DUAL STATOR WINDING INDUCTION MACHINE DRIVE

Alfred0 Mufioz-Garcia Thomas A. Lip0

Department of Electrical and Computer Engineering
University of Wisconsin - Madison
1415 Engineering Drive
Madison. WI 53706-1691USA
Tel.: (608) 262-0727 Fax: (608) 262-1267 email: alfredoQcae.wisc.edu - 1ipoQengr.wisc.edu

Abstract- A new dual stator winding induction ma-

chine drive is described in this paper. The new ma-
chine consists of a squirrel cage induction motor hav-
ing two stator windings wound for 2 and 6 poles which
are simultaneously fed from two independent variable
frequency power supplies. The proposed drive offers
the advantage of simultaneously producing two inde-
pendent electromagnetic torques which can be con-
trolled to produce any desired combination of output
torque. Since each torque component is controlled by
an independent set of stator currents a minimum o p
erating frequency can be set for each stator winding
and still achieve exactly zero or near zero speed and
rated torque operation. This feature is specially im-
portant to minimize the negative impact of the stator
resistance influence at low speed operation. The drive
is well suited for either constant volts per Hz (V/f) or
field oriented (FO) operation.
The occurrence of circulating harmonic currents,
common to traditional dual stator machines, is com-
pletely eliminated by the dissimilar pole number in
each stator winding. Stator winding

Fig. 1. Dual stator winding distributions.

I. INTRODUCTION
pletely eliminated. Since the output torque corresponds
The use of a common magnetic structure shared by two to the algebraic sum of two independent torques, the sta-
sets of stator windings was first introduced in the late tor frequency is no longer determined uniquely by the
1920’sas a means to increase the power capability of large rotor speed and the slip frequency but also by the added
synchronous generators [l]. Since then, dual stator ma- variable of a second torque component. From the per-
chines have been used in many other applications, ranging spective of control theory this adds one degree of freedom
from synchronous machines with AC and DC outputs [2] to the system.
and as current source inverters [3] to large pumps, com- Additionally, by having two independent torque compo-
pressors and rolling mills driven by induction motors, to nents the low frequency operation, including stand still,
name a few. They have also been used to improve the can be improved. This is particularly important for con-
reliability at the system level [4], [5]. stant volts per hertz operation at zero speed, where the in-
Dual stator machines are normally constructed by fluence of the stator resistance becomes dominant. In the
“splitting” the stator winding into two displaced but iden- proposed machine zero speed operation does not translate
tical windings [6]. However, this results in mutual cou- into zero excitation frequency, hence significantly reduc-
pling between the stators, causing circulating harmonic ing the influence of the Ir voltage drop.
currents [7] as well as coupling between the electromag-
netic torques produced by each stator winding [8]. This
11. DUALSTATORWINDING
IM DRIVE
means of splitting the winding is a major drawback be-
cause the circulating currents add extra stator losses and The stator of the proposed machine is constructed by
demand larger semiconductor device ratings. dividing the original three phase winding into two sepa-
The dual stator winding induction machine proposed rate three phase windings wound for a dissimilar number
in this paper differs from the “standard” dual stator ma- of poles. Any combination of dissimilar pole number could
chine in that the two stator windings are wound for a be used, however, to better utilize the magnetic material,
dissimilar number of poles. This approach, in principle, avoid localized saturation and additional stator losses, it
eliminates the net magnetic coupling between the stator is found that the most advantageous configuration is a 2-6
windings and decouples the torques produced by each set pole combination. Such an arrangement is shown in Fig. 1.
of stator currents. Also, circulating harmonic currents, From the perspective of magnetic material utilization it is
due to the so-called mutual leakage coupling, are com- convenient to choose a pole number combination that, in

0-7803-4943-1/98/$10.000 1998 IEEE 601

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 5

I
0.85
- I
I a) Medium and high speed operation
I
I , ,

6 pole MMF peak magnitude (pu)

t
Fig. 2. 6 pole peak MMF for constant total peak MMF.

