Content uploaded by P.J. Tavner
Author content
All content in this area was uploaded by P.J. Tavner on Oct 14, 2019
Content may be subject to copyright.
The BDFM as a Generator in Wind Turbines
R. A. McMahon, X. Wang
and E. Abdi-Jalebi
Engineering Department
Cambridge University
Trumpington Street
Cambridge CB2 1PZ, UK
P. J. Tavner
School of Engineering
Durham University
Durham DH1 3LE, UK
P. C. Roberts
Scientific Generics Ltd.
Harston Mill, Harston
Camridge CB2 5GG, UK
M. Jagiela
Politechika Opolska ul.
Luboszycka 7
45-37 Opole, Poland
Abstract— The Brushless Doubly-Fed Machine (BDFM) is
attractive for use in wind turbines, especially offshore, as it
offers high reliability by virtue of the absence of brush-gear.
Critical issues in the use of the BDFM in this role at a system
level include the appropriate mode of operation, the sizing of
associated converter and the control of the machine. At a machine
level, the design of the machine and the determination of its
ratings are important. Both system and machine issues are
reviewed in the light of recent advances in the study of the BDFM,
and preliminary comparisons are made with the well-established
doubly fed wound rotor induction generator.
I. INTRODUCTION
The generation of electrical power from wind energy is
a proven technology. There is already a substantial installed
capacity in a number of countries and new capacity is being
continually added. Various designs of wind turbines have been
proposed but horizontal axis, three-bladed machines with a
capacity from 600 kW to 3MW or above is the currently
preferred option. In these machines a variety of architectures
is used, including direct drive, in which the turbine is directly
coupled to a low speed generator, and indirect drive, with the
turbine coupled through a gearbox to a high speed generator.
There is only one large manufacturer of direct drive turbines
and they have a significant installed capacity, especially in
northern Europe. However, the majority of wind turbines in
the world employ the indirect drive architecture. In these
machines, generation is normally from a slip-ring induction
machine. As the gearbox gives an increased shaft speed, a four
or six pole machine can be used. The stator of the generator
is connected directly to the fixed voltage, fixed frequency
electricity grid and the rotor is supplied through the slip rings
with a variable voltage, variable frequency supply generated
from a converter.
This form of double feed enables generation to take place
over a range of turbine speeds and provided that the range
of speeds is moderate, the converter rating need only be a
fraction of the total generator output thereby keeping the
system cost low yet retaining a reasonable level of reliability.
The converter rating is generally about one third of the rating
of the generator, allowing speed variations of ±33%. Varying
the voltage supplied to the rotor can be used to manage the
reactive power flow from the generator. Control schemes have
also been developed to enhance the response of the system
to changing wind speed and to accommodate varying grid
conditions.
However, there are drawbacks to the use of slip ring
induction generators, particularly the additional cost and bulk
of a machine which incorporates slip rings and the need to
maintain brush-gear and to replace brushes on a regular basis.
Studies have shown that the reliability of large wind turbines
is improving but that faults with generator and converter sub-
assemblies contribute significantly to turbine failure rates [1].
Further studies have shown that problems with brush-gear
are a significant issue in wind turbine operation and that the
problem will be more severe in machines deployed offshore
where additional wind resources are available [2].
The Brushless Doubly-Fed Machine (BDFM) is an in-
teresting alternative to the slip ring induction motor which
eliminates the need for brush-gear. The BDFM, alternatively
known as the self-cascaded machine, is one of a class of
doubly fed machines which includes the slip ring induction
machine [3]. The key to the machine is the use of two stator
windings of different pole numbers, chosen so that there is no
direct coupling between them, in combination with a special
form of rotor which can couple both fields. This form of
self-cascaded machine was patented by Lydall in 1903 [4]
and following improvements by Hunt [5] enjoyed a degree of
commercial success.
Burbridge and Broadway re-examined the machine and
proposed new concepts in the design of the rotor winding
and, importantly, proposed the use of double feed, that is the
connection of one stator to the fixed frequency mains and
the supply of the second stator with an converter giving a
variable voltage, variable frequency supply [6]. With double
feed, the machine can act in a synchronous mode with a shaft
speed related to the two excitation frequencies. Later work by
Wallace, Spee and others led to the development of the BDFM
in its contemporary form; indeed the term BDFM is due to
them. Their interest was very much focused on the use of the
BDFM as a variable speed generator for wind generation [7]
and, to a lesser extent, on variable speed drive applications
such as pumping [8].
