calculate the particular solution of eqs. (2.45) up to (2.47 ) inclusive, compatible with the assumptions
41
d, q; t
= s, r; v = e, 0),
(3.10)
where the superscripts c have obvious meanings, we split up the matrix occurring in the substitution ofeqs. (3.10) into (2.45) according to
-1 ]-1[COS
[ +YS -1 .]-1
[.]
XCdse ] [Y S
- xys .
X:as. = _+ 1 Ys
. - x~s
dre
XY.
Y.
s
X
[
XC ar
-XY. + sc Y.
x:XCdSOaso ] =
[ X dro
Xcaro
-XYs
1 Ys
. - x~s
- XYo
Yo - S
-XYo +SC Yo
tac]
= _
cos tac
.1.
(3.13)
Substitution of eqs. (3 .11, 3.12) into the torque expressions (2.46) yields finally
sCxy
sCxy
.lac + __0 sin 2 .lac
.1 .
2
.1 0
2'
1
+
ay
2 (y
Y
)
SC
xy
SCxy
fl o" =
s ~ - ~ __e __0 sin tac cos ta c.
Yo
Y. L1 e .1 0
x
C
= - -' cos 2
(3. 14)
(3.16)
The constant erepresents a relative measure of the slip resistance ro such, th at
o <; e <; + l. Comparing eqs. (3 .14, 3.15, 3.16) with (3.4, 3.5 ), the torque
equations (3.14) are, rewritten in terms of the new constants,
42
(3.11
fl
(3.18)
.A~
definition
- :~
.,..Ao( "
~---:--""'
T
0(
~---tf-+'-";";"--';:"!---+-+-"';; ~ (e.r)~2H(rle.<)'
~~~.. <.)............L
ft"
(r<;y,).2fMC~
c ............ ..
(3.19)
.-N-"
t (f ~'/t)~2H(I/ft:.')
-f
A*
(t:.'1f)"H(fIt:.<)
.-..............
..................... -1
~o
r
0.2
0.1
......
~ .,J"'/
.....!
0.0
0.1
......
.~
O.l
43
and where, in eq. (4.2 ), the relevant dimensionless steady state mechanical
torque is represented by the functional form t5 (Q). Using the notations adapted,
we now express the original differential eqs. (2.1 8) conforming to
xXasdS
~dr
X ar
1[1
IIQ
[ +
YS1
- xy ,.
.
X ar
- 1 - XY s
Ys
- XY s
Yr
- 1+Q
- XY r l - Q
- Xdr
Yr
J-
X ds
X as
X dr
(4.4)
X a ,
44
the steady state torque regarding the particular solution. Introducing the
new perturbation variablesYut and w by
Xut
xut"+Yul> (u
d, q; t = s, r), Q = Qc+ w,
(4.6)
[j"1 [
j qS
jdr
jQr
_ Ilw
Ys
- 1
+ 1
Ys
- XY s
. ][y", 1
- XY s
.
-sc+ w + xqr:
- XY r
Yr
sC- w
.
- XY r
Yr
Xd"
oe
xQ /+Y Qn -Xclr" -Ydn - XQS c, x d."
Yas
Ydr
Yq r
'
(4.8)
D () -
Ys+
- 1
+1
Ys+
- XYr
XQre
- XY .
- XY s
- se
XQre
Yr+
+
se
- XY'r
Yr+ - Xclre
e
c
-Xci" - XQ Xds e oe+ Il
= O.
(4.9)
Here, in eq. (4.9), oe is a constant, equalling (ao jaQ )nc. Using eqs. (3.2,3.3) ,
we easily show the equivalence of (4.9 ) with
- 1 - XY s
- ;.cY r
Ys+
- XY r
Yr+ }' - se
-XY r + se Yr+ xY rjtJ
+ xYr + se -Yr oC+ Il
YS+
+ 1
D ()
0,
(4.10)
daries of the stable regions, we can use the HURWITZ criteria of stability, and
we put
D (2)
(4.12 )
al = 2Il(Ys+Yr) + bc,
a2 = Il[(Ys+Yr) 2+ sc2+ 1 + 2aYsYr] + 2b c(Ys+Yr) + (xYrjLJ )y"
a3 = 2Il[sc2Ys+Yr+aYsYr(Ys+Yr) ] + bc[(Ys+Yr) 2+ SC2+ 1 + 2ay,ys] +
+ (XYr jLJ ) [Y r2- sC2+ (1+ a)YsYr],
(4.1 3)
0, 1,2,5),
al a o
a3 a2 al ao
as a4 a3 a2
as a4
> 0.
(4.14)
and constitutes the generaUy known steady state condition of stability. Still,
there are four conditions left and it is highly surprising that the steady state
expression is generally accepted as the only non-trivial one existing. For here,
without any excessive effort, we see that the condition al > 0, equivalently - bc
> 2Il (Ys+Yr) , offers a boundary independent of SCfor the admissible negative
value ofthe local mechanical damping coefficient. We stipulate that it is quite
weU possible to have al <
without violating the steady state condition
bc > 0/-lcjoQc. Ana1ogous considerations apply to the condition a2 > 0, but
then in dependence on Sc. However, if bc > 0, and this wiU be considered as a
norm al case, al and a2 will be positive. Further, directing attention at the condition al a4 - aoas > 0, equivalently
46
C
UI(Ys
+ Yr) + <5 ,ll
c
1 (<5 - oflcjoQc), lllJ + 2<5 cYr [1 +aYs(Ys+Yr) J+ (xy/ /IJ) [1 +aYs(Ys+2Yr) J+ 2sc7 s5" - xYrjIJ )
1>0, (416)
.
(4.17)
'I bo
b2 bI bo
's b4 ba b2
bs b4
'3
2P(Ys+Yr) ,
P
2PYr+Ys+Yn p[l + (Ys+Yr) 2J + 1, 2P (Ys+Yr) ,
p
py/+ l
, 2PYr+ Ys+Yn p[l +(Ys+Yr) 2J + l > ,
Yr
py/+ l
F(p) - 4r,.3[1 + (Ys+Yr) 2Jp3 + 2Yr[1 - 2YsYr-2Yr 2- (Ys+Yr) 4JPZ + [2Ys+ 3Yr- (2Ys+Yr) (Ys+Yr) 2JP + Ys+Yr > O.
(4.20)
The inequality (4.20) does not always hold. Hence, even in quite a normal case
c
= 0, SC = 0, representing the particular solution to the unloaded motor, the
stability of the solution depends essentially on II too. Now, by means of graphs
of F (p ) versus p at different sets of val ues of the pair (y., Yr) , we determined the
appearance and the values of the real roots of F(p) = 0, and, in the appropriate
cases, the sign ofF(p) ne ar the roots. In this way we could tra ce the regions of
values of the triples (y., Y,., p) giving rise to an unstable final state of operation.
We visualized this interdependence in fig. 7. In this figure, the parametrie
<5
47
(4.1 9)
48
SC
Si = 0.4
3
0.2
~0.2
0.4
0.6
0.4
0.6
0.'
SC
pa: 0.6
o.B
0.2
o.~
0.4
0.6
0.1
-b=0.8
0
0
0.2
0.4
0.6
0.8
0.2
(je =
0.4
0.6
0.'
o.
of IJ lie somewhere on the intervals (0.1; 0.50) and (10; 50) respectively. Further, generally, Ys and y" do not differ very much. Rence, with a view to figs.
