Design of Flux-Switching Permanent Magnet

Machine Considering the Limitation of Inverter and

Flux-Weakening Capability

Wei Hua, Ming Cheng Z. Q. Zhu, D. Howe

Department of Electrical Engineering Department of Electronic and Electrical Engineering
Southeast University University of Sheffield
Nanjing 210096, China Sheffield, United Kingdom
huawei1978@seu.edu.cn, mcheng@seu.edu.cn Z.Q.Zhu@sheffield.ac.uk, D.Howe@sheffield.ac.uk

Abstract—In this paper, a general design procedure for the flux- necessary to combine the machine design parameters and the
switching permanent magnet (FSPM) machine with different inverter limitation to find the optimal match which satisfies the
topologies is proposed. Firstly, a 3-phase 12-stator-tooth/10- performance requirement in the design stage [10].
rotor-pole topology is introduced and its operation principle is
described. Then, the basic design method, including The purpose of this paper is to propose a general design
determination of the stator, rotor and magnet dimensions, is procedure for FSPM machine with different topologies,
proposed under the constraints of some dimensions and electrical without specific application background. Firstly, a 3-phase,
parameters. The winding turns are obtained by iteratively 12/10 (12-stator-tooth/10-rotor-pole) topology and operation
solving the base-speed and current equations simultaneously. The principle of the FSPM machine is introduced. Then, the basic
influence of current density and slot packing factor on the flux- design method, including initial determination of the stator and
weakening capability is investigated in the design stage to satisfy rotor dimensions, is proposed with some constraints of
the torque and wide-speed operation requirement. Further, the dimensions and electrical parameters to compare with a 12/8-
output power equation, consequently, the sizing equation is pole doubly salient PM (DSPM) motor [11]. In the
derived for determining the initial dimensions of a FSPM determination of winding turns, the parameters (slot packing
machine, in which the traditional D2sila is replaced by D2sola. and armature current density) that will affect the inverter
Hence, the power density of different topologies can be compared limitation and flux-weakening capability are taken into
directly by the equation. It reveals that the 3-phase 12/10-pole account. In addition, in order to obtain a general design
topology can offer higher power and power density than those of
procedure for any FSPM machine topologies without specific
the 2-phase 8/6-pole machine by ~11% in theory under the same
application background, a general output power equation,
conditions. The experiments on the prototype motor verify the
performance predictions. consequently, the sizing equation is derived. The traditional
D2sila is replaced by D2sola. The original design data obtained
Keywords-Brushless machines; doubly-salient; finite element from a special design procedure is re-substituted into the sizing
(FE) analysis; flux switching; permanent magnet; design. equation to verify the accuracy of the sizing equation. Hence, it
can be used to determine the initial dimensions satisfying the
specification of individual requirement and to directly compare
I. INTRODUCTION
the power density of different topologies. Both finite element
Flux-switching permanent magnet (FSPM) machine is a re- calculations and experiments verify that the machine exhibits
emerging PM brushless machine having magnets and bipolar PM flux-linkage and sinusoidal phase back-EMF
concentrated windings in the stator with doubly-salient waveform.
structure [1]. The initial research has revealed that the machine
has the advantages of high power/torque density, high II. TOPOLOGY AND OPERATION PRINCIPLE
efficiency, and high flux-weakening capability, favorable for
cooling and high-speed operation. However, the available A. Topology
papers on the development of the FSPM machine are focused
on performance analysis rather than design [2]-[9]. Although Fig. 1 shows the cross-sections of a 3-phase, 12/10-pole
the design of a 3-phase FSPM motor for aircraft application is FSPM machine. It can be noted that the rotor of the machine is
proposed in [8], it is mainly about the selection of magnet similar to that of a switched reluctance (SR) motor. In addition,
material due to its special application. The general design the concentrated windings, also similar to those in a SR motor,
principle, e.g. the choices of the machine initial dimensions and are employed, which lead to low copper consumption and low
parameters are unavailable with the given torque or power copper loss due to short end-windings. In the FSPM machine a
requirement. In addition, since the machines are supplied by concentrated coil is wound around the two adjacent teeth with a
power electronic inverter, which results in that the voltage and piece of permanent magnet (PM) in the middle. The stator
current limitations will constrain the machine performance, it is contains 12 segments of “U”-shape magnetic cores, between

