152 CES TRANSACTIONS ON ELECTRICAL MACHINES AND SYSTEMS, VOL. 2, NO.

1, MARCH 2018

Fast Design Method of Variable Flux

Reluctance Machines
L.R. Huang, J.H. Feng, S.Y. Guo, J.X. Shi, W.Q. Chu, and Z.Q. Zhu, Fellow, IEEE

Abstract—In this paper, a fast design method is developed Rso Stator outer radius
based on a combination of analytical and finite element (FE) Te Electromagnetic torque
methods for variable flux reluctance machines (VFRMs). Firstly,
Ts, Tr, Tc Synchronous/reluctance/cogging torques
the feasibility of using analytical method in optimization under
unsaturated condition is confirmed. Then, by applying the FE Wa, Wb, Wc Winding functions of phases A, B and C
method, the influence of magnetic saturation is considered. Wf Winding function of field winding
Compared with the unsaturated case, the optimal split ratio for Wt Stator tooth width
magnetically saturated case is increased by 1~1.2 times, the βs, βr Stator and rotor slot opening ratios
optimal rotor pole arc ratio varies within 0.33~0.44, and the stator
θ Mechanical angle
pole arc ratio remains the same. Based on this, the optimal
structural parameters can be initially set by analytical method θs Stator slot pitch
and then refined by the FE method. Due to the fast speed of λ Split ratio
analytical method, less variable counts and narrowed variation Λr1 Magnitude of fundamental rotor radial
ranges, the proposed method is significantly faster than the permenace component
conventional pure FE based global optimization. Finally, the
Λs, Λr Stator and rotor permeance functions obtained
proposed method is used for optimizing the 6-stator-slots VFRMs
having different numbers of rotor poles. The
by single-side saliency model
6-stator-slot/7-rotor-pole (6s/7r) VFRM is found to have the μ0 Vacuum permeability
highest torque density. It is prototyped and tested to verify the I. INTRODUCTION
analyses.

Index Terms—Analytical method, optimal design, torque

density, variable flux reluctance machine.
D UE to the increasing concerns on the price of rare-earth
magnet material and the risk of demagnetization in
permanent magnet (PM) machines [1-3], many magnetless
NOMENCLATURE machines, including induction machines (IMs) [4],
rotor-wound-field synchronous machines (RWFSMs) [5-6],
As Total stator slot area
switched reluctance machines (SRMs) [7-8], synchronous
F Magnetomotive force (MMF)
reluctance machine (SynRMs) [9-10], vernier reluctance
Fa, Ff MMFs of armature and field currents machines (VRM) [11], stator-wound-field flux switching
Fs Modulated MMF machines (SWFFSMs) [12-13], and variable flux reluctance
Fsa, Fsf Modulated MMFs of armature and field machines (VFRMs) [14-18] have been extensively
currents investigated.
g0 Airgap length VFRMs are developed in [14] and [15]. Fig. 1 shows the
hs, hr Yoke thickness of stator and rotor configurations of two typical VFRMs, i.e.
Ia, Ib, Ic Currents of phases A, B and C 6-stator-slot/4-rotor-pole (6s/4r) and 6s/5r VFRMs. They have
If Current of field winding doubly-salient structure, which are similar to that of switched
Irms, Idc RMS values of armature and field currents reluctance machine (SRM), and two sets of concentrated
kT Torque coefficient windings, i.e., AC armature and DC field windings. Apart from
Lstk Machine stack length the advantages inherited from SRMs, such as robust rotor and
Pcu Total copper loss of VFRM compact windings, VFRMs show significantly smaller torque
Pcua, Pcuf Copper losses of armature and field currents ripple and acoustic noise [16], and more flexible rotor pole
Rsi Stator inner radius number selection [17-18] than the SRMs. Moreover, the
stator-located winding structure avoids the requirement of
slip-rings/brushes and the heat can be easily dissipated from the
This article was submitted for review on 26, January, 2018.
L.R. Huang and Z.Q. Zhu (corresponding author) are with the Electrical
stator. All these merits extend the application of VFRMs.
Machines and Drives Group, University of Sheffield, Sheffield S1 3JD, U.K. In order to obtain the highest torque density, design methods
(e-mail: lhuang18@sheffield.ac.uk; z.q.zhu@sheffield. ac.uk). of VFRMs are important. In [19], the torque density of a
J.H. Feng, S.Y. Guo, J.X. Shi and W.Q. Chu are with the CRRC Zhuzhou
48s/40r VFRM is proved to be closely related to the rotor pole
Institute Co. Ltd, Shidai Road, Shifeng District, Zhuzhou, Hunan, China
(e-mail: fengjh@csrzic.com; guosy@csrzic.com; shijx2@csrzic.com; arc and rotor tooth height. In [20], four 6-stator-pole VFRMs
wenqiang_chu@dynexsemi.com;).
HUANG et al: FAST DESIGN METHOD OF VARIABLE FLUX RELUCTANCE MACHINES 153

with 4, 5, 7 and 8-rotor poles are globally optimized. It is Regarding the stator and rotor radial permeance, both of
concluded that the maximum torque is achieved when the 3 DC
DC 2
optimal rotor pole arc to pole pitch ratio is around 1/3 and the B1
C1
Stator
stator pole arc is equal or slightly smaller than rotor pole arc. Rotor
Then, a weighted evaluation function is introduced in [21] to
take the torque density, torque ripple, cogging torque, power