7 and 8, norm al induction motors under norm al operating conditions have
stabie steady states if oe = 0, 0 < se < sk, i.e. operating as locally unloaded
motors. As to locally loaded motors, i.e. with respect to the cases 0 < oe, eqs.
(4.1 3,4.14) offer the mathematical back-groundfor numerical calculationof other
stability diagrams. We expect those diagrams to be analogous to the diagrams
given by fig. 8, but to be of a more or less shrunk type as by 0 < oe additional
49
" ,z,;:",,,.,,.,r...
..
(5.3)
to constitute the perturbation torque of the machine i, (i = 1,2). We assign
to the constants IIi and to the symbo1s {J/ analogous meanings as pointed out
in the preceding section as to the quantities II and (Jc respectively. Moreover,
as we need the linear parts of the differential equations only, we shall indicate
the presence of one or more terms of the second or higher degree in the perturbation variables by . .. only. Then, on the one side we have,
/-l i - /-l /
(5.4)
IIiW i+ {J / W i+ . ,
(5.5)
(5.6)
+ ... , (i =
1,2). (5.7)
Further, in order to use eqs. (2.41, 5.7) for the construction of the differential
equations in terms of the p erturbation variables Yu ti , W i and fJ, we determine in
advance,
s = l -() = l - Q"- we = sC- i(w l + w2 ),
cos ia = cos (ia" - fJ) = cosia"+ fJ sin ia c + ... ,
(5.8)
sin ia = sin(ta"- fJ) = sinlac- fJ coslac+... ,
(J = Wo = l(W1 -W 2) = - i.
IIiW i
-(XarD'dsi-XdrD'asi-XqsD'dri+XdsD'qri+{J/W i )
Y.,
- 1-H wl-w2),
jasl
+ 1+l(W1-
jarl
-XYr
W2), y .,
Yr
Yr
jarl
jas2
jas2
jdr2
jar2
xqS 1"+Ya sl
.
sIn
1
"2
a" -
Xdrl"+Ydrl
Xqr 1c+Yqrl
c
Xds2 + Y as2
c
XqS2 +YqS2
c
Xr 2 + Ydrl
c
Xqr2 +Yqr2
cos i a C
- sin ! a C
+ fJ sin ! a
+ fJ cos i
+ ....
(5.9)
aC
51
Ys-1
+1
Ys
-se
Yr
jas2
Yr
J
i
jas2
Yr
-1
+y, 1
jar2
Yr
+XaSl"
- XSl
"
+ sina"
-coSa"
+Xarl"
-X dr l
-XYr
Yr
+S"
-XYs
-Se
Yr
Ya Sl
Ya Sl
Y dr l
"
Yarl
" + sinac
Yd S2
+cosa"
Ya S2
- XaS2
+XS 2"
-XYs
YB
: -XY r
jr2
+ ....
(5.10)
yd r2
+Xar 2"
-X ar2 c
Ya r2
215 2c
W l
W 2
-1
f3
The set (5.l 0) constitutes an autonomous set of differentia1 equations, and,
consequently, the stabi1ity of motion of the particular solution under investigation depends on1y on the character of the roots of the characteristic equation
to the (11 x lI) matrix to be formed in an obvious way from the square matrix
occuring in (5 .10). Now, our parametric domain being nine-dimensiona1,
(y., Yn x, ro /ro + rn IIl' II2' bI"' 15 2" , a" acting as the nine parameters), a gener al investigation on the boundaries of the stability regions will be too 1engthy a task,
even if carried out by computer techniques, and - if done - will not re sult in any
surveyable information. Nevertheless, we push our investigation slightly
further. At first, we remember that, although we assumed the motors of the
system to be equivalent, generally III -:;t. II2' as Pi has the meaning of the po1ar
moment of inertia of the rotor i and the 10ad i reduced to the rotor shaft. The
10ads may differ markedly. However, in many app1ications, especially where
52
loads are driven by gearings, the difference between the loadings affects the
difference between the effective polar moments of inertia only slightly. By the
effective pol ar moment of inertia of a loaded motor, we mean the quantity of
rotor and relevant load reduced to the rotor shaft. On the contrary, the
difference between the loadings essentially affects the difference between the
functions <5 l c (tJ l ) and <5 2"(tJ 2 ). In fact, the electrical transmission shaft just has
the task to receive markedly different steady state torques <5 l "(tJ l ), <5 2 " (tJ 2 ) ,
handling those torques under the isochronous speed tJ l = tJ 2 = tJ". Thus,
we may accept III = II2 = II as a good approximation. Then it is to advantage
to proceed to the perturbation variables W e and W o defined by (5.1 ), and to
the variables Yutv, (v = e, 0), given by
Yute = t(YUt1 + Yut2 ) ,Yuto = t(Yun-Yut2)' (u = d, q; t = s, r);
(5.11 )
the definitions (5.11 ) are in accordance with eqs. (2.42). In addition, we shall
use the quantities
<5.0 = t(<5 1"+ <5 2") , <5 0" = t(<5 l "- <5 2").
(5.12)
In terms of the v-quantities, (v = e, 0, where e refers to even and 0 refers to
odd), the system equivalent to the system (5.10) appears as
jdse
Ys
- 1 -"Ys
jase
+1
Ys
-"Ye
jdso
jaso
- "Ys
- Se
Ys
ja,o
jaro
Yo
+S"
- "Yo
"
+ sintac
"
C
C
+ Xa ro" -X dro - XqSO
-Xaso C
c
+ XdsO
Yo
Yd se
Yq se
+Xqre"
Ydre
-Xare C
Y ore
C
- Xos e
+ Xa se"
C
+Xqro
-X dro C
Yd so
+ cos -~aC
Yo so
+ ....
(5.13)
Yaro
yqro
We
<5 0 C
<5 eC
Wo
-1
f3
53
As to the quantities Y v , x utvC, reference is made to eqs. (2.42, 2.43, 3.11, 3.12).
The characteristic equation to be studied, follows in an obvious way from
(5.13). We note that the determinant ofthe square matrix of (5.13) equa1s the
constant term of the characteristic po1ynomia1 of the eleventh degree in , and
does not contain the parameter Il. Hence, the necessary condition of this
term being positive in order that the characteristic equation will be a HURWITZ po1ynomia1, constitutes the normal steady state condition for stabie
operation of our system. To derive this condition directly from eq. (5.13)
wou1d be a tedious task. Now we know the steady state va1ues of induction
motors to dep end on1y slightly on y ., provided the stator constant accepts
norm al va1ues on1y. Hence, let us calcu1ate the determinant in question at
Ys = o. Approximating the ana1ysis by means of the assumption TB = 0
(henceYB= 0) is a usua1 procedure [57,58]. In this way, using eqs. (3.11,3 .12,
5.13 ), we determine the va1ue of the determinant equa1 to
2
Xy e
XY02) [ C (
SC2_ y /
(je - XYe LI/ cos
( Llo -
T.
_ (SCXYe
Llo
+ SCLleXYo) [(j 0
Y02. 2.1 ) ]
aC + x yo SC2_
Ll02 sm 2ac +
2
2
(SCXYe _ SCxy 0 ) sin ac] sin aC
Ll e2
Ll02
.
21
(5.14)
Ofle") _
oQc
Lle
C _
C
Oflo ) sin aC (5 15)
oQc
.
instead of (5.15). Hence, under the norm al condition C> 0, we attain the steady
state stabi1ity condition,
0flo"
0flo c
(joc- oQc .
