A Novel Ring Permanent Magnet Flux Reversal

Machine for a Direct-Drive Wind Generator

Anupam Sharma and S. Sashidhar
School of Electrical Sciences
Indian Institute of Technology Goa
Goa – 403 401, India
[email protected]

Abstract—Flux Reversal Machines (FRMs) have simple and

rugged construction with high power density and fault tolerance
capability making it suitable for applications such as roof-top
wind and aircraft generators. Further, in an FRM, the Permanent
Magnets (PMs) are placed on the stator, and hence the magnets
are not subjected to centrifugal forces. However, the major
drawbacks of FRM are low power factor, voltage regulation
and significant leakage flux. In this paper, a novel ring outer
rotor FRM with Inner Bridge (IB) is proposed to address these
limitations for a direct-drive roof-top wind generator. A Magnetic
Equivalent Circuit (MEC) has been developed for a ring FRM
to account for the leakage flux. Further, the proposed FRM is Figure 1. 2-D cross-section of the single phase PM FRM
compared with a conventional 6/14 outer rotor FRM using 2-D
Finite Element (FE) analysis and the results are presented. (left) and its flux density and contour plot obtained from
Index Terms—Permanent Magnets, Flux Reversal Machine 2-D FEM analysis (right) [2].
(FRM), Reluctance Machines, Finite Element Analysis.

I. I NTRODUCTION
flux leakage in the aligned position. In [3], a three-phase
Roof-top wind power applications require the generator 6/8 FRM is proposed with hetero-polar stator poles. This
to be compact with high power to weight ratio due to geometry reduces the leakage flux between the adjacent
lack of sufficient floor area for its installation. Further, the poles. However, the same is not claimed in the paper. The
power rating of the generators used in these applications mutual inductance is reduced to improve the fault tolerance
is generally a few kW. Thus, PM generators are preferred capability. It is claimed that FRMs with more phases tend
over the Doubly-Fed Induction Generators (DFIGs) in these to have improved performance. Further, it is also claimed
applications [1]. DFIGs are relatively inefficient at lower that higher number of poles may increase the power density
power levels compared with the PM generators. The FRM and help to achieve a stable operation, but would result in
is one such PM generator which is claimed to have high increased core losses and more frequent switching of power
power density, suitable for wind power generators. Unlike electronic devices, thereby increasing the cost and overall
conventional PM Synchronous Generators (PMSGs), the losses in the control circuit.
magnets in an FRM are mounted on the stator which is In [4], a double stator FRM is proposed to increase the
stationary, and hence the magnets are not subjected to torque density. The double stator utilizes the inner space of
centrifugal forces. The rotor is salient and resembles the rotor the FRM. However, it introduces an additional air-gap ad
of a Switched Reluctance Machine (SRM), which is rugged more flux leakage. A consequent-pole PM FRM to reduce the
in construction. flux leakage between the adjacent north and south poles is
In [2], a single-phase 2/3 Flux Reversal Generator (FRG) presented in [5]. The influence of various PM arrangements
with Sm-Co magnets is proposed. A flux-MMF diagram on the performance of a PM FRM is presented in [6]. Various
technique is used to analyze the generator, which is claimed PM arrangements such as NS-NS, NS-SN, NSNS-SNSN and
to be superior over generalized machine modelling technique. NSNS-NSNS are compared and analyzed. It is claimed that
The generator is claimed to have low inductance variation the torque of a FRM is produced by dominant harmonics
with the rotor position and current. However, the issues of and that the optimal number of rotor poles is around 14 for a
flux leakage, cogging torque and efficiency improvement 12 stator slot FRM. Further, it is also claimed that the FRM
are not addressed. The 2-D cross-sectional view of the 2/3 with four PMs on each stator tooth has higher torque and
single phase FRG, and the corresponding flux contour and efficiency than that of the FRM with two PMs, particularly at
flux density plots from 2-D FE analysis are shown in Fig. low speeds.
1. From the flux plot, it is observed that there is significant In [7], a novel FRM with Halbach consequent-pole magnets