DC4

DC1
A2

A1
factor, field winding voltage fluctuation and copper
Armature
consumption into account during the optimization of stator Field winding
winding
wound field synchronous machine. However, in all these C2
B2
DC

existing works, the optimization purely relies on the finite

5 6
DC

element (FE) based global optimization method, which is (a)

known to be time consuming. C3
DC
2
D
In this paper, a fast optimization method is developed by the B1
C1

Stator
synergy of analytical and FE methods. In Section II, the Rotor
feasibility of using analytical method in machine optimization
under magnetically unsaturated condition is verified. In Section

DC4

DC1
A2

A1
III, the influence of magnetic saturation on the optimal
structural parameters is investigated by FE analyses. Based on Armature Field winding
winding
this, a fast optimization method is developed in Section IV. It is C2
B2
DC

then employed to optimize the 6-stator-pole VFRMs having

5 6
DC

different numbers of rotor poles to verify its capability. Finally, (b)

the experimental validation on a 6s/7r VFRM is presented in Fig. 1. Cross sections and winding configurations of the 6s/4r and 6s/5r
Section V. VFRMs. (a) 6s/4r VFRM. (b) 6s/5r VFRM
3
FEA Analytical
Stator radial permeance (10-3)
II. ANALYTICAL OPTIMIZATION METHOD
2.5
In this section, the optimization by analytical method is β s = 0.44
2
discussed. Since the analytical method can only be applicable
to the linear case (the permeability of cores is set to infinity), 1.5
β s = 0.67
β s = 0.56
the influence of magnetic saturation will be investigated later in 1
Section III.
0.5 β s = 0.78
A. Analytical torque calculation model
0
Based on the Lorentz force law, the instantaneous torque 20 0 30 10
40 50 60
expression of VFRMs is given by (1). The detailed derivation Mechnical angle (deg.)
procedure can be found in [22]. Fig. 2. Distributions of analytically and FE predicted stator radial permeances
2π for different stator slot opening ratios (Rsi=22.5mm, g0=0.5mm, θs=60deg.).
Te ( t ) =
− Rsi Lstk ∫ Fs (θ , t ) Λ r (θ , t ) dFs (θ , t ) (1)
0 TABLE I
where Rsi is the radius of stator inner surface; Lstk is the machine MAIN SPECIFICATIONS OF 6S/4R AND 6S/5R VFRMS
stack length; θ is the mechanical angle in the stator reference Parameter Unit
VFRM
6s/4r 6s/5r
frame; Λr is the rotor radial permeance function obtained by
Stator outer diameter mm 90
salient rotor and slotless stator model; and Fs is the “Modulated Airgap length mm 0.5
MMF” defined by Total copper loss W 30
=Fs (θ , t ) F (θ , t ) g 0 Λ s (θ ) µ0 (2) Turns per coil (AC/DC) - 144/144
Split ratio - 0.5 0.52
where g0 is the airgap length; μ0 is the vacuum permeability; Λs Stator pole arc deg. 27 24
(θ, t) is the stator radial permeance obtained by smooth rotor Rotor pole arc deg. 34.6 26
and slotted stator model; and F(θ, t) is the MMF generated by
1.2
the armature and field windings, i.e.
F= (θ , t ) Fa (θ , t ) + Ff (θ , t ) (3) 1
Output torque (Nm)

where Fa (θ, t) and Ff (θ, t) are the MMF functions [23] of 0.8
armature winding and field winding, respectively. They can be 0.6
6s/5r
deduced by the product of corresponding winding functions 6s/4r
and excitations, i.e. 0.4

 Fa (θ , t) = Wa (θ ) I a ( t ) + Wb (θ ) I b ( t ) + Wc (θ ) I c ( t ) 0.2 FEA Analytical

 (4)
 Ff (θ , t) = W f (θ ) I f ( t ) 0
120 1800 240 60 300 360
where Wa (θ), Wb (θ), Wc (θ) and Wf (θ), Ia (t), Ib (t), Ic (t) and If (t)
Rotor position (Elec. deg.)
are the winding functions and currents of phase A, phase B, Fig. 3. Torque waveforms of optimized 6s/4r and 6s/5r VFRMs predicted by
phase C and field windings, respectively. analytical and FEA methods (Pcu=30W).
154 CES TRANSACTIONS ON ELECTRICAL MACHINES AND SYSTEMS, VOL. 2, NO. 1, MARCH 2018

them can be calculated by the analytical model for single-sided Split ratio Lever arm Rg
saliency motor [24]. Taking the stator radial permeance for
example, its distribution under one stator slot pitch is Slot area Excitation Fs
Stator pole arc
) µ0  g0 + g (θ )
Λ s (θ=
 Stator peameance
  π Rsi sin (θ ) sin ( β sθ s 2 − θ 2 ) Stator yoke thickness
  , θ ∈ [ 0, β sθ s ] (5)
 ( )  2 sin ( β sθ s 4 ) cos (θ 2 − β sθ s 4 )
g θ = Global saturation
Rotor yoke thickness
  0 , θ ∈ [ β sθ s , θ s ]
 
Rotor pole arc Rotor permeance Λr
where βs is the stator slot opening ratio and θs is the slot pitch.
Fig. 4. Relationship between optimization parameters and torque equation
A comparison between analytically and FE predicted stator variables.
radial permeance distributions over one stator slot pitch for
different stator slot opening ratios is presented in Fig. 2. Good 0.8
0.8
0.8
accuracy is found for the analytical method.

