SCI Volume 28 Issue 6 Pages 3363-3378
SCI Volume 28 Issue 6 Pages 3363-3378
SCI Volume 28 Issue 6 Pages 3363-3378
KEYWORDS Abstract. This study proposes a special topology of transvers ux permanent magnet
Cogging torque; generator, which bene ts from low Permanent Magnet (PM) costs due to use of cheap
Direct drive wind Ferrite PMs. In this structure, only one PM per phase is used to ensure easy manufacturing.
turbine; Despite the above-mentioned advantages of the topology, it is subject to some disadvantages
Ferrite permanent such as unbalanced voltage, high demagnetization risk of the PM, and high cogging torque.
magnet; Aiming to solve these problems and maintaining the advantages of the topology, a new
Magnetic equivalent structure is proposed. The stator of this structure contains two series connected coils that
circuit; eliminate all the even-order harmonics and balance the voltage waveform. This trick reduces
Permanent magnet the armature reaction as well as demagnetization of the PMs signi cantly. Moreover, the
demagnetization; rotor teeth of the proposed structure are skewed, causing a signi cant reduction in cogging
Transverse ux torque. The output results of the proposed structure are compared with those of the original
permanent magnet one in terms of harmonic components, cogging torque pro le, and PM demagnetisation.
machine.
© 2021 Sharif University of Technology. All rights reserved.
Figure 7. Magnetic eld crossing the conductor A1 in di erent rotor positions for basic topology: (a) Initial position and
(b) one pole pitch movement.
3368 A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378
Table 1. Speci cation of the Single Permanent Magnet-Transverse Flux Machines (SPM-TFMs).
Parameter Symbol Value Unit
Rated power Pout 1.5 kW
Rated speed nm 500 rpm
No. of phases m 3 |
No. of poles P 36 |
No. of rotor teeth Nrotor 18 |
Outer diameter of the rotor Dor 195 mm
Inner diameter of the rotor Dir 80 mm
Length of one stator core L 80 mm
Number of armature turns per phase N 86 |
PM coercive force Hc {285 kA/m
PM remanent ux density Br 0.38 T
Here, B1 is the main component of B(x) . The The total ux linkage of the basic topology could
value of B1 for both topologies is as follows: be calculated as follows:
8 Z =2
<B1
> = 4Bmax Ws
2 ( Ws ) cos( 2 ); Ws 6= NP NP Lr Bm
'= L B(x) dx = : (16)
(13) 2 r =2
>
:
B1 = 2Bmax ; Ws = By substituting Eqs. (15) and (16) into the gen-
By substituting Eq. (13) into Eq. (12), the average eral equation of induced voltage (V = Nd'=dt), the
value of the electromagnetic torque for both topologies RMS value of the Back-EMF can be calculated:
can be derived as follows (in the case of Ws = ): p
V = 2fe NP Lr Bm ; (17)
mP Dg Im Lr Bmax
Tave = cos('): (14) where fe is the electrical frequency of armature. For
the proposed topology with N=2 turn per phase in
3.2. Back-EMF equation each coil, the induced voltage per coil is half of the
To derive the Back-EMF equation of the TFPMs, voltage in basic topology. Therefore, through the
the e ect of stator segments must be included in the series connection of the coils, the induced voltage (main
voltage equation such as the proposed equation in component) is equal for both topologies.
[1,27]. In this paper, the PM ux density distribution By using the proposed design equation, two di er-
in the air gap is modeled by the proposed graph given ent topologies of SPM-TFMs are designed for a small-
in Figure 9. The ux density distribution of PM in scale direct drive wind turbine. As stated earlier, the
the air gap could be written in furrier series as in selected power and speed for the case study turbine are
Eq. (15) where the rst component of the magnetic considered as 1.5 kW and 500 rpm, respectively. The
ux density is calculated in Eq. (13) by replacing Bmax speci cations of the designed SPM-TFMs are listed in
with Bmax Bmin . Table 1.
x In order to make a comparison between the
B(x) = Bm cos : (15) topologies, all the speci cations such as rotor dimen-
sions, PM type, machine length, and machine outer
diameter are supposed to be equal in the case of
both topologies. To validate the design equations, the
average torque and the main component of the Back-
EMF are derived using Finite Element Analysis (FEA)
and analytical approaches. Figure 10 shows the torque
waveform of the SPM-TFM resulting from FEA versus
rotor position based on the dimensions given in Table
1. The average value of the FEA result and the torque
value of Eq. (14) are also given in Figure 10. Upon
comparing the average torque value resulting from FEA
Figure 9. The model of Permanent Magnet (PM) ux with the analytical equation of Eq. (14), an acceptable
density distribution in air gap. coincidence is produced that has only 6% divergence.
