The current issue and full text archive of this journal is available on Emerald Insight at:

[Link]

A robust
A robust numerical method for numerical
drawing the fluid flux-barrier method

geometry in the inner and outer

rotor radial flux synchronous
reluctance machines based on the Received 22 March 2022
Revised 24 August 2022

finite difference method 2 September 2022

Accepted 2 September 2022

Ali Jamali Fard

Electrical Machines and Transformer Research Laboratory (EMTRL),
Amirkabir University of Technology, Tehran, Iran and Department of Electrical
Engineering, Amirkabir University of Technology, Tehran, Iran, and
Mojtaba Mirsalim
Electrical Machines and Transformer Research Laboratory (EMTRL),
Amirkabir University of Technology, Tehran, Iran; Department of Electrical
Engineering, Amirkabir University of Technology, Tehran, Iran and
Engineering Department, St. Mary’s University, San Antonio, Texas, USA

Abstract
Purpose – Rotor shape optimization is crucial in designing synchronous reluctance machines (SynRMs)
because the machine performance is directly proportional to the rotor’s magnetic saliency ratio. The rotor
geometry in synchronous reluctance machines is complex, and many geometrical parameters must be
optimized. When fluid flux-barrier geometry is desirable, using analytic equations to prepare the rotor
geometry for finite element analysis could be tedious. This paper aims to provide a robust numerical
procedure to draw the fluid flux-barrier geometry in transversally laminated radial flux inner and outer rotor
.
SynRMs by directly solving the magnetic vector potential equation using the finite difference method.
Design/methodology/approach – In this paper, the goal is to have a robust procedure for drawing the
rotor geometry for an arbitrary number of slots (Ns), poles (p) and flux-barrier layers (Nfb). Therefore, this
paper targeted several combinations to investigate the performance of the proposed algorithm. The MATLAB
software is used to implement the proposed algorithm. The ANSYS Maxwell software is used for counterpart
finite element simulation to check the correctness of the results derived by the proposed method.
Findings – Several inner and outer rotor SynRMs considering a different number of poles and a different
number of flux-barrier layers per pole are studied to investigate the performance of the proposed algorithm.
Results corresponding to each case are presented, and it is shown that the method is robust, flexible and fast
enough, which could be used for the generation of the rotor geometry for the finite element analysis
effectively.
Originality/value – The value of the proposed algorithm is its simplicity and straightforwardness in its
implementation for the preparation of the rotor geometry with the desired fluid flux-barrier layer curvature
resolution suitable for the finite element analysis. The procedure presented in this paper is based on the ideal
magnetic loading concept, and in future works, a similar idea could be used for linear and axial flux SynRMs. COMPEL - The international
journal for computation and
Keywords Synchronous reluctance machines (SynRMs), Magnetic vector potential (MVP) equation, mathematics in electrical and
electronic engineering
Finite difference method (FDM), Electrical machines © Emerald Publishing Limited
0332-1649
Paper type Research paper DOI 10.1108/COMPEL-03-2022-0123
COMPEL 1. Introduction
Synchronous reluctance machines (SynRMs) are gaining more attention in applications requiring
high dynamics, high torque density and fault-tolerant capabilities, such as traction and household
appliances (Bianchi et al., 2022; L
opez-Torres et al., 2018). Robust rotor structure, absence of rotor
joule loss, low cost and convenience in manufacturing are the advantages of a SynRM (Cai et al.,
2016; Wang et al., 2015). As the SynRM rotor is made of iron lamination, no brushes/slip-rings and
expensive permanent magnets are needed (Cai et al., 2014, 2016). Although electrical machines
with rare-earth permanent magnets have high performance, the high cost and the insufficiency of
rare-earth permanent magnet materials will be obstacles to further development in applications
such as electric vehicles or hybrid electric vehicles (Liu et al., 2016; L
opez-Torres et al., 2018). Line-
start synchronous reluctance motors are also a suitable alternative to induction motors in fixed-
speed direct-on-line applications such as pumps and fans (Baka et al., 2021; Hu et al., 2022).
In a SynRM, the total developed electromagnetic torque is the reluctance type. Therefore,
the rotor shape optimization is crucial in designing SynRMs because the machine
performance is directly proportional to the rotor’s magnetic saliency ratio. In the case of an
improper design, a SynRM presents some drawbacks, such as a high torque ripple and a
low-power factor (Bianchi et al., 2022; Cai et al., 2014). The rotor geometry in SynRMs is
complex, and many geometrical parameters are involved (Bacco and Bianchi, 2019). Several
flux-barrier layers per pole should be taken into account to achieve a good magnetic saliency
ratio (Wang et al., 2017; Guan et al., 2016). This, in turn, makes the rotor optimization process
a difficult task because in an optimization problem, when the number of decision variables is
high, the number of objective function calls increases.
Figure 1 shows some possible rotor topologies to generate reluctance torque for a
continuous rotation. As you can see, these structures aim to generate an appropriate non-
uniform permeability distribution in the rotor to generate reluctance torque. Figure 1(a)
represents the rotor geometry of a switched reluctance machine. In this machine, the rotor
magnetic anisotropy is achieved by salient poles. Figures 1(b) and 1(c) show two topologies
possible for SynRMs. The SynRM’s smoothly rotating field distinguished it from switched
reluctance machine (Miller et al., 1991).
The structure shown in Figure 1(b) is called transversally laminated anisotropy (TLA)
structure because the staking direction is axial. The staking of the rotor shown in Figure 1(c) is in
the radial direction, and it is known as axially laminated anisotropy (ALA) structure. The most
appealing feature of the ALA rotor construction is the opportunity to significantly increase the
number of flux-barrier layers per pole, making it possible to significantly boost the magnetic
saliency of the rotor. This elevated saliency, in turn, increases the machine’s power factor and,
potentially, its efficiency by lowering the required stator current. Unfortunately, the ALA