This work was supported by National Natural Science Foundation of

China under Project 50377004, and the Specialized Research Fund for the
Doctoral Program of Higher Education of China under Project 20050286020.
2403
1-4244-0365-0/06/$20.00 (c) 2006 IEEE
which 12 pieces of PMs are inset. The magnets are magnetized
circumferentially in alternative opposite directions. Compared
to the conventional PM machines having magnets in the rotor,
the placement of both PMs and windings in the stator is
favorable for cooling and is desirable for the aerospace and EV
applications where the ambient temperature of the machine
may be high.
The major speciality of the stator configuration can be more
easily seen from Fig. 1(b). Due to symmetry, only half of the
stator, i.e. 6 stator teeth and 6 coils are discussed here. At the
position shown in Fig. 1(a), the PM flux of phase A reaches
maximum. The polarity of flux in coils A1 and coil A2 is (a)
identical with slight difference in their amplitudes due to
magnetic circuit differences. The relative position between the
PMs and coils can be described as: A1+_N_B1−_S_C1−_
(b)
N_A2+_S_B2−_N_C2− (where N/S denotes the polarity of
magnets and +/− denotes the opposite polarity of stator coils). Figure 1. Topology of FSPM machine. (a) Cross-section. (b) Configuration
of PMs and windings.
Obviously, for a FSPM machine, each phase consists of 4 coils
in series. Since coils are wound around two stator teeth, when
the rotor rotates 18 mechanical degree anticlockwise from the
position shown in Fig. 1(a), both fluxes of coils A1 and A2 will
change from positive to negative simultaneously while the
amount of flux is unchanged, i.e. A1−_N_B1+_S_C1+_
N_A2−_S_B2+_N_C2+. Thus, the phase flux-linkage in a
FSPM machine is bipolar. Consequently, the FSPM machine
may exhibit high torque density, as shown later. Besides, the
magnetic circuit of FSPM machine is short, which may result (a) (b)
in low iron loss in the stator back-iron. ψm

B. Operation Principle θ
In Fig. 2(a), the rotor pole aligns with one of two stator
teeth over which a coil is wound and the PM flux which is e
linked in the coil goes out of the coil and into the rotor tooth.
When the rotor moves forward to align with the other stator θ
tooth belong to the same coil as shown in Fig. 2(b), the PM i
flux linked goes out of the rotor tooth and into the stator tooth,
keeping the same amount of flux-linkage whilst reversing the
polarity, i.e., realizing the “flux-switching”. Consequently, as θ
the rotor moves, the flux-linkage in the windings will change
periodically, as shown in Fig. 2(c). (c)
Figure 2. Operation principle of FSPM machine. (a) “Flux-switching”
Based on the analysis above, the PM flux-linkage induced principle. (b) The ideal phase PM flux-linkage, back-EMF and armature
in the phase windings is assumed to be ideal sinusoidal versus current.
rotor position. Hence, the induced back-EMF in the phase
winding is also sinusoidal. When the alternating currents i are
applied to the corresponding windings in phase with the PM
flux-linkage ψm (back-EMF e), the motoring torque can be
produced. Because the waveforms of flux-linkage and thus the
back-EMF are essentially sinusoidal [2], [4], it makes the
FSPM machine an excellent candidate for brushless AC drive
operation. In addition, since the armature reaction flux in a
FSPM machine does not pass through the magnets, the
irreversible demagnetization withstand capability of the Figure 3. Determination of machine dimensions in original design.
magnets is high, which makes it particularly suitable for flux-
weakening operation. Moreover, flux-focusing can be utilized III. SPECIFIC DESIGN UNDER CONSTRAINTS
and low cost magnets, such as ferrites, may be employed [2],
although in which the demagnetization withstand capability of A. Dimension Design
the magnets is inferior due to the low magnet performances. For the fair comparison, the preliminary design of a 12/10-
Hence, in this paper, NdFeB is chosen to produce the magnetic pole FSPM motor is constrained some dimensions and
field as the excitation source.

