1808
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 6, NOVEMBER/DECEMBER 2012
Design of a Flux-Switching Electrical Generator
for Wind Turbine Systems
Javier Ojeda, Marcelo Godoy Simões, Guangjin Li, and Mohamed Gabsi
Abstract—This paper proposes a parametric optimization of
a flux-switching electrical machine customized for a wind turbine application with a typical operating range for average and
low-power wind energy sites. Statistics of wind resources are
taken into consideration for the machine design for definition
of the turbine power envelope. Both copper and iron losses for
three different machine designs are evaluated. A very important
consideration taken in this design is the elimination of gearbox
requirements for coupling to the turbine. Although the developed approach makes the machine somewhat voluminous, the
overall performance is highly improved because a direct-drive
flux-switching electrical generator becomes very competitive for
small-scale wind turbines. The design methodology presented in
this paper will support widespread application of small-scale wind
turbines for rural systems, farms, and villages. This paper concludes by demonstrating that a very cost-effective distributed wind
system can be approached with this design.
Index Terms—AC motor drives, electric machines, wind energy,
wind power generation.
I. I NTRODUCTION
T
HIS paper proposes, for the first time, the utilization of
a flux-switching electrical machine (used in the past in
motoring mode) [1]–[9], as a wind system generator. The focus
here is specifically for situations with predominantly low power
needs, and with low-wind-speed patterns, as might be found in
inland rural areas.
A custom-designed approach must be followed, in order to
consider the best operating range for low wind velocities, overall optimization, and long-term cost return. In addition, a machine designed for low speed may overpower at higher speed,
and mechanical coupling of turbines to generators with gearbox
decreases efficiency and impacts overall reliability [10], [11].
A typical wind turbine system has its operating range as
shown in Fig. 1. Permanent-magnet generators and self-excited
induction generators have been used for low-power wind turbines, squirrel-cage induction generators are usually used for
Manuscript received December 19, 2011; revised April 11, 2012 and
June 19, 2012; accepted June 27, 2012. Date of publication October 4, 2012;
date of current version December 31, 2012. Paper 2011-EMC-777.R2, approved for publication in the IEEE T RANSACTIONS ON I NDUSTRY A PPLICA TIONS by the Electric Machines Committee of the IEEE Industry Applications
Society.
J. Ojeda, G. Li, and M. Gabsi are with Systèmes et Applications des
Technologies de l’Information et de l’Energie, École Normale Supérieure de
Cachan–University Paris Sud 11–Centre National de la Recherche Scientifique,
UniverSud Paris, 94230 Cachan, France (e-mail: ojeda@satie.ens-cachan.fr;
guangjin@satie.ens-cachan.fr; gabsi@satie.ens-cachan.fr).
M. G. Simões is with the Department of Electrical Engineering and Computer Science, Colorado School of Mines, Golden, CO 80401-1887 USA
(e-mail: mgs@mines.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIA.2012.2221674
Fig. 1.
Control range for a typical wind turbine.
medium-power wind turbines, and doubly fed induction generators or synchronous machines are used for high-power wind
turbines; very small wind turbine generators are often based
on inexpensive permanent-magnet generator machines [12] and
typically do not have optimized operation in low-wind-speed
range. For these reasons, the flux-switching machine (FSM)
may be the best candidate in the special case of low-power lowwind-speed situations.
II. T URBINE P OWER C ONTROL
Large-scale wind turbines frequently have optimized speed
and/or pitch control, but the market for small-scale wind turbines does not accommodate expensive solutions. Gearboxes
are typically used in high-power wind turbines, but they impact
low-power wind turbines for mainly two reasons: 1) They have
low efficiency because of the viscous loss (1% of the rated
power per gearbox stage, proportional to the mechanical speed)
[13], and 2) they are the major factor (at least 19%) of the downtime in wind turbine generations due to maintenance needs [14].
Therefore, an optimized electric generator should be designed
in order to have the best efficiency for a low-wind-velocity
operating range, and a direct-drive system would be preferable.
Small wind turbines have considerable cost constraints in
order to be competitive in rural systems and applications for
farms and villages. Typically, small turbine generators are
permanent-magnet machines, not optimized for capturing wind
energy in the low-wind-speed range (up to about 7 m/s). The
use of gearbox impairs the machine capability to further produce power, since the friction losses are extremely significant.
Therefore, no turbine control methodology will impact the
best operation of the generator in low speed (since most of
the turbine control systems only optimize the aerodynamic
performance). Therefore, wind energy considerations must be
0093-9994/$31.00 © 2012 IEEE
OJEDA et al.: DESIGN OF A FLUX-SWITCHING ELECTRICAL GENERATOR FOR WIND TURBINE SYSTEMS
1809
Fig. 4. Losses for a high-power doubly fed induction generator.