Figure 1.
Some possible rotor
topologies for radial-
flux reluctance
machines
structure is vulnerable to elevated iron losses in the rotor laminations close to the air gap, limiting A robust
its efficiency advantages. Perhaps even more importantly, the significant manufacturability numerical
challenges associated with the axially laminated rotor construction have seriously hindered its
commercialization opportunities (Pellegrino et al., 2016).
method
This work focuses on TLA structure (more common because of the manufacturing
feasibility). For a TLA rotor structure, several geometries are possible (Figure 2).
Segmented, circular and fluid flux-barrier geometries are explained in Pellegrino et al. (2016).
In general, there are two ways to specify the geometry of a flux-barrier layer [however, a
combination is also possible (Bianchi et al., 2016)]:
(1) predefined methods (segmented, circular, etc.); and
(2) methods based on the magnetic (fluid) flux lines.

In the first method, the designer defines a fully constrained geometry for flux-barrier layers.
In this case, adequate geometrical constraints should be considered to avoid infeasible cases
during the rotor shape optimization process. Other reasonable conditions, which will not
influence the machine performance, are also necessary to decrease the number of design
parameters to simplify the optimization process (Guan et al., 2016). Figures 2(a) and 2(b)
depict two models of this category.
In the second method, the geometry of rotor flux-barrier layers is determined by knowing
the information of magnetic flux lines. In this case, flux-barrier layers are parallel with
magnetic flux lines to achieve a high rotor magnetic saliency ratio. A closed-form expression
of the magnetic flux lines inside the rotor can be derived from the conformal mapping theory
and the Joukowski air-flow potential formulation (Pellegrino et al., 2016). However, the
generation of the rotor geometry using complex analytic equations of the Joukowski method
is complicated. This formulation was originally developed to describe the fluid flow paths
channeled by two infinite plates forming an angle p/p, and with a plug of radius, centered

Figure 2.
Different rotor
geometries of
synchronous
reluctance machines
COMPEL into the origin of the reference frame. However, this method does not provide a good insight
for an electric machine designer.
In this paper, we introduce a new methodology for calculating the geometry of the
flux-barrier layers. The proposed method is based on the ideal magnetic loading
concept in alternating currents (AC) machines and a numeric method [finite difference
method (FDM)] instead of the analytic methods to draw the rotor geometry with a
desirable geometry resolution.