(jo"
osc .
1
) 1
(5.17)
C >
a C SIn aC =
a C SIn aC
(- - e 1
_C +2-1
(jc _~
oQc
(j c+ ~
osc
As long as (j1 c and (j2 c assume nonnegative va1ues on1y, we may state for the right
hand side of eq. (5.17),
(5.18)
54
the last expression being the absolute value of the slope of the isotache at the
point (s", a") in the torque-torque diagrams of fig. 6, whereas the slope in
question has the sign of the angle a", (la"1 < n /2). Therefore, incorporating all
normal loads by the conditions bi" ' 0, (i = 1,2), the steady state stability
condition becomes
(5.19)
With respect to eq. (5.19), it is natural to determine the slope ofthe isochones.
Direct computation yields,
(~::)s" =
G+e) ~
"
cotg
a".
(5.20)
Hence, by means of eqs. (5.19, 5.20), we can write the stability condition
(5.19) into the more involved form
I
eT-- -e c -e + e
-"1
")'
" (' OftO
"
aft." s"
(5.21)
The result (5.21 ) differs principally from the stability condition obtained by
HANNAKAM
[49], vi;;;.
(5.22)
the condition (5.22) will be found by the assumption that the set
",u 0
t
0.2
o.t
."
0.0
o.t
.6
oS"
0.2
-+-+-.;:"
..,.e.9 to
0.3
up into two separate factors. One factor resembles eq. (4.9) and offers the
characteristic equation of the "even" part, representing the characteristic
equation of a single asynchronous motor having Ye as rotor constant, IJ as
normalized dimensionless polar moment ofinertia and CJe" as the local mechanical
damping coefficient. In the preceding section, this characteristic equation was
thoroughly studied. The second factor reads
-XQse C
y.+ - 1 - "Y.
+ 1 Ys+
- "Ys +xdse c
-SC
Yo+
-"Yo
- "Yo +sc Yo+
+xQre" -Xdre C -XQeC +xdse" CJe"+IJ?
- 1
+1
= 0.
(5.24)
56
-1 -Xqs "
y.+.
+1 ys+. +xas." +1
+Xqr." - Xar" o."+I!
-1
.
2lIys +0."
II
0." + (1 + a)Ys"y.2 JLl., lI+ 2y so." + "y.2 JLl., 2lIys+ o.o
, "y.2 JLl.
, 0.0+ (1 + a)Ys"Ye2JLl.
> O.
(5 .26)
To grasp the character of (5.26) under normal conditions we use lI "y.2 JLl.,
bringing
o."{ (2lIys + oe") [0."+ 1( + a)y s"Ye 2JLl.] +[[2} > [[2,,2Ys"Ye2JLl .,
(5.27)
0."
> x7s"Ye2JLl .
(5.28)
(5 .29)
into (5.28), then we have the condition
SC
>
,,2YsYe
+ ,,2YsYe
(5.30)
57
CHAPTER 11
ANALOG COMPUTATION
-s
=
=
=
-XQS - XY sXdr
- H (r: ),
Xd s+Y sXQS
- XY sXqr
- XY rXds
+ Y rXdr- SXqr
- xy ,.xqs+ sx dr + y"xqr
y sx ds
(6.2)
1
II (Xd rXqS-XdSXq,.)-( bCjll ) ( l - s)
Xu t(O) = $(0)
0, (u
d, q; t = s, r).
59
.~
.,
.~
.2
:~
I_n
2-J-2:~
'--25
5-6-2~
,...-----l.
3~t
'l
fa
.,
8"9~1()-11
r-_~.c::::==~-3.
:1
'
Il-13
.~
1't-iS-IS
7::::::D.---t.
., -"
,s~
11
~
.
rr"
6,;JI"
=-ftIoo 8
1
+IOoV
2111
18--""'" r - - - - - - - - - - \ i
Uil i! !.!
iX
16
+.ooV 21 H 292.6j
l.
XiX
'---;D--~'
25"
misslOn shaft, we used eqs. (2.41) and the torque expressions (2.35) supplemented by normal mechanical torque dependences according to (6.1 ). The
complete system reads:
60
- iaSI
- iqSI
- iarl
- i qrl
=
=
=
=
ta,
H
YS(XqSI -~Xqrl) +(l-t)xasl- (.) sin ia,
Yr(Xdrl- ~Xa.l) + Y o(Xar2- ~Xas2) - (l-)xqn,
Yr(xqrl - ~XqS I) + Yo(xqr2- ~XqS 2) - (l - )xarl ,
Ys (Xas2 - ~Xdr2)
- iaS2 =
- i qS2 =
- idr2 =
- iqr2 =
..
..
Ys(xqS2- ~Xqr 2 )
(6.3)
2 = +i
= H 2 + l ), t = H 2 - l ),
Ys = rs/awo(ls+ m.,) , Yr = (ro+rr) /awo(lr + mr), Ya =
ro/awo(lr+ mr) .
~:D--I(,Jr~
IC
-T-"
= ro/r,. = t [ (I /e
o(Je
1].
61
t =
(6.4)
H()2-()1),
62
I"
O,l
/'"
0,2
63
0,3
0,3
0,5
1,0
cf~o,015
0,1
o,s
~
0,3
0,3
0,5
64
X.O,6
0.5
05
Finally we have to ex amine the influence of the constants Yso Yr' Herewith we
arrive at our main object with respect to slowly varying states ofmotion. Therefore, we reproduce the final states of motion only although the results have
been obtained by simulating the behaviour of the induction motor when
started from the initial condition of rest. Fig. 18 shows some results. We note
that in cases of small electrical damping the "mainly electrical transients"
have not died out when the final states are approached. We know that it is not
allowed to separate transients from other states of motion in our nonlinear
problem. But the graphs belonging to normal IJ values reproduced so far,
affirm the statement that the "mainly electrical transients" cause the violent
torque pulsations at low speeds (j. Therefore, interpreting the graphs of fig. 18
with respect to the influence of Ys> Yr as to slowly varying states of motion, we
disregard the ringing transients discussed. Now, inspecting fig. 18 and realizing
th at each specimen of the curves approaches the slowly varying states from
a different "initiai" condition, we see th at Ys affects the slowly varying states only
slightly. On the contrary the influence of Yr is essential.
65
Tr. Z
i"
0,'
0,1
"
DJ
','
0,'
O,S
_0,1
- 0,.1:
-O,~
-,.
i"
0,+
-0,1
'.1
.5
_,5
',5
',1
66
O.
~=O,2
~=D,l
r,. :D,~
IC:z/o
1'C=IO
IC=IO
f'
t,.=o,S
X~/O
1,1
-0,1
-0,1
r,.:rO,2
IC =20
1,1
IJ
1.1
1,1
1,1
IJ
-0,'
-0,1
~:O,2
!'
IC=30
;;.
-0,1
_0, 1
~:D,.