Average torque (Nm)

Average torque (Nm)

0.75
0.77
0.75
Similarly, the rotor radial permeance function can also be 0.7 0.7 0.73
deduced but is not presented here for simplicity. 0.65 0.65
0.6
0.7
0.6
After obtaining the permeance and MMF distributions, the 40
Sta 35 3 50
40
Sta 35 3 50 0.67
tor 0 45 tor 0 45
instantaneous torque of VFRMs can be predicted analytically. (M tooth 25
ech w
. de idth 20 30 Ro
40
35 tor pole a
rc
g.)
(M tooth 25
ech w
. de idth 20 30 Ro
40
35 tor pole a
e
rc
g.) 0.63
h. d e g.) h. d
For verification, the 6s/4r and 6s/5r VFRMs are chosen as
g.) (Mec
(Mec
0.6
(a) (b)
examples. Their main specifications are listed in Table I. Fig. 3
0.9
compares the analytically and FE predicted torque waveforms. 0.9 0.9

Average torque (Nm)

Average torque (Nm)
Good agreement is observed, indicating the accuracy of using 0.8 0.8
0.85

analytical method in torque estimation under magnetic 0.7 0.7 0.8

0.6 0.6
unsaturation condition 0.5
35
0.5
35
0.75

Sta 30 2 35 Sta 30 2 0.7

B. Optimal AC/DC ratio for maximum torque tor 5
(M tooth 20 25
30
rc
tor 5
(M tooth 20 25
30
rc
35
ech w 20 tor pole a ech w 20 tor pole a 0.65
. de idth 15 15 Ro g.) . de idth 15 15 Ro
Further, by substituting (2) and (3) into (1), the torque g.)
(Mec
h. d e g.)
(M e c h. de
g.)

0.6
equation can be divided into three components: synchronous (c) (d)
Fig. 5. Variation of average torques against stator and rotor pole arcs for 6s/4r
torque Ts, reluctance torque Tr and cogging torque Tc, as shown and 6s/5r VFRMs (Split ratio=0.5). (a) Analytical results of 6s/4r VFRM. (b)
in (6). FE results of 6s/4r VFRM. (c) Analytical results of 6s/5r VFRM. (d) FE results
2π Λ (θ , t ) dF (θ , t ) of 6s/5r VFRM.
2

Te ( t ) =
− Rsi Lstk ∫ r sa
− ① Tr
0 2 dθ 1.2
2π Λ (θ , t ) dFsf (θ , t )
2

Rsi Lstk ∫ r
− ② Tc (6) 1
0 2 dθ
d  Fsa (θ , t ) Fsf (θ , t ) 
Per unit value

0.8
Rsi Lstk ∫ Λ r (θ , t ) 
2π
③ Ts
0 dθ 0.6
Optimal value
where Fsa and Fsf are modulated MMFs of armature and field 1st rotor radial permeance harmonic
0.4
currents, respectively. 6s/4r-Average torque
Based on the harmonic analysis, the torque principle of 0.2 6s/5r-Average torque
VFRMs is comprehensively illustrated with the concept of 0
magnetic gearing effect in [22]. It is found that the average 0.3 0.1 0.4 0.2
0.5 0.6
torque of VFRMs is mainly generated by synchronous torque Rotor pole arc ratio
Fig. 6. Variation of 1st rotor radial permeance harmonic and average torques of
and can be concisely expressed by:
6s/4r and 6s/5r VFRMs against rotor pole arc ratio under linear condition.
Te _ avg ≈ Ts _=
avg kT Rsi Lstk N r Λ r1 I rms I dc (7)
where kT is a coefficient determined by the stator radial TABLE II
CONSTRAINTS FOR OPTIMIZATION
permeance and winding functions; Λr1 is the magnitude of the Parameter Symbol Unit Value
1st rotor radial permeance harmonic; Irms and Idc are the rms Stator outer radius Rso mm 45
values of armature and field currents, respectively. Stator/rotor yoke thickness hs / hr mm ≥5
Assuming the total copper loss (the sum of armature copper Airgap length g0 mm 0.5
Total copper loss Pcu W 30
loss Pcua and field copper loss Pcuf) of VFRM, Pcu, is a constant, Turns per coil (AC/DC) na/nf - 144/144
i.e. Stack length Lstk mm 25
2
Pcu = Pcua + Pcuf = I rms Ra + I dc2 R f (8)
TABLE III
where Ra and Rf are the total resistances of armature and field OPTIMAL STRUCTURAL PARAMETERS FOR 6S/4R AND 6S/5R VFRMS UNDER
windings, respectively. LINEAR CONDITION
Then, the torque equation (7) can be rewritten as: Split ratio Stator tooth arc ratio Rotor tooth arc ratio
(Analytical/FEA) (Analytical/FEA) (Analytical/FEA)
Te _ avg = ke Rsi Lstk N r Λ r1 I rms (P
cu
2
− I rms Ra ) R f (9) 6s/4r 0.46 / 0.46 0.46 / 0.45 0.44 / 0.44
In order to maximize the average torque, dTe_avg/dIrms=0: 6s/5r 0.48 / 0.48 0.4 / 0.4 0.44 / 0.44
HUANG et al: FAST DESIGN METHOD OF VARIABLE FLUX RELUCTANCE MACHINES 155