Several BDFMs have been constructed in recent years,
including the 182 frame size machine used by Brune et al.
[7], the 160 frame size machine reported by Williamson and
Ferreira [9] and a similar size machine by Roberts et al. [10].
To be presented at European
Power Electronics Conference,
Portoroz, Slovenia,
August, 2006
The largest machine appears to be the 100 kW 12/8-pole
machine reported by R¨uncos et al. [11] but few details are
given. Nevertheless, only recently has attention been given to
issues such as the design of the machine, its ratings and its
operation as a system component. In this paper these, and
other relevant issues, are explored for the use of the BDFM
as a replacement for doubly fed induction generators for wind
turbines, particularly in an offshore environment.
II. BASIC SYSTEM CONFIGURATION
A BDFM can operate in several modes but the synchronous,
or doubly fed, mode is used for controlled variable speed
operation. In this arrangement, shown in Fig. 1, one winding,
the power winding, is connected directly to the mains or
grid. The other winding, the control winding, is supplied
with variable voltage at variable frequency from a converter
connected to the mains or grid. Details can be found in [9],
[10].
BDFM
frequency converter
Fractionally rated
p1p2
3φgrid, 50Hz
3φvariable frequency
Fig. 1. BDFM system configuration
Stator and rotor quantities are shown for the synchronous
mode in Fig. 2. The shaft angular velocity is given by
ωr=ω1+ω2
p1+p2
(1)
where ω1and ω2are angular frequencies of the supplies to
the power winding (p1pole pairs) and the control winding (p2
pole pairs) respectively. and slips s1,s2for the two windings
can be defined as
s1ω1−p1ωr
ω1
(2)
s2ω2−p2ωr
ω2
(3)
A further relationship for the so-called natural speed ωn,
that is the synchronous speed when the control winding is fed
with dc, is given by
ωn=ω1
p1+p2
(4)
p1p2
ω1
p1
ω2
p2
ωr1ωr2
V1V2
ωr
f1=ω1
2πf2=ω2
2π
Stator 1 Rotor Stator 2
Fig. 2. BDFM synchronous mode of operation
III. EQUIVALENT CIRCUIT FOR THE BDFM
The analysis of the performance of the BDFM as a generator
is greatly aided by the use of an equivalent circuit, and a
form due to Roberts et al. is shown in Fig. 3 [10]. Quantities
are shown referred to the power winding and iron losses are
neglected. The circuit is valid for all other modes of operation,
including the synchronous mode. Most parameters are as found
in a standard induction machine but the referred rotor reac-
tance L
rwill be relatively larger than that for a standard cage
rotor machine. The design constraints on the rotor generally
lead to a larger harmonic, or differential, leakage component
than normal and this component is likely to be the largest
component of rotor leakage inductance. Furthermore, with
complex rotors, such as the nested loop design, the presence
of multiple sets of independent rotor circuits means that the
bulk equivalent circuit parameters associated with the rotor
change with rotor frequency, but these changes are normally
not significant.
I1I
2
V1
s2
s1
V
2
I
r
R1
Vr1Vr2
R
2s2
s1
R
r/s1
jω1L1jω1L
2
jω1L
r
jω1Lm1jω1L
m2
N1:1 1:N2
V2
I2
Zr
Fig. 3. BDFM Referred Per-Phase Equivalent Circuit
Vr1Vr2
I1I
2
I
r
V1,ω
1
s2
s1
V
2
R1R
2s2/s1
R
r/s1
jω1L
r
jω1Lm1jω1L
m2
Fig. 4. Alternative Referred Per Phase Equivalent Circuit
In general, values for parameters can be found by analysis,
numerical simulation or experimental determination. In the
case of the BDFM not all the parameters can be found from
terminal measurements so an alternative, but electrically equiv-
alent, form of circuit has been proposed, shown in Fig. 4 [10].
The equivalent circuit as shown assumes that the saturation
of the iron circuit, if it occurs, does not significantly affect
parameter values. The parameters are defined in Table I.
