963

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

Synchronous Drive Performance

of Brushless Doublv-Fed Motors
4

Ruqi Li, Renk Spke, Alan K. Wallace, and G. C. Alexander

Abstract-This paper presents recent studies of the steadystate performance of the brushless doubly-fed machine for adjustable-speed drive applications. The equivalent circuit and
basic performance equations are presented followed by a proven
technique for the solution of the least restrained form of the
equations. Model predictions and experimental results are given.
The studies show that the machine acts like both a synchronous
motor, with power factor adjustable by control winding excitation, and a slip-ring induction machine with slip power recovery
and a low converter rating. The model is suitable for development of control schemes and performance index optimization.

CONTROLLED
FREOUENCY

bidirectional

UTILITY

S m Y

1 __

I. INTRODUC~ON

single-frame self-cascaded induction machine, which

is capable of operating in both an induction mode
and a form of synchronous mode, has attracted the attention of many researchers since the concept was first developed by Hunt at the turn of this century [11-[61. This type
of machine evolved, both physically and conceptually,
from two induction machines with cross-connected rotor
phase-windings on the same shaft, thus avoiding the use
of brush gear. The advent of power electronic converters
capable of adjustable frequency, adjustable voltage and
bidirectional power-flow operation has revived interest in
the self-cascaded induction machine [7], [81. This interest
is promoted by the demonstrated adjustable speed drive
(ASD) capability in which one of the stator windings is
supplied via a converter of a rating significantly smaller
than that of the machine. This configuration, shown
schematically in Fig. 1, is now referred to as a brushless
double-fed machine (BDFM).
To avoid direct transformer coupling, it is essential that
the two windings on the stator have different numbers of
poles. Although many pole-number combinations are possible, laboratory demonstration machines have, so far,
concentrated on six-pole power windings and two-pole
control windings. Although this is a wide ratio of pole
numbers, the 6/2 pole configuration can be demonstrated
to be the minimum combination to avoid unbalanced
magnetic pull (UMP) on the rotor. In general, UMP is
produced if the difference between the pole numbers of
the two stator systems is two. This feature presents a
design problem of leakage reactance minimization which

Paper IDC 92-3, approved by the Industrial Drives Committee of the

IEEE Industry Applications Society for presentation at the 1992 IAS
Annual Meeting, Manuscript released for publication February 25, 1994.
The authors are with the Department of Electrical and Computer
Engineering, Oregon State University, Corvallis, OR 97331.
IEEE Log Number 9402632.

I
..

!I i i c

.,

.. .. .

!\

Y.'

Fig. 1. Schematic of BDFM ASD

has been addressed in a recent paper [91. As indicated in

Fig. 1, the rotor has a specialized cage structure with the
number of identical sections, or nests, equal to the sum of
the pole-pairs of the two stator windings. Specialized
single stator winding connections have been developed for
induction motors to produce two magnetic fields of different pole numbers. Notably the connections of Creedy
[2] and Dahlander [lo] can be used to give pole ratios of
3 : 1 or 2 : 1, respectively. For pole-changing machines
these types of stator windings enable more effective use of
slot space and hence increase available conductor area.
However, for the BDFM, it has been shown that parallel
winding imbalance results in the lower pole-number connection giving rise to internal circulating currents, causing
both additional losses and instability in certain areas of
the speed range. In consequence, two isolated three-phase
stator windings are preferred as they enable a better
overall performance compromise and can be individually
rated depending upon the drive application.
Analysis and performance prediction of the cascade
connection of two machines can be undertaken by the
appropriate connection of two induction motor equivalent
circuits. Although this approach is not rigorously applicable to the self-cascaded cage-rotor machine, it has been
used by several investigators [2]-[61. To overcome this
deficiency and also to enable the simulation of nonsynchronous behavior, a detailed dynamic model of the
BDFM was developed [ll]and utilized in several studies
[121-[14]. This highly detailed model enables in-depth
predictions of the effects of machine geometry and connections but is completely inappropriate for ASD studies