the steady state, will tend to produce a nearly trapezoidal

MMF distribution. This type of distribution is most effi-
ciently obtained by choosing the number of poles in the
ratio 1:3. On the other hand, the magnetizing inductance
varies inversely proportional to the square of the number
of poles, hence a machine with a large number of poles
results in low power factor and reduced efficiency. In ad-
dition, to achieve a sinusoidal winding distribution the
stator winding must be distributed among several slots,
therefore, for a given stator inner diameter, the number b) Low speed operation
of slots per pole decreases in proportion to the number of
poles. Also, for a given rotor speed, the stator frequency Fig. 3. General control strategy.
increases directly proportional to the number of poles.
This translates into additional losses, in the machine and (V/f) as well as vector control or field orientation (FO).
in the power converter, further reducing the efficiency. All
The basic control method is to generate two torque com-
these factors suggest that the maximum number of poles
mands such that combined they produce the required out-
should be kept to a minimum, hence the best combination
put torque. By choosing adequate current commands
is to choose 2 and 6 poles.
the two individual torques can be added or subtracted,
The total MMF distribution in the airgap corresponds hence providing an extra handle to control the excitation
to the sum of the MMF’s produced by each stator wind- frequency. TWOdistinct modes of operation are consid-
ing. To avoid the presence of highly saturated points and, ered: low speed where the two torques are subtracted and
at the same time, fully utilize the magnetic core it is desir- medium to high speed, where the torques are added. This
able to maintain the total peak flux density distribution is shown in Fig. 3.
equal to that created by a two pole winding acting alone. A further discussion of the control method will be de-
Figure 2 shows the peak magnitude of the 2 pole MMF, layed until after a derivation of the machine model is con-
as a function of the 6 pole peak MMF, for a constant ducted and it will be presented in IV.
total peak MMF. The optimum distribution corresponds
to choosing a 6 pole MMF equal to approximately 40%
111. MACHINE
MODEL
that of the 2 pole distribution. In this case the total MMF
and the 2 pole MMF have the same peak amplitude, thus Given the particular characteristics of the machine con-
preserving the saturation level. sidered in this study it is critical to clarify the interac-
The rotor of the proposed machine corresponds to that tion between the actual rotor current distribution and the
of a standard squirrel cage type. This construction guar- MMF produced by each stator winding. The interaction,
antees that both stator current distributions will simulta- if any, between the currents in the two sets of stator wind-
neously couple with the rotor flux to produce the desired ings must be studied as well. For this purpose a detailed,
torque. yet simple, dynamic model of the machine will be devel-
Because of the decoupling effect produced by the dif- oped. The coupled magnetic circuit theory and complex
ference in pole number, the proposed dual stator machine space vector notation will be used through out the deriva-
behaves as two independent induction machines that are tion. This technique is chosen because of its generality
mechanically coupled through the shaft. Therefore, all and the great deal of simplification that is achieved. The
the known control techniques used in induction machine following general assumptions are made:
drives are also applicable to the dual stator winding ma- 0 negligible saturation,

chine. This includes both scalar constant volts per hertz 0 uniform airgap,

602

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 -Y : 'X The magnetizing inductance Lmsi is known to be: [12]