TAB L E I
DEFINITION OF PARAMETERS
Parameters Power wind-
ing
Control
winding
Rotor
Resistance R1R
2R
r
Inductance L1L
2L
r
Magnetizing
inductance Lm1Lm
2-
Turns ratio to rotor N1:1 N2:1 -
The equivalent circuit can be simplified for a core or ideal
BDFM. Noting that the rotor reactance is likely to dominate
the overall rotor impedance leads to a core BDFM as shown
in Fig. 5(a). In an ideal BDFM, the rotor impedance will be
zero and that leads to the model in Fig. 5(b). These simplified
models are useful in deriving certain benchmark results.
(a) (b)
I1I
2I
2
I1
V1,ω
1
s2
s1
V
2
s2
s1
V
2
V1,ω
1
jω1L
r
Fig. 5. BDFM Core Model and Ideal Model
From Fig. 5(a) the equation for the synchronous component
of BDFM torque, T, can be derived:
T=3|V1V
2s2|
ωn|ω1L
rs1|sin δ(5)
where δis the load angle.
The widely quoted relationship relating the power in the
control winding, P2, to that in the power winding, P1,is
obtained from the circuits in Fig. 5.
P2=P1
ωr
ωn
(6)
IV. PARAMETER EXTRACTION
Parameters for the equivalent circuit can be calculated
or measured experimentally. If calculated it is desirable to
confirm values experimentally. Some parameters are relatively
straightforward to measure, for example the resistances of the
two stator windings by dc measurements, although allowance
must be made for temperature changes. The magnetizing
reactances of the two stator windings can be measured by
driving the BDFM externally at the appropriate synchronous
speeds. The extraction of further parameters is not easy but
a method based on fitting to measured torque speed curves
is given by Roberts et al. [10]. The referred rotor resistance
and reactance can be obtained as well as the ratio of the turns
ratios for the two couplings of different pole number.
A complete set of parameters is then available for the
equivalent circuit of Fig. 4. Unfortunately, measurements at
the machine terminals do not allow the independent determi-
nation of the two stator leakage reactances. However, if rotor
quantities can be measured, then they can be determined. Abdi
Jalebi et al. have shown how this is possible using wireless
transmission of current measurements by Bluetooth [12]. The
ability to separate parameters has the advantage of being
able to compare stator leakage reactances with manufacturers’
estimates.
Parameters can also be obtained by calculation and this is
particularly useful for the inductances and rotor resistance.
Williamson et al. provide a method of machine analysis
based on the harmonic decomposition of machine mmfs [13].
Although the method was described in the context of a
machine with a nested loop rotor, the method of analysis
is applicable to other rotor designs. An alternative approach
based on coupled circuits was described by Wallace et al.
[14] and this was developed and generalized by Roberts
to enable parameter values to be found for BDFMs with
different rotor configurations [15]. Kroitzsch and Riefenstahl
have also recently applied coupled circuit analysis to the study
of cascaded machines, including the BDFM [16].
Finite element methods can also be used to analyse electrical
machines and they can take into account saturation. The time
stepping approach has been used by Ferreira et al. [17] and
more recently by Jagiela et al. [18]. However, this type of
numerical analysis does not yield parameter values directly
and so far have not been used to generate parameter values.
In addition, there are some special issues in the BDFM relating
to the field pattern which are described later.
V. G ENERATOR OPERATION
The BDFM is operated as shown in Fig. 1. Under these
conditions the power winding voltage and frequency are fixed.
The control winding frequency is set by the shaft speed by
equation (1). The real power flow is determined by the me-
chanical input power, i.e. the torque multiplied by the shaft’s
angular velocity, less losses in the machine. The remaining
variable is the control winding voltage which controls the flow
of reactive power, as in a conventional synchronous machine.
However, the flow of VArs is subject to an amplification factor,
that is the ratio of VArs generated in the power winding to the
input VArs on the control winding, which falls with increasing
speed deviation. The flow of VArs has a major effect on the
rating of the machine and associated converter or, put another
way, the real power that a machine of fixed rating can handle
is influenced by the reactive power regime.