J'C=JO
0,2-
0,1
0.8
-0,1
0,9
~I
1,/
iJ
9
-O~I
67
Results obtained by the analog computation technique from eqs. (6.3) using
the block diagram of fig. 12, are shown in figs. 19, 20. The graphs in fig. 19
refer to the case e = I N 5, Ys = 0.1, Yr = 0.5, IJl = IJ 2 = 16, bIc = 0.32; we
changed b2c by equal steps for each new run. The sequence of graphs illustrates
how the system by loading more and more the motor 2, finally fails to produce
the torques PI and P 2 in proportion to bIc and b2c In that case, the isochronous
speed = ( 2 + I ) = I = 2 is not reached; here, motor 1 exerts the torque
PI at a norm al slip value, whereas motor 2 runs at a slip value higher than the
slip Sk at break down torque. Hence, the system no longer works as a true
electrical transmission shaft. We have found the system to be stabie up to
68
()2 C
'.1
'.1
We examined the oscillating motions illustrated in fig. 20 also from eqs. (6.4),
using the corresponding block diagram of fig. 13. Comparison of the limitcycles being found from the approximated eqs. (6.4) and those from the exact
set (6.3) again yields a justification of the statement that with respect to slowly
varying states of motion, the value of Ys is of small influence as long as all
parameters involved lie on "norm al" intervals. Hence, we may conclude that
the parameter Ys will be the very quantity to construct approximating theories.
In the following chapter we shall develop the methods.
69
CHAPTER III
S =
(u = d, q; t = s, r),
(9.1 )
to constitute the general solution. We substitute eqs. (9.1 ) into the system
(2.18,4.2 ), hence into
70
0,
(9.3 )
(9.4)
(9.5 )
] =
XY r
[Y r -So] [x'l1'o]
+ so Yr
Xq,o
+ [~(I1'o]
Xqro
(9.6)
With respect to the first order eqs. (9.4), the same substitution leads to
'l ~ .
1
.
-XYr
-1
+ 1
- XY r
Yr
. -XYr + so
(9.7)
(9.8)
where, for the description of the final states of motion, again we retain the
71
constant part of the solution (9.8) only. Then the remaining parts of the
set (9.7) can be simplified to
d, q; t;=0, 1) ,
"
2
p = IIY r , d =
"
<5
Si ( '
0 1) , -j
i= - , t = ,
c
Yr.
Yr
Yr
dy = d----;
dy
Yr"
"
= -d
)I,
(9.10)
"
(9 .13)
Elimination of X ao from the set (9.13) gives
Now, provided
I I I1 +o2- ~:'I
;o",'
(9.15)
app1ies, eq. (9.l4) offers the means to construct an approximate phase diagram
by the method of isoclines [65], originating from LINARD'S studies [66].
Certainly the condition (9.15) does not ho1d if
(9.16)
Hence, constructing a phase diagram with the help of eq. (9.14) using the
condition (9.15), we know that some region near the 1ine o' = o(l +02)
fai1s to give reliab1e information.
72
(9.17)
Further, in the field of isoclines of fig. 21, we have drawn the 1ine o' = o
(1 + 02 ) . We see the isoclines to change the slop es rather abruptly in the
neighbourhood of this 1ine. Hence, if we exclude such a region, the isoclines
give rise to phase curves spiralling smooth1y to the stabie point. We have met
this type of motion in the results obtained by ana10g computation in chapter 11;
in this respect the work of BLOM [67] must be mentioned particu1arly.
\
\
\
I
I
+1
1-~
I I I ~e I
I +.8I I +.6I I +.4I
\ \ \
\
\
\ \
\ \
\ \
1-.~Y,.
\
\
\
I I \
-~o
-.8
-.6
l-.',.
\
/ I I
/ I
,--// /
Let us once more use eq. (9.14) in the case d = 0, and, instead of approximating the equation by the condition (9. 15), let us consider the region where
02 1, hence -0.1 .;;; o .;;; +0.1. Then eq. (9.14) can be approximated by
73
+~
o, =
p o
0,
(9.18)
having a solution
o
(9.19)
Thus, near the origin, the phase curves have the character of a normal damped
motion and the motion is criticaly damped if p = 4, hence IlYr 2 = 4%. Here,
we refer to fig. 18; apparently the case Il = 15, Yr = 0.5, % = 0.9 having
IlYr 2 J4% R::::i 1.04 is criticaly damped, perfectly checking our conclusion.
The graphs constructed by the approximated method of isoclines compare
rather weIl with the results obtained by analog computation techniques.
Now, sometimes use is made of quasi stationary methods, wh ere - in the case
of the induction motor treated here - the expression of the electrical torque
arises from the steady state expression, thus from the Klosz-formula (3.4),
there by taking s for SC and fl for flc. Thus the relevant quasi stationary investigation starts by studying
fl
= -Ils+<5(I - s)
Sk
Sk
(9.20)
- + 28 +-
Obviously, the differential eq. (9.20) must fail to describe the damped harmonic
oscillation near the stabIe, particular solution. In relation to this quasi stationary investigation, our graphical method gained information by the fact that
our treatment brought in not only the slip variabIe sitself, but - and of course
in addition to the term Ils containing the derivative of s - terms where s, S" and,
unfortunately, the third derivative of s also occur.
As we used the zero order eq. (9.11) only, the influence of Ys has disappeared
totally. An investigation ofthe graphical solution ofthe second order eq. (9.1 2)
is possible, but hardly worth the effort. As our main object emerges in the
stability problems of the electrical transmission shaft, we dis miss the solution
of the second order eq. (9.12).
~dSe
~qse J
=
l -
+ t
+ 1 .1.1 _ 1_
]!a.
+]a . - 1
+ 1
- t .
Xdso
i qso
l l
dse
X qse ]
X
.ta .
sin ta
COS
X dso
X qSO
(10.1 )
X dso =
X qSO
ta,
= sin T cos
X qse =
(10.2)
as first integrals. Hence, rejecting the terms having period neariy one in T,
we accept
X dse
= 0,
X qse
= - cos
ta,
Xdso
= + sin
ta,
X qSO
= 0,
(10.3)
to offer the approximated expressions for slowly varying states of motion even
if Ys
0, but Y s is normal. Substitution of (10.3) into the remaining parts of
(2.45, 2.46, 2.47) yields
i dre ]
~qre
- s
Ye
+s
Ye
Yo
- s
+s
Yo
X qro
sin ta,
Xdro
i qro
fte =
-Xd re
cos
ta -
(10.4)
ft o =
- X dro
cos
ta -
X qre
sin
ta.
TVY eYo,
s
-v-'
= -ft v ,
ft v
Y eY o
X uv
"
= -X
urv
, (u = d,q; v = e, 0 )
(10.5 )
~d< J
_
X qe
-
Xdo
-
X qO
ile
II
= -
i de cos
-
1
ta -
i qo sin
ta, il o =
(10.6)
i do cos
ta -
i qe sin
ta.
75
Here, our notation of the standard mechanical torque expressions (5.3, 5.4),
(10.7)
has to be brought in accordance with the new quantities (10.5). Using Ys
and
(2.47 ): a
(2.44) : ft.
(3 .15 ) : Sk
= 0
(10.8)
it foIlows
- "ft-. -- ft.
I -rO
/.
rO
(10.9)
apply; moreover, we h ave dropped the higher order terms being represented
b y the dots in (10.7). Note that the particular solution of eqs. (10.6, 10.9)
agrees with the description laid down in eqs. (3 .10) up to (3 .18) inclusive,
but then taken at Ys = O. We need this particular solution furtheron; it reads
in fuIl
'
+ (e /x) ]- l.
la",
(10.11)
(10.12)
Xd'
Xa.
Xao
Xao
(10.13)
-xqo sin ia = f
-
1( J )f( ) .