where Nsi is the total number of turns per phase of each

winding set and P is the number of poles. Llsi represents
the total per-phase self leakage inductance of each winding
and it can be calculated by traditional methods [14], [15].
The sub-matrices Ls12 and LSz1account for the mutual
leakage coupling between the two stator windings. In gen-
eral, the leakage flux can be divided into slot, end wind-
ing, belt and zig-zag components and each one of them
will contribute to the self and mutual leakage inductance.
For simplicity, however, the mutual leakage due to the zig-
zag and belt leakage components will be neglected and it
-2 will be assumed that they only contribute to the self leak-
:2 pole winding 0 :6 pole winding
age. Therefore, it will be assumed that only the slot and
Fig. 4. Dual stator winding induction machine. end winding components contribute to the mutual leak-
age. Furthermore it will be assumed that the end winding
stator windings sinusoidally distributed, leakage varies as the slot leakage [8].
no electrical interconnection between stators, and
negligible inter-bar current B. Slot Leakage Inductance
It is also assumed that the two stator windings are To study the mutual leakage term let us consider the
wound for 2 and 6 poles respectively and that one stator is winding distributions shown in Fig. 5. They correspond
displaced with respect to the other by a fix but arbitrary
angle <. The main stator ( 2 pole) will be referred to as
to a) fractional pitch due to the displacement between <
stators, 60" phase belt, and b) fractional pitch, 30" phase
the abc windings and the secondary stator, having 6 poles, belt. Note that, since the two set of windings have two
will be labeled as syz windings. The rotor of the machine and six poles respectively, their pitch angles a1 and a2 are
is a standard squirrel cage type. The proposed dual stator in the ratio 612. In Fig. 5-a, defining pl and p2 as the pitch
winding machine is schematically shown in Fig. 4. of the abc and zyz windings respectively, for a variation
of 6 between zero and 20" the corresponding pitch factors
A . Stator Flux vary as 819 < p l < 1 and 213 < p2 < 1.
As shown by Alger [15] and Lip0 [16], the slot leak-
Since both stators are sinusoidally distributed in space age can be divided into self leakage and mutual leakage.
but wound for a different number of poles (and electri- The self leakage represents that part of the flux produced
cally isolated) there is no mutual coupling due to main by the in-phase current component (i.e., slots with coil
flux between them [9], [lo]. However, since both windings sides belonging to the same phase). The mutual leak-
share common slots and are in close proximity, there is a age accounts for the leakage flux due to having conduc-
common leakage flux linking them. This gives rise to the tors from different phases sharing common slots. In gen-
so-called mutual leakage coupling [8], [111. eral, for a two layer winding the self, (Lsls), and mutual,
The total flux linked by the stator windings and due (Lslm),components of the slot leakage inductance can be
only to the stator currents isabc and isxyzcan be written, expressed, as a function of the pitch p , by
in matrix form, as +
L S ~=S LIT LIB %(P)LITB + (5)
Lslm = km(P)LlTB (6)
where LIT and LIB are the slot leakage inductances as-
where sociated to the coils in the top and bottom halves of the
slots. They are calculated for the case of unity pitch and
do not depend on winding pitch. The term L ~ T B repre-
sents the mutual inductance between coils in the top and
bottom halves of the slot. The quantities k , and km are
called slot factors and they correspond to proportionality
constants that depend on the pitch.
For the winding distribution of Fig. 5-a, the phase belt
Lsl and Ls2 represent the self inductance matrices of the slot leakage flux associated to phase a, when 819 < p l < 1,
abc and cgz windings respectively. They are of the form: is
i91 &da = ( L l T I + h 3 l ) i a s - LlTBlkml (pl)(ibsGcs)
+
~ l s i Lmsi -&
2 +2LlTB2izs - (2 - Icm2(P2))LlTB2(izs+iys)
(7)
Llsi + Lmsi -&.Rw.i The equations applying to the remaining phases of the abc
-+i Llsi + Lmsi winding can be found by symmetry and the corresponding

603

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 A z -x Y C z -x Y B z -x Y
-z x -Y -B -z x -Y -A -z x -Y -C

A -ZX - Y -B Z-XY C - Z X - -A Z -X Y B -ZX - Y -C Z-XY

-B Z -X Y c -Z X - Y -A Z -X Y B -Z X-Y -c Z -X Y

A similar reasoning yields the slot leakage inductance ma- Labc =(LI,~+ $ ~ + ~ L~m s l~e j ( )e ~ +ii r6l ) ~*(15)~
~
trix of the q z winding

1
-1 where n is the number of rotor bars, 6 is 112 the angle

[+ + + between rotor bars and the complex vector currents L1

+rT2
Ls12 = -LlTB2 1 km2 1 km2 1 km2
1 +-l
km2 1 + k m z 1 +-l
km2 and irl are defined by
isl = $ (ius + gibs + a2i,,)
i 1 []
0 (16)
LIT2 +
LlBZ -3km~L1~~2
LlB2 -3kmZLlTB2 O
LIT2 -k LlB2 (9) *
ITl = ;2;[ 1 4 42 a * . -bn-l ] (17)
The slot factor kml must be zero when p l = 1 and one
when p l = 213. Hence when 213 < p l < 1 ZTn
kml = 3 - 3 p l (10) with = and 4 = e j 2 r / n .
ej2r/3 The vector
Similarly, km2 must be zero when p2 = 1 and one when [irl i r 2 .- -&IT
represents the instantaneous rotor cur-
p2 = 819. Hence when 819 < p2 < 1 rents defined according to Fig. 6.
km2 = 9 - 9Pz (11) A similar analysis for the q z winding yields [13]
If the winding distribution of Fig. 5-b is considered in-
stead, the results are similar and the structure of (8) and XsZy+ = (L1s2 + t ~ m s 2i )s 2
(9) remain unchanged [13]. 2n sin E6 (18)
rN*l~ , , z e j f (ep+6-c)ir2
+