An illustration of the control of reactive power is given in
Fig. 6. The power winding VArs have been calculated using
the core model of Fig. 5 (a) and the equivalent circuit of Fig. 4
for generating at 750 rpm with a driving torque of 25 Nm. The
machine is the frame size 180 machine described in [10]. The
progressive reduction to zero of the lagging VArs drawn by the
power winding and the subsequent rise in VArs exported as
the control winding voltage is increased is seen. The effect of
stator impedances and magnetizing reactances make the actual
control winding voltage greater than that predicted from the
core model.
−1.5 −1 −0.5 00.5 1
−2.5
−2
−1.5
−1
−0.5
0
0.5
Power winding VArs (kVAr)
Control winding VArs (kVAr)
Equiv. Cct. Model of Fig. 4
Core model (Fig. 5(a))
Experimental
Fig. 6. The control winding VArs changing with the power winding VArs.
The power winding voltage is fixed at 120 Vrms.NegativesignoftheVArs
stands for absorbing VArs from the grid; positive sign of the VArs stands for
providing VArs to the grid.
Although the ability to export VArs to the grid appears
attractive, it does require the VA ratings of the machine to
be increased, assuming a constant power output. The increase
in the rating of the power winding depends on the load power
factor and the increase in the rating of the control winding will
reflect the parameters of the particular machine. The effect
on the control winding is dependent on the speed deviation
from natural speed as a result of the VAr amplification effect.
This derating is likely to be modest up to a speed deviation
of say 25%, but above a deviation of say 75% the derating is
severe and in reality more than one VAr on the control winding
may be needed to provide one VAr on the power winding.
The effect of VAr generation on available power generating
capacity is shown in Fig. 7. For the same conditions as in
Fig. 6, viz. a shaft speed of 750 rpm and a driving torque
of 25 Nm, the power output of the BDFM relative to the
maximum power is plotted as a function of control winding
voltage. Plots using the core model of Fig. 5(a) and the
equivalent circuit of Fig. 4 are shown. The figure shows that
the power handling of the machine is compromised by both
under and overexcitation of the control winding, as expected.
The degree of excitation has therefore to be chosen with care.
VI. CONVERTER ISSUES
Varying the control winding voltage affects the rating of the
inverter supplying the control winding as well as the rating of
the winding itself. The link between the grid and the control
winding will be a bi-directional converter. The machine-side
inverter will handle real power and reactive power and will
be matched to the control winding. In contrast, the line-
side inverter need only transfer real power, and therefore in
principal could have a lower rating. The ratings of the two
inverters are shown in Fig. 8 as a function of speed deviation.
The operating conditions are 120 Vrms on the power winding,
a load torque of 25 Nm and the control winding voltage is
40 60 80 100 120 140
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Control winding voltage (V)
Equiv. Cct. Model of Fig. 4
Core model (Fig. 5(a))
PBDFM
¯
B¯
Jωrlπd/2
Fig. 7. Dimensionless quantity of the ratio of the total BDFM power output to
the product of the BDFM magnetic loading, electric loading, ωr, and machine
volume. The ratio was calculated over a range of control winding voltages.
The power winding voltage was fixed at 120 Vrms, prime mover torque at
25Nm, shaft speed at 750rpm.
adjusted to give unity power factor on the power winding.
Note that this condition involves transfer of VArs to the power
winding to supply the magnetizing VArs. Note also that the
rating of the line side inverter is equal to the real power
transferred to the mains.
0 0.5 1 1.5 2
0
0.5
1
1.5
2
2.5
3
ωr
ωn
Sinv1experimental
Sinv2experimental
Sinv1simulation
Sinv2simulation
Inverter ratings (kVA)
Fig. 8. Inverter ratings. Sinv1is the machine-side inverter rating and Sinv2
is the line-side inverter rating
The results show that power factor correction, that is the
transfers of VArs to the power winding in this case, comes
at a substantial penalty in machine-side inverter rating, and
that penalty becomes larger the greater the speed deviation.
An alternative strategy is to supply some VArs to the grid
from the line side inverter and allow the power winding to
draw lagging VArs. Although this will increase the rating
of the line-side inverter, the control winding voltage can be
reduced, reducing both the ratings of the control winding and
the machine-side inverter. At each speed the total converter
rating can be minimized. A plot showing the minimized rating
compared to the rating used in Fig. 8 against a base of speed
deviation is shown in Fig. 9.