Fig. 2.
Fig. 3.
Wind speed distribution.
Rayleigh wind speed distribution.
taken in the proper electromagnetic and mechanical design of
an electric generator, and it seems that an FSM is the best
candidate for such application.
III. W IND E NERGY D ESIGN C ONSIDERATIONS
Energy captured by the shaft of a wind turbine must be
evaluated by the historical wind power intensity (in watts per
square meter) in order to access the economical viability of
the site. Therefore, it is appropriate to define local wind power
as proportional to the distribution of speed occurrence, and
a statistically based design should consider the different sites
where the turbines will be installed. Thus, with the same annual
average speed, very distinct wind power characteristics may
affect the optimal design of the generator. Fig. 2 shows a typical
curve of wind speed distribution for a given site. If wind speed
is lower than 3 m/s (denominated by “calm periods”), the power
becomes very low for extraction of energy, and the system is
usually stopped. Therefore, calm periods will determine the
necessary time for energy capture. Power distribution varies
according to the intensity of the wind and with the power
coefficient of the turbine. Then, a typical distribution curve of
power is assumed to have the form as in Fig. 3. Sites with high
average wind speeds do not have calm periods, and there is not
much need of storage. However, high wind speed may cause
structural problems in the system or in the turbine.
Although wind resources can be described by the Weibull
distribution, in practice, it is more convenient to use the
Rayleigh distribution given by
v −( vc )2
2
e
.
(1)
h(v) =
c
c
Such function is shown in Fig. 3 where the “factor c” is defined
as the scale factor, related to the number of days with high wind
speeds. The higher is “c,” the higher is the number of windy
days. Therefore, such parameter represents the statistical nature
of wind speed for most practical cases.
The vertical axis in Fig. 2 is given in percentage of hours/
year per meter/second. For optimally designing an electrical
generator, the random nature of wind distribution in a particular
site is considered in order to design the best operating range,
defining electrical characteristics such as machine frequency
and voltage ratings. The majority of losses occurs because a
gearbox is used for matching generator speed with the turbine
speed. Therefore, this paper considers the elimination of gearbox, in order to significantly contribute for an overall optimal
system, which eventually is consistent with the volumetric
needs of ferrite-based magnets, as discussed in the next section.
IV. W IND G ENERATOR P ERFORMANCE C ONSIDERATIONS
An electrical generator used for a wind turbine system has
efficiency imposed by three main issues, as shown in Fig. 4:
1) stator losses; 2) converter losses; and 3) gearbox losses.
Stator losses are considered in this paper by the proper design of
the machine for the right operating range; the converter losses
are given by the proper design of the power electronic circuits
(ON-state conduction losses of transistors and diodes, plus their
frequency-proportional switching losses, are not considered in
this paper, because they must be approached in a specific power
electronic topology that best fits the final design). The third
main factor responsible for a noticeable power loss is the use of
a gearbox, as shown in Fig. 4 for a typical wind turbine system.
It can be considered that mechanical viscous losses due to a
gearbox are proportional to the operating speed, as indicated by
Pgear = Pgear,rated
η
ηrated
(2)
where Pgear,rated is the loss in the gearbox at rated speed
(on the order of 3% of the rated power), η is the rotor speed
(in revolutions per minute), and ηrated is the rated rotor
speed (in revolutions per minute). Losses in the gearbox dominate the efficiency in most wind turbine systems, and simple
calculations show that a significant annual power dissipation in
the generator system is due to the gearbox.
From the full energy available in the wind, just part of it
can be extracted for energy generation, quantified by the power
coefficient Cp . The power coefficient is the relationship of
the possible power extraction and the total amount of power
contained in the wind. The turbine mechanical power can be
given by
Pt =
(Cp ρAV 3 )
2
(in kg · m/s).
(3)
1810
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 6, NOVEMBER/DECEMBER 2012
The air density ρ can be corrected by the gas law ρ = P/RT )
for every pressure and temperature of the place with the following expression:
ρ = 1.2929
273 P
T 760
(4)
where P is the atmospheric pressure (in millimeters of mercury) and T is the Kelvin absolute temperature. At normal
conditions (T = 296 K and P = 760 mmHg), the value of ρ is
1.192 kg/m3 , and at T = 288 K (15 ◦ C) and P = 760 mmHg,
it is 1.225 kg/m3 . One can consider the temperature decrease of
approximately 1 ◦ C for every 150 m. The influence of humidity
can be neglected. If just the altitude h (in meters) is known
(10 000 ft = 3048 m), the air density can be estimated by the
two first terms of a series expansion like
ρ = ρ0 e −(
0.297
3048 h
) ≈ 1.225 − 1.194 × 10−4 h.