, of() da ,d [of( ) sma
. ]
2 e
sma+ta-di+tdi T
,
1
-Xdocos 2 a=
-xqe sin ia =
+ H1 Je)f(l J) sin a +
(10.14)
ile = Hf()+f(l J)J - Hf() - f(l J)J cos a_ ~ [.lP() +P( l J) _ .lP() - P(l J) cos a]
di 2
,
ilo
(10.15)
Fe
Fo
Fe()
Fo()
=
=
Fp = Fp()
=
=
(10.16)
il.
- _ F'
fto p sm a -
oFe da
a
di +
d [oFo . ]
di
sm a .
(10.17)
(10.18)
a" =
dw
w da'
a' =
w, di = wda' ilvs =
(10.19)
+ !Dow+Fe-Focos a =
il.',
(10.20)
In the 1ines preceding eq. (5 .11 ) we pointed out that normally no will be
small compared with n e. Assuming n o*- 0, we expect a kind of interference
phenomenon, reflected by the first two terms out of the leading terms of the
set (10.20). Such a phenomenon complicates the description severely and
adds 1ittle to the understanding of the main behaviour as to stability of the
electrieal transmission shaft. Besides, constructing an equivalent set by means
of eqs. (10.20) by evaluating the first (second) equation of the set minus
n o/n e times the second (first) equation of the set, taking n o <; 0.1 n e, estimating carefully the relevant terms of the resulting set, we can conclude that
n o= 0 offers a fairly good approximation if n o <; 0.1 n e. Hence, for practical
applications, we take n o= 0 exactly, dismissing the interference phenomenon.
To reduee the complexity of eqs. (10 .20) onee more by one step, we note that
has the sense of the average slip (divided by Sk) of the two motors involved.
Therefore, we expect th at exerts only relatively small deviations from some
mean value of as long as the departure from the isochronism of the relevant
rotors is not too large. Wetranslate this consideration mathematically by
treating d /da as zero in all terms of (10.20) but in the term n ew d /da of
the first equation of the set, as n e is a relatively large quantity. In this way
we attain the simplified results
78
- eW da
d
II
+"21
lil-
"2
dw
e W da
[~U o - 2FeFo
. ]
- - Slna
+"21
W+
Fe- Focosa
- ,
/-l:
(10.21 )
[U~ e + T
oFe + T
oFo COS a] w+ F
_p sm a /-l:.
If we treat as a constant, then the second equation of the set (10.21) may
be classified as a LINARD'S equation [66] and resembles the equation of a
synchronous motor investigated by COPPEL [5] considerab1y. Because of the
smooth behaviour of Fe, Fo, Fp as to on the interval of where instabilities can
be expected [43], we shall take the approximation to be constant in this
equation. As to the functions Fe, Fo, Fm we refer to fig. 22. Once the approximated equation has been solved, meaning that we have found a solution
w = w(a, ), we can use the first equation of (10.21 ) to determine = (a)
in approximated terms. Now, with respect to the second eq. (10.21 ), let us
define the quantities
Q =
r"
1='/2
f(el, 1('1e,), {, ~,
+6
o~
F"
I('/t:l
__
~~
__
----
---
~
I(e)
_____
__
___
it -~
3
Fo
./= 'ILt
f(z:>,rW~),Fe,~, ~
/('1~)
(t:)
_2
79
(10.23)
fh
We add to eq. (10.23) that in our normal cases Idvl 1, (v=e,o), hence
If(a) I 1. In the two following sections, we shall study the nonlinear eq.
(10.23), describing the essence of the dynamical behaviour of slowly varying
states of motion of the electrical transmission shaft.
Q2 (a)
Q2 (0) - 4 sin 2 ia
(11.1 )
2nfh
! f(~) Q (~) M.
o
(11.2)
The meaning of (11.2 ) is obvious: as a crosses the interval (0,2n) once, then
the work do ne by the driving torque fh equals the work absorbed by the
damping torque. Let
f m = sup lf (a) I, Qm
= sup IQ (a) I,
(11.3 )
then, from (11.2), we have 0 <: Ifhl <: fmQm, and consequently
a
o <:
(11.4)
M.
+ 2M.
(11.5 )
2nl-L
f f(~)VQ2 (0) -
H d~,
4 sin 2
( 11.6)
H,
(11. 7)
2 jQ (0) ;
f
o
(8 jk) f (de+do- 2do sin
H d~ =
,,/2
( 11.8)
The integral (11.8) can be expressed in terms of K (k ) and E (k), the complete
elliptic integrals of the first and of the second kind respectively. Indeed, from
(11.8) follows [68]
1- k2
2nfL = (8 jk ) [ (d.+ do) E -2do 3k2
1- 2k2]
+ 2do~
E ,
(11.9)
8 1k
2k2
[(de+ ido)
~f
(de-ido)
~;].
(11.10)
high speed SJ(a) wiU throw the system out of isochronism. Here we meet the
crucial point why the stability investigation based on the HURwITz-method
with respect to the linearized part of the nonlinear differential equations, fails
to draw practical boundaries as to safe, stable operation as demonstrated in an
elegant manner by MALKIN [70] with respect to another case.
12. LiJnit-Cycles
The eq. (10.23) admits another solution than the solution being considered
in the preceding section. Here, we ask for a solution such, that a varies periodicaUy in time between the extreme values amin and amax bounded by
- n < amin <; a <; amax <
n. For, if lal surpasses n, pole slipping will
foUow and apparently the considerations of the preceding section apply fuUy.
Let us start the investigation by studying the symmetrical case where the
difference torque vanishes, hence, where !-l = O. Then, because of symmetry
(12.1)
Equivalent to (10.23) with vanishing !-l is
SJ2(a)
SJ2(0) -
4sin
la -
2/ f(~)SJ(~) d~.
(12.2)
( 12.3)
(12.4)
/ f(~)SJ(~) d~
(12 .5)
0,
SJ2(0) - 4 sin 2 !a-2jmSJmam <; SJ2(a) <; SJ2(0) - 4 sin 2 !a+2jmSJmam, (12.6)
and for dv (hence fm) sufficiently small it follows the periodicity condition
from (12.5, 12.6), correct up to the first order in dv ,
82
HM =
O.
(12.7)
H=
sin
~;
d.+ do cos
(12.8)
0 = 4k 2
of
COS2
rp
Vl-kZ sin 2 rp
drp.
(12.9)
,,/2
rp
.
drp
k sm 2 rp
COS2
VI
,,/2
,,/2
2
0
VI - k2'sm 2 rp
2
)
=
=
(12.10)
(12.12)
We rewrite (12.12) in accordance with (10.22), yielding
(~
o ok
o ok'
(12.13)
"*
!h = 0 because of !h =
aK _ (d -.id ) aE - 0 . G _ 1 aE - ~_.