0094-9994/94$04.00 0 1994 IEEE

IEEE TRANSACTIONSO N INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

964

related to system performance and control. Consequently

the appropriate transformation processes were developed
to produce a more conventional two axis model 1151. The
development of this latter model demonstrated that, starting from the basic stator coils and rotor cage loops,
equivalent stator phases and rotor phases can be deduced with impedances determinable directly from machine geometry. Finally, it is possible to develop steady
state equivalent circuits from the two axis model to represent different modes of BDFM [16].
Appropriate analytical tools are thus available to investigate the perceived potential advantages of the BDFM
drive compared with existing ASDs, namely: (i) full range
speed control with a significantly smaller converter than
an induction motor of the Same rating; (ii) controllable
power factor and low harmonic distortion of the utility
supply; (iii) robust machine construction; and (iv) operation as an induction motor in the event of converter
failure.

POWER SYSTEM FIELDS

Pp pole-pairs

CONTROL SYSTEM FIELDS

Pc pole-pairs

ROTOR
FIELDS

OF BDFM OPERATION
11. FUNDAMENTALS

Analysis of the BDFM makes the following assumptions: (i) the magnetic circuit is linear; (ii) both iron losses
and stator winding harmonic effects are neglected; and
(iii) the supplies to the machine, from both the utility
supply and the power electronic converter, are three-phase
balanced sinusoids.
A. Basic Principles

The presence of two stator fields of different pole

numbers, different frequencies and possibly different sequences, and the resulting rotor fields produced by their
interaction, make the BDFM analysis significantly more
complicated than either induction or conventional synchronous motors. Fig. 2 shows the fundamental rotating
field relationships of the machine in the doubly-fed mode.
In response to the stator fields, rotor currents are produced at frequencies of

+/- denotes sequence of f,

c / > denotes f, c or > fMU
Fig. 2. Relative directions of rotation.

the natural speed of a machine having ( P p + Pc) pole

pairs. For the control sequence opposing the power sequence, speeds below the natural speed result and the
machine is said to be in the subnatural synchronous
mode, until (when f c = f,) standstill is reached. If the
control sequence is the same as the power sequence
hypernatural speeds are produced. However, when

no rotor currents are induced from either of the stator

fields and stable operation is impossible in a narrow band
f,, = < f p - P,f,,,) and fRc = ( f , + P c f m ) . (1) around the speed
For the resulting rotor fields to induce emfs in the power
winding at frequency f p only and in the control windings
at f c only it is required that
At all other speeds above f,,, synchronous operation is
fRP =fRC*
(2) possible.
It is convenient to define three relative slips as follows:
This is referred to as the synchronous condition. The
(a)
the slip of the rotor with respect to the field of the
controllable synchronous speed of the rotor is, in consepower windings
quence

(3)

s,

fP

- PPfm

.
7

(4)

fP

for the sequence directions defined in Fig. 2. Usually the

power frequency, f p , will be constant (from the utility
supply) and the control frequency, f c , can be adjusted in
both positive and negative ranges to meet the requirements of the load. Excitation of the control winding by a
dc source (i.e., f c = 0) produces synchronous operation at

(b) the slip of the control winding field with respect to

the rotor

965

LI et al.: SYNCHRONOUS DRIVE PERFORMANCE OF BRUSHLESS DOUBLY-FED MOTORS

(c) the slip of the control winding field with respect to

the power winding field (also referred to as total slip)

B. Steady-State Electrical Equations

Reference [151 developed performance equations for all
the operational modes of the BDFM. For the synchronous
mode, which is of the greatest interest for ASD applications, the performance analysis is most simply expressed
in the rotor d-q axis domain, on a per axis basis, by the
equivalent circuit shown in Fig. 3; a slight modification
[15] is necessary for the single condition of dc excitation.
The true physical quantities of the stator windings (Le.,
their magnitudes and frequencies are obtainable from the
inverse of the transformation processes that were used to
develop the two-axis model.
The general synchronous performance equations are:

.
,'C

Fig. 3. Equivalent circuit.

Substituting from (9) into (6) and (7) gives

and, for fc Z 0
VC

S =
or, for fc

($+ jxc)ic jxcRiR

-

(6a)
(7)

0
(7a)
Hence at any given speed in ASD operation the excitation
of the control winding can be used to influence the power
factor of both the power and control windings. The term

and

o = jXfRIp- jXcRiC+

:(

+jxRI,.