( P − I 2 R )1/2 −  alleviating the magnetic saturation. Compared with the linear

ke Rsi Lstk N r Λ r1 N r  =
cu rms a
dTe _ avg dI rms = 0 case, the optimal split ratio will be increased by 1~1.2 times
 2 −1/2 
I R ( P −
 rms a cu rms aI 2
R )  depending on the load condition.
2 (10) 0.65
⇒ P=
cu 2 I rms R
=a 2 Pcua 6s/4r-nonlinear 6s/4r-linear
⇒ Pcuf =Pcu − Pcua =Pcua
0.6 6s/5r-nonlinear 6s/5r-linear
Therefore, the largest average torque/copper loss can be

Split ratio
obtained when the copper losses of field and armature windings 0.55
are the same, i.e. Pcua=Pcuf. Further, if the field and armature nonlinear
windings share the same slot area and turns number, the optimal 0.5 linear
AC/DC ratio is Irms/Idc=1. This conclusion is confirmed by FE
in [15] and will be used for all the optimizations in this paper. 0.45
0 10 20 30 40 50 60 70 80 90 100 110 120 130
C. Analytical optimization method
Copper loss (W)
From the torque expression (1), it can be seen that Lstk, Fs and (a)
Λr are three variables directly related to output torque. Hence, 1.3
6s/4r-nonlinear
five structural parameters, i.e., split ratio, stator and rotor pole 6s/4r-linear
arcs, stator/rotor yoke thicknesses, need to be optimized due to 1.2 6s/5r-nonlinear

Split ratio (p.u.)

6s/5r-linear
their close relationships with the aforementioned three
variables, as shown in Fig. 4. Since the core saturation is 1.1
neglected for analytical method, the stator/rotor yoke
thicknesses are kept minimum according to the mechanical 1
requirement and will be optimized later with the FE method. Rated condition
With the design constraints listed in Table II, the optimal 0.9
10 20 30 40 50 60 70 80 90 100 110 120 130
0
split ratio and stator/rotor pole arc ratios can be obtained with
Copper loss (W)
parametric calculation using the analytical model, as shown in (b)
Fig. 5. Finally, the globally optimized split ratio, stator and Fig. 7. Variation of optimal split ratio against copper loss. (a) Optimal value. (b)
rotor pole arc ratios by analytical and FE methods are compared Per unit value
in Table III. Good agreements are found between analytical and
FE results, indicating the feasibility of using analytical method As ↓ I ∝ As ↓ T↓
for machine design under linear condition. λ↑ Saturation ↓ T↑
Rro ↑
Further, it is worth noticing that the optimal rotor pole arc
ratio is found to be 0.44 for all VFRMs under linear case. This Fig. 8. Relationship between split ratio and output torque.
is mainly due to the fact that the average torque of VFRMs is B. Stator pole arc ratio
proportional to the fundamental rotor permeance component The influence of load and magnetic saturation on the optimal
Λr1 (see equation (7)), and Λr1 peaks when the rotor pole arc stator pole arc ratio βs is illustrated in Fig. 9.
ratio is 0.44, as shown in Fig. 4. It is noted that the optimal βs is almost independent of load
III. INFLUENCE OF MAGNETIC SATURATION condition. As shown in Fig. 10, an increase of βs will lead to
In this section, the magnetic saturation is taken into account smaller slot area and larger tooth width Wt, both of which will
by using the nonlinear FE method. Several 6s/4r and 6s/5r alleviate the stator saturation. However, a larger βs also means
VFRMs are designed under different copper loss levels with an increase in average airgap permenace Λs and more severe
pure global optimization method. The obtained optimal magnetic saturation. Hence, βs has insignificant influence on
structural parameters are compared to those calculated by saturation. Its optimal value is almost constant with the increase
analytical method under linear case to investigate the influence of copper loss. In this case, the optimal value of βs for linear
of core saturation. case is also applicable in nonlinear case.
0.5
A. Split ratio
The influence of load and magnetic saturation on the optimal
Stator pole arc ratio

0.45
split ratio λ is shown in Fig. 7. It can be seen that the optimal
split ratio shows an upward trend with the increase of copper 0.4
loss. This can be explained by Fig. 8. On one hand, the increase
of λ will lead to a reduction in slot area As and electric loading 0.35
6s/4r-nonlinear 6s/4r-linear
when the copper loss is fixed. On the other hand, owing to the 6s/5r-nonlinear 6s/5r-linear
reduced current, the magnetic saturation will be alleviated.
0.3
Also, the rotor outer radius Rro becomes longer. A compromise 0
10 20 30 40 50 60 70 80 90 100 110 120 130
should be made between these two aspects and the split ratio Copper loss (W)
tends to increase under larger load condition for the sake of Fig. 9. Variation of optimal stator pole arc ratio against copper loss.
156 CES TRANSACTIONS ON ELECTRICAL MACHINES AND SYSTEMS, VOL. 2, NO. 1, MARCH 2018