2404
specifications to be the same as those of the prototype of 12/8- rotor electrical angle velocity ω and PM flux-linkage ψm can be
pole DSPM motor [11], including stator tooth number Ps, stator expressed as follows,
outer diameter Dso, rotor inner diameter Dri, active stack length
la, rated speed nr, dc-link voltage Udc, and current density Js as U max
listed in Table I. In the preliminary design stage, the optimal ( Lq I q ) 2 + ( Ld I d + ψ m ) 2 ≤ ( )2 (5)
ω
split ratio (Ksio=Dsi/Dso) is investigated to be 0.55 [9]. Since
When the machine is operated in the constant torque region
there are no recommended dimension choices from the
under the current limit of the inverter, the inverter is able to
available references, the stator tooth width βs, rotor pole width supply a voltage higher than the required voltage of the
βr, PM thickness in magnetization direction hpm, stator yoke machine, i.e., the machine operating current Ia and voltage Ua
thickness hys, and slot opening hslot are chosen to be identical in satisfy: Ia=Imax, Ua<Umax. In contrast, when the machine is
the original design as shown in Fig. 3, satisfying the operated above the base speed (nr=1500rpm), the inverter
relationship as follows, voltage limit is met, namely entering constant power region.
Then, the machine is operated under flux-weakening control,
πDsi πD so K sio i.e., Ia=Imax, Ua=Umax [13]. The maximum voltage Umax that the
β s = β r = h ys = hslot = h pm = = (1)
4 Ps 4 Ps inverter can supply is dependent on the dc-link voltage Udc and
Obviously, the stator tooth arc factor cs is 0.25 in this case. the PWM strategy adopted. Hence, the maximum fundamental
In addition, the rotor pole height hpr is chosen as 1/4 rotor outer phase voltage in peak value of the hysteresis PWM controller is
radius Rro, i.e.,
U max = 2 / π ⋅ U dc = 0.636U dc (6)
1 1 1 In this paper, Udc is constrained to be 440V, whilst the rated
h pr = Rro = ( Dsi − 2 g ) = ( Dso K sio − 2 g ) (2)
4 8 8 current Imax is limited to the corresponding current density of
where g is the airgap length. With the given Dso and g, the main 5A/mm2 in RMS value, although which is unknown here due to
dimensions of the machine can be determined primarily except the winding turns phase is not determined.
the winding turns per phase, which is relative to the
performance requirement and the limitation of inverter, and C. Winding Turns Calculation
will be carried out in the next section. The relationship between rated mechanical angular velocity
Ωb (rad/s), rated voltage of inverter Umax, rated current of
TABLE I. DESIGN SPECIFICATIONS OF 12/10-POLE FSPM MACHINE inverter Imax, the PM flux-linkage transformed into d-axis ψm,
Items Constraints and the dq-axes inductances Ld/Lq can be expressed as follows,
Rated speed, nr (rpm) 1500
dc-link voltage, Udc (V) 440 U max
Phase number, m 3 Ωb = (7)
Stator outer diameter, Dso (mm) 128 8ψ m Ld C − ( Ld + Lq )C 2
Active stack length, la (mm) 75 p r L2q I max
2
+ ψ m2 +
Rotor inner diameter, Dri (mm) 22 16( Lq − Ld )
Stator tooth number, Ps 12
RMS current density, Js (A/mm2) 5 where, C = ψ m ∓ ψ m2 + 8( Lq − Ld ) 2 I max
2
(“−” for Ld<Lq; “+”
Items Design data
Stator inner diameter, Dsi (mm) 70.4 for Ld>Lq); and Ω b = nr 2π / 60 . On the other hand, the
Airgap length, g (mm) 0.35 relationship between the rated current in peak value and the
Rotor pole number, Pr 10 machine electrical parameters can be expressed as,
Stator tooth arc, θs (degree) 7.5
Rotor pole arc, θr (degree) 7.5
Stator tooth height, hst (mm) 24.2 I max = 2 J s Aslot K p / N coil (8)
Rotor pole height, hpr (mm) 8.71
Phase winding turns, nph 280 where, Js, Aslot, Kp, and Ncoil are the rated current density
Magnet volume, Vm (mm3) 12*(4.6*28.8*75) (RMS), area of half slot, slot packing factor, and winding turns
Rated Torque, Tem (Nm) 13.01 per coil, respectively, between which only Ncoil is unknown and
Rated Power, Pem (Nm) 2036 the other parameters are pre-determined. Further, the open-
Rated peak phase voltage, Umax (V) 280 circuit field can be studied by finite element (FE) analysis, and
Rated peak phase current, Imax (A) 5.37
then the phase PM flux per turn Φm and the dq-axes permeance
B. The Limitation of Inverter per turn Λd/Λq can be obtained.
Since the FSPM machine is supplied by inverter, there are 2 Since ψm=Ncoil⋅Φm, Ld=Λd⋅N2coil and Lq=Λq⋅N2coil, once Ncoil
constraints caused by the supply limitations [12]-[14], i.e., is given, the corresponding ψm, Ld and Lq can be given. Hence,
through iteratively calculating (7) and (8) simultaneously, the
I d2 + I q2 ≤ I max
2
(3) demanded Ncoil can be found to match both (7) and (8), i.e.,
U d2 + U q2 ≤ U max
2
(4) Udc=440V, Js=5A/mm2, and nr=1500rpm, simultaneously.
Table I lists the original design data of the 12/10-pole FSPM
Neglecting the winding resistance Rph, the voltage ellipse machine under the constraints, satisfying the performance
equation including the dq-axes currents Id/Iq, inductances Ld/Lq, requirement aforementioned. It should be noted that the
relationships between the base speed Ωb, the current density Js,

2405
and the winding turns per coil Ncoil are expressed in (7) and (8). 30

If any two of them are determined, the third one can be 25 nr=1500rpm
calculated.