(9)

In these equations, as a result of the transformation

process, the phasor quantities are expressed at the slip
frequency of the rotor as defined in (4). The actual
applied voltages of the machine are generally of different
frequencies. However, in the reference frame of (6149)
they are of the same frequency and consequently require
a phase displacement angle p . Using the power winding
voltage as reference, the equivalent circuit voltages can be
written

XPRXCR

;(

.
'L

+JXR)

in (6a) is the equivalent of the open-circuit voltage of a

conventional synchronous motor.

C.Steady-State Torque Equation

The steady-state torque equation of the BDFM can be
expressed as a special case of the instantaneous electrical
torque of the general dq machine, which is
7,=PpMpR(iqpidR -

idpi,,)

+PcMCR(i,CidR - i d c i , ~ ) .

The d-axis currents, id, and id,, lead their q-axis counterparts, i,, and i,,, by 90" but have the same magnitude.
The d-axis current, id,, leads its q-axis counterpart, i,,
by 90" if the sequence of fc is the same as that of f,, and
vice versa. Hence, the steady-state (i.e., nonpulsating)
components of the torque equation become
in which Vpph and Vcph are the magnitudes of the actual
power and control winding phase voltages.

T,

2~,~,,1m(i,i,*)

+ ~ P , M , ~ I ~ ( Z ~ I , * } ,(12)

IEEE TRANSACIIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

966

where the asterisk denotes a conjugate value. Because of

the manner of solution of the electrical equations it is
convenient to express (12) in the form

T,

2 P p M p ~ { I p j z-~I~p r I ~ j )
2 P p M c ~ { z c j z-~z~c r z ~ i } (12a)

-k

in which the subscripts r and i denote real part and

imaginary part. The components of the total torque are
due to power winding, Tp,, and control winding, T,,.

D. Power Equations
The power balance equations of the power winding,
control winding and rotor are developed in the appendix.
Of particular significance is the power of the control
winding

POWER
WlNOlNG

(a)

I
ROTOR

CONTROL
WINDING

or
W,

fc

= rclz - -xCR(icjiRr
fP

- icriRj).

In the subnatural speed region, as defined by Fig. 2, fc

is positive. Typically, W, will become negative; i.e., power
will be returned from the control winding, in a manner
equivalent to slip-power recovery of a slip-ring motor, if
T,, is positive. If the control winding excitation is such as
to enforce a positive W,, Tc, will become negative and
machine efficiency will be reduced. Conversely, in the
hypernatural speed region, fc is negative and most
efficient operation using positive control torque needs
power input to the control winding. Power flow diagrams
for these conditions are shown in Figs. 4(a) and 4(b),
respectively.
Since

(b)

Fig. 4. Power flow diagrams: (a) typical for subnatural speeds and
(b) typical for hypernatural speeds.

Hence, by deduction from (21) the total mechanical power

and the rotor losses

111. SOLUTION
OF STEADYSTATE
MODEL

and W,,

TPe. 2mfm

which from (41, becomes

Also

which from ( 5 ) becomes

(14)

The solutions of the steady-state equations (6), (7), (91,

and (12) will depend on whether voltage or current control is employed by the converter supplying the control
winding and whether the machine is in motoring or generating mode. A general method of obtaining a solution has
been developed, which gives accurate numerical solutions
to the equations for the minimum of given operational
restraints. For both motoring and generating operation
the minimum of known conditions are typically the torque
[which is probably a function of rotor speed], the applied
voltage magnitudes and the control frequency [which is
dictated by the rotor speed in accordance with (3)].
When adjustable speed operation is provided by ac
excitation of the control winding system determination of
the steady-state performance requires the solution of a set
of nonlinear simultaneous equations. This is basically
because the angle p of (10) and (11) is not known but is a
function of torque and excitation voltage magnitude.
Hence the torque equation must be solved simultaneously

967

LI et al.: SYNCHRONOUS DRIVE PERFORMANCE OF BRUSHLESS DOUBLY-FED MOTORS

with the voltage equations. This is not always so: If one

more quantity is specified, as in the operation of the
BDFM as an automotive alternator [17], the following
procedure is avoidable.
Derivation of a solvable set of nonlinear equations
requires the expression of the voltage equation in rectangular form and separation into real and imaginary parts:

50

r
C

?!