(5)
Thus, the turbine torque is given by
Tt =
ρARV 2 CT
Pt
=
ω
2
(6)
where the torque coefficient is defined as CT = Cp /λ.
If S = 1 m2 and ρ = 1.2929 kg/m3 , the maximum potential
of wind can be obtained from the aforementioned equation
(without taking into account the aerodynamic losses in the
rotor, the wind speed variations in several points of the blade
sweeping area, the rotor type, and so on) as calculated by
P
= 0.5926 × 0.6464 × V 3 = 0.3831 V 3
A
(7)
where PA is the wind power per swept area (in watts per square
meter) and V is the wind speed (in meters per second).
For a constant and steady wind speed of 6 m/s, the available
shaft turbine power is 260 W. However, wind is never steady nor
constant, and such instantaneous power calculation is not useful
for sizing and economic studies. The random nature of wind
must be considered, and statistical analysis must be performed.
Assume, for instance, that a wind turbine site, with strong
winds, with periods of low wind and some really good average
speed can be defined as a Rayleigh probability density function.
Equation (8) shows the calculation of the average cubic wind
speed under Rayleigh conditions, and the shape factor (k) is
related to the average cubic wind speed as (9)
∞
3
∞
3
(v )avg =
v h(v)dv =
0
3 √
= k3 π
4
2
k = √ vavg
π
v 2
2v
v · 2 exp −
dv
k
k
3
0
Fig. 5. Typical Rayleigh wind probability density function for low wind speed
as found in inland and rural areas.
That is, under Rayleigh wind speed probability distribution
function, the average power extracted by the turbine shaft is
1.91 multiplied by the instantaneous power calculated by (7).
As an example, a Rayleigh distribution function with a cubic
average of 6 m/s, as shown in Fig. 5, gives 496 W for a 1-mradius blade, and of course, a larger diameter turbine is required
for the machine described in this paper.
Since power raises with the cubic growth of wind speed and
the square of turbine radius, it is expected that, for low-windspeed Rayleigh site conditions (on the order of 8-m/s average
cubic wind speed), a turbine with a radius on the order of 3 m
will be needed for the generator described in this paper. The
turbine mechanical design is not the scope of this paper and is
left for future research and development.
Therefore, the main design motivation of this paper is the
elimination of the gearbox device [15]–[17]. While, for large
wind turbines, a good gearbox system can be employed, any
highly efficient gearbox is not usually employed in small wind
turbines, for economics reasons, and in such applications, the
gearbox expects higher losses than indicated in the literature,
because those small turbines must have a very competitive
low price.
V. FSM D ESIGN
The design of an electrical machine is often considered
from a given point defined by the maximum torque. In a first
approximation, one can consider that the external volume (or
the weight) of the machine is influenced by the maximum
torque for the considered application, which could be obtained
by the wind turbine. Equation (3) shows that the Pturbine and
their power coefficient Cp (function of the tip-speed ratio λ)
can be considered. The tip-speed ratio λ is defined as
λ=
(8)
(9)
and the relationship of the average cubic wind speed with the
average speed becomes
3
2vavg
3√
6
3
√
(v )avg =
π
= (vavg )3 ∼
= 1.91(vavg )3 . (10)
4
π
π
Ωr
v
(11)
where Ω is the generator rotational speed and r is the radius of
the blade.
Maximizing the Cp coefficient depends on the operating conditions. Therefore, in order to obtain a turbine power rated for
5 kW, the radius of the blade and the rotational speed (Ωb ) can
be derived accordingly. As an example of the design, considering an optimal blade tip-speed ratio of three, a 1-m blade is
obtained for a wind speed of 6 m/s, and a rotational speed of
OJEDA et al.: DESIGN OF A FLUX-SWITCHING ELECTRICAL GENERATOR FOR WIND TURBINE SYSTEMS
Fig. 6.
1811
Power envelope of the turbine; f (v) function.
30 rad/s is considered as the minimum speed. Then, for a 5-kW
generator, the maximum torque (Tmax ) might be computed
based on the rated angular speed, so for the maximum speed
(vmax ), the power is limited to the maximum power (Pmax )
shown in Fig. 6.