Torque (Nm)
20
5A/mm2
D. Improving the Flux-Weakening Capability 15 6A/mm2
7A/mm2
The speed of the machine can be derived as follows [13],
10 8A/mm2
2
9A/mm
U
Ω= (9) 5

p r (ψ m + Ld I d ) 2 + ( Lq I q ) 2 0

When U=Umax, Id =−Imax, and Iq=0, i.e., the armature current is 0 2000 4000 6000 8000
Speed (rpm)
10000 12000 14000 16000

fully used to counteract the PM flux-linkage and the current

component producing-torque Iq is zero. Then, the maximum (a)
operation speed that the motor can reach is: 6
nr=1500rpm
5
U max
Ω max = (10)
pr (ψ m − Ld I max )

Power (kW)
4
5A/mm2 6A/mm2
If ψ m ≤ Ld I max is satisfied, the speed can be infinite in 3 7A/mm2

theory. However, in most PM motors, this condition can not be 2 8A/mm2

satisfied. Thus, a factor Kfw is defined to reflect the flux- 2
9A/mm
weakening capability of the machine designed, i.e. 1

0
Ld I max 0 2000 4000 6000 8000 10000 12000 14000 16000
K fw = (11) Speed (rpm)
ψm
Obviously, the larger Kfw indicates the better flux- (b)
weakening capability. However, the results from FE analysis Figure 4. Steady-state performance (Kp=0.4) with different Js. (a) Torque-
indicate the PM flux Φm is significantly high due to the flux- speed characteristics. (b) Power-speed characteristics.
focusing effect, resulting in excellent torque capability. On the
other hand, the d-axis inductance is small due to the heavy
magnetic saturation. Hence, the flux-weakening capability of 25
the FSPM machine with NdFeB is inferior to the case of nr=1500rpm
adopting ferrite under the current density of 5A/mm2, in which 20
it can reach infinite speed [2]. Substituting (8), ψm=Ncoil⋅Φm, Kp=0.4
Torque (Nm)

Ld=Λd⋅N2coil into (11), it can be re-written as, 15 Kp=0.45

Kp=0.5
Kp=0.55
10 Kp=0.6
2
J s Aslot K p
(Λ d N coil )( 2 )
N coil 2 Λ d J s Aslot K p 5
K fw = = (12)
N coil Φ m Φm 0
From (12), it can be seen that Kfw can be improved by 0 1000 2000 3000 4000 5000 6000 7000
Speed (rpm)
reducing Φm, or (and) increasing Λd, Js, Aslot, and Kp. However,
generally it is a trade-off between the torque capability in the (a)
constant torque region and the flux-weakening capability in the 6
constant power region [12]-[14]. Due to Φm, Λd, and Aslot can 5
nr=1500rpm
not vary as the machine is designed, the variable parameters
Kp=0.4 K =0.45
can only be Js and Kp. Thus, in the followings, the investigation
Power (kW)

4 p
Kp=0.5
of improving the flux-weakening capability of the designed 3
Kp=0.55
FSPM machine is carried out in two ways. Kp=0.6
2
1) Adjustment of Js 1
Fig. 4 compares the steady-state performance versus current
0
densities with Kp equal to 0.4. Table II lists the corresponding
0 1000 2000 3000 4000 5000 6000 7000
performance. Obviously, with J s varying from 5-9A/mm2
Speed (rpm)
which can be easily realized by incorporating appropriate
(b)
cooling method, the rated speed (base speed) are strictly kept
1500rpm satisfying the requirement, whilst Kfw improves from Figure 5. Steady-state performance (Js=5A/mm2) with different Kp. (a)
0.475 to 0.855, consequently the flux-weakening region is Torque-speed characteristics. (b) Power-speed characteristics.