0 0.5 1 1.5 2
0
1
2
3
4
5
ωr
ωn
Minimized total rating
Experimental
Total inverter rating (kVA)
Total rating as in Fig.8
Fig. 9. Comparison of minimized total inverter rating and total inverter rating
showninFig.8
The strategy of using both inverters to supply VArs re-
duces the total converter rating. In practice, it is likely to
be convenient to make both inverters of equal size and the
optimization can be carried out to equalize the loadings. There
is also the possibility of optimizing converter ratings and
machine size as supplying the VArs on the power winding
side of the machine reduces the rating of the control winding,
permitting the use of a smaller machine or allowing a higher
output. Such an optimization would aim to minimize system
cost and the relative cost of machine and converter would be
needed as inputs. Finally it should be noted that the prototype
machine has somewhat larger resistances and reactances than
a production machine of similar rating and so the VArs and
losses are greater than normal. Furthermore, large machines
are proportionately less resistive than small machines.
VII. MACHINE DESIGN
A. Principles of operation
The design of the rotor is key to operating of a BDFM
as it must couple two fields. Lydell proposed a rotor with
two distinct windings, as in the stator. This approach leads
to a large rotor resistance and hence losses. Hunt realized
that conductors in the rotor were carrying currents producing
mmfs of opposite sign and he showed how a winding with
lower resistance could be devised. As a result, the self-
cascaded machine enjoyed some commercial success but the
winding was irregular and therefore expensive to manufacture.
Broadway and Burbridge re-examined the issue of rotor design
and proposed two classes of design, the nested loop design
and the progressive loop design. The nested loop design has
been generally adopted in recent designs [9] although the
practicality of progressive loop designs has recently been
shown. Both types of design have a regular pattern, simplifying
construction, but the relative sparsity of conductors brings
increased space harmonics. Analysing the detailed effects of
the space harmonics is not straightforward but an increased
rotor leakage inductance is one important consequence.
B. Air gap flux
The air gap flux is the resultant of three mmfs - those of
the two stator windings and that of the rotor. The resulting
air gap flux distribution contains fields of the pole numbers
of the two principal couplings, plus rotor space harmonics
and slotting harmonics. Space harmonics from the distributed
stator windings exist, but are small. A flux plot is shown
in Fig. 10, derived from finite element analysis. The flux
pattern is unconventional, being primarily the sum of two
asynchronous fields of different pole number. An instantaneous
plot, as in Fig. 10, gives an apparent 6-pole pattern. The
presence of two different air gap field components creates a
difficulty in determining the magnetic loading; the issue is
considered by McMahon et al. [19] and a simple closed form
of the quadrature sum of the two fields is shown to be a good
representation.
C. Determining the relative magnitude of the air gap fields
At first sight, the ratio of the amplitudes of the two principal
field components, B1and B2, might appear a free parameter,
but it can be shown that for maximum machine output there
is an optimum ratio, which can be approximated to
B2
B1
=3
p2
p1
(7)
Fig. 10. Flux pattern in the BDFM; Vs1=90Vrms, (power winding voltage),
f1=50hz, Vs2=64 Vrms, (control winding voltage), f2=30hz, ωr=83.78
rad/sec, t=0.2s
This translates into a rotor turns ratio of 3
(p1
p2)2and this
has important implications for rotor design. Obviously, with
normal distributed windings, it is relatively easy to achieve a
given turns ratio, but with nested loop and progressive loop
designs it is not so easy. The rotor turns ratio nris given by
nr=N1rKω1r
N2rKω2r
(8)
where N1rand N2rare turns for the 2p1and 2p2pole fields
respectively, with constant winding factors Kω1rand Kω2r.
For the nested loop type rotor with single turn loops, only
the winding factor can be changed. The following discussion
is based on a BDFM with one loop in each nest. Adjusting the
turns ratio by changing a winding factor essentially degrades
the coupling to a winding and increases referred resistances
and hence losses. If the ratio of the pole numbers approaches
unity, the desired winding factors become approximately equal
and the correct pitch for the loop can be calculated. However,
if the ratio of the pole numbers is large, as in the case with
the 2/6-pole combination, which is attractive as it offers the
highest natural speed, the nris not close to unity. For a 2/6-
pole machine, the required nris 0.48 so one winding factor
will necessarily be small and lead to a high referred resistance.