The machine chosen for this application is a permanentmagnet FSM [17] here defined as a flux-switching electric
generator (FSEG). This machine has many improvements when
compared to other permanent-magnet machines, such as the
use of a passive rotor and improved heat transfer. It has a
torque density similar to that of classical permanent-magnet
machines [18] and is used for applications such as aircraft
systems [19] and electric vehicles [20]. The FSEG is described
by the numbers of stator teeth (Ns ) and rotor teeth (Nr ). The
electrical frequency (fs ) is determined by
fs =
1
Nr Ω.
2π
Fig. 7. Optimization geometry.
(12)
For a low-speed generator (around 30 rad/s), the number of
rotor teeth has to be important in order to obtain a satisfying
electrical frequency; consequently, only generators with a high
number of rotor teeth have been selected. The structure choice
is given by a compromise between the couple (Ns and Nr ),
torque, torque ripple, and copper losses. The increase of Nr
affects directly the torque and also the iron losses. The torque
ripple is determined by Nr and Ns [7]. This paper shows
two structures preoptimized (i.e., chosen to obtain a good
compromise between electromagnetic torque, torque ripple, and
copper losses): 1) a flux-switching generator with 120 stator
teeth and 100 rotor teeth and 2) a flux-switching generator with
120 stator teeth and 140 rotor teeth. For both machines (FSEG
120/100 and FSEG 120/140), the permanent-magnet thickness
is defined (3 mm), and the magnetic flux density can be selected
from 0.4 T (ferrite), 0.8 T (bonded NdFeB), and 1.2 T (sintered
NdFeB). Then, geometry parameters for all structures are defined in Fig. 7.
αic is the thickness of the iron path, βr is the half thickness
of a rotor tooth, rext is the external radius, and rag is the airgap radius. These parameters are bounded in a limited range
of variations so as to guarantee a structure possible to be
built mechanically. Each structure is optimized by a parametric
procedure where the torque by active length and the torque
ripple are computed with a finite-element (FE) simulation. The
optimization procedure is shown in Fig. 8.
The optimization procedure is based on a random parametric
optimization. Parameters (Lactive , rext , rag , βr , and αic ) are
Fig. 8. Optimization scheme function.
selected according to a stochastic law (bounded by a minimum
and a maximum value), and a FE simulation is achieved.
From the computation of the required torque (which is the
main criterion), machines minimizing mass and losses are
selected. Correlations between parameters and output criteria
are achieved in order to select the three best machines that
satisfied the required torque and minimized the mass and losses.
The three best machines have been compared (one with
ferrite, another with bonded NdFeB, and a third one with sintered NdFeB). The active length Lactive is calculated from the
required torque and from the torque by active length simulated
by 2-D FE analysis. The copper loss (Pc ) is computed by
Pc (vmin ) = σc J 2 Vc (vmin )
(13)
where σc is the copper resistivity, J is the current density (in
this case, equal to 5 A/mm2 ), and Vc is the copper volume
which depends on the minimum wind velocity (vmin ). The
maximal torque is obtained for the minimum wind speed and,
thus, produces the maximal copper loss. For greater values of
rotational speed, copper loss can decrease or remain constant
according to the control strategy. Thus, in this study, the copper
loss is considered constant. The iron loss (Pi ) is computed by
Pi (Bs , fs , vmin , v) = k(Bs , v)Mt (vmin , v)
(14)
1812
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 6, NOVEMBER/DECEMBER 2012
Fig. 9. Copper and iron losses versus wind velocity.
Fig. 11.
Thermal simulation for 1.8 kW of copper losses.
Fig. 10. Three-dimensional view of the FSEG 120/100 with ferrite magnet.
where k is an iron loss coefficient [21] determined by iron flux
density (Bs ) and the operating frequency (which is determined
by the wind velocity) and Mi is the iron mass which depends on
the minimum and actual wind velocity. Copper and iron losses
are shown in Fig. 9. The final geometry with a ferrite magnet is
shown in Fig. 10.
A. Thermal Aspect and Demagnetization Issues
The thermal design is very important for this application
[21] because the generator will be enclosed in a nacelle, and
consequently, the cooling capability is strongly constrained.
The generator will be used in a very long duration cycle (many
hours per day), and the temperature rise must be carefully
limited. For a ferrite-magnet-based generator (the worst case
in terms of copper losses), a thermal simulation using a FE
software is achieved with the maximal copper losses (1.8 kW
corresponding to a vmin equal to 5 m/s), as shown in Fig. 11.
It is possible to observe that the maximal temperature is kept
below the 155◦ C level (corresponding to the insulation class F)
and lower than the Curie points of ferrite (300◦ C) and NdFeB
(310 ◦ C–400 ◦ C). The demagnetization issue is challenging for
designers of permanent-magnet motors [23]. The demagnetization field on the permanent magnet can depolarize the magnetic
moments and, then, cancel the remanent magnetization. In
order to avoid this issue, the demagnetization field is computed
on the permanent magnet by the FE method and is shown
in Fig. 12.