A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378 3369
4. 3D Non-linear MEC
Magnetic Equivalent Circuit (MEC) method is an ana-
lytical method for modeling electromagnetic machines
[31,32]. The MEC has been successfully developed for
di erent types of electrical machines in the literature
[33{39]. In TFPMs, due to the 3-Dimensional (3D) na-
ture of the ux path, 3D MEC is required. Compared
with the FEA, the MEC is a faster method for reducing
the computational time of the simulations. However, Figure 12. Cross sectional view of the proposed Single
it is usually less accurate than the FEA. Hence, to Permanent Magnet-Transverse Flux Machine (SPM-TFM)
increase the accuracy of the MEC, the iron saturation permeance network (excluding variable air gap
e ect is modeled in this study. permeances): (a) Front view and (b) side view.
3370 A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378
Figure 13. Parametric dimensions of the proposed Single Figure 14. Model of the variable air gap permeances.
Permanent Magnet-Transverse Flux Machine
(SPM-TFM): (a) Front view and (b) side view. air gap permeances are the most important parts of the
permeance network as energy conversion takes place
where t1 is the permeability of the iron which is a in the air gap region. Therefore, the more precise
variable depending on the working point in the iron B- the variable air gap permeance modeling, the more
H curve; c is the permeability of iron for the iron parts accurate the MEC results. The method of modeling
whose saturation e ect is not modeled (rotor yoke); 0 air gap permeances is based on the proposed method
is the permeability of air; and Dor and Dir are the of [31] in which the air gap permeance between one
outer and inner diameters of the rotor, respectively. rotor tooth and one stator tooth during the rotation is
The rotor leakage permeances shown in Fig- considered, as given in Figure 14. The corresponding
ure 13(b) are calculated through Eqs. (22){(24): parameters of Figure 14 are given in Eqs. (28) and (29):
C 0 Lr min(; Ts )
Y
GLr1 = 0 s ; (22) Gmax = f ; (28)
L 2Lr g
(D Ys )
2 j Ts j :
GLr2 = 0 or
p
Ln L L 2Lr ; (23) a=
p
; b=
Dor
(29)
m
Based on Figure 14, the permeance between two certain
D
GLr3 = 0 ir
p
Ln LL : (24) nodes of stator and rotor is equal to Gmax when the
angle between them () is within b and b. In this
m
condition, the maximum overlap between stator and
Each segment of the ring-shaped PM is modeled by a rotor tooth occurs, leading to maximum permeance
current source shunted by a leakage permeance as in between the teeth. Moreover, the permeance follows
Eqs. (25) and (26): a sinusoidal function when jbj < jj < jaj. Beyond
those limits, the permeance value is zero in Eq. (30) as
(Dor 2 Ys )2 (Dor 2 Ys 2Yr =3)2 follows:
'1 = Br ;
4p (25)
8
>>Gmax ;
> b>> b
>
>
>
>
GM 1 = >
>
> 1+cos( a bb )
<Gmax
> 2 ; a>>b
h i
P M (Dor 2 Ys )2 (Dor 2 Ys 2Yr =3)2 G= (30)
>
; >
> 1+cos( a+bb )
4p Lm
>
>
>Gmax 2 ; b>> a
(26) >
>
>
>
>
:
where Br and P M are the PM remanent ux density 0; elsewhere
and PM permeability, respectively.
The Magneto Motive Force (MMF) of each wind- 4.3. Iron saturation e ect
ing is modeled by a voltage source as follows: As mentioned earlier, to enhance the accuracy of the
Fu = Nu Iu ; (27) MEC method, modeling the iron saturation e ect
is required. This e ect is considered for the rotor
where Nu is the upper coil's number of turns and Iu is tooth permeance and stator permeance whose magnetic
the armature current conducted by the upper coil. ux density is of high value. The non-linear MEC
algorithm of the TFPM is presented in Figure 15. At
4.2. Variable air gap permeance the beginning, the machine physical parameters are
In MEC modeling for electrical machines, the variable inserted. Next, a function (B = f(H ) ) is tted on
A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378 3371
Figure 15. Non-linear Magnetic Equivalent Circuit (MEC) algorithm of the proposed Single Permanent
Magnet-Transverse Flux Machine (SPM-TFM).