U-

2
al

U
0
_-

F ( Y ) = 0,

(17)

where

y=

[Ipr,zP~,zL.r~zcl,zRr~zR~,Pll

is the vector of unknowns and

[ f,(Y f,(Y
1 7

1 7

f d Y ) 7 f J Y ) fs(Y
9

1 9

fdY

1 7

f7

(Y )I

is the function vector containing the seven scalar functions of the above equations.
It is found in practice that Newton's algorithm for
iterative approximation is adequate to obtain solutions to
(17), as follows. At the ith iteration the unknown vector
y") and the previous approximation y ( ' - and the difference A y ( ' - ' ) are related by
y ( ' ) = y('- 1 ) - A y ( ' - 1 ) ,
where

0
time (2s/div)
(C)

Fig. 5. Speed control: subnatural speed range.

The elements of the Jacobian

conditions. However, these solutions are usually recognizable as unrealistic.

Solution convergence rarely requires more than three
iterations, and enables the real and imaginary parts to be
recombined to illustrate basic phasor relationships.

IV. SIMULATION
AND EXPERIMENTAL
RESULTS
are readily determined from the equivalent circuit parameters and the imposed conditions. Initial values y(O) need
The experimental machine has a six-pole power windto be estimated to start the process; an estimate of pC0) ing, designed for 60 Hz operation, and a two-pole control
and solutions of (61, (7), and (9) provide these values. It is winding for excitation in the range 0-60 Hz. Open-loop
found that if P(") is close to the true solution convergence speed control, in accordance with (3), is shown in the test
is satisfactory. Inappropriate choice of PcO)may cause data of Fig. 5 for operation in the "subnatural" speed
convergence to solutions representing unstable operating range against a load of approximate constant power. The

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

968

control current

(5HZ)

12

20
24
phase control voltage, V
16

28

(a)

phase control voltage, V

-20

(b)

Fig. 6 . Representative phasor diagrams: (a) lagging power factor and

(b) unity power factor.

slow ramp of the control frequency can be followed in a

true synchronous mode: rapid adjustments and operation
in the hypernatural speed range corresponding to

r
3
0.8
Y

:0.4
L

cannot be achieved without closed-loop speed control.

The open-loop unstable regime given by the above expression is attributed to induction mode torques from both
the power windings and the control windings which combine to overcome the diminishing synchronous torques
given by (13). Above the naturally unstable speed, f m u ,
open-loop stability returns.
Solution of the performance equations given in Sect.
111, results in phasor diagrams of the form shown in Fig. 6
(in which some nonrepresentative magnitudes are used to
assist clarity). The phasor diagrams illustrate how an
increase of control current serves to improve motor power
factor, exactly analogous to synchronous condenser operation of a conventional synchronous machine. In order to
improve the power factor and extend the operation to
draw leading VARs is only possible if the expense of
larger rotor currents and associated losses as shown in
Fig. 7. These model predictions correlate well with test
data [181. The calculated input power requirement of the
BDFM in order to draw leading VARs is shown in Fig. 8.
The cost of operating the machine in this mode is unac-

12

20

16

24

28

phase control voltage, V

(C)

Fig. 7. Operation as synchronous condenser: (a) terminal currents, (b)

power phase angle, and (c) power relations.

ceptable. However, it should be emphasized that the experimental motor has a low rating ( - 5 LVA) and requires serious design investigations.
V. CONCLUSIONS
A solution technique has been developed which enables
predictions of BDFM performance in the steady-state
mode with minimum restrictions: the parameters of the
model are all readily determined from machine dimensions and operating conditions.

969

LI ef al.: SYNCHRONOUS DRIVE PERFORMANCE OF BRUSHLESS DOUBLY-FED MOTORS

phase angle of stator voltages in rotor frame

W
power
F ; y function vector; vector of unknowns

2.0

POWERRELATIONSHIPS
The per axis power equations of the BDFM are determined from the electrical performance equations of
Sect. I1 B as follows.
For the power winding:
APPENDIX:

I.5

I .o

a
a
>
s

W,

0.5

~,(v,Pjp*)x , ( ( r p + ~ X , ) Z ;+jxPRZRi:),
=

i.e.,

0
input power, kW

In this the first term is obviously the power winding loss

and the second term is the power which crosses the airgap
from the power winding, Wpa,.
Similarly for the control winding:

Fig. 8. Power requirements for power factor correction.