In both cases, the demagnetization field is smaller than
the permanent-magnet coercive field (−250 kA/m for the ferrite magnet and −900 kA/m for sintered NdFeB). Moreover,
the permanent magnets are located on the stator; therefore,
Fig. 12. Demagnetization field on a permanent magnet for a ferrite and a
sintered NdFeB configuration.
improved characteristics against demagnetization are possible [24].
B. Permanent Magnets and Power Range
In order to deal with the interaction between the permanent
magnet and the power range in order to maximize the captured
energy, the operating wind velocity range (vmin and vmax )
optimized for a function (Fig. 6) that uses a weighted mean
energy (W ) is considered for the three designed machines, as
proposed by
20m
s
f (v)η(v)ω2 (v)dv
W = vlim 20m
(15)
s
vlim ω2 (v)dv
where ω2 is the Rayleigh distribution of the wind with two
parameters and η is the generator efficiency calculated by
η(v) =
P (v) − Pc (v) − Pi (v)
.
P (v)
(16)
The efficiency was computed in Fig. 13 for a ferrite magnet
configuration and in Fig. 14 for a sintered NdFeB magnet
solution.
OJEDA et al.: DESIGN OF A FLUX-SWITCHING ELECTRICAL GENERATOR FOR WIND TURBINE SYSTEMS
1813
Fig. 13. Efficiency as function of vmin and vmax for ferrite magnet (maximum value of 77%).
Fig. 15. Weighted mean energy for ferrite magnet (maximum value of
1700 Wh for vmin = 6 m/s and vmax = 12 m/s).
Fig. 14. Efficiency as function of vmin and vmax for sintered NdFeB magnet
(maximum value of 83%).
The efficiency is limited by the copper loss for low wind
velocity (vmin ) and constrained by the iron loss for high wind
velocity (vmax ). Thus, the product f (v)η(v)ω2 (v) determines
the energy efficiency of the system. Consequently, the system
has to be designed for f (v) for the turbine and for η(v) for
the generator for a typical wind statistical distribution [for a
geographic region ω2 (v)]. Figs. 15–17 show the weighted mean
energy represented for the three machines considering an inland
Rayleigh wind distribution typical in France.
The energy capacity is limited by both copper and iron losses,
and the wind distribution emphasizes wind speed range where
it is maximized.
As shown in these figures, there is an optimal range for
maximum captured energy. Depending on the selection of the
permanent magnet and on the typical wind distribution, the
optimal range is different. One can notice that the machine
with ferrite magnets has to be oversized in order to satisfy the
specifications, and consequently, the losses are more significant
than for a NdFeB magnet selection. Thus, the optimal range
for the ferrite magnet is smaller and the captured energy is
decreased, and a comparison has been made, as shown in both
Figs. 18 and 19.
Fig. 16. Weighted mean energy for bonded NdFeB magnet (maximum value
of 2100 Wh for vmin = 5 m/s and vmax = 12 m/s).
Fig. 17. Weighted mean energy for sintered NdFeB magnet (maximum value
of 2300 Wh for vmin = 5 m/s and vmax = 13 m/s).
1814
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 6, NOVEMBER/DECEMBER 2012
TABLE I
C OMPARISON OF M ASS AND E NERGY C OST
FOR THE D ESIGNED M ACHINES
Fig. 18. Captured energy for a large operating range.
TABLE II
C OMPARISON OF C HOSEN O PTIMAL D ESIGN V ERSUS
A C ERTAIN M ACHINE IN THE L ITERATURE
Fig. 19. Captured energy for a small operating range.
Both Figs. 18 and 19 show the energy captured area (power
extracted by the turbine). This area is bounded by the maximum
power from the turbine (Pmax ) and the system losses; such area
can be calculated by
vdesign
Eca (vmin , vmax ) =
(P (v)−Pc (v)−Pi (v)) dv
vmin
vdesign
Eca (vmin , vmax ) =
(P (v)−Pc (v)−Pi (v)) dv. (17)
vmin
The turbine power [P (v)] is given by Fig. 6. Copper and
iron losses are given by (13) and (14), respectively. The wind
velocity and vdesign define where P (v) equals the sum of
losses (no captured energy). Fig. 18 shows that a large power
range (corresponding to a small value of vmin ) leads to an
overdesigned generator with correspondent higher losses. Thus,
the energy captured is positive only for few values of wind
speed. Despite a larger power range, the energy captured area
is smaller. In Fig. 19, the power range is small and vmin is high,
and for this configuration, the energy captured area is higher
than that of the previous case. Consequently, maximizing the
power range area is not recommended for a wind turbine generator design. Therefore, the methodology proposed in this paper
shows the required compromise between acceptable losses and
the turbine power range. Table I shows the optimization results
for the designed three machines.