Assuming that the end winding leakage behaves as the

slot leakage, the mutual inductance matrix Ls12 is pro- where
portional to the second term in (8). Clearly, the mutual 2 .
inductance matrix Ls21 will be proportional to the first L2 =3 (zxs +ai,, +a2izs) (19)
term in (9). and
Given the symmetric structure of the matrices Ls12 and
Ls21 and assuming no neutral connection, it is apparent
that after applying the space vector definition
r?.slabc= $ (Asla + d s l b -k @2Aslc) (12)
their contribution to the total flux vector will be zero. Note that because of the different number of poles con-
This result is very important since it proves that both sidered in this case it is necessary to define a new rotor
stator windings are fully decoupled and the total flux current space vector irz. The physical meaning of this
linked by the stator windings can be written as new definition can be understood by recognizing that the
Asabc = Lslisabc + L s r l i r (13) equivalent electrical angle between adjacent rotor bars,

604

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 Rotor lcmv tor flux is defined by the vector transformation
TXri 1

where P is given by the number of poles of the stator

winding. Setting P = 2 in (25) and applying it to (24)
yields the rotor flux referred to the 2 pole stator winding.
It is clear that the first two terms of (24) correspond to
those of the single stator machine and they are given by
1171
+ +
2Lb(1 - cosf&) 2Le y a r ) irl
" d
(26)
-1Endring Lr1
and
Fig. 6. Rotor current definition.

referred to the P pole xyz winding, is now $er instead of where Lb and Le represent the bar and end-ring segment
8,. Also note that the actual rotor bar currents in (17) inductance respectively and g is the airgap length.
and (20) are the same. The last term of (24) represents the contribution due
Finally, the stator voltage equations for both stator to the secondary stator winding currents. Following the
windings are obtained by differentiating the flux linkages same technique, it can be shown that
(15) and (18) and adding the resistive voltage drops. This
procedure yields
vsl = T S l L + (Lis1 + & w 1 ) P L
+-2n sin d Lmalej(er+a)(p + j wr ) -rl
i
(21)
However, for P > 2, both summations in (28) are identi-
for the abc winding and cally zero, hence ArS2is identically zero, regardless of the
v,, = TS2iS2 + (Lls2 + 4Lms2) PL2 value of L2, and the total rotor flux, in the 2 pole winding
subspace, corresponds to the sum of (26) and (27).
(22)
+=~,,~ej+(er+a-t) (P+ j + r ) ir2 Finally, taking the time derivative of the rotor flux vec-
tor and adding the resistive voltage drop gives the rotor
for the xyz winding. The complex vector voltages y,, and voltage equation in the subspace defined by the 2 pole
vs2 are winding. This yields
2
5 1 = 5 (vas +
avbs -k a2'%s) -0 = rrllrl+ LrlPir1
(23) (29)
2
L 2 = 5 (vm + a v y s +
a2vz,) + w L m s l e - i ( e r + 4 (p - j w r ) -si
i
with rrl = 2Re -I-2% (1 - cosa,).
It is important to notice that the stator current islde- A similar analysis for the P pole winding reveals that,
pends only on the applied voltage and the rotor cur- in this case, the contribution of the second term in (24)
rent irl. Similarly the current ia2depends only on the is zero and the rotor voltage equation in the subspace
applied voltage s2 and the rotor current This result defined by the P pole winding is
verifies the known fact that, for a sinusoidally distributed
winding, there only exists coupling between current dis- - +
0 = rr2ir2 Lr2~ir2
tributions of the same number of poles.
where rr2 represents the equivalent rotor resistance and
C. Rotor Flux Lr2 is the equivalent rotor inductance in the P pole sub-
The rotor flux can be divided into three components, space, yielding
one due to the rotor currents i, and two due to the stator Tr2 = 2Re +
2% [1 - cos ( $ a r ) ] (31)
currents isabc and iszyz.In matrix form

Note that although the instantaneous rotor current dis-

tribution simultaneously contain two components of dif-
ferent frequencies and pole number, each stator field is
capable of interacting only with that part of the rotor
In general, the complex vector representation of the ro- field with the "correct" number of poles. This is true

605

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 *
J isdrl I m

I I
2 pole winding (uk)

4%)~~ bqdr2 j(+)(m2- %)&qdr2

Y qh2

P pole winding (xyz)