2406
 TABLE II. PARAMETERS OF FSPM MACHINE (Udc=440V, Kp=0.4) ratio is widened from 2.1 to 4.3, by increasing Kp, although
Js (A/mm2) 5 6 7 8 9 which inevitably results in more difficult for winding. In
addition, it can be found that due to the limitation of slot
Ncoil (turn) 70 67 64 61 59
packing factor, the effect in this case is inferior to the case of
Imax (A) 5.365 6.727 8.216 9.851 11.458 increasing Js.
Id (A) -0.218 -0.327 -0.465 -0.636 -0.830
In summary, the flux-weakening capability can be
Iq (A) 5.361 6.719 8.203 9.831 11.428
effectively improved by increasing Js and/or Kp if acceptable.
nr (rpm) 1495 1502 1506 1510 1490
nmax (rpm) 3156 4027 5413 7932 13594
E. Summary
nmax/nr 2.111 2.681 3.594 5.253 9.123
In this section, the specific design procedure of the FSPM
ψ m (Wb) 0.136
0.162 0.155 0.148 0.141 machine for a normal 3-phase variable-speed motor drive
Ld (mH) 14.308 13.108 11.960 10.865 10.165 system can be summarized as follows.
Lq (mH) 15.533 14.230 12.984 11.796 11.035
Firstly, the main dimensions of stator and rotor as well as
Kfw 0.475 0.570 0.665 0.760 0.855 magnets are obtained by (1) and (2) with the volume
Copper diameter (mm) 0.871 0.890 0.911 0.933 1.071 requirement, e.g. the stator outer diameter Dso and the stack
Windings length (m) 141.48 135.42 129.35 123.29 119.25 length la are given.
Rph (Ω) 1.349 1.236 1.127 1.024 0.958 Secondly, the performance requirement, e.g. the base speed
Copper loss Pcu (W) 58.240 83.865 114.150 149.09 188.70 nr, the dc-link voltage Udc, the peak value of airgap flux
23.462 density Bg (which determines the peak value of the flux linked
Tem (Nm) 13.011 15.619 18.230 20.844
in the windings, consequently the torque capability as
Pem (W) 2036.4 2455.8 2875.2 3296.9 3661.8
discussed in the section IV), the inverter conducting frequency
(which is determined by the rotor pole number Pr), is satisfied
from calculating the number of windings turns by (7) and (8)
since the static characteristic per turn, such as the PM flux Φm,
TABLE III. PARAMETERS OF FSPM MACHINE (Udc=440V, Js=5A/mm2) the dq-axes inductances per turn (permeances), can be obtained
Kp 0.45 0.5 0.55 0.6 based on the FE analysis.
Ncoil (turn) 68 66 65 63
Thirdly, the slot packing factor and/or the current density
Imax (A) 6.214 7.113 7.945 8.942
Id (A) -0.283 -0.360 -0.442 -0.542
may be adjusted to agree with the specific ratio of maximum
Iq (A) 6.207 7.104 7.933 8.926 speed to the base speed requirement (nmax/nr) whilst the base
nr (rpm) 1502 1509 1491 1496 speed is kept unchanged.
nmax (rpm) 3663 4327 5147 6409
nmax/nr 2.439 2.867 3.452 4.284 IV. GENERAL DESIGN WITHOUT CONSTRAINT
ψ m (Wb) 0.157 0.152 0.150 0.145
Ld (mH) 13.502 12.720 12.337 11.589 The above design of 12/10 FSPM machine is discussed
Lq (mH) 14.658 13.809 13.393 12.582 according to the dq-axis modeling to obtain the winding turns
Kfw 0.535 0.594 0.653 0.713 under the constraints of volume and some electrical parameters
Copper diameter (mm) 0.937 1.003 1.060 1.125 to ease for comparative study with a 12/8-pole DSPM machine.
Windings length (m) 137.44 133.40 131.38 127.33 In addition, the 12/10-pole topology is a 3-phase topology for
Rph (Ω) 1.131 0.959 0.846 0.728 variable-speed drive system, in which the torque capability in
Copper loss Pcu (W) 65.52 72.80 80.08 87.36 the constant torque region and the flux-weakening capability in
Tem (Nm) 14.640 16.271 17.903 19.536
the constant power region are the two important performances.
Pem (W) 2302.8 2570.3 2795.5 3061.0
However, except operated as a motor, the FSPM machine can
also be used as a generator. Additionally, it is possible to
significantly extended from 3156rpm to 13594rpm, i.e., the employ other phase number, e.g. 2-phase. Hence, to derive a
speed-regulation ratio (nmax/nr) effectually increases from 2.1 to general design procedure of FSPM machine with any topology
9.1 as shown in Table II, which indicates that with the normal and without special application background, a general output
slot packing factor of 0.4, by increasing the current density, the power equation, consequently, a sizing equation is necessary to
FSPM machine can be designed to satisfy different wide-speed be derived to determine the initial dimensions for the engineers
operation request whilst the base speed of 1500rpm is strictly to design a FSPM machine with only power or torque
kept. requirement.