the iron B-H curve. Therefore, the function of iron 4.4. MEC results
permeability ( = f(H ) =H ) is calculated. Next, an The total ux linkage of the TFPM in the no-load
initial working point for the iron part is assumed (H0 , condition is extracted by both FEA and MEC methods
B0 , 0 ). Afterwards, the MEC matrixes are solved and shown in Figure 16. Moreover, the net output
and a new working point is determined for the iron voltage of the machine in the no-load condition is
part (H1 , B1 , 1 ). If the new ux density (B1 ) is close calculated by both methods and presented in Figure 17.
enough to the assumed value of ux density (B0 ), then As can be seen, the maximum ux linkage resulting
the results of this step are saved and the rotor position from MEC is about 0.006 Wb. Turn higher than
is updated and the next step begins. Otherwise, the the FEA result. Consequently, the MEC method is
initial working point is wrong; instead, a new working employed to calculate the peak of induced voltage to
point is assumed. This process is repeated until the be about 10 V higher than the FEA (Figure 17).
initial and nal values of ux density (B0 and B1 ) The three-phase cogging torque is also calculated by
get close enough. The parameter B determines how both methods and presented in Figure 18. Unlike the
much the two ux densities should get close. The ux linkage and induced voltage, the MEC method is
parameter H determines the change step of the iron subject to noticeable errors (12%) in cogging torque
working point. The smaller the above values (B and calculation.
H ), the more accurate the saturation modeling, yet at According to no-load analysis, it is deduced that
the cost of longer processing time. Consequently, these the MEC method has acceptable accuracy in calcu-
two values must be chosen wisely. lating voltage and ux linkage, although it may not
3372 A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378
Figure 16. Comparison of the net ux linkage in Figure 19. Comparison of full load voltages of the
proposed Single Permanent Magnet-Transverse Flux
Magnetic Equivalent Circuit (MEC) method and the Machine (SPM-TFM) resulting from Magnetic Equivalent
Finite Element Analysis (FEA) (no-load condition). Circuit (MEC) method and the Finite Element Analysis
(FEA).
5. Results and comparisons
To achieve a better insight into the magnetic character-
istic of the topologies, simulations have been conducted
based on FEA. The magnetic ux density distribution
in one pole pair of the basic topology is depicted in
Figure 20(a). As can be seen, when the rotor teeth are
aligned with the U-shaped cores, the I-shaped cores are
aligned with the rotor slots. Thus, the main portion of
the ux passes through the U-shaped core and results
Figure 17. Comparison of the net Back-EMF of the in maximum ux linkage. Figure 20(b) represents
proposed Single Permanent Magnet-Transverse Flux the magnetic ux density distribution in the proposed
Machine (SPM-TFM) in Magnetic Equivalent Circuit
(MEC) method and the Finite Element Analysis (FEA). topology. Similarly, in the basic position, most of the
PM uxes are linked to the upper coil, while the lower
coil has minimum ux linkage.
According to Figure 9 and Eq. (15), the leakage
magnetic ux density (Bmin 0 ) under the rotor slot
will cause a reduction in the machine rating. The
leakage ux can be reduced by increasing the spacing
between stator parts which would result in increasing
the machine diameter. Therefore, a compromise must
be made between minimizing Bmin 0 and the machine
diameter.
In order to compare the SPM-TFMs performance,
the output voltage in both no-load and full-load
Figure 18. Cogging torque of the proposed Single conditions is extruded. Moreover, demagnetization
Permanent Magnet-Transverse Flux Machine (SPM-TFM) characteristics of the machines are analyzed.
resulting from Magnetic Equivalent Circuit (MEC)
method and the Finite Element Analysis (FEA).
almost the same behavior until 150% of the nominal in PM shape is not necessary, anymore. Moreover,
current. At higher currents, the proposed topology skewing the rotor teeth is a simple and cost-e ective
shows better performance. The di erence is mainly procedure in comparison with skewing the PMs.