Fig. 7. Complex vector model of the dual stator winding

squirrel cage induction machine. Sped & Toque

not only in an average basis but also instantaneously. It Fig. 8. Proposed control scheme for constant V/f operation.
is well known that sinusoidally distributed windings only
couple with fields wound for the same number of poles, controlled by adjusting the frequency of the 6 pole wind-
however the rotor cage is clearly not a sinusoidal wind- ing. By keeping the frequency above this pre-set limit the
ing and one might expect that the presence of two su- influence of the stator resistance is minimized, hence sim-
perimposed flux distributions would give rise to pulsating plifying the control. In this mode the two MMF's rotate
torques. However, this is not the case for the proposed asynchronously but because of the reduced frequency the
dual stator winding machine. An equivalent circuit, using additional losses caused by saturation should be minimal.
d-q notation, is shown in Fig. 7 In the medium to high speed range the negative effect
Neglecting saturation, the electromagnetic torque can of the stator resistance is not a concern and the frequen-
be expressed as the partial variation of the co-energy with cies are kept in the ratio 1:3. This constraint guarantees a
respect to position [9] nearly trapezoidal flux distribution and the torque is con-
trolled by adjusting the magnitude of the applied voltages.
The trapezoidal shape, in turn, allows for slightly greater
two pole flux than when only the two pole winding is ex-
which can be written as the separate sum of the torques cited thereby producing slightly more torque per ampere.
produced by each set of stator currents

A . Constant V/f Control

Substituting the corresponding matrices and carrying out The operation and control will be explained with the
the differentiation yields help of Fig.3. As shown in this figure there are two dif-
ferent operating modes. For high speed both stators are
fed with voltages of the same frequency. ,This produces
the torque-speed curves of Fig.3-a. The output torque
for a given rotor speed corresponds to the algebraic sum
of the torques TIand T2 produced by each of the sta-
where we have used P = 6. Since irl and ir2are orthogo- tor. The torque produced by each winding is controlled
nal vectors the two torque components can be controlled by adjusting the magnitude of the stator voltages.
independently by the stator currents.
When both stators are fed with different frequencies the
result is that shown in Fig.3-b. By fixing the frequency
AND SPEEDCONTROL
IV. TORQUE of one of the stators, say abc, the total output torque can
Since the proposed machine behaves as two indepen- be adjusted by controlling the frequency (and voltage)
dent induction machines, mechanically coupled through supplied to the q z winding f2. In the case shown in
the shaft, all the known control techniques used in induc- Fig.3-b an increase in torque requires an increase in f2
tion machine drives are also applicable to the dual stator and vice versa. In this case one of the stator windings
winding machine. (abc) operates in the motoring region while the other ( q z )
In general there are two distinct modes of operation, operates as a generator. Note that this operating mode
the low speed (i.e., frequencies below a few hertz) and the corresponds to the one required to operate at zero speed
medium to high speed range. In the low speed range the and that the torque can be controlled tiom zero to rated
goal is to maintain the frequency of the 2 pole winding value. A simplified block diagram of the proposed control
above a minimum level (about 3 Hz) and the torque is scheme is shown in Fig. 8.

606

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 N

s
0

Fig. 9. Proposed control scheme for field oriented operation.

B. Vector Control
A simplified block diagram of the proposed control
scheme is given in Fig. 9. As in the constant V/f method
-E - I I I
2 pole toque I
the vector control is divided into two operating regions: i-i
a high speed range defined by frequencies above a mini-
mum frequency fmin and a low speed range for frequencies
below f m i n . For the high speed region the proposed con-
trol method is quite straight forward dividing the output
torque among the two windings as to yield similar stator Y

currents and a nearly trapezoidal flux distribution. In the 0

low speed range a negative torque command is given to 0.0 0.2 0.4 0.6 0.8 1.0
Time (sec)
the secondary winding hence increasing the torque pro-
duced by the primary winding which yields an increased Fig. 10. Full matrix and complex vector model simulation
stator frequency. The goal is to maintain the primary results for a free acceleration run at 60 Hz (fz = 180
stator frequency at a constant value equal to fmin. Hz). Complex vector and full matrix model traces are
superimposed.
V. SIMULATION
RESULTS
VII. CONCLUSIONS
To prove the correctness of the complex space vector
model of the dual stator machine a complete set of simu- A new type of dual stator winding induction machine
lations has been carried out. The results from the space has been presented. The most relevant characteristic
vector model are compared to those obtained from a full of the new machine is that the two stator windings are
matrix model of the machine. Fig. 10 shows both set of wound for a dissimilar number of poles. This feature
results superimposed. It is clear that both simulations yields two independent torques that can be controlled as
are identical hence proving the validity of the complex desired to produce the net output torque. In addition, the
vector model. As shown in Fig. 10 the rotor currents con- problem of high circulating harmonic currents common to
tain two different frequencies dictated by the frequency of dual stator induction drives is also eliminated. The ma-
each of the stator currents and the rotor mechanical speed. chine is especially well suited for both constant V/f and
Although the rotor currents simultaneously produce two sensorless FO control.
field distributions that rotate at different speeds, because By implementing an adequate torque control scheme it
of the different number of poles and the sinusoidal char- is possible to operate at stand still and, at the same time,
acteristic of the stator windings, they do not give rise to limit the minimum frequency at which the 2 pole stator
harmonic torques. winding operates. This is an important feature since it
Simulation results using the control techniques pro- reduces the impact of the stator resistance voltage drop at
posed in the previous section are shown in figures 11 and low frequencies and greatly simplifies the implementation
12, hence proving the feasibility of the method. of sensorless control algorithms.