The use of multiple loops is more favourable, although the
analysis is not straightforward. To aid analysis, Roberts has
shown that, to a good approximation, a nest of loops can be
represented by an electrically equivalent single circuit [15].
In addition, alternative rotor windings such as the progressive
loop approach may give better performance.
D. Balancing the magnetic and electric loadings
It is well known in machine design that the magnetic and
electric loadings need to be balanced to obtain maximum
output from a given frame size machine. This procedure is
more difficult in a BDFM as there are two principal air gap
flux components present. The rotor design is likely to be the
most challenging as the smaller diameter compared to the
stator restricts the space available. Making the rotor teeth
larger allows a higher magnetic loading to be achieved but this
takes place at the expense of rotor slot area and hence electric
loading. At the same time there must be enough back iron to
carry the flux without saturation, allowing for the presence of
a shaft, and the problem is more critical as the pole numbers
of the fields are reduced.
A view has to be taken of the acceptable current density in
the rotor conductors, which can generally be higher than the
stator as the rotor can run hotter. Set against this is the uneven
distribution of current in the loops within one nest in the nested
loop type of winding; the progressive loop winding is better
in this respect. At present there is not a simple procedure
for rotor design but an iterative design scheme to optimize
machine rating has been proposed [20], [21].
Similar considerations apply to the stator. The windings
are generally multi-turn coils, and normal slot fill factors will
apply. Also, experience suggests that the two stator windings,
which of course occupy the same slots, must be mutually
insulated. Again there is no simple design procedure to balance
the tooth width and slot areas, and also enough stator back
iron must be provided. The iterative design procedure used
for the rotor can be applied to the stator. It is worth noting
[22] that some economy in stator conductors could be obtained
by ingenious windings, as for example used by Hunt, but the
benefit of electrically isolated windings is more valuable.
VIII. MACHINE RATING
The rating S of a BDFM at natural speed is given by
|S|=π2
√2(d
2)
2
l¯
B¯
J|ω1|
p1
1
(1 + 1
nr)1+( nrp2cos φ
p1cos(φ+δ))2
(9)
where cos φis the power factor of the power winding, ¯
Bthe
magnetic loading, ¯
Jthe electric loading, d the rotor diameter
and l the stack length.
As has been shown earlier, careful management of real
and reactive power flows will enable the machine to generate
an amount of real power approaching the theoretical rating,
assuming that resistive losses are low. For the frame size 180
4/8-pole machine reported by the authors, realistic values of
electric and magnetic loadings are 31.9 kA/m and 0.33 T re-
spectively. The reduced magnetic loading ¯
Bis a consequence
of premature saturation of rotor teeth and there seems no
practical barrier to a ¯
Bof 0.45 T or higher in a commercial
machine. The natural speed is 500 rpm on a 50 Hz supply. The
rating of the machine is then 76 Nm at 500 rpm, equalling
4 kW. Under these conditions the load angle δis about 40
degrees and the pull out torque is 117 Nm, using parameter
values published in [10].
This performance can be compared to that of a wound rotor
induction machine. The rating of such a machine is given by
Sim =π2
√2(d
2)
2
lωs
p¯
B¯
J(10)
where ωsis the synchronous angular frequency.
A direct comparison to a 4/8-pole BDFM is a wound rotor
induction machine with a synchronous speed of 500 rpm, i.e.
a 12-pole machine. The output of the BDFM is about 20%
less for the same volume of active materials. However, when
the removal of brush gear is allowed for, the BDFM may in
fact be smaller and cheaper.
IX. MACHINE CONTROL
The BDFM is a synchronous system but its control is
complicated by the presences of unstable regions, as first noted
by Spee et al. [23]. Li et al. from the same research group
first investigated the open loop stability of the system [24].
Various control algorithms for the BDFM have been published,
for example by Zhou et al. [25]. The realization that a d-q
model of the machine in the rotor reference frame could be
rotated into synchronism with the stator currents, represented a
significant step forward [26]. This realization was generalized
and used to present a general eigenvalue stability analysis in
[15]. Recent investigations into the implementation of control
schemes are also given in [15].