Precise open-loop synchronous speed control is realized

by adjustment of the frequency of the control winding
voltage and current.
Torque of the machine and power factor of the power
winding are controlled by the magnitude of the control
winding current, in a manner comparable to that of a
conventional synchronous machine.
The rating of the control winding, and the power electronic converter, is a significantly small fraction of the
rating of the machine but is a function of the required
operating conditions of speed and power factor.
The solution technique is appropriate to examine methods of efficiency improvement either as an ASD or power
factor compensation.

W,

se(Vci:)= rcI: + swCag,

(19)

se(-jXCRiRjC*).

(20)

in which
WCag=

Finally, for the rotor there is a power balance equation of

the form

I.e.,

VI. NOMENCLATURE

A. Subscripts and Operators

P; C
R; m
r; i
e; ag
!Re; Im

power winding; control winding

rotor electrical; rotor mechanical
real part; imaginary part
electrical; air gap; mechanical
real part of; imaginary part of

B. Parameters

v; z

phasors of voltage; current

currents: instantaneous; rms
frequency
f
slip
S
P
number of pole pairs
r ; X winding resistance; reactance at power frequency
T ; T torques: instantaneous; steady-state
M
mutual inductance
i; I

which can be written as

thus the total power crossing the airgap can be apportioned between rotor loss and mechanical output.

REFERENCES
L. J. Hunt, A new type of induction motor, Joumal IEE (London),
vol. 39, pp. 648-667, 1907.
F. Creedy, Some developments in multispeed cascade induction
motors, Joumal IEE (London),vol. 59, pp. 511-532, 1921.
B. H. Smith, Synchronous behavior of doubly-fed twin-stator
induction machine, IEEE Truns. Power Apparut. Syst. vol. PAS-88,
pp. 1227-1236, 1967.
A. R. Broadway and L. Burbidge, Self-cascaded machine: a low
speed motor or high frequency brushless alternator, h o c . IEF
(London),vol. 117(7), pp. 1277-1290, 1970.
A. Kusko and C. B. Somah, Speed control of a single frame
cascade induction motor with slip-power pump back, IEEE Trans.
Ind. Appl., vol. LA-14(2), pp. 97-105, 1978.

IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 30, NO. 4, JULY / AUGUST 1994

F Shlbata and K Taka, Speed control system for brushless

cascade induction motors in control range of S, > 1 and S, > 1,
IEEE Winter Pouer Meeting 1986, Paper 86WM226-5
H K Lauw, Characteristics and analysis of the brushless doublyfed machine, report for Bonneville Power Administration, Contract N o 79-85BP24332-Mod 4, June 1989.
A K Wallace, R Spte, and H K Lauw, The potential of
brushless doubly-fed machines for adjustable speed drives, IEEE
U S Pulp and Paper Technical Conference Record, pp 45-50, 1990
A K Wallace, R SpCe, and G C Alexander, The brushless
doubly-fed machine Its advantages, applications and design methods, IEE Sath International Conference on Electncal Machines and
Dnues, Conference Publication 376, IEE (London), pp 511-517,
1993
G H Rawcliffe, Induction motors old and new, h o c Roy Soc
London, vol A362, pp 145-178, 1978
A K Wallace, R SpCe, and H. K Lauw, Dynamic modeling of
brushless doublv-fed machines, IEEE U S Annual Meeting
- Confer.
ence Record, p p 329-334, 1989.
R. SpCe, A. K. Wallace, and H. K. Lauw, Performance simulation
of brushless doubly-fed adjustable speed drives, IEEE U S Annual
Meeting Conference Record, pp. 738-743.
A. K. Wallace, P. Rochelle and R. SpCe, Rotor modeling and
development for brushless doubly-fed machines, International
Conf. on Electrical Machines, Cambridge, MA, pp. 54-59, 1990.
P. Rochelle, R. SpCe, and A. K. Wallace, The effect of stator
winding configuration on the performance of brushless doubly-fed
machines in adjustable speed drives, IEEE U S AnnuaI Meeting
Conference Record, pp. 331-337, 1990.
R. Li, A. K. Wallace, R. SpCe, and Y. Wang, Two-axis mode
development of cage-rotor, brushless doubly-fed machines, IEEE
Trans. Energy Conc., vol. 6(3), pp. 453-460, 1991.
R. Li. A. K. Wallace, and R. SpCe, Dynamic simulation of
brushless doubly-fed machines, IEEE Trans. Energy Con.,
pp. 445-452.
V. Javadekar, D. K. Ravi, R. Spte, and A. K. Wallace, A
variable-speed automotive alternator, Fourth European Conference on Power Electronics And applications, vol. 4, pp. 99-103, 1991.
R. Li, A. K. Wallace, and R. SpCe, Determination of converter
control algorithms for stable brushless doubly-fed drives using
Floquet and Lyapunov techniques, IEEE Power Electronic Specialists Conference Record, pp. 571-577, 1991.