due to three reasons: To do so, the rotor teeth in the proposed ma-
chine are skewed at di erent angles and the pro le of
First, the armature ux density distribution in the the electromagnetic torque is extracted and shown in
air gap for the proposed topology is symmetric, Figure 28. As can be seen, changing the skew angle
while in the case of the basic topology, it is not has a considerable e ect on the value of average torque
symmetric (Figure 8); and torque ripple. Since the result of changing skew
Second, the stator leakage permeances are di erent angle does not have a linear e ect on machine torque
in two topologies; characteristic, a sensitivity analysis of the skew angle
Third, the current waveform of the basic topology on the average torque and torque ripple is conducted
contains even order harmonics. In other words, at and the result is presented in Figure 29. Regarding the
the equal rms current, the fundamental component mechanical constraints in the machine design, skewing
of the current waveform is higher than that of the at 7 degrees will be operational. By increasing the
basic one, which results in higher torque production. skew angle from zero to 7 degrees, the average torque
keeps dropping almost linearly and achieves a 14.4%
According to three main advantages of the pro- drop in the nal degree. E ect of increasing the skew
posed topology (cancelling all even order harmon- angle on the torque ripple is not linear. As can be
ics, better PM demagnetization behavior, and better seen from Figure 29, torque ripple goes through an
torque production), the proposed topology can be the oscillation from 4 to 7 degrees. It is a fact that the
recommended candidate for a small-scale direct drive minimum torque ripple belongs to 7 degrees of the
wind turbine. Since the electromagnetic torque of the
generator plays an important role in the start-up and
vibration level of a wind turbine, reducing the torque
ripple of the structure is desired. To analyze the torque
pro le of the proposed machine, the pro le of the
cogging torque versus the rotor position is extracted
and presented in Figure 27. As can be seen, both
of topologies have a noticeable cogging torque (18%)
which will a ect the generator output characteristic.
In order to minimize the torque ripple of the
structure and ensure an optimized design, a method
based on skewing the rotor teeth is developed. The
skew of the rotor poles in typical topologies [17,18]
requires special shaping of the PMs which is not a cost-
e ective solution. In case of the SPM-TFM topology,
the rotor pole skew could be applied by skewing only Figure 28. Torque waveform of the Single Permanent
the rotor teeth. Consequently, any special changing Magnet-Transverse Flux Machine (SPM-TFM) in di erent
skew angles.
Figure 27. Cogging torque comparison for Single Figure 29. Torque characteristics of the Single
Permanent Magnet-Transverse Flux Machine (SPM-TFM) Permanent Magnet-Transverse Flux Machine (SPM-TFM)
topologies. in di erent skew angles.
3376 A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378
16. Nasiri-Zarandi, R., Ghaheri, A., and Abbaszadeh, K. 28. Anglada, J.R. and Sharkh, S.M. \Analytical calcula-
\Cogging torque reduction in U-core TFPM genera- tion of the torque produced by transverse ux ma-
tor using di erent halbach-array structures", In 2018 chines", IET Electric Power Applications, 11(7), pp.
International Symposium on Power Electronics, Elec- 1298{1305 (2017).
trical Drives, Automation and Motion (SPEEDAM),
Amal Coast, Italy, pp. 1153{1158 (2018). 29. Anglada, J.R. and Sharkh, S.M. \Analysis of trans-
verse ux machines using a virtual mutual inductance
17. Green, J. \Development of high-performance generator approach", IEEE Transactions on Energy Conversion,
for wind turbines", Final Report, Stage II, National 33(2), pp. 465{472 (2017).
Wind Technology Center, Boulder, USA (2009). 30. Alam, F.R. and Abbaszadeh, K. \Magnetic eld
18. Gieras, J.F., Hamilton Sundstrand Corp. \Transverse analysis in eccentric surface-mounted permanent-
ux machine", U.S. Patent 7,830,057 (2010). magnet motors using an improved conformal mapping
method", IEEE Transactions on Energy Conversion,
19. Kastinger, G. \August. Design of a novel transverse 31(1), pp. 333{344 (2015).
ux machine", In Proc. ICEM, Brugge, Belgium
(2002). 31. Ostovic, V., Dynamics of Saturated Electric Machines,
pp. 112{368, Springer Science & Business Media,
20. Liu, C., Lei, G., Ma, B., Wang, Y., Guo, Y., and Zhu, Berlin, Germany (2012).
J. \Development of a new low-cost 3-D ux transverse
ux FSPMM with soft magnetic composite cores and 32. Sudho , S.D., Power Magnetic Devices: A Multi-
ferrite magnets", IEEE Transactions on Magnetics, Objective Design Approach, John Wiley & Sons, New
53(11), pp. 1{5 (2017). Jersey, USA, pp. 264{355 (2014).