2) Adjustment of Kp A. Output Power and Torque Equations

Similarly, keeping the constant Js=5A/mm2, the steady- According to Fig. 2(c), assuming the phase PM flux-
state performance versus Kp varying from 0.4 to 0.6, which is linkage and consequently phase back-EMF are pure sinusoidal,
normally acceptable in most machines, is shown in Fig. 5 and a sinusoidal armature current are injected into phase winding in
Table III. Obviously, the flux-weakening capability can also be phase with back-EMF to produce unidirectional power
enhanced from 0.475 to 0.713. Consequently, speed-regulation

2407
(torque). Hence, neglecting the resistance, the input power P1 Additionally, the aforementioned optimal split ratio of
can be expressed as, FSPM machine is 0.55 [6], i.e., K sio = Dsi / Dso = 0.55 . Hence,
(21) can be re-written in the term of D2sola instead of D2sila as
m T
P1 =
T 0 ∫
e(t ) ⋅ i (t )dt
(13)
follows,

m T 2π 2π m 2π 3 Pr
=
∫E m sin( t ) ⋅ I m sin( t )dt = E m I m As B g Dso2 l a nr c sη
2
P2 = K s K d K sio (22)
T 0 T T 2 120 Ps
where m is the phase number, Em and Im are the magnitude of
Consequently, the power density ξp=P2/mvol can be easily
sinusoidal phase back-EMF and current, respectively. If η is expressed as,
denoted as the efficiency of the machine, the output power P2
can be expressed as,
P 2π 2 Pr
ξp = 2
= 2
K s K d K sio As Bg nr c sη (23)
P2 = ηP1 = ηmE m I m / 2 (14) πD l / 4
2
so a 30 Ps
On the other hand, Em satisfies the relationship as follows where, mvol is the volume of the machine, equal to πD2sola/4.
(supposed to be operated as a motor, thus the minus sign “−”is
Neglecting the copper loss and iron loss, the output torque
neglected),
T2 equals to electromagnetic torque Tem, i.e.,
dψ m dΦ m dθ dΦ m
e= = n ph = n ph ωr (15) P2 P2 2π 2 Pr
dt dθ dt dθ T2 = = = 2
K s K d K sio As Bg Dso2 l a c sη (24)
where nph is the winding turns per phase. Whilst, according to
ωr nr π / 30 4 Ps
Fig. 2(c), the PM flux Φm versus the rotor position θ, can be and the torque density ξT=T2/mvol yields,
expressed as,
T2 P
ξT = = 2π r K s K d K sio
2
As Bg c sη (25)
2π πDso l a / 4
2
Ps
Φ m = −Φ peak cos( θ) (16)
θ cr Obviously, (22)-(25) indicate that the output power (torque)
where Φpeak is the peak value of flux in a period of rotor pole and power density (torque density) are independent of phase
pitch, which equals θcr=2π/Pr, thus (16) yields, number m and winding turns nph, but are proportional to the
ratio of rotor pole number to stator tooth number Pr/Ps, the
Φ m = −Φ peak cos(Prθ ) (17) airgap flux-density at no-load Bg and the armature current
electrical loading As. Hence, a 3-phase, 12/10 FSPM machine
Substituting (17) into (15) yields, can offer higher power and power density than those of a 2-
phase, 8/6 FSPM machine by ~11% in theory under the same
e = n phω r Pr Φ peak sin( Prθ ) = E m sin( Prθ ) (18)
conditions. Hence, the design of any topologies of FSPM
Then, the peak value of phase back-EMF Em satisfies, machine can be performed by a general procedure with only
the power or torque requirement as illustrated as follows.
πDsi
E m = n phω r Pr Φ peak = n phω r Pr K d B g l a cs (19)
Ps B. General Design by Output Equation
where Kd is a leakage factor, Bg is the peak flux density at no- In this section, the derived output power (torque) equation
load aforementioned in section III (E), cs is the stator tooth arc will be validated by the example of the designed 12/10-pole
factor, which equals 0.25 here. FSPM machine in section III. Based on (23) and (25), the
sizing equation can be expressed as,
In addition, if a rotor skewing is adopted to alter the
waveforms of back-EMF to be more sinusoidal, a skew factor P2
π D so2 l a = (26)
is defined as follows, K s = cos( δ ) , where δ is rotor 2π Pr
3

2θ cs
2
K s K d K sio As B g n r c sη
120 Ps
skewing angle and the stator tooth pitch θ cs = 2π / Ps .
T2
Whilst the peak value of sinusoidal armature current Im yields, Dso2 la = (27)
2π Pr 2
K s K d K sio As Bg c sη
2