RESULTS
VI. EXPERIMENTAL REFERENCES
[l] P. L. Alger, E. H. Freiburghouse and D. D. Chase, "Double wind-
A 5 Hp prototype is currently under construction and ings for turbine alternators", AIEE Transactions, Vol.49, January
the experimental results will be reported shortly. 1930, pp. 226-244.

607

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.
 ’ 1.0 1.8 2.6 3.4 4.2 5.0
Time (sec)

Fig. 12. Direct field orientation at stand still.

191 D. W. Novotny and T. A. Lipo, Vector Control and Dynamics of

AC Drives, Clarendon Press-Oxford, 1996.
P. W. Franklin, “A theoretical study of the three phase salient
pole type generator with simultaneous AC and bridge rectified S. A. Nasar, “Electromechanical energy conversion in nm-
D C output”, IEEE Transactions on Power App. and Systems, Vol. winding double cylindrical structures in presence of space har-
PAS-92, No. 2, March/April 1973, pp. 543-557. monics”, IEEE Transactions on Power App. and Systems, Vol.
T. Kataoka, E. H. Watanabe and J. Kitano, UDynamic control PAS-87,No. 4, April 1968, pp. 1099-1106.
of a current-source inverter/double-wound synchronous machine R. Schiferl, UDetailed analysis of a siz phase synchronous ma-
system for AC power supply”, IEEE Transactions on Industry chine with AG and D C stator connections”, Ph.D. Dissertation,
Applications, Vol. IA-17, No. 3, May/June 1981, pp. 314-320. Purdue University, West Lafayette, Indiana, 1982.
J. R. Fu and T . A. Lipo, uDwturbance f i e operation of a mul- G. R. Slemon, Electric Machines and Drives, Addison-Wesley
tiphase current regulated motor drive with a n opened phase”, Publishing Co., 1992.
IEEE Transactions on Industry Applications, Vol. 30, No. 5, S e p A. Muiioz-Garcia, “Analysis and control of a dual stator winding
tember/October 1994, pp. 1267-1274. squirrel cage induction machine for high performance drives”,
J. C. Salmon and B. W. Williams, “A split-wound induction mo- Preliminary Ph.D. Thesis proposal, University of Wisconsin-
tor design to improve the reliability of PWM inverter drives”, Madison, 1997.
IEEE Transactions on Industry Applications, Vol. IA-26, No. 1, A. S. Langsdorf, Theory of Alternating Current Machinery, Sec-
JanuarylFebruary 1990, pp. 143-150. ond Edition, McGraw-Hill, 1955.
E. F. Fuchs and L. T. Rosenberg, “Analysis of a n alternator with P. L. Alger, Induction Machines, Gordon and Breach Science Pub-
two displaced stator windings”, IEEE Transactions on Power App. lishers, Second Edition, 1970.
and Systems, Vol. PAS-93, No. 6, November/December 1974, pp. T. A. Lipo, Introduction to AC Machine Design, University of
1776-1786. Wisconsin-WisPERC, 1996.
K. Gopakumar, V. T. Flanganathan and S. R. Bhat, “Split-phase A. Muiioz-Garcia and T. A. Lipo, “Complez vector model of the
induction motor operation from P W M voltage soume inverter”, squirrel cage induction machine including instantaneous rotor
IEEE Transactions on Industry Applications, Vol. 29, No. 5, S e p b a r currents “, IEEE Industry Applications Society 1998 Annual
tember/October 1993, pp. 927-932. Meeting, St. Louis Missouri, 1998, in press.
T. A. Lipo, “A d-q model for a h phase induction machines”, In-
ternational Conference on Electric Machines, Athens, Greece, 1980,
pp. 860-867.

608

Authorized licensed use limited to: UNIVERSITY OF STRATHCLYDE. Downloaded on May 20,2020 at 00:45:12 UTC from IEEE Xplore. Restrictions apply.