The level of flux density is lower for a ferrite magnet machine. Therefore, the total mass is increased so as to obtain
the same performance. However, in this case, the captured
energy is less than those in the other two cases. The best
set of characteristics in terms of mass, captured energy, and,
consequently, the mass–energy ratio is reached for the sintered
NdFeB (1.2 T) magnet configuration due to the high efficiency
of the permanent magnet. However, when cost–energy ratio
is computed using cost market data (as for the year 2012)
for all components (iron, copper, and permanent magnet), the
cost–energy ratio criteria show a different scenario where the
ferrite magnet configuration shows the smallest ratio. Although
mass is the highest one, the price of the ferrite magnets is
about 20 times smaller than the price of NdFeB magnets,
showing that the total cost is smaller than that of the NdFeBmagnet-based design. Neodymium is a strategic material, and
its price will most probably increase in the future. Thus, it is
expected that the cost–energy ratio will further increase in the
near future. Although the mass–energy ratio is the smallest for
ferrite magnets, it is indeed still a good alternative when compared to rare-earth permanent-magnet options. Table II shows
a comparison of the generator with bonded NdFeB magnet
and a high-torque generator taken in the literature for similar
application [23]. The machine selected in the literature has the
same electromagnetic characteristics (torque and speed) and the
same type of permanent magnet with the machine designed in
this paper and can be used for a fair comparison.
The flux-switching electrical generator designed in this paper, which considered the operating range for a typical wind
distribution, operates in low speed compatible with the linear
wind velocity and turbine radius (eliminating the need of
gearbox), and is constructed out of low-cost standard ferrite
magnets, is indeed a promising solution for wind turbine applications.
OJEDA et al.: DESIGN OF A FLUX-SWITCHING ELECTRICAL GENERATOR FOR WIND TURBINE SYSTEMS
1815
R EFERENCES
Fig. 20. Typical power electronic topology.
C. Control Strategy
For the control point of view, the FSEG is equivalent to an interior permanent-magnet machine. Therefore, previous control
methodologies used for this kind of machine [17], [20] can be
used in the drive system. Fig. 20 shows a diagram for a power
electronic topology [26] of the FSEG for a grid-connected wind
turbine, as well as stand-alone dc output without connection
to the grid [27]. The authors expect to report in the future
the required power-electronic-based control for flux-switching
electrical generators.
A flux-switching electrical generator can be connected to a
small-scale wind turbine with fixed attack angles of the blades.
They are connected to either the distribution grid or a dc load
through power electronic interfaces. The control must be based
on the load flow which acts on the turbine rotation. As the
rotor speed changes according to the wind intensity, the speed
control of the turbine has to command low speed at low winds
and high speed at high winds, so as to follow the maximum
power operating point. The maximum power tracking requires
a hill-climbing-type controller or maybe a fuzzy-logic-based
controller for commanding the speed of the generator [3], [4],
but the discussion of these controllers is out of the scope
of this paper.
VI. C ONCLUSION AND PATH F ORWARD
This paper has presented the detailed design and parametric
optimization of a flux-switching electrical machine, customized
for a wind turbine operating range typical of low-wind-velocity
sites. The statistics of wind resources were taken into consideration in order to define the power envelope for the turbine
considering copper plus iron losses for three machine designs.
Three machines were detailed for configurations with a ferrite
magnet and two NdFeB magnets in order to extract the energy
potential of each configuration. Although the NdFeB magnet
has a greater index energy/mass, regarding the market evolution and the difficulty to obtain rare-earth magnet, the ferrite
configuration is a serious alternative for the future. This paper
has shown that such a direct-drive flux-switching electrical
generator can be very competitive and possibly the best solution
for small-scale wind turbines, typically used in rural systems,
small farms, and villages.
[1] B. J. Chalmers and E. Spooner, “An axial-flux permanent-magnet generator for a gearless wind energy system,” IEEE Trans. Energy Convers.,
vol. 14, no. 2, pp. 251–257, Jun. 1999.
[2] H. Polinder, F. F. A. van der Pijl, G.-J. de Vilder, and P. J. Tavner,
“Comparison of direct-drive and geared generator concepts for wind
turbines,” IEEE Trans. Energy Convers., vol. 21, no. 3, pp. 725–733,
Sep. 2006.