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.
 Φg

Φl
ℜs
ℜr Φm
ℜl

Figure 2. 2-D cross-sectional view of an outer rotor ring PM

FRM with IB (left) and corresponding flux plot obtained
from 2D-FEM analysis (right).
Figure 3. Magnetic equivalent circuit of a FRM with two
Table I. Calculation of number of rotor poles in an FRM.
rotor poles and one pole-pair of magnet.
Number of Rotor Poles Number of Pole-Pairs Spatial Field
2 1 1
3 2 1
4 2 2 modelling in the magnetic equivalent circuit of an FRM to
6 4 2
14 or 16 15 1 increase its efficiency and power factor. Magnetic Equivalent
13 or 17 15 2 Circuit (MEC) approach can be used to determine the amount
of leakage flux in an FRM. In order to formulate the MEC, a
ring outer rotor PM FRM, which is having the entire air-gap
embedded in the slot opening is proposed. This arrangement circumference with PM including slot opening is considered
not only reduces the flux leakage, but also increases the as shown in Fig. 2. This FRM has simplest structure with
back-EMF since the PMs no longer influence the permeance two rotor poles and the formulation of an MEC would be
of coil windings, thereby improving the power factor and simple and effective. Stator teeth are connected to form an
torque density. In [8], a quasi-Halbach PM FRM with magnets inner bridge (IB) as shown in Fig. 2. The corresponding MEC
mounted on or inserted into the stator are compared with a is shown in Fig. 3. The relationship between number of rotor
conventional FRM. PM FRM with magnets mounted on or poles Nr , the number of pole-pairs on the stator Pm and the
inserted in the stator is claimed to be suitable for direct-drive spatial field distribution pattern p is given in Eq. (2). The
applications. It is stated that with the use of Halbach PMs corresponding combinations of these parameters are tabulated
in the stator slot opening, torque ripple will reduce while in Table-I.
average torque will increase. For a two-pole spatial field distribution, the number of rotor
In this paper, a novel ring PM outer rotor FRM is presented poles Nr can be computed as given in [10],
for a direct-drive roof-top wind generator application. The
design criteria of the proposed FRM is presented in Section-II. Nr = Pm ± p. (2)
The results obtained from 2-D FE analysis and its comparison
with a conventional FRM [9] are presented in Section-III and The number of rotor poles also decide the operating frequency
Section-IV, respectively. The conclusions are presented in f of an FRM, which can be computed as
Section-V.
N × Nr
f = . (3)
60
II. M AGNETIC E QUIVALENT C IRCUIT D ESIGN OF A
N OVEL PM R ING FRM Where, N is the speed of rotation in revolution per minute
Leakage flux and the corresponding leakage factor kf (rpm) and f is the frequency in Hz. The ring FRM can be
significantly affect the analytical design of an FRM, due to modified to conventional FRM by incorporation of spacer
more leakage flux. Kf is chosen to be 0.45 in a 2 pole, 3 slot between the permanent magnet pole-pairs, in which case, Eq.
(2/3) single-phase FRG [2]. This is the ratio of maximum flux (2) will not be applicable. The air-gap flux φg and leakage
linking a stator phase φp to the magnet flux φm . flux φl can be calculated as

φp FP M <r
kf = . (1) φg = . (4)
φm <eq
Due to considerable flux leakage, the 2/3 single-phase FRG
efficiency is reported to be 68 % at 1800 rpm [2]. Hence, FP M <l
φl = . (5)
efforts have to be made to reduce the flux leakage and its <eq

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.
Figure 4. 2-D cross-sectional view of a conventional 6/14 Figure 6. 2-D cross-sectional view of a novel 6/14 ring PM
PM FRM front view (left) and side view (right). FRM with IB front view (left) and side view (right).