Ruqi Li (S90-M92) was born in Beijing, China.

He received the B.S. degree from the Wuhan
Institute of Hydraulic & Electrical Engineering,
Wuhan, China, in 1982, the M.S. and the Ph.D.
degrees from Oregon State University in 1988
and 1991, respectively. From 1982 to 1985 he
was with the Department of Electrical Engineering, Beijing University of Agricultural Engineering. He worked at the University of California,
Berkeley, and Power Ten, Inc., Sunnyvale, California, as a research and design engineer from
November 1991 to October 1992. Currently he is with the Electrical
Division, Failure Analysis Associates, Inc., Menlo Park, California. His
areas of interest in1dude adjustable speed drives, power electronics and
power systems.

Rene Sp6e (S84-M88-SM92) was born in

Stuttgart, Germany He attended the University
of Stuttgart and Oregon State University, where
he received the M S and Ph.D degrees in 1984
and 1988, respectively.
In 1988 he joined the Department of Electrical and Computer Engineering at Oregon State
University, where he is currently an Associate
Professor His areas of interest include power
electronic systems, electric machines, ac adlustable speed drives, and utility applications of

power electronics.
Dr Sp&eis a member of the Industrial Drives and electric Machines
Committees of the IEEE Industry Applications Society He is the
recipient of an IEEE-IAS Prize Paper Award

Alan K. Wallace (M78-SM84) was born in

Sheffield, England, and received the B.Eng. and
Ph.D. degrees in electrical power engineering
from the University of Sheffield in 1963 and
1966, respectively.
From 1966 to 1967 he worked with Imperial
Chemical Industries on the application of digital
computers to process control. In 1967 he joined
the University of Nottingham and taught electrical machine design and power system analysis
until 1974. From 1974 to 1984 he was engaged in
design and development activities in the ground transportation h&ry
of Canada. He worked with Spar Aerospace of Toronto and Canadair
Services and was Manager of Power Distribution for the Urban Transportation Development Corporation located in Kingston, Canada. In
1984 he joined the Department of Electrical and Computer Engineering
at Oregon State University, Corvallis. His interests are primarily in
adjustable-speed drives, linear motor applications to transportation, and
power electronic applications.
Dr. Wallace is a member of the Institution of Electrical Engineers
(London, UK) and is a Registered Professional Engineer in the Province
of Ontario. From 1988-1991 he was an associate editor of the IEEE
TRANSACTIONS
ON POWERELECTRONICS.

Gerald C. Alexander (M51) received the B.S.

degree from Oregon State College in 1951, the
Sc.M. degree from the Massachusetts Institute
of Technology in 1961, and the Ph.D. degree
from the University of California at Berkeley in
1973. His industrial and academic experience
has ranged from analysis of radio and radar
systems, electromagnetics and antennas, through
power systems, power electronics, and design
and control of machines. He has been on the
faculty of the Department of Electrical and
Computer Engmeering at Oregon State University since 1955 as well as
consulting as a licensed Professional Engmeer in the State of Oregon.
His current activities are focused on parameter estimation of electrical
machines operating under load, data acquisition systems, and instrumentation and high-frequency transients in power systems.