As πDsi 4 Ps
I m = K i I rms = 2 (20) Referring the original design data from FE analysis in Table I
2mn ph
for a 12/10 FSPM machine, which include P2=2036W, Pr=10,
Substituting (19), (20) and ωr=2π nr/60 into (14) yields, Ps=12, Ks=1 (unskewed rotor), Kd=0.92 [6], Ksio=0.55 [9],
As=28800A/m, Bg=2.25T [5], nr=1500rpm, cs=0.25, and
2π 3 Pr η=0.85, the active machine volume can be obtained from (26),
P2 = K s K d As B g D si2 l a nr c sη (21) i.e.,
120 Ps

2408
 0.2

0.15

PM flux-linkage (Wb)
0.1

0.05

-0.05

-0.1

-0.15

-0.2
0 9 18 27 36
Rotor position (mech.)
Figure 7. FE predicted phase PM flux-linkage of 12/10-pole FSPM motor.
250
200
150

(a) 100

Back EMF (V)

50
0
-50
-100
-150
-200
-250
0 9 18 27 36
Rotor position (mech.)

(a)

(b)
Figure 6. Prototype of the designed 12/10-pole FSPM motor. (a) Stator
without windings. (b) Stator with windings and unskewed rotor.

Dso2 = 12 *10 −4 m 3 (28)

Substituting la=75mm, the stator outer diameter can be
obtained as Dso=126mm, which is very close to 128mm in the
original design, indicating the output power equation and the
sizing equation is accurate for initial machine design.
In summary, the general design procedure of the FSPM
machine with any topology and without application
background can be described as follows:
(b)
Firstly, according to requirements of power (torque),
efficiency, stator/rotor topology, electrical loading, etc, the
volume of the machine can be determined primarily by (26) or
(27).
Secondly, based on (1) and (2), the stator, rotor and
magnets dimensions can be obtained.
Thirdly, by the FE analysis, the static analysis at no-load,
including the airgap flux-density Bg, PM flux, back-EMF per
turn at the base speed e0 (since the winding turns are not
determined), winding inductances per turn, and resistance per
turn, etc can be performed.
Fourthly, according to the dc-link voltage Udc and adopted
current control strategy, the maximum value of the
fundamental component of phase voltage Umax, which the (c)
inverter can supply, is determined. Figure 8. Phase back-EMF of 12/10-pole FSPM motor at 1200rpm. (a) FE
predicted. (b) Measured (100V/division). (c) Harmonics analysis.