Figure 7. PM arrangement in a conventional FRM (left) and

proposed ring PM FRM (right).
Figure 5. 2-D cross-sectional view of a novel 6/14 ring PM
FRM front view (left) and side view (right).
ring FRM thus increases the air-gap flux density and thereby
improves the torque density (torque/volume) of the motor.
Where, FP M is the magneto-motive force (MMF) of PM, The number of stator slots and rotor poles considered
which is computed as
φ m <m Table II. Design parameters of the conventional PM FRM
FP M = . (6) [9] considered for comparison with the novel PM ring FRM.
2
The various reluctance elements including the leakage reluc- Description Value
Number of stator/rotor poles 6/14
tance are computed as Number of magnet-pairs per stator pole 2
Ls,r,l,m Air-gap thickness (mm) 0.5
<s,r,l,m = . (7) Magnet thickness (mm) 2.5
µ As,r,l,m Stator/Rotor pole arc (Mech. degrees) 48/4.87
Outer diameter of stator/rotor (mm) 200/270
Here, Ls,r,l,m and As,r,l,m are the mean length and area Number of turns per phase 132
of cross-section of stator, rotor, leakage path and magnet, Stack length (mm) 150
respectively. The computed values of the leakage path and Shaft diameter (mm) 50
magnet reluctances are 1690399 A/Wb and 39867 A/Wb,
respectively. It is found that the average air-gap flux density of are 6 and 14, respectively. Conventional PM FRM, has two
a ring FRM shown in Fig. 2 is 0.253 T through the analytical pole-pairs of PMs per stator pole, which corresponds to 4
MEC design. magnets per stator pole that can be converted to ring FRM
by incorporation of one magnet per stator pole. Total number
III. 2-D FEM A NALYSIS OF THE P ROPOSED N OVEL R ING of pole-pairs will be 15 for a two-pole spatial field distribution
FRM from Eq. (2).
The 2-D cross-sectional view of a conventional FRM and The proposed novel ring PM FRM has an inner bridge
proposed novel PM ring FRM are shown in Figs. 4, 5 & 6, (IB) between the stator teeth as shown in Fig. 8. This design
respectively. The proposed ring PM FRM has the combined modification will reduce the cogging torque of the novel ring
advantages of both conventional stator pole mounted PMs as PM FRM. The design parameters of the proposed 6/14 Ring
well as the PMs inserted in the slot-opening of the stator. PM outer rotor FRM are tabulated in Table-II. An analytical
The comparison of PM arrangement in a conventional and method is used for the selection of various design parameters
proposed ring FRM is shown in Fig. 7. In the proposed ring of the FRM. A 6/14 conventional PM outer rotor FRM is
FRM, the PMs are placed above the slot-opening to reduce considered for comparison with the proposed PM ring FRM.
the flux leakage in a conventional FRM. The proposed PM For the sake of comparison, the magnet volume is main-

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.
Figure 8. PM arrangement in a novel ring PM FRM with Figure 10. 2-D FEM flux density plot of the novel ring PM
inner bridge. FRM.

Figure 9. 2-D FEM flux density plot of the conventional Figure 11. 2-D FEM flux density plot of the novel ring PM
6/14 FRM. FRM with inner bridge.

tained the same in both the FRMs. It is to be observed that, for as given in [13],
the ring PM arrangement to have same magnet volume as that D
lg = . (8)
of conventional stator pole mounted PMs, the magnet thickness 1000
has to be reduced. As the air-gap flux density Bg is directly Here, D is the rotor outer diameter, which is 270 mm for
proportional to magnet thickness, this design modification both the conventional FRM and novel ring PM FRM. lg is
would reduce the air-gap Bg . However the air-gap thickness is thus evaluated to be 0.27 mm. However, due to the design
maintained same in both the FRMs for the sake of comparison. constraints the chosen value of lg is 0.5 mm.
It is to be noted that, the effective air-gap (air-gap + magnet
thickness) is lower in the proposed PM ring outer rotor FRM, IV. C OMPARISON OF P ROPOSED N OVEL R ING FRM WITH
and hence this compensates for the reduction in Bg due to C ONVENTIONAL FRM USING FEM A NALYSIS
reduction in magnet thickness. The flux contour and flux density plot of the conventional
Various analytical methods are available in literature to and novel ring FRM obtained from 2-D FE analysis are shown
calculate air-gap flux density Bg in the air-gap in PM ma- in Fig. 9. It is observed that the ring arrangement results in
chines [11]. In small size generators, radially magnetized ring more air-gap flux density as compared to conventional FRM.
magnets can be employed, where the magnets can be modelled The average air-gap flux density Bg in the proposed (without
as a cylindrical plane with surface charge density [12]. In this IB) and conventional FRM are 0.5 T and 0.473 T, respectively.
paper, Nd-Fe-B PMs are used to form the ring with residual It is observed that the incorporation of an inner bridge
induction Br of 1.2 T. The main design variables of the (IB) between the stator teeth further increased the average
proposed FRM are length of magnet, length of the air-gap and air-gap flux density Bg in the proposed ring PM FRM to
number of rotor poles. The air-gap length lg can be calculated 0.544 T. It is to be observed that, the value of Bg obtained