[3] M. G. Simoes, B. K. Bose, and R. J. Spiegel, “Fuzzy logic based intelligent control of a variable speed cage machine wind generation system,”
IEEE Trans. Power Electron., vol. 12, no. 1, pp. 87–95, Jan. 1997.
[4] M. G. Simoes, B. K. Bose, and R. J. Spiegel, “Design and performance
evaluation of a fuzzy-logic-based variable-speed wind generation system,” IEEE Trans. Ind. Appl., vol. 33, no. 4, pp. 956–965, Jul./Aug. 1997.
[5] U.S. Dept. Energy, Wind Powering America, Washington, DC. [Online].
Available: http://www.windpoweringamerica.gov/
[6] E. Morgan, M. Lackner, R. Vogel, and L. Baise, “Probability distributions
for offshore wind speeds,” Energy Convers. Manage., vol. 52, no. 1,
pp. 15–26, Jan. 2011.
[7] P. Nielsen, “Offshore wind energy projects, feasibility study guidelines,”
in Proc. SEAWIND-Altener Proj.-Feasib. Study Guidelines (EMD), 2003,
pp. 1–39.
[8] B. Multon, O. Gergaud, H. B. Ahmed, X. Roboam, S. Astier, B. Dakyo,
and C. Nichita, “Etat de l’art dans les aerogenerateurs electriques,” in
Proc. Extrait du Rapport de Synthese ECRIN, 2002, pp. 1–52.
[9] E. Muljadi and C. Butterfield, “Pitch-controlled variable-speed wind turbine generation,” IEEE Trans. Ind. Appl., vol. 37, no. 1, pp. 240–246,
Jan./Feb. 2001.
[10] M. Pinilla and S. Martinez, “Optimal design of permanent-magnet directdrive generator for wind energy considering the cost uncertainty in raw
materials,” Renewable Energy, vol. 41, no. 1, pp. 267–276, May 2012.
[11] P. Tavnet, G. Van Bussel, and F. Spinato, “Machine and converter reliabilities in wind turbines,” in Proc. 3rd IET Int. Conf. Power Electron., Mach.
Drives, Mar. 2006, pp. 127–130.
[12] B. Palle, M. G. Simoes, and F. A. Farret, “Dynamic simulation and
analysis of parallel self-excited induction generators for islanded wind
farm systems,” IEEE Trans. Ind. Appl., vol. 41, no. 4, pp. 1099–1106,
Jul./Aug. 2005.
[13] J. R. Cotrell, “A preliminary evaluation of a multiple generator drivetrain configuration for wind turbines,” in Proc. 21st ASME Wind Energy
Symp., 2001, pp. 345–352.
[14] W. Qiao, X. Yang, and X. Gong, “Wind speed and rotor position sensorless
control for direct-drive PMG wind turbines,” IEEE Trans. Ind. Appl.,
vol. 48, no. 1, pp. 3–11, Jan./Feb. 2012.
[15] A. Grauers, “Design of direct-driven permanent-magnet generators
for wind turbines,” Ph.D. dissertation, Chalmers Univ. of Technol.,
Gothenburg, Sweden, 1996.
[16] M. Dubois, H. Polinder, and J. Ferreira, “Comparison of generator topologies for direct-drive wind turbines,” in Proc. IEEE Nordic Workshop
Power Ind. Electron., 2000, pp. 22–26.
[17] E. Hoang, H. B. Ahmed, and J. Lucidarme, “Switching flux permanent
magnet polyphased synchronous machines,” in Proc. EPE Conf., 1997,
pp. 1–6.
[18] W. Hua, Z. Zhu, M. Cheng, Y. Pang, and D. Howe, “Comparison of fluxswitching and doubly-salient permanent magnet brushless machines,” in
Proc. ICEMS, 2005, vol. 1, pp. 165–170.
[19] Y. Amara, E. Hoang, M. Gabsi, M. Lecrivain, and S. Allano, “Design and
comparison of different flux-switch synchronous machines for an aircraft
oil breather application,” Eur. Trans. Elect. Power, vol. 15, no. 6, pp. 497–
511, Nov./Dec. 2005.
[20] A. Zulu, B. C. Mecrow, and M. Armstrong, “A wound-field threephase flux-switching synchronous motor with all excitation sources on
the stator,” IEEE Trans. Ind. Appl., vol. 46, no. 6, pp. 2363–2371,
Nov./Dec. 2010.
[21] Cogent Electrical Steel, Surahammar, Sweden. [Online]. Available: http://
www.sura.se
[22] G. Li, J. Ojeda, E. Hoang, M. Lecrivain, and M. Gabsi, “Comparative
studies between classical and mutually coupled switched reluctance motors using thermal-electromagnetic analysis for driving cycles,” IEEE
Trans. Magn., vol. 47, no. 4, pp. 839–847, Apr. 2011.