2409
 Fifthly, with the fundamental component of phase voltage power density than those of the 2-phase 8/6 machine by ~11%
Umax and the induced back-EMF per turn e0 in the windings, the in theory under the same conditions. Finally, the experiments
phase winding turns nph can be determined primarily. on the prototype of the 12/10-pole FSPM motor is compared
with the predicted result from FE analysis and studied by
Sixthly, adjusting the current density or slot packing factor, harmonics analysis, indicating that the PM flux-linkage and
the torque capability and the flux-weakening capability of the back-EMF waveforms of FSPM machine are essentially
FSPM machine can both be adjusted. sinusoidal with negligible harmonics, which is most suitable
Seventhly, by optimization of the original design, the for brushless AC operation.
machine can be designed to satisfy the specifications finally.
ACKNOWLEDGMENT
V. EXPERIMENTS VALIDATION Wei Hua thanks the University of Sheffield for providing a
As mentioned above, the design is based on the operation one-year visiting studentship.
principle, assuming the FSPM machine is essentially a PM
synchronous motor with the sinusoidal PM flux-linkage and REFERENCES
back-EMF. Hence, the validation of the static characteristics,
[1] S. E. Rauch and L. J. Johnson, “Design principles of flux-switching
especially the back-EMF waveform is necessary. Fig. 6 shows alternators,” AIEE Trans. Vol. 74III, pp. 1261-1268, 1955.
the stator and unskewed rotor. Fig. 7 is the predicted phase PM [2] E. Hoang, A. H. Ben-Ahmed, J. Lucidarme, “Switching flux permanent
flux-linkage of the 12/10-pole FSPM motor. Obviously, the magnet polyphased synchronous machines,” Proc. 7th Eur. Conf. Power
waveform is close to ideal sinusoidal distribution versus the Electron. and Applicat., vol. 3, pp. 903-908, 1997.
rotor position. In addition, the waveform reveals the bipolar [3] E. Hoang, M. Gabsi, M. Lecrivain, and B. Multon, “Influence of
magnetic losses on maximum power limits of synchronous permanent
polarity as analyzed above. magnet drives in flux-weakening mode,” Proc. IEEE Ind. Applicat.
Fig. 8 is the corresponding phase back-EMF waveform at Society Annu. Conf., vol. 1, pp. 299-303, 2000.
[4] Z. Q. Zhu, Y. Pang, Howe, D., S. Iwasaki, R. Deodhar, and A. Pride,
1200rpm, including the predicted and measured results. “Analysis of electromagnetic performance of flux-switching permanent
Obviously, the measured result is in a good agreement with the magnet machines by non-linear adaptive lumped parameter magnetic
predicted one in both the waveform and peak value. Further, circuit model,” IEEE Trans. Magn., vol. 41, no. 11, pp. 4277-4287,
Fig. 8(c) shows the harmonics analysis on the back-EMF 2005.
waveform. The harmonic components are compared with the [5] W. Hua, Z. Q. Zhu, M. Cheng, Y. Pang, and D. Howe, “Comparison of
flux-switching and doubly-salient permanent magnet brushless
fundamental one, from which it can be seen that the harmonic machines,” Proc. 8th International Conf. on Electrical Machines and
components are so small due to the THD (total harmonic System, vol. 1, 2005, pp. 165-170.
distortion) is only 2.136%, that they can be neglected in the [6] W. Hua, M. Cheng, X. Zhu and J. Zhang, “Investigation of end-effect in
mathematics modeling, confirming that the back-EMF brushless machines having magnets in the stator with doubly salient
waveform is sinusoidal. structure,” Proc. International Magnetic Conf. (Intermag2006), vol. 1,
2006, pp. 197.
[7] Z. Q. Zhu, Y. Pang, W. Hua, M. Cheng, D. Howe, “Investigation of end-
VI. CONCLUSION effect in PM brushless machines having magnets in the stator”, Journal
of Applied Physics, vol. 99, no.8, 08R319, April 2006.
In this paper, the basic topology and operation principle of [8] Y. Amara, E. Hoang, M. Gabsi, M. Lecrivain, and S. Allano, “Design
the FSPM machine is introduced firstly by assuming that the and comparison of different flux-switch synchronous machines for an
PM flux-linkage waveform is bipolar and sinusoidal aircraft oil breather application,” Proc. 2nd IEEE International Conf. on
distribution in terms of rotor position. Then, the special design Signals, Systems, Decision and Information Technology, pp. 26-26,
procedure of a 12/10-pole FSPM motor under the constraints of 2003.
[9] W. Hua, M. Cheng, X. Zhu and J. Zhang, “Optimal design of flux-
inverter limitation and some dimensions as well as electrical switching permanent magnet machine based on FE analysis,” Proc. The
parameters, is proposed step by step, including the initial stator, 12th Biennial IEEE Conference on Electromagnetic Field Computation
rotor and magnets dimensions, winding turns, which are (CEFC2006), pp. 333, 2006.
calculated by iteratively solving the base-speed and current [10] S. Huang, J. Luo, F. Leonardi, T. A. Lipo, “A general approach to sizing
equations according to dq-axis theory. The influence of current and power density equations for comparison of electrical machines,”
IEEE Trans. Ind. Applicat., vol. 34, no. 1, pp. 92-97, 1998.
density and slot packing factor on the flux-weakening [11] M. Y. Lin, M. Cheng, E Zhou, “Design and analysis of a new 12/8-pole
capability is investigated in the design stage. doubly salient permanent-magnet motor,” Proc. 6th International Conf.
on Electrical Machines and System, vol. 1, 2003, pp. 21-25.
Next, to obtain a general design procedure of FSPM [12] T. M. Jahns, “Flux-weakening regime operation of an interior permanent
machine without topology and application background magnet synchronous motor drive,” IEEE Trans. Ind. Applicat., vol. IA-
limitation, a general output power (torque) equation, 23, pp. 681–689, 1987.
consequently, a sizing equation is derived to determine the [13] S. Morimoto, Y. Takeda, and T. Hirasa, “Current phase control methods
initial dimensions for the engineers to design a FSPM machine. for permanent magnet synchronous motors,” IEEE Trans. Power
electron., vol. 5, pp. 133-139, 1990.
In the sizing equation, the traditional D2sila is replaced by [14] Z. Q. Zhu, Y. S. Chen, and D. Howe, “Online optimal flux-weakening
D2sola. Hence, the power (torque) density of different machine control of permanent-magnet brushless ac drives,” IEEE Trans. Ind.
topology can be compared directly by the equation. It reveals Applicat., vol. 36, no. 6, pp. 1661–1668, 2000.
that the 3-phase, 12/10 topology can offer higher power and

2410