Considerable research has gone into control schemes for
the BDFM; the most promising control scheme to date is a
rotor flux orientated scheme [26]. The rotor flux orientated
scheme (and other schemes, see for example [15]), depend on
complex theoretical models, some involving the linearization
of the inherently non-linear BDFM.
There is a need for further work to develop the rotor
flux orientated control scheme, and other control schemes,
into usable practical schemes. For example, how the control
scheme will respond when issues of grid-load interaction has
not yet been considered, and whether the schemes will be
effective for a wide range of machine designs, particularly
those prone to considerable unstable regions. At that stage
meaningful modelling of the expected behaviour of the BDFM
and its associated converter in a wind turbine application can
be undertaken.
X. CONCLUDING REMARKS
Recent advances in the study of the BDFM allow more
complete consideration of the application of this machine to
variable speed generation in wind turbines. However, work re-
mains to be done in devising a simpler method of determining
the balance of electric and magnetic loadings in the machine,
especially in the rotor. Further investigation of rotor windings
would be also valuable. The control of the machine remains
a challenge and, as well as further development of practical
control schemes, fundamental studies of the influences which
determine the unstable regions of operation are needed.
BDFMs are relatively slow speed machines, and the highest
speed machine, the 2/6-pole, is the counterpart of an 8-
pole wound rotor induction machine. Nevertheless, both the
4/8-pole and 2/6-pole BDFMs with natural speeds of 500
and 750 rpm respectively are practical contenders for use
in indirect drive wind turbines, especially bearing in mind
the balance between generator speed and the gearbox ratio.
However, the great prize which the BDFM offers is the
elimination of brush-gear, an objective which has been the
goal of electrical engineers for many years. The next step is a
comparative costing of a BDFM based system using the design
procedures outlined in this paper.
REFERENCES
[1] P. J. Tavner, J. Xiang, and F. Spinato, “Improving the reliability of
wind turbine generation and its impact on overall distribution network
reliability,” in Proc. IEE CIRED Conf., Turin, 6-9 June 2005.
[2] DOWEC, “Estimation of turbine reliability figures within the dowec
project,” Dutch Offshore Wind Energy Converter Report Nr. 10048,10
2003.
[3] B.Hopfensperger and D.J.Atkinson, “Doubly-fed a.c. machines: classifi-
cation and comparison,” in Proc. 9th. European Conf. Power Electronics
and Applications, August 2001.
[4] Siemens Brothers & Co. Ltd. and F. Lydall, “Improvements in polyphase
induction motors,” British Patent No.: 16839, July 1902.
[5] L. J. Hunt, “The ‘cascade’ induction motor,” Institution of Electrical
Engineers, Journal, pp. 406–434, 1914.
[6] A. R. W. Broadway and L. Burbridge, “Self-cascaded machine: a
low-speed motor or high frequency brushless alternator,” Proceedings,
Institution of Electrical Engineers, vol. 117, pp. 1277–1290, 1970.
[7] S. Bhowmik, R. Sp´ee, and J. H. R. Enslin, “Performance optimization
for doubly fed wind power generation systems,” IEEE Trans. Industry
Applications, vol. 35, no. 4, pp. 949–958, July/August 1999.
[8] R. Li, R. Sp´ee, A. K. Wallace, and G. C. Alexander, “Synchronous
drive performance of brushless doubly-fed motors,” IEEE Transactions
on Industry Applications, vol. 30, no. 4, pp. 963–970, July/August 1994.
[9] S. Williamson and A. C. Ferreira, “Generalised theory of the brushless
doubly-fed machine. part 2: Model verification and performance,” IEE
Proceedings - Electrical Power Applications, vol. 144, no. 2, pp. 123–
129, 1997.
[10] P. C. Roberts, R. A. McMahon, P. J. Tavner, J. M. Maciejowski, and
T. J. Flack, “Practical equivalent circuits and parameter estimation for
the brushless doubly-fed (induction) machine (BDFM),” Proc. IEE B -
Elec. Power App., 2004.
[11] F. R¨uncos, R. Carlson, A. M. Oliveira, P. Kuo-Peng, and N. Sadowski,
“Performance analysis of a brushless doubly-fed cage induction gener-
ator,” in Proc. 2nd Nordic Windpower Conf., 2004, 1-2 March.