As ↓ I ∝ As ↓ Saturation ↓ arc ratios are obtained in Figs. 13(a) and (b), respectively. On
βs ↑ Wt ↑
one hand, the average airgap permeance Λr0 drops with the
decrease of βr. Hence, the magnetic saturation is alleviated. On
Λs ↑ Saturation ↑
the other hand, the decrease of βr will lead to a drop in Λr1, as
Fig. 10. Relationship between stator pole arc ratio and core saturation.
well as the average torque. Considering both aspects, the
C. Rotor pole arc ratio optimal βr will drop from 0.44 to 0.33. Moreover, the more
The influence of load and magnetic saturation on the optimal severe the saturation is, the smaller the optimal rotor pole arc
rotor pole arc ratio βr is illustrated in Fig. 11. The optimal value ratio will be.
shows a downward trend with the increase of copper loss. Its D. Stator/rotor yoke thickness
variation range is between 0.33~0.44.
0.55
Compared with the linear condition, the stator and rotor yoke
6s/4r-nonlinear 6s/4r-linear thicknesses are expected to increase when the cores are
0.5 6s/5r-nonlinear 6s/5r-linear saturated. The optimal yoke thicknesses can be obtained by
Rotor pole arc ratio

global optimization method, as shown in Fig. 14.

0.45 7

Stator yoke thickness (mm)

6s/4r-nonlinear
0.4 nonlinear
6.5 6s/4r-linear
linear 6s/5r-nonlinear
0.35 6 6s/5r-linear

0.3 nonlinear
5.5
0 10 20 30 40 50 60 70 80 90 100 110 120 130 linear
Copper loss (W) 5
Fig. 11. Variation of optimal rotor pole arc ratio against copper loss.
4.5
Λr0 ↓ Saturation ↓ T↑ 0 10 20 30 40 50 60 70 80 90 100 110 120 130
βr ↓ Copper loss (W)
Λ r1 ↓ T↓ (a)
Fig. 12. Relationship between rotor pole arc ratio and output torque. 9
6s/4r-nonlinear
Rotor yoke thickness (mm)

1.2 8 6s/4r-linear
Rotor radial permeance (p.u.)

βr=1/6 βr=1/3 βr=1/2 6s/5r-nonlinear

1
7 6s/5r-linear
0.8
6 nonlinear
0.6 linear
5
0.4

0.2 4
0 10 20 30 40 50 60 70 80 90 100 110 120 130
0 Copper loss (W)
0 60 120 180 240 300 360 (b)
Rotor position (Mech. deg.) Fig. 14. Variation of optimal stator and rotor yoke thicknesses against copper
(a) loss. (a) Stator yoke thickness. (b) Rotor yoke thickness.
0.7
IV. FAST OPTIMIZATION METHOD
Rotor radial permeance (p. u.)

0.6
0.5 Based on the revealed influence of magnetic saturation on
the optimal structural parameters, a fast optimization method is
0.4
developed by combining analytical and FE methods.
0.3 DC component
Linear optimal value
0.2 1st component A. Conventional optimization method
0.1 For conventional method, the torque calculation relies fully
on FE method. The global optimization module of ANSYS
0
0.40.2 0.5 0.3 0.6 Maxwell 15.0 can be used to get the optimal specification. The
Rotor pole arc ratio procedure is:
(b) Step 1: The parametric optimization is firstly used to get the
Fig. 13. On-load rotor radial permenace distributions and variations of dc and
1st harmonics against rotor pole arc ratio for 6s/4r VFRM (Pcu=30W). (a) Rotor
rough variation range of all the structural parameters.
permeance distributions. (b) Variations of dc and 1st components. Step 2: All five structural variables are further globally
optimized using the genetic algorithm.
This is mainly due to the influence of βr on the rotor In this case, 2~3 days are usually required.
permeance distributions, as shown in Fig. 12. By using the
on-load permeance calculation method developed in [22], the B. Proposed fast optimization method
rotor permeance distributions of different rotor pole arc ratios The procedure of proposed fast optimization method is:
and the variations of the dc and 1st components with rotor pole Step 1: Set the rotor pole arc ratio as 0.44 (optimal value
HUANG et al: FAST DESIGN METHOD OF VARIABLE FLUX RELUCTANCE MACHINES 157

obtained from the analytical model) and stator/rotor yoke

thicknesses as minimum values for mechanical consideration,
the stator pole arc ratio and split ratio are optimized using
analytical method.
Step 2: Globally optimize split ratio (1~1.2 times of linear
optimal value), rotor pole arc ratio (0.33~0.44) and stator/rotor
yoke thicknesses, whereas the stator pole arc ratio is fixed as
the obtained optimal value of Step 1.
By using the fast optimization method, only 6~12 hours are (a) Prototype
required to obtain the optimal results, which are much shorter
than the conventional method. The low time consumption is
mainly benefited from the fast calculation speed of analytical
method, less variable counts and narrowed variation ranges, as
shown in Table IV.
TABLE IV Load machine Torque transducer VFRM
COMPARISON OF CONVENTIONAL AND PROPOSED METHODS
Step Conventional method Proposed method
Parametric optimization Analytical optimization (b) Test rig
Split ratio Split ratio Fig. 16. Prototype and test rig of 6s/7r VFRM. (a) Prototype. (b) Test rig.
I Stator pole arc ratio Stator pole arc ratio
TABLE V
Rotor pole arc ratio Rotor pole arc ratio (fixed to 0.44)
MAIN SPECIFICATIONS OF PROTOTYPE 6S/7R VFRM
(3 variables, 1~2 hours) (2 variables, 1~2 mins)
Parameter Value Parameter Value
Global optimization Global optimization
Number of phases 3 Stator outer diameter 90mm
Split ratio Split ratio (1~1.2 times of step I)
DC-bus voltage 48V Airgap length 0.5mm
Stator pole arc ratio Stator pole arc ratio (identical to step I)
Rated speed 400rpm Split ratio 0.56
II Rotor pole arc ratio Rotor pole arc ratio (0.33~0.44)
Rated power 70W Stator pole arc ratio 0.31
Stator yoke thickness Stator yoke thickness
Rated torque 0.82Nm Rotor pole arc ratio 0.366
Rotor yoke thickness Rotor yoke thickness
Stack length 25mm Turns per coil (AC/DC) 144/144
(5 variables, 2~3 days) (4 variables, 6~12 hours)