33. Ojaghlu, P., Vahedi, A., and Totoonchian, F. \Mag-
21. Husain, T., Hasan, I., Sozer, Y., Husain, I., and netic equivalent circuit modelling of ring winding
Muljadi, E. \Design of a modular E-core ux concen- axial ux machine", IET Electric Power Applications,
trating transverse ux machine", IEEE Transactions 12(3), pp. 293{300 (2017).
on Industry Applications, 54(3), pp. 2115{2128 (2018).
34. Tong, W., Wang, S., Dai, S., Wu, S., and Tang,
22. Bastawade, P., Chaudhari, B.N., Ugale, R.T., and R. \A quasi-three-dimensional magnetic equivalent
Pramanik, A. \Analytical and FEA based analysis circuit model of a double-sided axial ux permanent
of homopolar poly-phase transverse ux machine", magnet machine considering local saturation", IEEE
In 2016 IEEE International Conference on Power Transactions on Energy Conversion, 33(4), pp. 2163{
Electronics, Drives and Energy Systems (PEDES), 2173 (2018).
Trivandrum, India, pp. 1{6 (2016).
35. Horvath, D.C., Pekarek, S.D., and Sudho , S.D. \A
23. Liu, G., Shao, M., Zhao, W., Chen, Q., and Mo, scaled mesh/nodal formulation of magnetic equivalent
L. \Cost reduction of a new fault-tolerant Halbach circuits with motion", IEEE Transactions on Energy
permanent magnet machine using ferrite magnet", Conversion, 34(1), pp. 58{69 (2018).
IEEE Transactions on Magnetics, 50(11), pp. 1{4
(2014). 36. Mignot, R.B., Espanet, C., Chamagne, D., and Martin,
T. \Modeling of an axial ux pm motor using a 3D
24. Zhao, X. and Niu, S. \Design of a novel consequent- magnetic equivalent circuit", In 2014 IEEE Vehicle
pole transverse- ux machine with improved permanent Power and Propulsion Conference (VPPC), Coimbra,
magnet utilization", IEEE Transactions on Magnetics, Portugal, pp. 1{9 (2014).
53(11), pp. 1{5 (2017).
37. Johnson, M., Gardner, M.C., and Toliyat, H.A. \A
25. Ahmed, A., Wan, Z., and Husain, I. \Permanent mag- parameterized linear magnetic equivalent circuit for
net transverse ux machine with overlapping stator analysis and design of radial ux magnetic gears-Part
poles", In 2015 IEEE Energy Conversion Congress and I: Implementation", IEEE Transactions on Energy
Exposition (ECCE), Montreal, Canada, pp. 791{798 Conversion, 33(2), pp. 784{791 (2017).
(2015).
38. Hasan, I., Husain, T., Sozer, Y., Husain, I., and
26. Oskarsd \A general description and com-
ottir, M.O. Muljadi, E. \Analytical modeling of a double-sided
parison of horizontal axis wind turbines and vertical ux concentrating E-core transverse ux machine with
axis wind turbines", Doctoral Dissertation, School pole windings", In 2017 IEEE International Electric
of Engineering and Natural Sciences, University of Machines and Drives Conference (IEMDC), Miami,
Iceland, Iceland, pp. 130{365 (2014). USA, pp. 1{7 (2017).
27. Nasiri-Zarandi, R., Ajamloo, A.M., and Abbaszadeh, 39. Hasan, I., Husain, T., Uddin, M.W., Sozer, Y., Husain,
K. \Proposing the output equations and 3-D MEC I., and Muljadi, E. \Analytical modeling of a novel
modeling for U-Core TFPM generators", In 2018 transverse ux machine for direct drive wind tur-
International Symposium on Power Electronics, Elec- bine applications", In 2015 IEEE Energy Conversion
trical Drives, Automation and Motion (SPEEDAM), Congress and Exposition (ECCE), Montreal, Canada,
Amal Coast, Italy, pp. 292{297 (2018). pp. 2161{2168 (2015).
3378 A. Mohammadi Ajamloo et al./Scientia Iranica, Transactions D: Computer Science & ... 28 (2021) 3363{3378