2. Proposed method
2.1 Magnetic vector potential equation
A four-pole inner rotor SynRM and a ten-pole outer rotor SynRM are shown in Figure 3.
Flux lines illustrated in this figure are calculated due to the excitation of the stator
winding when there is no flux-barrier layer in the rotor. The flux lines pattern inside the
solid rotor could be used as an indication of the optimal flux-barrier geometry. Flux
lines are contour lines of the magnetic vector potential (MVP). Because the maximum
variation in the value of the MVP occurs on a path perpendicular to the flux lines, we
have the maximum magneto-motive force drop on the same path. Therefore, flux-
barrier edges should be in parallel with flux lines (as shown in Figure 2b) to minimize
the d-axis flux lines distortion and effectively block the q-axis ones with a beneficial
effect on the rotor anisotropy (Bianchi et al., 2022).
As you can see in Figure 3, for both inner and outer rotor machines, for the
derivation of flux-barrier layers geometry, we should solve the MVP equation in a
hollow cylinder as shown in Figure 4. For a magnetic domain with a linear
permeability and without free currents, the MVP equation is as follows (Jackson,
1999):

r2 A ¼ 0 (1)

In our case, the focus is on one axial cross-section of the machine. Therefore, the problem is a
two-dimensional problem, and we only have Az component; consequently:

r2 A z ¼ 0 (2)

Equation (2) is Laplace equation which in a cylindrical coordinate system is as follows

(Bastos and Sadowski, 2003):

Figure 3.
Flux lines due to the
excitation of the
stator winding
considering a solid
rotor
 A robust
numerical
method

Figure 4.
Solid rotor geometry
and its dimensions

@ 2 Az 1 @Az 1 @ 2 Az @ 2 Az
þ þ 2 þ 2 ¼0 (3)
@r2 r @r r @ u2 @z

Here r, u and z are radial, circumferential and axial components of the coordinate system. Az
is a scalar field. Because the value of Az along the axial direction is constant, its derivative
with respect to z equals 0. Therefore, we have:

@ 2 Az 1 @Az 1 @ 2 Az
þ þ 2 ¼0 (4)
@r2 r @r r @ u2

A z is a function of r and u. Equipotential flux lines are contour lines of this function.
Therefore, for the calculation of flux lines in the hollow cylinder, at first, we should
calculate A z (r, u) by solving equation (4) and then finding its contour lines.

2.2 Boundary conditions

The geometry of our interest is a hollow cylinder in which two boundary conditions should
be specified. The first one is on its inner surface, and the second one is on its outer surface.
The solution variable Az is calculated by knowing the value of the normal and tangential
components of the flux density on boundaries. The flux density is equal to the curl of the
MVP, mathematically:

B¼rA (5)

The curl of A in the cylindrical coordinate system is as follows:




COMPEL
^r
1
@
r^u ^z

@ @

r
@r

rA¼  (6)
@u @z
Ar rAu Az

Because the only non-zero component of A is Az, we have:

^r r^u ^z

1
@ @ @
¼ 1 @A ^r  @A
r
@r
r @u @r
B¼rA¼ 
z z ^u (7)
@ u @z
0 0 Az

According to equation (7), the radial component of flux density is equal:

1 @Az
Br ¼ (8)
r @u

In a radial flux AC machine, the interest is having a sinusoidal distribution of the flux
density in the middle of the air gap. For an ideal p-pole machine, the radial component of
flux density as a function of circumferential coordinate in the middle of the air gap is equal
to:
 
p
Br ðuÞ ¼ Bp  sin u (9)
2

where Bp is the peak value of the flux density in the middle of the air gap. Combining
equations (8) and (9), we have:

Figure 5.
Air gap radius
  
p 1 @Az A robust
Br ðuÞ ¼ Bp  sin u ¼ (10)
2 Rg @ u numerical
method
where Rg is the air gap radius as shown is Figure 5. For inner rotor machines, the air gap
radius is almost equal to Rro, and for the outer rotor machines, it is almost equal to Rri. If we
integrate equation (10), then we can calculate Az as follows:
 
2 p
Az ¼   Rg  Bp  cos u (11)
p 2

Equation (11) is used for determining the value of A z on the outer surface of inner
rotor machines and the inner surface of outer rotor machines. The second boundary
condition is the zero-vector potential boundary condition. It is used to confine the
magnetic flux inside the rotor. Table 1 summarizes boundary conditions for inner and
outer rotor machines.