C. Optimization of VFRMs with different stator/rotor pole
number combinations
Further, the developed fast optimization method is applied to
the optimization of the 6-stator-slot VFRMs with 2~20 rotor
pole numbers according to the constraints listed in Table I.
Their torque capabilities are shown in Fig. 15. Good agreement
is found between the results of conventional and proposed
methods while the proposed method can obtain the optimal
design in much shorter time. Moreover, the 6s/7r and 6s/11r
VFRMs are found to have the highest torque density.
Considering the fact that 6s/7r VFRM has a much lower
electrical frequency than 6s/11r VFRM under the same rotating
speed, 6s/7r is the preferred stator/rotor pole combination under
investigated specification.
1.2
1 and turns number, the RMS current of armature winding is
Average torque (Nm)
It can be seen that both the armature and field windings are
concentrated types and wounded on all the stator teeth. During
the experiment, the field winding is excited by a DC supply
while the armature winding is connected to an inverter.
When the machine is operating at open-circuit and only the
field windings are excited, the phase back-EMF can be
measured, as shown in Fig. 17. Two different field currents are
tested and good agreement can be found between FEA and
experimental results. Although some harmonics can be
observed from the back-EMF waveforms, they are close to
sinusoidal for the 6s/7r VFRM.
Regarding the torque performance, the static torque is
measured when both armature and field windings are excited by
DC current with the following relationship: Idc=Ia=-2Ib=-2Ic.
Since the armature and field windings share the same slot area

equal to that of field winding to achieve the maximum torque.

0.8 Again, the measurements match well with the FEA results, as
0.6 can be seen from Fig. 18.
Conventional method Then, the average torque of the prototype is also measured
0.4
under different load condition, as presented in Fig. 19. It can be
Proposed method
0.2 found that good agreement is found between FEA and
0 measured results under low copper loss condition. When the
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 copper loss is around rated condition, the measured average
Rotor pole number torque is slightly smaller than the FEA prediction. This is
Fig. 15. Variation of the average torque against rotor pole numbers for mainly due to the flux leakage of end windings.
6-stator-slot VFRMs with different rotor pole numbers (Pcu=30W).
Finally, the torque ripple performance of the prototype is
V. EXPERIMENTAL VERIFICATION measured when copper loss is 20W, as presented in Fig. 20. The
For experimental verification, a 6s/7r VFRM is prototyped, torque profile is not fluctuating in ideal periodic way. This is
as shown in Fig. 16. Its main specification is listed in Table V. mainly due to the measurement error and disturbance from the
158 CES TRANSACTIONS ON ELECTRICAL MACHINES AND SYSTEMS, VOL. 2, NO. 1, MARCH 2018

load machine. Nevertheless, the peak-to-peak values of VI. CONCLUSION

measurement and FEA prediction match with each other, In this paper, a fast optimization method is developed for
verifying the feature of small torque ripple in 6s/7r VFRM [17]. VFRMs by a combination of analytical and FEA methods. The
capability and accuracy of analytical method in torque
2
FEA estimation under linear situation is firstly confirmed. The
optimal rotor pole arc of linear case is found to be 0.44 for all
Phase back-EMF (V)

Measurement
1
the VFRMs. Then, the influence of saturation effect on optimal
Idc= 1.3A structural parameters is revealed. It is found that, depending on
0
the load condition, (1) the split ratio will increase by 1~1.2
times of the linear optimal value; (2) the rotor pole arc ratio will
-1
Idc = 2.6A decrease and vary within 0.33~0.44; (3) the stator pole arc will
remain the same as its linear optimal value. Based on this, a fast
-2
0 60 120 180 240 300 360
optimization method is developed. Its capability is verified on
Rotor position (Elec. deg.) the 6-stator-slot VFRMs having different rotor poles. Finally, a
6s/7r VFRM is prototyped and tested to verify the analyses.
Fig. 17. FEA predicted and measured phase back-EMFs at 400rpm for 6s/7r
VFRM.
REFERENCES
1
FEA [1] A. M. EL-Refaie, “Fractional-slot concentrated-windings synchronous
permanent magnet machines: opportunities and challenges,” IEEE Trans.
Static torque (Nm)