[12] P. C. Roberts, E. Abdi-Jalebi, R. A. McMahon, and T. J. Flack, “Real-
time rotor bar current measurements using bluetooth technology for a
brushless doubly-fed machine (BDFM),” in Int. Conf. Power Electronics,
Machines and Drives, vol. 1. IEE, March 2004, pp. 120–125.
[13] S. Williamson, A. C. Ferreira, and A. K. Wallace, “Generalised theory
of the brushless doubly-fed machine. part 1: Analysis,” IEE Proceedings
- Electrical Power Applications, vol. 144, no. 2, pp. 111–122, 1997.
[14] A. K. Wallace, R. Sp ´ee, and H. K. Lauw, “Dynamic modelling of
brushless doubly-fed machines,” in Conference record of the IEEE
Industry Application Society Annual Meeting. San Diego, CA: IEEE,
1989.
[15] P. C. Roberts, “A study of brushless doubly-fed (induction) machines,”
Ph.D. dissertation, University of Cambridge, 2004, available from
http://www-control.eng.cam.ac.uk/˜pcr20.
[16] J. Kroitzsch and U. Riefenstahl, “The inverter controlled cascade ma-
chine as a variable speed brushless generator,” in Proc. 2th Int. Conf.
Power Electronics, Machine and Drives, Edingburgh, Uk, 31March -2
April 2004.
[17] A. C. Ferreira and S. Williamson, “Time-stepping finite-element analysis
of brushless doubly fed machine taking iron loss and saturation into
account,” IEEE Transactions on Industry Applications, vol. 35, no. 3,
pp. 583–588, 1999.
[18] M. Jagiela, M. Lukaniszyn, P. Tavner, and J. Bumby, “Formation
of circuit equations for inclusion in 2-d time-stepping finite element
analysis of rotating machines,” in Proc. Int. Conf. Fundamentals of
Electrotechnics and Circuit Theroy, Ustron, Poland, May 2005.
[19] R. A. McMahon, P. C. Roberts, X. Wang, and J. Tavner, “Performance
of the bdfm as a generator and motor,” Proc. IEE B - Elec. Power App.,
pp. 289–299, March 2005.
[20] X. Wang, P. C. Roberts, and R. A. McMahon, “Studies of inverter ratings
of bdfm adjustable speed drive or generator systems,” in Proc. 6th Int.
Conf. Power Electronics and Drive Systems, Kuala Lumpur, Malaysia,
December 2005.
[21] X. Wang, R. A. McMahon, P. C. Roberts, and P. J. Tavner, “Optimisation
of bdfm stator design using an equivalent circuit model and a search
method,” in Proc. IEE 3th Int. Conf. Power Electronics, Machines and
Drives, Dublin, Ireland, April 2006.
[22] P. Rochelle, R. Sp´ee, and A. K. Wallace, “The effect of stator winding
configuration on the performance of brushless doubly-fed machines in
adjustable speed drives,” in Conference record of the IEEE Industry
Application Society Annual Meeting. Seattle,WA: IEEE, October 7-12
1990, pp. 331–337.
[23] R. Sp´ee, A. K. Wallace, and H. K. Lauw, “Performance simulation of
brushless doubly-fed adjustable speed drives,” in Conference record of
the IEEE Industry Application Society Annual Meeting. San Diego,
CA: IEEE, 1989.
[24] R. Li, A. Wallace, and R. Sp ´ee, “Determination of converter control
algorithms for stable brushless doubly-fed drives using floquet and
lyapunov techniques,” in Proc. Power Electronics Specialists Conference
(PESC). IEEE, 24-27 June 1991, pp. 571–577, cambridge, MA.
[25] D. Zhou, R. Sp´ee, G. C. Alexander, and A. K. Wallace, “A simplified
method for dynamic control of brushless doubly-fed machines,” in
Conference record of the 22nd Industrial Electronics, Control and
Instrumentation Conference (IECON) 1996, vol. 2. IEEE, 5-10 August
1996, pp. 946–951, taipai, Taiwan.
[26] D. Zhou, R. Sp´ee, and G. C. Alexander, “Experimental evaluation of a
rotor flux oriented control algorithm for brushless doubly-fed machines,”
IEEE Transactions on Power Electronics, vol. 12, no. 1, pp. 72–78, 1997.