2.3 Mesh
Figure 6 shows an example of the solution mesh. In this example, the number of mesh layers
in the radial direction equals six. The mesh nodes are equally spaced in both radial and

Boundary Inner rotor Outer rotor

 
Inner surface Az = 0 2 p Table 1.
Az ¼   Rri  Bp  cos u
  p 2 Boundary conditions
Outer surface 2 p Az = 0 for inner rotor and
Az ¼   Rro  Bp  cos u
p 2 outer rotor machines

Figure 6.
An example of the
solution mesh
COMPEL circumferential directions. The number of mesh layers in the radial direction is Nr, and the
number of mesh layers in the circumferential direction is Nt. Therefore, the total number of
mesh nodes equals Nr  Nt. After generating the mesh, we should assign an index to each
mesh node. Indexing starts from innermost to outermost mesh layers by moving
counterclockwise on each mesh layer. The coordinate of the first mesh node is (Rri, 0). The
mesh indexing pattern is shown in Figure 7.
The accuracy and smoothness of calculated flux lines depend on the values of Nt and Nr.
To properly determine the values of Nt and Nr, we define the rotor geometry resolution
(RGR) parameter. The RGR equals the maximum radial or circumferential distance between
two adjacent mesh nodes. The circumferential angular distance between two adjacent nodes
of a mesh layer is equal to:

2p
Du ¼ (12)
Nt

The maximum circumferential distance between two adjacent mesh is for two mesh nodes
which belong to the outermost layer, and it is equal to:

Rro  Du (13)

The distance between two adjacent mesh nodes in the radial direction is equal to:

Rro  Rri
Dr ¼ (14)
Nr  1

According to equations (13) and (14), the RGR is equal to:

RGR ¼ maxðRro  Du; DrÞ (15)

RGR is one of the MATLAB program inputs. After setting the value of RGR, we can
calculate Nt and Nr as follows:
 
2p 2pRro
Rro  Du # RGR ) Rro  # RGR ) Nt ¼ Ceil (16)
Nt RGR

The ceil function is used to fix the value of Nt. Similarly, for Nr, we have:

Figure 7.
Indexing pattern of
mesh nodes
  
Rro  Rri Rro  Rri A robust
Dr # RGR ) # RGR ) Nr ¼ Ceil þ1 (17)
Nr  1 RGR numerical
method
By knowing the values of Nt and Nr, we can calculate Dr and Du, and finally, we can
calculate the coordinates of each mesh node. The value of radial coordinate of i-th mesh
layer nodes is equal to:

ri ¼ Rri þ i  Dr; i ¼ 0; 1;    ; ðNr  1Þ (18)

And the value of circumferential coordinate of the j-th node of a mesh layer is equal to:

uj ¼ j  D u; j ¼ 0; 1;    ; ðNt  1Þ (19)

The relationship between mesh node index and its coordinates is as follows:

Node index ¼ i  Nt þ j þ 1 () ðri ; uj Þ ¼ Node coordinates (20)

2.4 System of equations

As described in the mesh section, the total number of mesh nodes equals N r  N t.
Already we know the value of A z for innermost and outermost mesh layers from
boundary conditions. Therefore, the total number of unknowns is equal to
N r2  N t . We should have N r2  N t linearly independent equations to calculate the
value of Az for all inner mesh nodes uniquely. Figure 8 represents an internal mesh
node with its neighboring points.
There are three differential terms in equation (4). The first term could be discretized
using the second-order difference equation (Morton and Mayers, 2005):

@ 2 Az AFront  2AInner þ ARear

’ z z z
(21)
@r 2
ðDrÞ 2

The second term could be approximated by central difference method:

Figure 8.
An internal mesh
Notes: (1) Inner; (2) rear; node and its
neighboring points
(3) front; (4) left; (5) right
COMPEL 1 @Az 1 AFront  ARear
 ’ Inner  z z
(22)
r @r r 2Dr

where rInner is the radius of inner point. The third term is discretized like the first term:

1 @ 2 Az 1 ALeft  2AInner þ ARight

 ’  z z z
(23)
r 2
@u 2
ðrInner Þ2
ð D uÞ2

Therefore, for each inner mesh node, we can write following equation:

c1 AInner
z þ c2 ARear
z þ c3 AFront
z þ c4 ALeft
z þ c5 ARight
z ¼0 (24)

where:
8
>
> 2 2
>
> c1 ¼  
>
> ðDrÞ 2
ðrInner Þ2 2
>
>  ð D uÞ
>
>
>
> 1 1
>
> c2 ¼ 
>
> 2  rInner  Dr
>
> ðDrÞ2
>
>
< 1 1
c3 ¼ þ (25)
>
> ðDrÞ2 2  rInner  Dr
>
>
>
>
>
> ¼
1
>
> c4
>
> ðrInner Þ2  ðDuÞ
2
>
>
>
>
>
> 1
> c5
> ¼
: ðrInner Þ2  ðDuÞ2