0.5 Measurement Ind. Electron., vol. 57, no. 1, pp. 107–121, Jan. 2010.
[2] I. Boldea, L. N. Tutelea, L. Parsa, and D. Dorrell, “Automotive electric
0 propulsion systems with reduced or no permanent magnets: an overview,”
Idc = 1.3A IEEE Trans. Ind. Electron., vol. 61, no. 10, pp. 5696-5711, Oct. 2014.
[3] D. Dorrell, L. Parsa, I. Boldea, “Automotive electric motors, generators,
-0.5 and actuator drive systems with reduced or no permanent magnets and
Idc = 2.6A
innovative design concepts,” IEEE Trans. Ind. Electron., vol.61, no.10,
-1 pp.5693-5695, Oct. 2014.
0 60 120 180 240 300 360 [4] D. G. Dorrell, A. M. Knight, L. Evans, and M. Popescu, “Analysis and
Rotor position (Elec. deg.) design techniques applied to hybrid vehicle drive machines—Assessment
of alternative IPM and induction motor topologies,” IEEE Trans. Ind.
Fig. 18. FEA predicted and measured static torques when (Idc=Ia=-2Ib=-2Ic) for Electron., vol. 59, no. 10, pp. 3690–3699, Oct. 2012.
6s/7r VFRM. [5] C. Rossi, D. Casadei, A. Pilati, and M. Marano, “Wound rotor salient pole
synchronous machine drive for electric traction,” in Conf. Rec. IEEE IAS
1 Annu. Meeting, 2006, pp. 1235–1241.
FEA [6] W. Chu, Z.Q. Zhu, J. Zhang, X. Liu, D. Stone, and M. Foster,
0.8
Average torque (Nm)

Measurement
“Investigation on operational envelops and efficiency maps of electrically
excited machines for electrical vehicle applications,” IEEE Trans. Magn.,
0.6 vol. 51, no. 4, pp. 8103510, Apr. 2015.
[7] K. Kiyota and A. Chiba, “Design of switched reluctance motor
0.4 competitive to 60 kW IPMSM in third generation hybrid electric vehicle,”
IEEE Trans. Ind. Appl., vol. 48, no. 6, pp. 2303–2309, Nov./Dec. 2012.
0.2
[8] A. Chiba, Y. Takano, M. Takeno, T. Imakawa, N. Hoshi, M. Takemoto,
and S. Ogasawara, “Torque density and efficiency improvements of a
0 switched reluctance motor without rare earth material for hybrid vehicles,”
0 5 10 15 20 25 30 IEEE Trans. Ind. Appl., vol. 47, no. 3, pp. 1240–1246, May/Jun. 2011.
Copper loss (W) [9] Z. Azar and Z. Q. Zhu, “Performance analysis of synchronous reluctance
Fig. 19. FEA predicted and measured average torques under different copper machines having nonoverlapping concentrated winding and sinusoidal
loss for 6s/7r VFRM. bipolar with DC bias excitation,” IEEE Trans. Ind. Appl., vol.50, no.5, pp.
3346-3356, 2014.
0.6 [10] S. Ooi, S. Morimoto, M. Sanada, and Y. Inoue, “Performance evaluation
of a high power density PMASynRM with ferrite magnets,” IEEE Trans.
0.5 Ind. Appl., vol. 49, no. 3, pp. 1308–1315, May/Jun. 2013.
0.4 [11] S. Tabi, A. Tounzi and F. Piriou, “Study of a stator current excited vernier
Torque (Nm)