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.
 Flux Linkage (Wb)
Flux Density |B|

Figure 14. Comparison of phase flux linkage in conventional

and proposed PM ring FRM obtained from 2D-FEM
Figure 12. Comparison of air-gap flux density in conven- analysis.
tional and proposed ring PM FRM obtained from 2D-FEM
analysis.
Cogging Torque (N-m)

EMF (V)

Figure 15. Comparison of EMF per phase in conventional

and proposed PM ring FRM obtained from 2D FEM
Figure 13. Comparison of cogging torque in conventional analysis.
and proposed PM ring FRM obtained from 2D-FEM
analysis.
and proposed FRM as shown in Fig. 14. It can be observed
from Fig. 14 that the flux linkage is more in the case of PM
from FEM analysis is in close agreement with that obtained ring FRM.
from analytical MEC approach. Further it is also observed that The EMF per phase of the conventional FRM and the
the inter-magnet flux leakage is lower in the proposed FRM proposed PM ring FRM with and without IB obtained from
with IB compared with the conventional FRM. The use of 2D FEM analysis are shown in Fig. 15. It can be observed
ring magnet eases the manufacturing of an FRM stator with from the figure that, the shape of the EMF waveform of the
magnets, unlike in the case of a stator pole mounted PM FRM PM ring FRM is much closer to a sinusoid compared with the
where the magnets have to be glued on to the surface of the other FRMs. Further, the peak-peak magnitude of the EMF
stator pole. The 2D-FEM flux density plot of the novel ring waveform of PM ring FRM is higher compared to the other
PM FRM with IB is shown in Fig. 10. FRMs. This validates the claims of reduced leakage flux. The
The magnitude and shape of the air-gap flux density important performance parameters obtained from 2D FEM
waveform varies due to stator PMs, slot opening and the analysis of the conventional and proposed PM ring FRM are
interaction between stator and rotor slotting as reported in [14]. tabulated in Table-III.
Further, rotor pole arc and PM pole arc greatly influence the
air-gap flux density, which can be observed from Fig. 12. The Table III. Comparison of performance parameters of
proposed ring FRM has fewer harmonics in the air-gap flux conventional and proposed FRM obtained from 2D FEM
density waveform as compared to the conventional FRM. analysis.
The cogging torque comparison is carried-out for the con- Parameter Conventional FRM Proposed FRM
ventional and proposed FRM which is shown in Fig. 13. It Cogging Torque (peak-peak) 20.32 N-m 11.54 N-m
is to be observed that the incorporation of IB in ring PM Average air-gap flux density 0.47 T 0.54 T
FRM reduces the cogging torque significantly [15]. The flux EMF/phase (peak–peak) 486.78 V 492.49 V
linkage is observed to be sinusoidal for both the conventional