When calculating the coefficients of equation (24) using equation (25), we can
construct the matrix equation (26). The left-hand matrix (stiffness matrix) in equation (26) is
sparse, and we used MATLAB sparse matrix package to assemble and solve it. The
code is pretty fast and can generate the rotor geometry with the desired accuracy
instantly.
Figure 9 shows an internal mesh node, with its corresponding relationship as depicted in
equation (26). After solving matrix equation (26), we can have the value of Az for each mesh
nodes. Therefore, one can calculate the value of Az for any arbitrary point inside the rotor by
two-dimensional interpolation:

Figure 9.
An internal mesh
node, the
corresponding
equation for this
mesh node is shown
in equation (26)
 A robust
numerical
method

(26)

In equation (26), the value of Az for the innermost mesh layer nodes is equal to:

Aiz ¼ AzIBC;i ; i ¼ 1; 2;    ; Nt (27)

where IBC stands for inner boundary condition. For inner rotor machines, we have:

AzIBC;i ¼ 0 (28)

And based on the ideal magnetic loading concept, for outer rotor machines, we have:
 
2 p
AzIBC;i ð
¼   Rri  Bp  cos  i  1  Du Þ (29)
p 2

Similarly, the value of Az for the outermost mesh layer nodes is equal to:

AðzNr 1ÞNt þi ¼ AOBC;i

z ; i ¼ 1; 2;    ; Nt (30)

where OBC stands for outer boundary condition. For inner rotor machines, we have:
 
2 p
AOBC;i
z ¼   R ro  B p  cos  ð i  1 Þ  D u (31)
p 2

And for outer rotor machines:

AOBC;i
z ¼0 (32)

2.5 Per-unit model

Figure 10 shows the rotor of a four-pole SynRM. In this example, the number of
flux-barrier layers per pole is three. Geometrical parameters are as follows
(Collection I):
COMPEL fRri ; Rro ; wtrib ; wrrib1 ; wrrib2 ; wrrib3 ; wry1 ; wry2 ; wry3 ; wry4 ; db1 ; db2 ; db3 g (33)

where:
wtrib = width of the tangential rib;
wrribk = width of the k-th radial rib (k = 1,2,. . ., Nfb);
wryk = width of the k-th rotor yoke (flux-carrier) (k = 1,2,. . ., Nfb þ 1); and
dbk = depth of the k-th flux-barrier (k = 1,2,. . ., Nfb).
The width of the tangential rib is constant for all flux-barrier layers. However, the width of
radial ribs could be different. Although the number of rotor geometrical parameters is equal
to 13 for this example, the number of independent variables is equal to 12 because we have
the following constraint:

wry1 þ wry2 þ wry3 þ wry4 þ db1 þ db2 þ db3 ¼ Rro  Rri (34)

The constraint shown in equation (34) could be restated in another form by definition of the
air insulation ratio (Kair):
X 3 X 3
dbi dbi
i¼1 i¼1
Kair ¼ ¼ (35)
X
4 X
3 X
3
wryi þ dbi wry þ dbi
i¼1 i¼1 i¼1

where wry is the total rotor yoke thickness in alignment with the rotor q-axis:

wry ¼ wry1 þ wry2 þ wry3 þ wry4 (36)

The per-unit rotor model was introduced in Jamali Fard and Mirsalim (2022) to
simplify the optimization process and geometrical calculations. One benefit of
working with the per-unit rotor model is scaling. When optimization is done for a
specific motor rating, it can be scaled up or down for other ratings with little effort.