reluctance machine,” IEEE Trans. Energy Convers., vol. 21, no. 4, pp.
0.3 823-831, 2006.
[12] A. Zulu, B.C. Mecrow, and M. Armstrong, “A wound-field three-phase
0.2 flux-switching synchronous motor with all excitation sources on the
FEA Measurement stator,” IEEE Trans. Ind. Appl., vol.46, no.6, pp. 2363-2371, Nov./Dec.
0.1
2010.
0 [13] J.T. Chen, Z.Q. Zhu, S. Iwasaki, and R. Deodhar, “Low cost
0 60 120 180 240 300 360 fluxswitching brushless AC machines,” in Proc. Vehicle Power and Prop.
Rotor position (Elec. deg.) Conf. (VPPC2010), pp. 1-6, Sep. 1-3, 2010.
[14] T. Fukami, Y. Matsuura, K. Shima, M. Momiyama, and M. Kawamura,
Fig. 20. FEA predicted and measured torque ripples for 6s/7r VFRM (Copper “Development of a low-speed multi-pole synchronous machine with a
loss=20W). field winding on the stator side,” in Proc. Int. Conf. Elect. Mach., Rome,
Italy, 2010, pp. 1–6.
HUANG et al: FAST DESIGN METHOD OF VARIABLE FLUX RELUCTANCE MACHINES
[15] X. Liu, Z. Q. Zhu and Z. P. Pan, “Analysis of electromagnetic torque in
sinusoidal excited switched reluctance machines having DC bias in
excitation,” in Proc. Int. Conf. Elect. Mach., Sept. 2012, pp. 2882-2888.
[16] X. Liu, Z. Q. Zhu, M. Hasegawa, A, Pride, and R. Deodhar, “Vibration
and noise in novel variable flux reluctance machine with DC-field coil in
stator,” in Proc. Int. Conf. Power Electron. and Motion Control, Jun.
2012, pp.1100-1107.
[17] X. Liu and Z. Q. Zhu, “Electromagnetic performance of novel variable
flux reluctance machines with DC-field coil in stator,” IEEE Trans.
Magn., vol. 49, no. 6, pp. 3020-3028, Aug. 2012.
[18] X. Liu and Z. Q. Zhu, “Stator/rotor pole combinations and winding
configurations of variable flux reluctance machines,” IEEE Trans. Ind.
Appl., vol. 50, no. 6, pp. 3675–3684, Nov. 2014.
[19] T. Fukami, Y. Matsuura, K. Shima, M. Momiyama, and M. Kawamura,
“A multi-pole synchronous machine with nonoverlapping concentrated
armature and field winding on the stator,” IEEE Trans. Ind. Electron., vol.
59, no. 6, pp. 2583–2591, Jun. 2012.
[20] J. T. Shi, X. Liu, D. Wu, and Z. Q. Zhu, “Influence of stator and rotor pole
arcs on electromagnetic torque of variable flux reluctance machines,”
IEEE Trans. Magn., vol. 50, no. 11, pp. 1-4, Nov. 2014.
[21] J. Shaofeng, Q. Ronghai, and L. Jian, “Design considerations and
parameter optimization of stator wound field synchronous machines
based on magnetic the gear effect,” in Proc. ECCE, Sept. 2015, pp.
5195-5202.
[22] L.R. Huang, J.H. Feng, S.Y. Guo, J.X. Shi, W.Q. Chu, and Z.Q. Zhu,
“Analysis of torque production in variable flux reluctance machine,”
IEEE Trans. Energy Convers., vol. 32, no.4, pp. 1297-1308, Apr. 2017.
[23] T. Lipo, Analysis of Synchronous Machines. CRC Press, 2012.
[24] B. Gaussens, O. Barriere, E. Hoang, J. Saint-Michel, P. Manfe, M.
Lecrivain, and M. Gabsi, “Magnetic field solution in doubly slotted airgap
of conventional and alternate field-excited switched-flux topologies,”
IEEE Trans. Magn., vol. 49, no. 9, pp. 5083-5096, Sep. 2013.
L.R. Huang received the B.Eng. and M.Sc.
degrees in electrical engineering from
Zhejiang University, Hangzhou, China, in
2012 and 2015, respectively. Since 2015, he
has been working toward the Ph.D. degree
in the Department of Electronic and
Electrical Engineering, University of
Sheffield, U.K.
His major research interests include design and application of
reluctance machines and permanent magnet machines.
J.H. Feng (S’06) received his B.S. and M.S.
degrees in Electrical Machinery Control
from Zhejiang University, China in 1986
and 1989, respectively, and Ph. D. degree in
Control Theory and Control Engineering
from Central South University, China in
2008. Since 1989, he has been with CRRC
Zhuzhou Institute Co. Ltd., Zhuzhou, China,
where he is presently the Vice President and Chief Technology
Officer. He has published a number of journal and conference
proceedings papers. His research interests are modeling,
control, and communication of electrical systems, rail networks
and high-speed trains. He is also a Guest Professor in
Southwest Jiaotong University, Tongji University and Central
South University.
159
S.Y. Guo is a professorial senior engineer.
She got graduated from Central South
University in December 1981, and serves as
the chief technical expert in CRRC Zhuzhou
Institute Co., Ltd. in the field of R&D of the
electric machine systems for railway
locomotive and electrical vehicle
applications.
J.X. Shi received the B. Eng. and M. Sc.
degrees in electrical engineering from South
China University of Technology,
Guangzhou, China in 2010 and 2013,
respectively. Since 2013, he has been with
CRRC Zhuzhou Institute Co., Ltd.
His major research interests include design
and application of permanent magnet
machines for electrical vehicle applications.
W.Q. Chu (SM’16) received the B. Eng.
and M. Sc. degrees in electrical engineering
from Zhejiang University, Hangzhou, China
in 2004 and Huazhong University of
Science and Technology, Wuhan, China in
2007, respectively, and the Ph.D. degree in
the electronic and electrical engineering
from The University of Sheffield, UK, in
2013.
From 2007 to 2009, he was with Delta Electronics (Shanghai)
Co. Ltd. From 2012 to 2014, he was a postdoctoral research
associate with The University of Sheffield. Currently, he is a
principal design engineer with CRRC Zhuzhou Institute Co.
Ltd. His major research interests include the design and
analysis of novel machines for wind power and electrical
vehicle applications.
Z.Q. Zhu (M’90–SM’00–F’09) received the
B.Eng. and M.Sc. degrees in electrical and
electronic engineering from Zhejiang
University, Hangzhou, China, in 1982 and
1984, respectively, and the Ph.D. degree in
electrical and electronic engineering from
The University of Sheffield, Sheffield, U.K.,
in 1991.
Since 1988, he has been with The University of Sheffield,
where he is currently a Professor with the Department of
Electronic and Electrical Engineering, Head of the Electrical
Machines and Drives Research Group, Royal Academy of
Engineering/Siemens Research Chair, Academic Director of
Sheffield Siemens Wind Power Research Centre, Director of
Sheffield CRRC Electric Drives Technology Research Centre.
His current major research interests include the design and
control of permanent-magnet brushless machines and drives for
applications ranging from automotive to renewable energy.