(d.+ .id)
3 0 ak
3 0 ak - ,.. - + aK - d _ J.3.d0 '
(13.1)
aE -- 0 ,... G -- I + aK
aE --"23 (d.+ d0 )
(d,+3Id0 ) aK
ak -32d0 ak
( 13.2)
In eqs. (13.1 ) and (13.2), the modulus k has different meanings. Thus, for
(13.1) the definition (11.7) - i.e. k = 2jQ(0) - applies, whereas for (13.2) the
definition (12.8) - i.e. k = sin t am - applies. By means of
sin iam
2jQ(0)
(13.3)
G(a m)
G(2 sin-Ik)
1+
~;,
a rn
in case we are
(13.4)
we no te that G(am) is a very smooth function in am and that G (am) possesses the
attractive properties 0 .;;;; G(a",) .;;;; 1. Moreover, aG jaa", = 0 at a", = 0,
n, and a2 Gjaa m 2 = 0 at am ~ 102 degrees. The elementary function
(13.5 )
approximates G(a m) very weIl, H (am ) having the properties O.;;;; H (am) ';;;; + 1
as -n ';;;; am ';;;; +n and the properties aHjaa m = 0 at am = 0, n, and
a2 H jaa m 2 = 0 at am = nfV3 ~ 104 degrees. For comparison of G as to H we
84
HJG.
t B
6
__
~~~
30
~~
____
90
refer to fig. 23. Let us return to eqs. (13.1, 13.2). Then, pulling-out ofisochronism is possib1e, if
O<
- 3
(13.6)
In cases where neither (13.6) nor (13.7) applies, the system must be stable in
the large or totally unstable. Hence, using (13.6, 13.7), we can extend the
typical behaviour of the electrical transmission shaft with the occurrence of
stability in the large and the occurrence of instabi1ity. Then, in the symmetrical case, we can classify this behaviour for 0 < do and do < 0 as follows:
o<
do
(13.8)
85
d. <
= or if 0 < - do < d. <
hen ce generally ifO < d. + do < +=.
< 0<
+=,
(13.9)
Vnder the last condition, i.e. if 0 < d. + do < +=, the system is stahle in the
large. We add that 0 < -tdo < d. < - do gives rise to limit-cycles, whereas
the stahility at 0 < d. < - td o will he violated hy sufficiently high impact
loads. By means of eqs. (10.22, 10.13), we can show that the general consequence of (13.9) is rem\.rkahly simple and instructive: the positively symmetrically loaded electrical transmission shaft can he loaded up to the point,
where each of the motors handle their own maximum torque.
86
CHAPTER
IV
MEASUREMENTS
87
es and er can
. (14.3)
As to rxc occurring in (14.1 ) we need the definition eq. (3, end eh. I ) for
and eq. (3.2) yielding
rx
YS(Xas"-UXdr")
Ys(xqSC-uxqr")
0
aYsYr(SC- aYsYr) -scYs(scYs+Yr)
sCy s(sc-ay sY r) + ay sy r(scy s +Y r)
LlC=
2+
(14.4 )
(sC-aY sYr)
+ (SCYs +Yr ) 2.
By inspection of eqs. (14.3, 14.4) we see that the amplitudes of the harmonie
time-variable components ntrtigtc juo, (g=a,b,c; t=s,r; ns= l, nr= n), of
N Rlcjuo appearing in the left-hand side of (14.1) are given by
(14.5)
2n ra dlans and
4n ra dlans,
. C1ags Z.at C b Y 3
CI
Zct
agsZatCb Y 3
w h ereas t h e current Zbt
(t = s, r). Hence, it will be sufficient to determine the phase-lag of ias c to
Uas as ia." constitutes the actual line current and phase current of phase as,
whereas (compare eqs. 2.8,2.10, 2.12 )
uas = t(ucs.as-Uas.bs)
VI U
o cos (T+<P+~)
( 14.6)
acts as the phase driving voltage. With respect to i ar C, we can not determine a
phase angle as to Ua. because of the different frequencies of their harmonical
time dependency. However, if we transform the actual rotor current to the
88
reference frame where the stator currents are measured and defined, a definition of such a phase angle is possib1e.
From eqs . (14.1, 14.3, 14.4) it follows
'"1 - V2
rslas
o-
"3
Ys
n) +
(+ lP +::'6:)} .
s"(sc-aYsYr) + aYr(S"Ys+Yr) .
[J "(S"2+ a2y/)]'t.
sm r
(14.7)
s"(s"-aYsYr)
+ aYr(s"Ys+Yr)
---+---,:.~'-'---'--,:....,..,..~'--'-~
s"(S"Ys+Yr) - aYr(S"- aYsYr)'
(14.8)
r gS
r:
r:
(14.10)
In the complex plane the point assemblage rgs"(s") constitutes the well-known
OSSANNA-diagram; we note th at the expression (14.11 ) is simpier than (14.10)
and therefore we shall use the point assemblage Zo."(sc) to make the measuring resu1ts into an impedance diagram, strongly recommended for instance
by CANA Y [73]. Let us normalize the impedance Zg .," by the d efinition
7
zos" = '\-os" rs =
+ 1..
Yr + jsc
y.<aYr +'
JSc,
(14.12)
then we have
j
,,2
Ys
ays
c _ ,,2
S"Yr
a Real part Zo." _ . c _
Real part Zos - - C2+ 2", 2'
aSC
- 0 If s - aYn
ar r
Ys S
Zo."(O)
1+-, zo."(=)
Zo s"(aYr)
ays
,,2
1+ - , IZos"(O)-zgs"(=) 1 = - ,
(j- +-j)
1-2 -+ i
ays
Ys ays
(14.13)
,,2
2 + HZ ysC(=)+ZgSC(0)].
ays
89
.x 2
u ~ -----'"
ud's
po s. re a l axis
Zg Sc ,
The impedance circle diagram on diameter [.(;g.o (0) , .(;g.o( 00)] is shown in
fig. 24; the diagram possesses attractive features compared to the OSSANNAdiagram [73]. In addition to the notes of CANAY we remark that in the impedance diagram, the points representing measurements at normal slip values
have a large distance to the origin and cover a relatively large arc of the
locus SC. Hence, to establish the whole circle, we do not need measurements
for the determination of Zg sc(l) ; we shall avoid such measurements especially
because of the doubtful information regarding the influence of iron losses as to
the fictitious copper losses once SC < Sk . We regard such an information even
as unreliable if SC = I, hence at blocked rotor. An aid to an accurate construction of the impedance circle from measurements is offered by the property,
that every line running parallel to the straight line through .(;osc (O), .(;g.o( oo),
is a linear point assemblage of uniform scale in sc, correlated with the impedance circle by projective means through a ray-pencil in Z9 8C (0) [74].
We performed the measurements at an asynchronous slip ring motor, constructed for 220 V 6 /380 V Y, 8 hp /5.9 kW, n = 1450, 50 cycles, 22 / 12.7 A,
90
40 j
0 --0-
-.01
theory
'~
SC
o ~
- .0133
s~ meos.
.01
"'-. 0133
30j
25j
.02
20j
half-plane
.0265
-.03
.03
15j
10j
.0 5
.0667
-20
-15
o
I
-10
10
I
-5
20
30
10
.06
15
40
20
50
Zgs" .
(14.14)
91
We see that rs = 1.39 / .680 = 2.04 times the actual stator phase resistance.
Now, as rs concerns a series loss-resistance as to the constant power supply
voltage, this fact confirms the statement that normally the iron losses equal
the copper losses. The rotor resistance rr can be determined from fig. 25 if
either igr"(sc) or Mc(sc) is measured in absolute values. Here, we shall determine the fictitious rotor phase resistance rr from the additional information
obtained from the measurements regarding the electrical transmission shaft.