Figure 10.
The rotor of a four-
pole synchronous
reluctance machine
with three flux-
barrier layers per pole
and its geometrical
parameters
The same logic is used in this work. The rotor yokes per-unit variables are defined as A robust
follows (Jamali Fard and Mirsalim, 2022): numerical
8 method
> wry2
>
> gwry2 ¼ #1
>
> wry1
>
>
< wry3
gwry3 ¼ #1 (37)
>
> wry2
>
>
>
> wry4
>
: gwry4 ¼ w # 1
ry3

And per-unit variables related to the depth of flux-barrier layers are (Jamali Fard and
Mirsalim, 2022):
8
>
> wry2
< gdb2 ¼ w # 1
>
ry1
(38)
>
> wry3
: gdb3 ¼ wry2 # 1
>

Regarding definitions of Kair, wry and per-unit variables, we can rewrite Collection I in the
following form (Collection II):

fRri ; Rro ; wtrib ; wrrib1 ; wrrib2 ; wrrib3 ; Kair ; gwry2 ; gwry3 ; gwry4 ; gdb2 ; gdb3 g (39)

All variables in Collection II are independent. The advantage of using Collection II instead of
Collection I is avoiding geometrical errors when changing the rotor dimensions during the
optimization. Because variables introduced in Collection II are per unit, we have a good
indication of their values. The following matrix equation shows the relationship between
per-unit values and actual values:
2 32 3 2 3
1 1 1 1 0 0 0 wry1 ð1  Kair Þ  ðRro  Rri Þ
6 76 7 6 7
6 gwry2 1 0 0 0 0 0 76 wry2 7 6 0 7
6 76 7 6 7
6 0 gwry3 1 7 6 7 6 7
6 0 0 0 0 76 wry3 7 6 0 7
6 76 7 6 7
6 0 0 g 1 0 0 0 7 6 w 7 ¼ 6 0 7
6 wry4 76 ry4 7 6 7
6 76 7 6 7
6 0 0 0 0 1 1 1 76 db1 7 6 Kair  ðRro  Rri Þ 7
6 76 7 6 7
6 0 0 gdb2 1 0 7 6 7 6 7
4 0 0 54 db2 5 4 0 5
0 0 0 0 0 gdb3 1 db3 0
(40)

Because flux lines are contour lines of the function Az(r, u), we should have an adequate
methodology to calculate the contour line passing through a desired point inside the rotor to
draw the flux-barrier layers. For example, for the innermost flux-barrier layer of an inner
rotor motor, we should find two contour lines passing through these two base points
(cartesian coordinates are given):
 8  
COMPEL >
> wrrib1
>
< Point 1 ¼ Rri þ w ry1 ;
2
  (41)
>
> wrrib1
>
: Point 2 ¼ Rri þ wry1 þ db1 ;
2

We used MATLAB software to implement the proposed method. The inputs are variables of
Collection II, and outputs are segmented polylines describing the geometry of flux-barrier
layers. The smoothness of these polylines is determined by the proper setting of the value of
RGR.

3. Results and discussion

A simulation is performed to investigate the correctness of the solutions generated by the
proposed method. A four-pole inner rotor SynRM is considered, and a sinusoidal flux
density distribution with the peak value equal to 0.7 T (Tesla) is assumed on its outer

Figure 11.
Calculation of flux
lines inside the solid
rotor for an inner
rotor, four-pole
machine by the finite
element method
 A robust
numerical
method

Figure 12.
Magnetic vector
potential distribution
inside the solid rotor
for an inner rotor,
four-pole machine by
proposed method

Figure 13.
Calculation of the
magnetic vector
potential on the
candidate circle
inside the solid rotor
for an inner rotor,
four-pole machine by
the finite element
method
COMPEL surface. Figure 11 illustrates the finite element analysis results. We used the ANSYS
Maxwell software for the finite element analysis. Figure 11(a) shows the MVP distribution
inside the rotor. As one can see, the maximum value of MVP is equal to 0.0112 Wb/m.
Figure 11(b) represents the flux density vectors on the outer surface of the rotor. As can be
seen, the peak value of the radial component of the flux density is equal to 0.7 T. However,
the maximum value of the flux density amplitude could be higher than 0.7 T. Figure 11(c)

Figure 14.
Comparison of the
magnetic vector
potential waveform
derived by the
proposed method and
the waveform
calculated by the
finite element
analysis software
depicts the variation of the radial component of the flux density versus the circumferential A robust
component. As we expected, it is purely sinusoidal, and its peak value is equal to 0.7 T. numerical
Figure 12 shows the MVP distribution calculated by the proposed method for the same
case shown in Figure 11. As it is seen, the result is close to Figure 11(a) which validates the
method
correctness of solutions generated by the proposed method. The maximum value of MVP by
the proposed method is equal to 0.0112 Wb/m which is the same as what we calculated
using ANSYS Maxwell software.
For a better comparison, we considered a candidate circle in the middle of the rotor
[Figure 13(a)] and calculated the value of MVP on it. Figure 13(b) shows the MVP
distribution on the candidate circle, and Figure 13(c) shows the MVP waveform as a function
of circumferential coordinate. Figure 14(a) represents the candidate mesh points of the