[23] S. Ruoho, E. Dlala, and A. Arkkio, “Comparison of demagnetization
models for finite-element analysis of permanent-magnet synchronous machines,” IEEE Trans. Magn., vol. 43, no. 11, pp. 3964–3968, Nov. 2007.
[24] M. Cheng, K. T. Chau, and C. C. Chan, “Design and analysis of a new
doubly salient permanent magnet motor,” IEEE Trans. Magn., vol. 37,
no. 4, pp. 3012–3020, Jul. 2001.
1816
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 48, NO. 6, NOVEMBER/DECEMBER 2012
[25] Alxion Automatique Productique, High Torque Motors, Wind Turbine
Generator, Colombes, France. [Online]. Available: http://www.alxion.
com/
[26] M. Rolak, R. Kot, M. Malinowski, Z. Goryca, and J. T. Szuster, “Design
of small wind turbine with maximum power point tracking algorithm,” in
Proc. IEEE Int. Symp. Ind. Electron., 2011, pp. 1023–1028.
[27] G. Joos and J. Belanger, “Real-time simulation of a wind turbine generator coupled with a battery supercapacitor energy storage system,” IEEE
Trans. Ind. Electron., vol. 57, no. 4, pp. 1137–1145, Apr. 2010.
Javier Ojeda was born in Buenos Aires, Argentina,
in 1980. He received the Ph.D. degree in electrical
engineering from the Ecole Normale Supérieure de
Cachan (ENS Cachan), Cachan, France, in 2009.
In 2010, he was a Postdoctoral Fellow with
Tsinghua University, Beijing, China. Since
September 2010, he has been an Assistant Professor
with the Systèmes et Applications des Technologies
de l’Information et de l’Energie, ENS Cachan–
University Paris Sud 11–Centre National de la
Recherche Scientifique, UniverSud Paris, Cachan.
His main research topics are active damping of switched reluctance machine
motors for high-speed applications, machine noise modeling, and fault
diagnosis in electrical machines for vehicular applications.
Marcelo Godoy Simões received the B.S. and M.S.
degrees from the University of São Paulo, São Paulo,
Brazil, in 1985 and 1990, respectively, the Ph.D.
degree from The University of Tennessee, Knoxville,
in 1995, and the D.Sc. degree (Livre-Docência) from
the University of São Paulo in 1998.
He is currently an Associate Professor with the
Department of Electrical Engineering and Computer
Science, Colorado School of Mines (CSM), Golden,
where he has been establishing research and education activities in the development of intelligent
control for high-power electronic applications in renewable and distributed
energy systems and is currently the Director of the Center for the Advanced
Control of Energy and Power Systems. Since joining CSM, he has been
involved in activities related to the control and management of smart grid
applications.
Dr. Simões was a recipient of the National Science Foundation CAREER
award “Intelligent Based Performance Enhancement Control of Micropower
Energy Systems” in 2002. He was the Chair for the IEEE Industry Applications
Society Industrial Automation and Control Committee and the Vice-Chair for
the IEEE Industrial Electronics Society Technical Committee on Smart Grids.
Guangjin Li was born in Xiaogan, China, in 1984.
He received the Ph.D. degree in electrical engineering from the Ecole Normale Supérieure de Cachan
(ENS Cachan), Cachan, France, in 2011.
He was a Research Assistant with the Systèmes
et Applications des Technologies de l’Information
et de l’Energie, ENS Cachan–University Paris
Sud 11–Centre National de la Recherche Scientifique, UniverSud Paris. He is currently a Research
Associate with the Electrical Machines and Drives
Group, The University of Sheffield, Sheffield, U.K.
His main research interests include the design and the thermal and faultbased analysis of switched reluctance machines and other permanent-magnet
machines.
Mohamed Gabsi received the Ph.D. degree in electrical engineering from the University of Paris VI,
Paris, France, in 1987, and the Habilitation À Diriger
des Recherches from the University of Paris XI,
Orsay, France, in 1999.
Since 1990, he has been with the Electrical Machine Team, Systèmes d”Energies pour le Transport
et l’Environnement, Systèmes et Applications des
Technologies de l’Information et de l’Energie, École
Normale Supérieure de Cachan–University Paris
Sud 11–Centre National de la Recherche Scientifique, UniverSud Paris, Cachan, France, where he is currently a Full Professor
and the Director of the Electrical Engineering Department. His research interests include switched reluctance machines, vibrations and acoustic noise, and
permanent-magnet machines.