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.
 V. C ONCLUSIONS [5] Y. Wang, M. Cheng, M. Chen, Y. Du, and K. T. Chau, “Design of high-
torque-density double-stator permanent magnet brushless motors,” IET
A novel PM ring FRM with inner bride (IB) is proposed Electric Power Appl., vol. 5, no. 3, pp. 317–323, 2011.
for a direct-drive roof-top wind generator, by making suitable [6] Y. Gao, R. Qu, D. Li, J. Li, and G. Zhou, “Consequent-pole flux-reversal
design modifications to a conventional 6/14 outer rotor FRM. permanent-magnet machine for electric vehicle propulsion,” IEEE Trans.
Appl. Supercond., vol. 26, no. 4, pp. 1–5, 2016.
The FRM by default has the magnetic gearing effect making [7] R. Q. H. Huang, D. Li. and K. Xie, “A double stator flux reversal
it suitable for direct-drive low speed applications. It is claimed machine with halbach consequent pole in slot opening,” in Proc. of
that the incorporation of ring magnet with IB in the proposed Annual IEEE INTERMAG Conf., Singapore, April 2018, pp. 1–1.
[8] K. Xie, D. Li, R. Qu, Z. Yu, Y. Gao, and Y. Pan, “Analysis of a flux
PM ring FRM, improves the average air-gap flux density and reversal machine with quasi-halbach magnets in stator slot opening,”
reduces the cogging torque significantly. The inter-magnet and IEEE Trans. Ind. Appl., vol. 55, no. 2, pp. 1250–1260, 2019.
slot leakage flux are greatly reduced in the proposed PM [9] D. S. More, H. Kalluru, and B. G. Fernandes, “Outer rotor flux reversal
machine for rooftop wind power,” in Proc. of IEEE Industry Appl.
ring FRM, thereby improving the power factor. A MEC is Society Annual Meeting, Edmonton, Oct 2008, pp. 1–6.
developed for the estimation of the leakage flux and it is [10] G. Pellegrino and C. Gerada, “Modelling of flux reversal machine for
observed that the values of air-gap flux density obtained from direct drive applications,” in Proc. of 14th Eurpean Conf. on Power
Electronics and Appl., Birmingham, Sep. 2011, pp. 1–6.
FEM and MEC approach are in close agreement. [11] T. Yoon and D. K. Lieu, “A method to verify accuracy of predicted
magnetic orientation of a permanent ring magnet in a brushless dc
R EFERENCES motor,” IEEE Trans. Mag., vol. 43, no. 9, pp. 3638–3644, 2007.
[1] A. Sharma and S. Sashidhar, “Comparison of wind turbine generators [12] R. Ravaud, G. Lemarquand, V. Lemarquand, and C. Depollier, “Ana-
for roof-top wind power: Case study, issues and challenges,” in Proc. lytical calculation of the magnetic field created by permanent-magnet
39th Annual IEEE Region 10 TENCON Conf., Kochi, India, 2019, pp. rings,” IEEE Trans. Mag., vol. 44, no. 8, pp. 1982–1989, 2008.
592–597. [13] J. A. O. A. F. F. Filho and D. G. Dorrell, “Ringshaped surface-
[2] R. P. Deodhar, S. Andersson, I. Boldea, and T. J. E. Miller, “The mounted permanent magnet generators with modular stator for small
flux-reversal machine: a new brushless doubly-salient permanent-magnet wind turbines,” in Proc. of IEEE int. magnetics Conf. (INTERMAG),
machine,” IEEE Trans. Ind. Appl., vol. 33, no. 4, pp. 925–934, 1997. Singapore, April 2018, pp. 1–6.
[3] C. Wang, S. A. Nasar, and I. Boldea, “Three-phase flux reversal machine [14] G. Dajaku and D. Gerling, “Air-gap flux density characteristics of
(frm),” IEE Proceedings - Electric Power Appl., vol. 146, no. 2, pp. 139– salient pole synchronous permanent-magnet machines,” IEEE Trans.
146, 1999. Mag., vol. 48, no. 7, pp. 2196–2204, 2012.
[4] H. Li and Z. Q. Zhu, “Influence of magnet arrangement on performance [15] X. Zhu and W. Hua, “An improved configuration for cogging torque
of flux reversal permanent magnet machine,” in Proc. of Annual IEEE reduction in flux-reversal permanent magnet machines,” IEEE Trans.
IEMDC Conf., Miami, May 2017, pp. 1–6. Mag., vol. 53, no. 6, pp. 1–4, 2017.

Authorized licensed use limited to: University of Canberra. Downloaded on May 20,2021 at 12:33:55 UTC from IEEE Xplore. Restrictions apply.