Figure 15.
The rotor geometry of
the three inner rotor
synchronous
reluctance machines
considering the
different number of
flux-barrier layers
derived by the
proposed method
COMPEL proposed method for comparison of their values with the finite element analysis result. The
waveforms of both methods are given in Figure 14(b) in the same chart. As it is seen, we
have a good match between the two methods that validates the correctness of the proposed
method.
The advantage of the proposed method is in using the FDM to solve the MVP equation.
Because the mesh is predefined, the meshing time is negligible. By properly setting the value
of RGR, a fine and uniform mesh could be generated. Finally, contour lines can be calculated
with arbitrary smoothness.
The slot/pole combination selection is always a challenging problem in the design of
synchronous electric machines. This problem is even more difficult for SynRMs because the
number of flux-barrier layers is added to the slot/pole combination (Wang et al., 2022).
Several combinations should be studied to find a proper combination for a specific design.

Figure 16.
The rotor geometry of
the three outer rotor
synchronous
reluctance machines
considering the
different number of
flux-barrier layers
derived by the
proposed method
The proposed method should be able to generate the rotor geometry for any arbitrary A robust
combination. numerical
To evaluate the performance of the proposed method, we considered several p-Nfb
combinations for both inner and outer rotor SynRMs. Figure 15 shows the rotor geometries
method
derived by the proposed method of the three inner rotor SynRMs considering different
numbers of flux-barrier layers. The value of per-unit variables for generating these
geometries are as follows:
Kair ¼ 0:5; gwry2 ¼ gwry3 ¼ gwry4 ¼ gdb2 ¼ gdb3 ¼ 0:9 (42)
Figure 16 illustrates the rotor geometries of the three outer rotor SynRMs considering
different numbers of flux-barrier layers. According to the results, the method is robust
enough for the generation of different rotor geometries considering arbitrary p-Nfb
combination. After the geometry generation by the proposed method, further modifications
could be used to modify the geometry of the end flux-barrier.

4. Conclusion
A robust numerical procedure for drawing the fluid flux-barrier geometry in the inner and
outer rotor radial flux SynRMs was presented in this paper. The method is based on the
ideal magnetic loading concept and uses the FDM to solve the MVP equation. Several cases,
including the inner and outer rotor machines, were studied to investigate the algorithm’s
performance. According to the results, the method can generate the rotor geometry for an
arbitrary slot, pole and flux-barrier layer combination with the desired fluid flux-barrier
layer curvature resolution suitable for the finite element analysis. The advantage of the
proposed algorithm is its simplicity and straightforwardness in its implementation for the
generation of the rotor geometry suitable for optimization.