Those measurements follow in the next section.
ro =. '511
2.5
~
I
I
I
I
' .5
f
I
I
I
I
t\.12Y:::
1.0
I
/
12:S..
fo(
0.5
270
12a
. 1270 .124501225
1300
.e05
00
. 1079
1048
.11~()
." 75
.",
. 1027
~9.1
.""- - -
i4.
.p90~
.....
.- -'iJ
--~------{!)
.-
-~
. M"kgm
;norY
0.5
1195
1345
"360 .....
1375
"'-t.1250
/
/
1100
_\20e"80 ~"30
I ~~~"'2\'1190
1 '260.123~ 205
/
.m. ~
1.0
1.5
2.0
.1485
92
2.5
ro= .30Jl..
n,t /m
o
4SS
3~ 7
220
220
. S24
. 53 0
----,. . . . M"kgm
2.5
Fig. 27. Measuring results torque-torque diagram; ro
.30 Q.
93
Fig. 26 shows the steady state results ifro = .15 Q, fig. 27, ifro = .30Q and
fig. 28 if ro = .60 Q.
ro =
. 6011
2.5
M2 , kgm
t
n , t/m
-----. M1 , kgm
-0.5
2.5
-0.5
........
94
To
.60 fJ.
.207
th us rr
n if C= 1
650 t /m
300 t /m
o t /m
.567
.800
1.000
Y.
(>2
r., Q
1.01
2.01
3.14
.363
.179
.116
.175
.133
.158
(15.1)
95
REFERENCES
I. Die Entwicklung der Starkstromtechnik bei den Siemens-Schuckertwerke, Herausgegeben aus Anlass des 50 jhrigenJubilums der Siemens-Schuckertwerke AG, S 11- 25.
2. W. V. LYON and H. E. EGGERTON: Transient Torque-Angle Characteristics of Synchronous Machines, Trans. Am. Inst. El. Eng., 49, 1930.
3. Y. H. Ku: Acceleration Plane Method for Nonlinear Oscillations, Proc. Symp. Nonlinear Circuit Analysis, Pol. Inst. of Brooklyn, New York, N.Y. , Apr. 23, 24 ( 1953)
129- 153, especially 144-152.
4. B. L. KONSTANTINOV : The Problem of the Approximate Solution of O scillations of a
Synchronous Machine, El. Technology USSR, 2 ( 1959), 5/6 ( 1958) 224-229.
5. W. A. COPPEL: On the Equation of a Synchronous Motor, Quart. J. of Mech. and Appl.
Math. , 12, part 2 (May 1959) 242- 256.
6. K. P. Kovcs und I. Rcz : Transiente Vorgnge in Wechselstrommaschinen, Bd. 11,
Verlag Ungarischen Akademie der Wissenschaften, Budapest, 1959.
7. TH. LAffiLE: Moderne Methoden zur Behandlung nichtstationrer Vorgnge in elektrischen Maschinen, Bull, Schw. Ver., 41 ( 1950) Nr. 14,525- 536.
8. G. KRON: Non-Riemannian Dynamics of Rotating Electrical Machinery, .1. of Math.
and Physics, 3 ( 1934) Nr. 2, 103- 194.
9. G. KRON: Tensor Analysis of Ne tworks , Wiley, New York, Chapman and Hall, London
1939.
10. G. KR ON : The Application of Tensors to the Analysis of Rotating Electrical Machinery,
2nd ed., Gen. Electr. Rev., Schenectady, N.Y., 1942.
11. D. C. WHITE and H. H. WOODSON: Electromechanical Energy Conversion, M.I.T.,
Dep. of El. Eng. , Sept. 1957.
.
12. R. H. PARK: Two-Reaction Theory of Synchronous Machines I , Tram. A.I.E.E. , 48
( 1929) 716- 727.
13. L. P. VERETENNIKOV and D. V. VISELOV: The Problem of Diverse Writing of Different
Equations ofthe Gorev-Park type in variables (J), Abstr. from Pap. Publ. in Elektrishestvo,
11 ( 1959) 8-12.
14. L. P. VERETENNIKOV: Classification of differential equations of the Gorev-Park typ e,
Abstr. from Pap. Publ. in Elektrishestvo, 11 ( 1959) 13-20
15. J . GILTAY: Over ferromagnetische toestanden, Tijdschr. v. h. N ed. Radiogenootschap,
15 ( 1950) 253- 274.
16. H. SEIDEL: Ein Verfahren zur Analyse der magnetischen Vorgeschichte einer ferromagnetischen Messprobe mit HiJfe des Preisach-Diagrammes, Z. angew. Physik, 12 ( 1960)
493- 501.
17. I. I. TRESHCHEV: Investigation of Variable-Speed A. C. Machines, Abstr. from Pap.
Publ. in Elektrishestvo, 2 ( 1957 ) 18- 34.
18. J. A. STRATTON: Electromagnetic Theory, Mc Graw-Hill Book Comp. Inc., N.Y. 1941 ,
pg. II (3), pg. 64 ( 1.20 etc.).
19. F. OLLENDORFF: Technische Elektrodynamik, Bd. I, Berechnung magnetischer Felder,
Springer Verlag, Wien, 1952, S. 62, S. 71 (1.l5c).
20. Ref. [18]: pg. 121.
21. R ef. [11]: Ch. 4, Section 4.2.2. "Mechanical Force Due to Magnetic Field Coupling".
22. R ef. [19]: S. 286 (IV I, 18).
23. R. A. FRAZER and W. DUNCAN: Elementary Matrices, Ch. 111, Cambridge, Univ. Press,
1950.
24. E. MADELUNG: Die Mathematischen Hilfsmittel des Physikers, Dover Publications, N.Y. ,
1943, S. 88-90.
25. R ef. [10]: pg. 25 (Assumptions).
26. R ef. [6]: Einleitung, S. 7.
27. R ef. [11]: Ch. 6, pg. 21.
97
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47,
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
98
99
SAMENVATTING
101
102
CURRICULUM VITAE
Na het behalen van mijn einddiploma H.B.S. in 1931 aan de vierde H.B.S. te
's-Gravenhage, bezocht ik de Middelbare Technische School te Haarlem,
waarvan het einddiploma voor elektrotechniek in 1934 in mijn bezit kwam.
Gedurende het jaar 1934/ 1935 werd ik op de School voor Reserve Officieren
der Genie te Utrecht opgeleid tot officier van de verbindingsdienst. In 1935
liet ik mij inschrijven aan de Technische Hogeschool te Delft, behaalde in
mei 1937 het eerste deel van het candidaats-examen en aanvaardde in november 1937 een werkkring als reserve-officier bij de dienst van magazijnen en werkplaatsen der Genie te Utrecht; ik werd o.m. belast met het inkopen en keuren
van materiaal voor de verbindingsdienst. Uit deze werkkring werd ik op 1
juli 1940 door de bezetter ~nts1agen.
In november 1940, na een periode van drie jaar werkstudentschap mocht ik
het diploma van elektrotechnisch ingenieur in ontvangst nemen. Op dat tijdstip, en wel van oktober 1940 af, was ik reeds wer}<:zaam als ingenieur in algemene dienst bij de N.V. Algemene Signaal- en Telefoonbouw Onderneming.
In maart 1941 verwisselde ik van werkkring en tot februari 1948 verrichtte ik
- met een onderbreking van 1943/ 1945 - de taak van vooronderzoeker bij de
Octrooiraad te 's-Gravenhage, een taak die zich voornamelijk bewoog op het
gebied van de draaggolftelefonie. De jaren 1943J1945 bracht ik door in krijgsgevangenschap.