References
Bacco, G. and Bianchi, N. (2019), “Design criteria of flux-barriers in synchronous reluctance machines”,
IEEE Transactions on Industry Applications, Vol. 55 No. 3, pp. 2490-2498.
Baka, S., Sashidhar, S. and Fernandes, B.G. (2021), “Design of an energy efficient line-start two-pole
ferrite assisted synchronous reluctance motor for water pumps”, IEEE Transactions on Energy
Conversion, Vol. 36 No. 2, pp. 961-970.
Bastos, J.P.A. and Sadowski, N. (2003), Electromagnetic Modeling by Finite Element Methods, CRC
press.
Bianchi, N., Babetto, C. and Bacco, G. (2022), Synchronous Reluctance Machines: Analysis, Optimization
and Applications, IET, available at: [Link]
Bianchi, N., Bolognani, S., Carraro, E., Castiello, M. and Fornasiero, E. (2016), “Electric vehicle traction
based on synchronous reluctance motors”, IEEE Transactions on Industry Applications, Vol. 52
No. 6, pp. 4762-4769.
Cai, H., Guan, B., Xu, L. and Choi, W. (2014), “Optimal design of synchronous reluctance
machine”, COMPEL: The International Journal for Computation and Mathematics in
Electrical and Electronic Engineering, Vol. 33 No. 5, pp. 1569-1586, doi: 10.1108/COMPEL-
09-2013-0287.
Cai, S., Jin, M.-J., Hao, H. and Shen, J.-X. (2016), “Comparative study on synchronous reluctance and pm
machines”, COMPEL: The International Journal for Computation and Mathematics in Electrical
and Electronic Engineering, Vol. 35 No. 2, pp. 607-623, doi: 10.1108/COMPEL-12-2015-0447.
Guan, Y., Zhu, Z.Q., Afinowi, I.A.A., Mipo, J.C. and Farah, P. (2016), “Design of synchronous reluctance
and permanent magnet synchronous reluctance machines for electric vehicle application”,
COMPEL COMPEL: The International Journal for Computation and Mathematics in Electrical and
Electronic Engineering, Vol. 35 No. 2, pp. 586-606, doi: 10.1108/COMPEL-02-2015-0109.
Hu, Y., Chen, B., Xiao, Y., Shi, J., Li, L. and Li, X. (2022), “Rotor design and optimization of the three-
phase line-start synchronous reluctance motor”, 2019 22nd International Conference on
Electrical Machines and Systems (ICEMS), pp. 1-6.
Jackson, J.D. (1999), Classical Electrodynamics, American Association of Physics Teachers.
Jamali Fard, A. and Mirsalim, M. (2022), “Proposing a sequential iterative method based on a per-unit
model for rotor shape optimization in permanent magnet assisted synchronous reluctance
motors”, COMPEL – The International Journal for Computation and Mathematics in Electrical
and Electronic Engineering, Vol. 41 No. 5, pp. 1826-1848, doi: 10.1108/COMPEL-09-2021-0340.
Liu, X.L.Y., Liu, Z., Ling, T. and Luo, Z. (2016), “Analysis and design of a high power density
permanent magnet-assisted synchronous reluctance machine with low-cost ferrite magnets for
EVS/HEVS”, COMPEL – The International Journal for Computation and Mathematics in
Electrical and Electronic Engineering, Vol. 35 No. 6, pp. 1949-1964, doi: 10.1108/COMPEL-05-
2016-0233.
L
opez-Torres, C., Garcia, A., Riba, J., Lux, G. and Romeral, L. (2018), “Computationally efficient design
and optimization approach of PMA-SYNRM in frequent operating torque – speed range”, IEEE
Transactions on Energy Conversion, Vol. 33 No. 4, pp. 1776-1786.
Miller, T.J.E., Hutton, A., Cossar, C. and Staton, D.A. (1991), “Design of a synchronous reluctance motor
drive”, IEEE Transactions on Industry Applications, Vol. 27 No. 4, pp. 741-749.
Morton, K.W. and Mayers, D.F. (2005), Numerical Solution of Partial Differential Equations: An
Introduction, Cambridge University Press, Cambridge.
Pellegrino, G., Jahns, T.M., Bianchi, N., Soong, W.L. and Cupertino, F. (2016), The Rediscovery of
Synchronous Reluctance and Ferrite Permanent Magnet Motors: Tutorial Course Notes, Springer
Nature, Switzerland.
Wang, Y., Bacco, G. and Bianchi, N. (2017), “Geometry analysis and optimization of pm-assisted
reluctance motors”, IEEE Transactions on Industry Applications, Vol. 53 No. 5, pp. 4338-4347.
Wang, K., Zhu, Z.Q., Ombach, G., Koch, M., Zhang, S. and Xu, J. (2015), “Torque ripple reduction of
synchronous reluctance machines”, COMPEL: The International Journal for Computation and
Mathematics in Electrical and Electronic Engineering, Vol. 34 No. 1, pp. 3-17, doi: 10.1108/
COMPEL-11-2013-0366.
Wang, K., Zhu, Z.Q., Ombach, G., Koch, M., Zhang, S. and Xu, J. (2022), “Optimal slot/pole and
flux-barrier layer number combinations for synchronous reluctance machines”, 2013 Eighth
International Conference and Exhibition on Ecological Vehicles and Renewable Energies (EVER)
pp. 1-8.

Further reading
Woodson, H.H. and Melcher, J.R. (1968), Electromechanical Dynamics, John Wiley and Sons, Hoboken.

Corresponding author
Ali Jamali Fard can be contacted at: [Link]@[Link]

For instructions on how to order reprints of this article, please visit our website:
[Link]/licensing/[Link]
Or contact us for further details: permissions@[Link]