Gedurende de periode van februari 1948 tot tbruari 1955 was ik bij Prof. Dr.
C. J. GORTER werkzaam als wetenschappelijk ambtenaar op het Kamerlingh
Onnes Laboratorium van de Rijks Universiteit te Leiden, en had daar de
dagelijkse leiding van de afdeling voor elektronika. Tevens gaf ik van januari
1952 tot januari 1956 college aan natuurkundige candidaatsstudenten in de
elektrotechniek, in het bijzonder in de theorie ~n toepassing van elektronenbuizen, versterker- en oscillator-schakelingen, netwerk-analyse.
Op 1 februari 1955 volgde mijn aanstelling als wetenschappelijk hoofdambtenaar aan de Technische Hogeschool te Delft, en kreeg ik in de Afdeling
voor Elektrotechniek de praktische leiding van het onder Prof. Dr. Ir. J. P.
SCHOUTEN ressorterende laboratorium voor microgolftechniek. Na het verrichten van enkele praktische en theoretische onderzoekingen op dit vakgebied
werd mijn aandacht afgebogen naar dat van de niet-lineaire verschijnselen die
optreden bij niet-stationair draaiende elektrische machines. Uit de laatst genoemde onderzoekingen kwam deze dissertatie voort.
103
STELLINGEN
I
De door HANNAKAM afgeleide dimensieloze differentiaalvergelijkingen, die het
elektro-mechanische gedrag van een asynchrone sleepringmotor beschrijven,
bevatten 6 (zes) parameters; men kan dit aantal door geschikte parametercombinatie's tot 4 (vier) terugbrengen, waarvan er slechts 3 (drie) van essentieel belang ZIJn.
L. HANNAKAM: bergangsverhalten des Drehstromsschleifringlufers,
Teil I, Reg!. techn., Hft. 11 , 7. Jahrg. , 393- 398.
11
Voor het grafisch verwerken van de metingen van een belastingsproef aan
asynchrone machines, verdient het gebruik van het impedantiediagram de
voorkeur boven het OssANNA-diagram.
M. CANAY: Vorgedruckte Ortskurven und einfache Darstellung der
Theorie von Asynchronmachinen, Bull. SEV, Bd 51 (1960) Nr 25.,
1293- 1299.
111
De toepassing van galvanometers in elektronische schakelingen zou aanzienlijk bevorderd kunnen worden, door naast de tot nu toe gebruikelijke opgave
van de drie de betreffende galvanometer het meest kenmerkende grootheden
- zijnde de spanningsgevoeligheid, de stroomgevoeligheid en de tijdconstante een opgave te stellen van een groep van drie daarmede tezamen aequivalente
markante elektrische grootheden, en wel de zelfinductie, de capaciteit en de
parallel verlies weerstand die - geschakeld als parallelle resonantieketen - de
ingangsimpedantie van het elektromechanische meetsysteem vertegenwoordigen.
E. MEYER und C. MOERDER : Spiegelgalvanometer und Lichtzeiginstrumente, Akad. Verlagsges. Geest & Por tig K.-G., L eipzig, (1952 ), 45,5;
L. N. RIDENouR: Vacuum Tube Amplifiers, M .LT. Rad. Lab. Series 18,
Mc Graw-Hill Book C ., (1948), 11 - 16.
IV
De jongste ontwikkelingen van de versterker-transistortechniek, de fototransistortechniek en de draaispoel-galvanometertechniek, bieden grote mogelijkheden voor het realiseren van een nieuw type analogon (reken-)
machine; daartoe dient de toepassing van de genoemde technieken te worden
aangepast aan de te stellen opgave.
v
Voor het berekenen van schakelingen voor de drift-compensatie van elektronische gelijkspanningsversterkers, kan men met voordeel gebruik maken van
het begrip van de complexe steilheid van de anodestroom-gloeispanningskarakteristiek van de in de schakeling gebruikte elektronenbuizen.
M.I.T. Rad. Lab. Series, Vo!. 18, Vacuum Tube Amplifiers, Me GrawHili Book Comp., (1948), Seetion 11-12.
VI
Het bewijs, dat de rang van de incidentie-matrix, waarvan de elementen de
incidentie getallen van de gerichte takken en de knooppunten van een netwerk
weergeven, gelijk is aan het aantal (p) lineair onafhankelijke knooppuntsparen
van dat netwerk, en het bewijs, dat de determinant van elke pxp-deelmatrix
van de genoemde incidentie-matrix de absolute waarde 1 (n) bezit, kan men
op eenvoudige wijze leveren, door de indicering van de knooppunten en van
de takken van het netwerk op een doelmatige wijze te kiezen.
D. KNIG: Theorie der endliehen und unendliehen Graphen, Chelsea
Pub!. Comp. , New Vork, N.Y., (1950),9. kap., 8;
H. SEIFERT und W. THRELLFALL: Lehrbueh der Topologie, Chelsea
Publ. Comp., New Vork, N.Y., (1947), 21.
VII
Het door TELLEGEN afgeleide theorema, dat in een netwerkconfiguratie met t
takken voor de takstromen i. en de takspanningen li., (1' = 1, 2, ... , t), de
I
betrekking I i.v.
van de algemene topologische stelling, dat de rand van de rand van elk (k
simplex verdwijnt; (k :> 1).
+ 1)-
VIII
Voor aanpassingstransformatoren, waarbij de overdracht van signalen van
lage frequenties wordt beperkt door de eigen zelfinductie en de overdracht van
signalen van hoge frequenties wordt beperkt door de spreidingszelfinductie
van de transformator, kan het optimale transmissie-rendement het bedrag
{l-exp [ -(nj~) (2ja)'/'J) nimmer overschrijden: In de gegeven uitdrukking
stelt daarbij ~ voor de relatieve bandbreedte van het bandfilter waarin de
transformator met zijn die frequentieband beperkende zelfinducties is opgenomen, en stelt a voor de spreidingscofficint van de transformator; voorts
IX
De transmissie-cofficient van de elektromagnetische TEocgolf, die in een
rechthoekige golfgeleider de opstelling passeert van twee of meer dichtbij
elkaar geplaatste capacitieve schermen, kan - hoe klein de spleetopeningen
in de schermen ook zijn - de absolute waarde 1 (n) aannemen; voor relatief
kleine spleetopeningen kan men het gehele elektromagnetische veld tussen de
schermen met vrij grote nauwkeurigheid uit de beschikbare gegevens berekenen.
H. BOSMA: Appl. sci. Res., B, 7, 131- 144;
F. A. W. V. D. BURG: Appl. sci. Res., B, 7, 153- 183.
X
~
XI
Teneinde een voldoend aantal goede stafleden (wetenschappelijke medewerkers) blijvend aan het hoger onderwijs te binden, dient hun functie voldoende
perspectieven te bieden ten aanzien van zelfstandigheid, verantwoordelijkheid en salariring.
N. I. HALL: Proc. IRE (1957) 451- 454.
XII
De stellingen, die volgens art. 45 van het technische-hogeschoolstatuut door
. een promovendus ter verkrijging van de graad van doctor in de technische
wetenschap moeten worden geponeerd en volgens art. 49 van het technischehogeschoolstatuut naar behoren moeten worden verdedigd, dienen een nieuw
en creatief element te bevatten.
Stb. 594, (1958), Hoofdstuk V.